add amdgpu from linux 4.19.44 for recent AMD Radeon parts

committing now so this can be worked on in tree

Thanks to the OpenBSD Foundation for sponsoring this work and kettenis@
for helping.

234 files changed, 175435 insertions, 0 deletions

diff --git a/sys/dev/pci/drm/amd/amdgpu/ObjectID.h b/sys/dev/pci/drm/amd/amdgpu/ObjectID.h
new file mode 100644
index 00000000000..5b393622f59
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/ObjectID.h
@@ -0,0 +1,744 @@
+/*
+* Copyright 2006-2007 Advanced Micro Devices, Inc.  
+*
+* Permission is hereby granted, free of charge, to any person obtaining a
+* copy of this software and associated documentation files (the "Software"),
+* to deal in the Software without restriction, including without limitation
+* the rights to use, copy, modify, merge, publish, distribute, sublicense,
+* and/or sell copies of the Software, and to permit persons to whom the
+* Software is furnished to do so, subject to the following conditions:
+*
+* The above copyright notice and this permission notice shall be included in
+* all copies or substantial portions of the Software.
+*
+* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+* OTHER DEALINGS IN THE SOFTWARE.
+*/
+/* based on stg/asic_reg/drivers/inc/asic_reg/ObjectID.h ver 23 */
+
+#ifndef _OBJECTID_H
+#define _OBJECTID_H
+
+#if defined(_X86_)
+#pragma pack(1)
+#endif
+
+/****************************************************/
+/* Graphics Object Type Definition                  */
+/****************************************************/
+#define GRAPH_OBJECT_TYPE_NONE                    0x0
+#define GRAPH_OBJECT_TYPE_GPU                     0x1
+#define GRAPH_OBJECT_TYPE_ENCODER                 0x2
+#define GRAPH_OBJECT_TYPE_CONNECTOR               0x3
+#define GRAPH_OBJECT_TYPE_ROUTER                  0x4
+/* deleted */
+#define GRAPH_OBJECT_TYPE_DISPLAY_PATH            0x6  
+#define GRAPH_OBJECT_TYPE_GENERIC                 0x7
+
+/****************************************************/
+/* Encoder Object ID Definition                     */
+/****************************************************/
+#define ENCODER_OBJECT_ID_NONE                    0x00 
+
+/* Radeon Class Display Hardware */
+#define ENCODER_OBJECT_ID_INTERNAL_LVDS           0x01
+#define ENCODER_OBJECT_ID_INTERNAL_TMDS1          0x02
+#define ENCODER_OBJECT_ID_INTERNAL_TMDS2          0x03
+#define ENCODER_OBJECT_ID_INTERNAL_DAC1           0x04
+#define ENCODER_OBJECT_ID_INTERNAL_DAC2           0x05     /* TV/CV DAC */
+#define ENCODER_OBJECT_ID_INTERNAL_SDVOA          0x06
+#define ENCODER_OBJECT_ID_INTERNAL_SDVOB          0x07
+
+/* External Third Party Encoders */
+#define ENCODER_OBJECT_ID_SI170B                  0x08
+#define ENCODER_OBJECT_ID_CH7303                  0x09
+#define ENCODER_OBJECT_ID_CH7301                  0x0A
+#define ENCODER_OBJECT_ID_INTERNAL_DVO1           0x0B    /* This belongs to Radeon Class Display Hardware */
+#define ENCODER_OBJECT_ID_EXTERNAL_SDVOA          0x0C
+#define ENCODER_OBJECT_ID_EXTERNAL_SDVOB          0x0D
+#define ENCODER_OBJECT_ID_TITFP513                0x0E
+#define ENCODER_OBJECT_ID_INTERNAL_LVTM1          0x0F    /* not used for Radeon */
+#define ENCODER_OBJECT_ID_VT1623                  0x10
+#define ENCODER_OBJECT_ID_HDMI_SI1930             0x11
+#define ENCODER_OBJECT_ID_HDMI_INTERNAL           0x12
+#define ENCODER_OBJECT_ID_ALMOND                  0x22
+#define ENCODER_OBJECT_ID_TRAVIS                  0x23
+#define ENCODER_OBJECT_ID_NUTMEG                  0x22
+#define ENCODER_OBJECT_ID_HDMI_ANX9805            0x26
+
+/* Kaleidoscope (KLDSCP) Class Display Hardware (internal) */
+#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1   0x13
+#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1    0x14
+#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1    0x15
+#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2    0x16  /* Shared with CV/TV and CRT */
+#define ENCODER_OBJECT_ID_SI178                   0X17  /* External TMDS (dual link, no HDCP.) */
+#define ENCODER_OBJECT_ID_MVPU_FPGA               0x18  /* MVPU FPGA chip */
+#define ENCODER_OBJECT_ID_INTERNAL_DDI            0x19
+#define ENCODER_OBJECT_ID_VT1625                  0x1A
+#define ENCODER_OBJECT_ID_HDMI_SI1932             0x1B
+#define ENCODER_OBJECT_ID_DP_AN9801               0x1C
+#define ENCODER_OBJECT_ID_DP_DP501                0x1D
+#define ENCODER_OBJECT_ID_INTERNAL_UNIPHY         0x1E
+#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA   0x1F
+#define ENCODER_OBJECT_ID_INTERNAL_UNIPHY1        0x20
+#define ENCODER_OBJECT_ID_INTERNAL_UNIPHY2        0x21
+#define ENCODER_OBJECT_ID_INTERNAL_VCE            0x24
+#define ENCODER_OBJECT_ID_INTERNAL_UNIPHY3        0x25
+#define ENCODER_OBJECT_ID_INTERNAL_AMCLK          0x27
+
+#define ENCODER_OBJECT_ID_GENERAL_EXTERNAL_DVO    0xFF
+
+/****************************************************/
+/* Connector Object ID Definition                   */
+/****************************************************/
+#define CONNECTOR_OBJECT_ID_NONE                  0x00 
+#define CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I     0x01
+#define CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I       0x02
+#define CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D     0x03
+#define CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D       0x04
+#define CONNECTOR_OBJECT_ID_VGA                   0x05
+#define CONNECTOR_OBJECT_ID_COMPOSITE             0x06
+#define CONNECTOR_OBJECT_ID_SVIDEO                0x07
+#define CONNECTOR_OBJECT_ID_YPbPr                 0x08
+#define CONNECTOR_OBJECT_ID_D_CONNECTOR           0x09
+#define CONNECTOR_OBJECT_ID_9PIN_DIN              0x0A  /* Supports both CV & TV */
+#define CONNECTOR_OBJECT_ID_SCART                 0x0B
+#define CONNECTOR_OBJECT_ID_HDMI_TYPE_A           0x0C
+#define CONNECTOR_OBJECT_ID_HDMI_TYPE_B           0x0D
+#define CONNECTOR_OBJECT_ID_LVDS                  0x0E
+#define CONNECTOR_OBJECT_ID_7PIN_DIN              0x0F
+#define CONNECTOR_OBJECT_ID_PCIE_CONNECTOR        0x10
+#define CONNECTOR_OBJECT_ID_CROSSFIRE             0x11
+#define CONNECTOR_OBJECT_ID_HARDCODE_DVI          0x12
+#define CONNECTOR_OBJECT_ID_DISPLAYPORT           0x13
+#define CONNECTOR_OBJECT_ID_eDP                   0x14
+#define CONNECTOR_OBJECT_ID_MXM                   0x15
+#define CONNECTOR_OBJECT_ID_LVDS_eDP              0x16
+
+/* deleted */
+
+/****************************************************/
+/* Router Object ID Definition                      */
+/****************************************************/
+#define ROUTER_OBJECT_ID_NONE											0x00
+#define ROUTER_OBJECT_ID_I2C_EXTENDER_CNTL				0x01
+
+/****************************************************/
+/* Generic Object ID Definition                     */
+/****************************************************/
+#define GENERIC_OBJECT_ID_NONE                    0x00
+#define GENERIC_OBJECT_ID_GLSYNC                  0x01
+#define GENERIC_OBJECT_ID_PX2_NON_DRIVABLE        0x02
+#define GENERIC_OBJECT_ID_MXM_OPM                 0x03
+#define GENERIC_OBJECT_ID_STEREO_PIN              0x04        //This object could show up from Misc Object table, it follows ATOM_OBJECT format, and contains one ATOM_OBJECT_GPIO_CNTL_RECORD for the stereo pin
+#define GENERIC_OBJECT_ID_BRACKET_LAYOUT          0x05
+
+/****************************************************/
+/* Graphics Object ENUM ID Definition               */
+/****************************************************/
+#define GRAPH_OBJECT_ENUM_ID1                     0x01
+#define GRAPH_OBJECT_ENUM_ID2                     0x02
+#define GRAPH_OBJECT_ENUM_ID3                     0x03
+#define GRAPH_OBJECT_ENUM_ID4                     0x04
+#define GRAPH_OBJECT_ENUM_ID5                     0x05
+#define GRAPH_OBJECT_ENUM_ID6                     0x06
+#define GRAPH_OBJECT_ENUM_ID7                     0x07
+
+/****************************************************/
+/* Graphics Object ID Bit definition                */
+/****************************************************/
+#define OBJECT_ID_MASK                            0x00FF
+#define ENUM_ID_MASK                              0x0700
+#define RESERVED1_ID_MASK                         0x0800
+#define OBJECT_TYPE_MASK                          0x7000
+#define RESERVED2_ID_MASK                         0x8000
+                                                  
+#define OBJECT_ID_SHIFT                           0x00
+#define ENUM_ID_SHIFT                             0x08
+#define OBJECT_TYPE_SHIFT                         0x0C
+
+
+/****************************************************/
+/* Graphics Object family definition                */
+/****************************************************/
+#define CONSTRUCTOBJECTFAMILYID(GRAPHICS_OBJECT_TYPE, GRAPHICS_OBJECT_ID) (GRAPHICS_OBJECT_TYPE << OBJECT_TYPE_SHIFT | \
+                                                                           GRAPHICS_OBJECT_ID   << OBJECT_ID_SHIFT)
+/****************************************************/
+/* GPU Object ID definition - Shared with BIOS      */
+/****************************************************/
+#define GPU_ENUM_ID1                            ( GRAPH_OBJECT_TYPE_GPU << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT)
+
+/****************************************************/
+/* Encoder Object ID definition - Shared with BIOS  */
+/****************************************************/
+/*
+#define ENCODER_INTERNAL_LVDS_ENUM_ID1        0x2101      
+#define ENCODER_INTERNAL_TMDS1_ENUM_ID1       0x2102
+#define ENCODER_INTERNAL_TMDS2_ENUM_ID1       0x2103
+#define ENCODER_INTERNAL_DAC1_ENUM_ID1        0x2104
+#define ENCODER_INTERNAL_DAC2_ENUM_ID1        0x2105
+#define ENCODER_INTERNAL_SDVOA_ENUM_ID1       0x2106
+#define ENCODER_INTERNAL_SDVOB_ENUM_ID1       0x2107
+#define ENCODER_SIL170B_ENUM_ID1              0x2108  
+#define ENCODER_CH7303_ENUM_ID1               0x2109
+#define ENCODER_CH7301_ENUM_ID1               0x210A
+#define ENCODER_INTERNAL_DVO1_ENUM_ID1        0x210B
+#define ENCODER_EXTERNAL_SDVOA_ENUM_ID1       0x210C
+#define ENCODER_EXTERNAL_SDVOB_ENUM_ID1       0x210D
+#define ENCODER_TITFP513_ENUM_ID1             0x210E
+#define ENCODER_INTERNAL_LVTM1_ENUM_ID1       0x210F
+#define ENCODER_VT1623_ENUM_ID1               0x2110
+#define ENCODER_HDMI_SI1930_ENUM_ID1          0x2111
+#define ENCODER_HDMI_INTERNAL_ENUM_ID1        0x2112
+#define ENCODER_INTERNAL_KLDSCP_TMDS1_ENUM_ID1   0x2113
+#define ENCODER_INTERNAL_KLDSCP_DVO1_ENUM_ID1    0x2114
+#define ENCODER_INTERNAL_KLDSCP_DAC1_ENUM_ID1    0x2115
+#define ENCODER_INTERNAL_KLDSCP_DAC2_ENUM_ID1    0x2116  
+#define ENCODER_SI178_ENUM_ID1                   0x2117 
+#define ENCODER_MVPU_FPGA_ENUM_ID1               0x2118
+#define ENCODER_INTERNAL_DDI_ENUM_ID1            0x2119
+#define ENCODER_VT1625_ENUM_ID1                  0x211A
+#define ENCODER_HDMI_SI1932_ENUM_ID1             0x211B
+#define ENCODER_ENCODER_DP_AN9801_ENUM_ID1       0x211C
+#define ENCODER_DP_DP501_ENUM_ID1                0x211D
+#define ENCODER_INTERNAL_UNIPHY_ENUM_ID1         0x211E
+*/
+#define ENCODER_INTERNAL_LVDS_ENUM_ID1     ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_LVDS << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_TMDS1_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_TMDS1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_TMDS2_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_TMDS2 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_DAC1_ENUM_ID1     ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_DAC1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_DAC2_ENUM_ID1     ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_DAC2 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_SDVOA_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_SDVOA << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_SDVOA_ENUM_ID2    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_SDVOA << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_SDVOB_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_SDVOB << OBJECT_ID_SHIFT)
+
+#define ENCODER_SIL170B_ENUM_ID1           ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_SI170B << OBJECT_ID_SHIFT)
+
+#define ENCODER_CH7303_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_CH7303 << OBJECT_ID_SHIFT)
+
+#define ENCODER_CH7301_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_CH7301 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_DVO1_ENUM_ID1     ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_DVO1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_EXTERNAL_SDVOA_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_EXTERNAL_SDVOA << OBJECT_ID_SHIFT)
+
+#define ENCODER_EXTERNAL_SDVOA_ENUM_ID2    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_EXTERNAL_SDVOA << OBJECT_ID_SHIFT)
+
+
+#define ENCODER_EXTERNAL_SDVOB_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_EXTERNAL_SDVOB << OBJECT_ID_SHIFT)
+
+
+#define ENCODER_TITFP513_ENUM_ID1          ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_TITFP513 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_LVTM1_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_LVTM1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_VT1623_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_VT1623 << OBJECT_ID_SHIFT)
+
+#define ENCODER_HDMI_SI1930_ENUM_ID1       ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_HDMI_SI1930 << OBJECT_ID_SHIFT)
+
+#define ENCODER_HDMI_INTERNAL_ENUM_ID1     ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_HDMI_INTERNAL << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_KLDSCP_TMDS1_ENUM_ID1   ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1 << OBJECT_ID_SHIFT)
+
+
+#define ENCODER_INTERNAL_KLDSCP_TMDS1_ENUM_ID2   ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1 << OBJECT_ID_SHIFT)
+
+
+#define ENCODER_INTERNAL_KLDSCP_DVO1_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_KLDSCP_DAC1_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_KLDSCP_DAC2_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2 << OBJECT_ID_SHIFT)  // Shared with CV/TV and CRT
+
+#define ENCODER_SI178_ENUM_ID1                    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_SI178 << OBJECT_ID_SHIFT)  
+
+#define ENCODER_MVPU_FPGA_ENUM_ID1                ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_MVPU_FPGA << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_DDI_ENUM_ID1     (  GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_DDI << OBJECT_ID_SHIFT) 
+
+#define ENCODER_VT1625_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_VT1625 << OBJECT_ID_SHIFT)
+
+#define ENCODER_HDMI_SI1932_ENUM_ID1       ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_HDMI_SI1932 << OBJECT_ID_SHIFT)
+
+#define ENCODER_DP_DP501_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_DP_DP501 << OBJECT_ID_SHIFT)
+
+#define ENCODER_DP_AN9801_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_DP_AN9801 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY_ENUM_ID1         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY_ENUM_ID2         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_KLDSCP_LVTMA_ENUM_ID1   ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA << OBJECT_ID_SHIFT)  
+
+#define ENCODER_INTERNAL_UNIPHY1_ENUM_ID1         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY1_ENUM_ID2         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY2_ENUM_ID1         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY2 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY2_ENUM_ID2         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY2 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY3_ENUM_ID1         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY3 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY3_ENUM_ID2         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY3 << OBJECT_ID_SHIFT)
+
+#define ENCODER_GENERAL_EXTERNAL_DVO_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_GENERAL_EXTERNAL_DVO << OBJECT_ID_SHIFT)
+
+#define ENCODER_ALMOND_ENUM_ID1                  ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_ALMOND << OBJECT_ID_SHIFT)
+
+#define ENCODER_ALMOND_ENUM_ID2                  ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_ALMOND << OBJECT_ID_SHIFT)
+
+#define ENCODER_TRAVIS_ENUM_ID1                  ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_TRAVIS << OBJECT_ID_SHIFT)
+
+#define ENCODER_TRAVIS_ENUM_ID2                  ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_TRAVIS << OBJECT_ID_SHIFT)
+
+#define ENCODER_NUTMEG_ENUM_ID1                  ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_NUTMEG << OBJECT_ID_SHIFT)
+
+#define ENCODER_VCE_ENUM_ID1                     ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_INTERNAL_VCE << OBJECT_ID_SHIFT)
+
+#define ENCODER_HDMI_ANX9805_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_HDMI_ANX9805 << OBJECT_ID_SHIFT)
+
+/****************************************************/
+/* Connector Object ID definition - Shared with BIOS */
+/****************************************************/
+/*
+#define CONNECTOR_SINGLE_LINK_DVI_I_ENUM_ID1        0x3101
+#define CONNECTOR_DUAL_LINK_DVI_I_ENUM_ID1          0x3102
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID1        0x3103
+#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID1          0x3104
+#define CONNECTOR_VGA_ENUM_ID1                      0x3105
+#define CONNECTOR_COMPOSITE_ENUM_ID1                0x3106
+#define CONNECTOR_SVIDEO_ENUM_ID1                   0x3107
+#define CONNECTOR_YPbPr_ENUM_ID1                    0x3108
+#define CONNECTOR_D_CONNECTORE_ENUM_ID1             0x3109
+#define CONNECTOR_9PIN_DIN_ENUM_ID1                 0x310A
+#define CONNECTOR_SCART_ENUM_ID1                    0x310B
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID1              0x310C
+#define CONNECTOR_HDMI_TYPE_B_ENUM_ID1              0x310D
+#define CONNECTOR_LVDS_ENUM_ID1                     0x310E
+#define CONNECTOR_7PIN_DIN_ENUM_ID1                 0x310F
+#define CONNECTOR_PCIE_CONNECTOR_ENUM_ID1           0x3110
+*/
+#define CONNECTOR_LVDS_ENUM_ID1                ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_LVDS << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_LVDS_ENUM_ID2                ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_LVDS << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_eDP_ENUM_ID1                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_eDP << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_eDP_ENUM_ID2                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_eDP << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_I_ENUM_ID1   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_I_ENUM_ID2   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DUAL_LINK_DVI_I_ENUM_ID1     ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DUAL_LINK_DVI_I_ENUM_ID2     ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID1   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID2   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID3   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID4   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID5   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID5 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID6   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID6 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID1     ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID2     ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID3     ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID4     ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_VGA_ENUM_ID1                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_VGA << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_VGA_ENUM_ID2                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_VGA << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_COMPOSITE_ENUM_ID1           ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_COMPOSITE << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_COMPOSITE_ENUM_ID2           ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_COMPOSITE << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SVIDEO_ENUM_ID1              ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SVIDEO << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SVIDEO_ENUM_ID2              ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SVIDEO << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_YPbPr_ENUM_ID1               ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_YPbPr << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_YPbPr_ENUM_ID2               ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_YPbPr << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_D_CONNECTOR_ENUM_ID1         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_D_CONNECTOR << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_D_CONNECTOR_ENUM_ID2         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_D_CONNECTOR << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_9PIN_DIN_ENUM_ID1            ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_9PIN_DIN << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_9PIN_DIN_ENUM_ID2            ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_9PIN_DIN << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SCART_ENUM_ID1               ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SCART << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SCART_ENUM_ID2               ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SCART << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID1         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID2         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID3         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID4         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID5         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID5 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID6         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID6 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_B_ENUM_ID1         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_B << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_B_ENUM_ID2         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_B << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_7PIN_DIN_ENUM_ID1            ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_7PIN_DIN << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_7PIN_DIN_ENUM_ID2            ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_7PIN_DIN << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_PCIE_CONNECTOR_ENUM_ID1      ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_PCIE_CONNECTOR << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_PCIE_CONNECTOR_ENUM_ID2      ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_PCIE_CONNECTOR << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_CROSSFIRE_ENUM_ID1           ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_CROSSFIRE << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_CROSSFIRE_ENUM_ID2           ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_CROSSFIRE << OBJECT_ID_SHIFT)
+
+
+#define CONNECTOR_HARDCODE_DVI_ENUM_ID1        ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HARDCODE_DVI << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HARDCODE_DVI_ENUM_ID2        ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HARDCODE_DVI << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DISPLAYPORT_ENUM_ID1         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DISPLAYPORT_ENUM_ID2         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DISPLAYPORT_ENUM_ID3         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DISPLAYPORT_ENUM_ID4         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DISPLAYPORT_ENUM_ID5         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID5 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DISPLAYPORT_ENUM_ID6         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID6 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_MXM_ENUM_ID1                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_DP_A
+
+#define CONNECTOR_MXM_ENUM_ID2                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_DP_B
+
+#define CONNECTOR_MXM_ENUM_ID3                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_DP_C
+
+#define CONNECTOR_MXM_ENUM_ID4                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_DP_D
+
+#define CONNECTOR_MXM_ENUM_ID5                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID5 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_LVDS_TXxx
+
+#define CONNECTOR_MXM_ENUM_ID6                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID6 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_LVDS_UXxx
+
+#define CONNECTOR_MXM_ENUM_ID7                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID7 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_DAC
+
+#define CONNECTOR_LVDS_eDP_ENUM_ID1            ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_LVDS_eDP << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_LVDS_eDP_ENUM_ID2            ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_LVDS_eDP << OBJECT_ID_SHIFT)
+
+/****************************************************/
+/* Router Object ID definition - Shared with BIOS   */
+/****************************************************/
+#define ROUTER_I2C_EXTENDER_CNTL_ENUM_ID1      ( GRAPH_OBJECT_TYPE_ROUTER << OBJECT_TYPE_SHIFT |\
+                                                GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                ROUTER_OBJECT_ID_I2C_EXTENDER_CNTL << OBJECT_ID_SHIFT)
+
+/* deleted */
+
+/****************************************************/
+/* Generic Object ID definition - Shared with BIOS  */
+/****************************************************/
+#define GENERICOBJECT_GLSYNC_ENUM_ID1           (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_GLSYNC << OBJECT_ID_SHIFT)
+
+#define GENERICOBJECT_PX2_NON_DRIVABLE_ID1       (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_PX2_NON_DRIVABLE<< OBJECT_ID_SHIFT)
+
+#define GENERICOBJECT_PX2_NON_DRIVABLE_ID2       (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_PX2_NON_DRIVABLE<< OBJECT_ID_SHIFT)
+
+#define GENERICOBJECT_MXM_OPM_ENUM_ID1           (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_MXM_OPM << OBJECT_ID_SHIFT)
+
+#define GENERICOBJECT_STEREO_PIN_ENUM_ID1        (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_STEREO_PIN << OBJECT_ID_SHIFT)
+
+#define GENERICOBJECT_BRACKET_LAYOUT_ENUM_ID1    (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_BRACKET_LAYOUT << OBJECT_ID_SHIFT)
+
+#define GENERICOBJECT_BRACKET_LAYOUT_ENUM_ID2    (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_BRACKET_LAYOUT << OBJECT_ID_SHIFT)
+/****************************************************/
+/* Object Cap definition - Shared with BIOS         */
+/****************************************************/
+#define GRAPHICS_OBJECT_CAP_I2C                 0x00000001L
+#define GRAPHICS_OBJECT_CAP_TABLE_ID            0x00000002L
+
+
+#define GRAPHICS_OBJECT_I2CCOMMAND_TABLE_ID                   0x01
+#define GRAPHICS_OBJECT_HOTPLUGDETECTIONINTERUPT_TABLE_ID     0x02
+#define GRAPHICS_OBJECT_ENCODER_OUTPUT_PROTECTION_TABLE_ID    0x03
+
+#if defined(_X86_)
+#pragma pack()
+#endif
+
+#endif  /*GRAPHICTYPE */
+
+
+
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu.h
new file mode 100644
index 00000000000..813c0f6b29f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu.h
@@ -0,0 +1,1905 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#ifndef __AMDGPU_H__
+#define __AMDGPU_H__
+
+#include <linux/atomic.h>
+#include <linux/wait.h>
+#include <linux/list.h>
+#include <linux/kref.h>
+#include <linux/rbtree.h>
+#include <linux/hashtable.h>
+#include <linux/dma-fence.h>
+
+#include <drm/ttm/ttm_bo_api.h>
+#include <drm/ttm/ttm_bo_driver.h>
+#include <drm/ttm/ttm_placement.h>
+#include <drm/ttm/ttm_module.h>
+#include <drm/ttm/ttm_execbuf_util.h>
+
+#include <drm/drmP.h>
+#include <drm/drm_gem.h>
+#include <drm/amdgpu_drm.h>
+#include <drm/gpu_scheduler.h>
+
+#include <dev/wscons/wsconsio.h>
+#include <dev/wscons/wsdisplayvar.h>
+#include <dev/rasops/rasops.h>
+
+#include <kgd_kfd_interface.h>
+#include "dm_pp_interface.h"
+#include "kgd_pp_interface.h"
+
+#include "amd_shared.h"
+#include "amdgpu_mode.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_irq.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_ttm.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_gds.h"
+#include "amdgpu_sync.h"
+#include "amdgpu_ring.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_dpm.h"
+#include "amdgpu_acp.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+#include "amdgpu_vcn.h"
+#include "amdgpu_mn.h"
+#include "amdgpu_gmc.h"
+#include "amdgpu_dm.h"
+#include "amdgpu_virt.h"
+#include "amdgpu_gart.h"
+#include "amdgpu_debugfs.h"
+#include "amdgpu_job.h"
+#include "amdgpu_bo_list.h"
+
+/*
+ * Modules parameters.
+ */
+extern int amdgpu_modeset;
+extern int amdgpu_vram_limit;
+extern int amdgpu_vis_vram_limit;
+extern int amdgpu_gart_size;
+extern int amdgpu_gtt_size;
+extern int amdgpu_moverate;
+extern int amdgpu_benchmarking;
+extern int amdgpu_testing;
+extern int amdgpu_audio;
+extern int amdgpu_disp_priority;
+extern int amdgpu_hw_i2c;
+extern int amdgpu_pcie_gen2;
+extern int amdgpu_msi;
+extern int amdgpu_lockup_timeout;
+extern int amdgpu_dpm;
+extern int amdgpu_fw_load_type;
+extern int amdgpu_aspm;
+extern int amdgpu_runtime_pm;
+extern uint amdgpu_ip_block_mask;
+extern int amdgpu_bapm;
+extern int amdgpu_deep_color;
+extern int amdgpu_vm_size;
+extern int amdgpu_vm_block_size;
+extern int amdgpu_vm_fragment_size;
+extern int amdgpu_vm_fault_stop;
+extern int amdgpu_vm_debug;
+extern int amdgpu_vm_update_mode;
+extern int amdgpu_dc;
+extern int amdgpu_sched_jobs;
+extern int amdgpu_sched_hw_submission;
+extern uint amdgpu_pcie_gen_cap;
+extern uint amdgpu_pcie_lane_cap;
+extern uint amdgpu_cg_mask;
+extern uint amdgpu_pg_mask;
+extern uint amdgpu_sdma_phase_quantum;
+extern char *amdgpu_disable_cu;
+extern char *amdgpu_virtual_display;
+extern uint amdgpu_pp_feature_mask;
+extern int amdgpu_vram_page_split;
+extern int amdgpu_ngg;
+extern int amdgpu_prim_buf_per_se;
+extern int amdgpu_pos_buf_per_se;
+extern int amdgpu_cntl_sb_buf_per_se;
+extern int amdgpu_param_buf_per_se;
+extern int amdgpu_job_hang_limit;
+extern int amdgpu_lbpw;
+extern int amdgpu_compute_multipipe;
+extern int amdgpu_gpu_recovery;
+extern int amdgpu_emu_mode;
+extern uint amdgpu_smu_memory_pool_size;
+
+#ifdef CONFIG_DRM_AMDGPU_SI
+extern int amdgpu_si_support;
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+extern int amdgpu_cik_support;
+#endif
+
+#define AMDGPU_SG_THRESHOLD			(256*1024*1024)
+#define AMDGPU_DEFAULT_GTT_SIZE_MB		3072ULL /* 3GB by default */
+#define AMDGPU_WAIT_IDLE_TIMEOUT_IN_MS	        3000
+#define AMDGPU_MAX_USEC_TIMEOUT			100000	/* 100 ms */
+#define AMDGPU_FENCE_JIFFIES_TIMEOUT		(HZ / 2)
+/* AMDGPU_IB_POOL_SIZE must be a power of 2 */
+#define AMDGPU_IB_POOL_SIZE			16
+#define AMDGPU_DEBUGFS_MAX_COMPONENTS		32
+#define AMDGPUFB_CONN_LIMIT			4
+#define AMDGPU_BIOS_NUM_SCRATCH			16
+
+/* max number of IP instances */
+#define AMDGPU_MAX_SDMA_INSTANCES		2
+
+/* hard reset data */
+#define AMDGPU_ASIC_RESET_DATA                  0x39d5e86b
+
+/* reset flags */
+#define AMDGPU_RESET_GFX			(1 << 0)
+#define AMDGPU_RESET_COMPUTE			(1 << 1)
+#define AMDGPU_RESET_DMA			(1 << 2)
+#define AMDGPU_RESET_CP				(1 << 3)
+#define AMDGPU_RESET_GRBM			(1 << 4)
+#define AMDGPU_RESET_DMA1			(1 << 5)
+#define AMDGPU_RESET_RLC			(1 << 6)
+#define AMDGPU_RESET_SEM			(1 << 7)
+#define AMDGPU_RESET_IH				(1 << 8)
+#define AMDGPU_RESET_VMC			(1 << 9)
+#define AMDGPU_RESET_MC				(1 << 10)
+#define AMDGPU_RESET_DISPLAY			(1 << 11)
+#define AMDGPU_RESET_UVD			(1 << 12)
+#define AMDGPU_RESET_VCE			(1 << 13)
+#define AMDGPU_RESET_VCE1			(1 << 14)
+
+/* GFX current status */
+#define AMDGPU_GFX_NORMAL_MODE			0x00000000L
+#define AMDGPU_GFX_SAFE_MODE			0x00000001L
+#define AMDGPU_GFX_PG_DISABLED_MODE		0x00000002L
+#define AMDGPU_GFX_CG_DISABLED_MODE		0x00000004L
+#define AMDGPU_GFX_LBPW_DISABLED_MODE		0x00000008L
+
+/* max cursor sizes (in pixels) */
+#define CIK_CURSOR_WIDTH 128
+#define CIK_CURSOR_HEIGHT 128
+
+struct amdgpu_device;
+struct amdgpu_ib;
+struct amdgpu_cs_parser;
+struct amdgpu_job;
+struct amdgpu_irq_src;
+struct amdgpu_fpriv;
+struct amdgpu_bo_va_mapping;
+struct amdgpu_atif;
+
+enum amdgpu_cp_irq {
+	AMDGPU_CP_IRQ_GFX_EOP = 0,
+	AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE3_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE0_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE1_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE2_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE3_EOP,
+
+	AMDGPU_CP_IRQ_LAST
+};
+
+enum amdgpu_sdma_irq {
+	AMDGPU_SDMA_IRQ_TRAP0 = 0,
+	AMDGPU_SDMA_IRQ_TRAP1,
+
+	AMDGPU_SDMA_IRQ_LAST
+};
+
+enum amdgpu_thermal_irq {
+	AMDGPU_THERMAL_IRQ_LOW_TO_HIGH = 0,
+	AMDGPU_THERMAL_IRQ_HIGH_TO_LOW,
+
+	AMDGPU_THERMAL_IRQ_LAST
+};
+
+enum amdgpu_kiq_irq {
+	AMDGPU_CP_KIQ_IRQ_DRIVER0 = 0,
+	AMDGPU_CP_KIQ_IRQ_LAST
+};
+
+int amdgpu_device_ip_set_clockgating_state(void *dev,
+					   enum amd_ip_block_type block_type,
+					   enum amd_clockgating_state state);
+int amdgpu_device_ip_set_powergating_state(void *dev,
+					   enum amd_ip_block_type block_type,
+					   enum amd_powergating_state state);
+void amdgpu_device_ip_get_clockgating_state(struct amdgpu_device *adev,
+					    u32 *flags);
+int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev,
+				   enum amd_ip_block_type block_type);
+bool amdgpu_device_ip_is_idle(struct amdgpu_device *adev,
+			      enum amd_ip_block_type block_type);
+
+#define AMDGPU_MAX_IP_NUM 16
+
+struct amdgpu_ip_block_status {
+	bool valid;
+	bool sw;
+	bool hw;
+	bool late_initialized;
+	bool hang;
+};
+
+struct amdgpu_ip_block_version {
+	const enum amd_ip_block_type type;
+	const u32 major;
+	const u32 minor;
+	const u32 rev;
+	const struct amd_ip_funcs *funcs;
+};
+
+struct amdgpu_ip_block {
+	struct amdgpu_ip_block_status status;
+	const struct amdgpu_ip_block_version *version;
+};
+
+int amdgpu_device_ip_block_version_cmp(struct amdgpu_device *adev,
+				       enum amd_ip_block_type type,
+				       u32 major, u32 minor);
+
+struct amdgpu_ip_block *
+amdgpu_device_ip_get_ip_block(struct amdgpu_device *adev,
+			      enum amd_ip_block_type type);
+
+int amdgpu_device_ip_block_add(struct amdgpu_device *adev,
+			       const struct amdgpu_ip_block_version *ip_block_version);
+
+/* provided by hw blocks that can move/clear data.  e.g., gfx or sdma */
+struct amdgpu_buffer_funcs {
+	/* maximum bytes in a single operation */
+	uint32_t	copy_max_bytes;
+
+	/* number of dw to reserve per operation */
+	unsigned	copy_num_dw;
+
+	/* used for buffer migration */
+	void (*emit_copy_buffer)(struct amdgpu_ib *ib,
+				 /* src addr in bytes */
+				 uint64_t src_offset,
+				 /* dst addr in bytes */
+				 uint64_t dst_offset,
+				 /* number of byte to transfer */
+				 uint32_t byte_count);
+
+	/* maximum bytes in a single operation */
+	uint32_t	fill_max_bytes;
+
+	/* number of dw to reserve per operation */
+	unsigned	fill_num_dw;
+
+	/* used for buffer clearing */
+	void (*emit_fill_buffer)(struct amdgpu_ib *ib,
+				 /* value to write to memory */
+				 uint32_t src_data,
+				 /* dst addr in bytes */
+				 uint64_t dst_offset,
+				 /* number of byte to fill */
+				 uint32_t byte_count);
+};
+
+/* provided by hw blocks that can write ptes, e.g., sdma */
+struct amdgpu_vm_pte_funcs {
+	/* number of dw to reserve per operation */
+	unsigned	copy_pte_num_dw;
+
+	/* copy pte entries from GART */
+	void (*copy_pte)(struct amdgpu_ib *ib,
+			 uint64_t pe, uint64_t src,
+			 unsigned count);
+
+	/* write pte one entry at a time with addr mapping */
+	void (*write_pte)(struct amdgpu_ib *ib, uint64_t pe,
+			  uint64_t value, unsigned count,
+			  uint32_t incr);
+	/* for linear pte/pde updates without addr mapping */
+	void (*set_pte_pde)(struct amdgpu_ib *ib,
+			    uint64_t pe,
+			    uint64_t addr, unsigned count,
+			    uint32_t incr, uint64_t flags);
+};
+
+/* provided by the ih block */
+struct amdgpu_ih_funcs {
+	/* ring read/write ptr handling, called from interrupt context */
+	u32 (*get_wptr)(struct amdgpu_device *adev);
+	bool (*prescreen_iv)(struct amdgpu_device *adev);
+	void (*decode_iv)(struct amdgpu_device *adev,
+			  struct amdgpu_iv_entry *entry);
+	void (*set_rptr)(struct amdgpu_device *adev);
+};
+
+/*
+ * BIOS.
+ */
+bool amdgpu_get_bios(struct amdgpu_device *adev);
+bool amdgpu_read_bios(struct amdgpu_device *adev);
+
+/*
+ * Clocks
+ */
+
+#define AMDGPU_MAX_PPLL 3
+
+struct amdgpu_clock {
+	struct amdgpu_pll ppll[AMDGPU_MAX_PPLL];
+	struct amdgpu_pll spll;
+	struct amdgpu_pll mpll;
+	/* 10 Khz units */
+	uint32_t default_mclk;
+	uint32_t default_sclk;
+	uint32_t default_dispclk;
+	uint32_t current_dispclk;
+	uint32_t dp_extclk;
+	uint32_t max_pixel_clock;
+};
+
+/*
+ * GEM.
+ */
+
+#define AMDGPU_GEM_DOMAIN_MAX		0x3
+#define gem_to_amdgpu_bo(gobj) container_of((gobj), struct amdgpu_bo, gem_base)
+
+void amdgpu_gem_object_free(struct drm_gem_object *obj);
+int amdgpu_gem_object_open(struct drm_gem_object *obj,
+				struct drm_file *file_priv);
+void amdgpu_gem_object_close(struct drm_gem_object *obj,
+				struct drm_file *file_priv);
+unsigned long amdgpu_gem_timeout(uint64_t timeout_ns);
+struct sg_table *amdgpu_gem_prime_get_sg_table(struct drm_gem_object *obj);
+struct drm_gem_object *
+amdgpu_gem_prime_import_sg_table(struct drm_device *dev,
+				 struct dma_buf_attachment *attach,
+				 struct sg_table *sg);
+struct dma_buf *amdgpu_gem_prime_export(struct drm_device *dev,
+					struct drm_gem_object *gobj,
+					int flags);
+struct drm_gem_object *amdgpu_gem_prime_import(struct drm_device *dev,
+					    struct dma_buf *dma_buf);
+struct reservation_object *amdgpu_gem_prime_res_obj(struct drm_gem_object *);
+void *amdgpu_gem_prime_vmap(struct drm_gem_object *obj);
+void amdgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr);
+#ifdef notyet
+int amdgpu_gem_prime_mmap(struct drm_gem_object *obj, struct vm_area_struct *vma);
+#endif
+
+/* sub-allocation manager, it has to be protected by another lock.
+ * By conception this is an helper for other part of the driver
+ * like the indirect buffer or semaphore, which both have their
+ * locking.
+ *
+ * Principe is simple, we keep a list of sub allocation in offset
+ * order (first entry has offset == 0, last entry has the highest
+ * offset).
+ *
+ * When allocating new object we first check if there is room at
+ * the end total_size - (last_object_offset + last_object_size) >=
+ * alloc_size. If so we allocate new object there.
+ *
+ * When there is not enough room at the end, we start waiting for
+ * each sub object until we reach object_offset+object_size >=
+ * alloc_size, this object then become the sub object we return.
+ *
+ * Alignment can't be bigger than page size.
+ *
+ * Hole are not considered for allocation to keep things simple.
+ * Assumption is that there won't be hole (all object on same
+ * alignment).
+ */
+
+#define AMDGPU_SA_NUM_FENCE_LISTS	32
+
+struct amdgpu_sa_manager {
+	wait_queue_head_t	wq;
+	struct amdgpu_bo	*bo;
+	struct list_head	*hole;
+	struct list_head	flist[AMDGPU_SA_NUM_FENCE_LISTS];
+	struct list_head	olist;
+	unsigned		size;
+	uint64_t		gpu_addr;
+	void			*cpu_ptr;
+	uint32_t		domain;
+	uint32_t		align;
+};
+
+/* sub-allocation buffer */
+struct amdgpu_sa_bo {
+	struct list_head		olist;
+	struct list_head		flist;
+	struct amdgpu_sa_manager	*manager;
+	unsigned			soffset;
+	unsigned			eoffset;
+	struct dma_fence	        *fence;
+};
+
+/*
+ * GEM objects.
+ */
+void amdgpu_gem_force_release(struct amdgpu_device *adev);
+int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
+			     int alignment, u32 initial_domain,
+			     u64 flags, enum ttm_bo_type type,
+			     struct reservation_object *resv,
+			     struct drm_gem_object **obj);
+
+int amdgpu_mode_dumb_create(struct drm_file *file_priv,
+			    struct drm_device *dev,
+			    struct drm_mode_create_dumb *args);
+int amdgpu_mode_dumb_mmap(struct drm_file *filp,
+			  struct drm_device *dev,
+			  uint32_t handle, uint64_t *offset_p);
+int amdgpu_fence_slab_init(void);
+void amdgpu_fence_slab_fini(void);
+
+/*
+ * GPU doorbell structures, functions & helpers
+ */
+typedef enum _AMDGPU_DOORBELL_ASSIGNMENT
+{
+	AMDGPU_DOORBELL_KIQ                     = 0x000,
+	AMDGPU_DOORBELL_HIQ                     = 0x001,
+	AMDGPU_DOORBELL_DIQ                     = 0x002,
+	AMDGPU_DOORBELL_MEC_RING0               = 0x010,
+	AMDGPU_DOORBELL_MEC_RING1               = 0x011,
+	AMDGPU_DOORBELL_MEC_RING2               = 0x012,
+	AMDGPU_DOORBELL_MEC_RING3               = 0x013,
+	AMDGPU_DOORBELL_MEC_RING4               = 0x014,
+	AMDGPU_DOORBELL_MEC_RING5               = 0x015,
+	AMDGPU_DOORBELL_MEC_RING6               = 0x016,
+	AMDGPU_DOORBELL_MEC_RING7               = 0x017,
+	AMDGPU_DOORBELL_GFX_RING0               = 0x020,
+	AMDGPU_DOORBELL_sDMA_ENGINE0            = 0x1E0,
+	AMDGPU_DOORBELL_sDMA_ENGINE1            = 0x1E1,
+	AMDGPU_DOORBELL_IH                      = 0x1E8,
+	AMDGPU_DOORBELL_MAX_ASSIGNMENT          = 0x3FF,
+	AMDGPU_DOORBELL_INVALID                 = 0xFFFF
+} AMDGPU_DOORBELL_ASSIGNMENT;
+
+struct amdgpu_doorbell {
+	/* doorbell mmio */
+	resource_size_t		base;
+	resource_size_t		size;
+#ifdef __linux__
+	u32 __iomem		*ptr;
+#endif
+	bus_space_tag_t		bst;
+	bus_space_handle_t	bsh;
+	u32			num_doorbells;	/* Number of doorbells actually reserved for amdgpu. */
+};
+
+/*
+ * 64bit doorbell, offset are in QWORD, occupy 2KB doorbell space
+ */
+typedef enum _AMDGPU_DOORBELL64_ASSIGNMENT
+{
+	/*
+	 * All compute related doorbells: kiq, hiq, diq, traditional compute queue, user queue, should locate in
+	 * a continues range so that programming CP_MEC_DOORBELL_RANGE_LOWER/UPPER can cover this range.
+	 *  Compute related doorbells are allocated from 0x00 to 0x8a
+	 */
+
+
+	/* kernel scheduling */
+	AMDGPU_DOORBELL64_KIQ                     = 0x00,
+
+	/* HSA interface queue and debug queue */
+	AMDGPU_DOORBELL64_HIQ                     = 0x01,
+	AMDGPU_DOORBELL64_DIQ                     = 0x02,
+
+	/* Compute engines */
+	AMDGPU_DOORBELL64_MEC_RING0               = 0x03,
+	AMDGPU_DOORBELL64_MEC_RING1               = 0x04,
+	AMDGPU_DOORBELL64_MEC_RING2               = 0x05,
+	AMDGPU_DOORBELL64_MEC_RING3               = 0x06,
+	AMDGPU_DOORBELL64_MEC_RING4               = 0x07,
+	AMDGPU_DOORBELL64_MEC_RING5               = 0x08,
+	AMDGPU_DOORBELL64_MEC_RING6               = 0x09,
+	AMDGPU_DOORBELL64_MEC_RING7               = 0x0a,
+
+	/* User queue doorbell range (128 doorbells) */
+	AMDGPU_DOORBELL64_USERQUEUE_START         = 0x0b,
+	AMDGPU_DOORBELL64_USERQUEUE_END           = 0x8a,
+
+	/* Graphics engine */
+	AMDGPU_DOORBELL64_GFX_RING0               = 0x8b,
+
+	/*
+	 * Other graphics doorbells can be allocated here: from 0x8c to 0xef
+	 * Graphics voltage island aperture 1
+	 * default non-graphics QWORD index is 0xF0 - 0xFF inclusive
+	 */
+
+	/* sDMA engines */
+	AMDGPU_DOORBELL64_sDMA_ENGINE0            = 0xF0,
+	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE0     = 0xF1,
+	AMDGPU_DOORBELL64_sDMA_ENGINE1            = 0xF2,
+	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE1     = 0xF3,
+
+	/* Interrupt handler */
+	AMDGPU_DOORBELL64_IH                      = 0xF4,  /* For legacy interrupt ring buffer */
+	AMDGPU_DOORBELL64_IH_RING1                = 0xF5,  /* For page migration request log */
+	AMDGPU_DOORBELL64_IH_RING2                = 0xF6,  /* For page migration translation/invalidation log */
+
+	/* VCN engine use 32 bits doorbell  */
+	AMDGPU_DOORBELL64_VCN0_1                  = 0xF8, /* lower 32 bits for VNC0 and upper 32 bits for VNC1 */
+	AMDGPU_DOORBELL64_VCN2_3                  = 0xF9,
+	AMDGPU_DOORBELL64_VCN4_5                  = 0xFA,
+	AMDGPU_DOORBELL64_VCN6_7                  = 0xFB,
+
+	/* overlap the doorbell assignment with VCN as they are  mutually exclusive
+	 * VCE engine's doorbell is 32 bit and two VCE ring share one QWORD
+	 */
+	AMDGPU_DOORBELL64_UVD_RING0_1             = 0xF8,
+	AMDGPU_DOORBELL64_UVD_RING2_3             = 0xF9,
+	AMDGPU_DOORBELL64_UVD_RING4_5             = 0xFA,
+	AMDGPU_DOORBELL64_UVD_RING6_7             = 0xFB,
+
+	AMDGPU_DOORBELL64_VCE_RING0_1             = 0xFC,
+	AMDGPU_DOORBELL64_VCE_RING2_3             = 0xFD,
+	AMDGPU_DOORBELL64_VCE_RING4_5             = 0xFE,
+	AMDGPU_DOORBELL64_VCE_RING6_7             = 0xFF,
+
+	AMDGPU_DOORBELL64_MAX_ASSIGNMENT          = 0xFF,
+	AMDGPU_DOORBELL64_INVALID                 = 0xFFFF
+} AMDGPU_DOORBELL64_ASSIGNMENT;
+
+/*
+ * IRQS.
+ */
+
+struct amdgpu_flip_work {
+	struct delayed_work		flip_work;
+	struct work_struct		unpin_work;
+	struct amdgpu_device		*adev;
+	int				crtc_id;
+	u32				target_vblank;
+	uint64_t			base;
+	struct drm_pending_vblank_event *event;
+	struct amdgpu_bo		*old_abo;
+	struct dma_fence		*excl;
+	unsigned			shared_count;
+	struct dma_fence		**shared;
+	struct dma_fence_cb		cb;
+	bool				async;
+};
+
+
+/*
+ * CP & rings.
+ */
+
+struct amdgpu_ib {
+	struct amdgpu_sa_bo		*sa_bo;
+	uint32_t			length_dw;
+	uint64_t			gpu_addr;
+	uint32_t			*ptr;
+	uint32_t			flags;
+};
+
+extern const struct drm_sched_backend_ops amdgpu_sched_ops;
+
+/*
+ * Queue manager
+ */
+struct amdgpu_queue_mapper {
+	int 		hw_ip;
+	struct rwlock	lock;
+	/* protected by lock */
+	struct amdgpu_ring *queue_map[AMDGPU_MAX_RINGS];
+};
+
+struct amdgpu_queue_mgr {
+	struct amdgpu_queue_mapper mapper[AMDGPU_MAX_IP_NUM];
+};
+
+int amdgpu_queue_mgr_init(struct amdgpu_device *adev,
+			  struct amdgpu_queue_mgr *mgr);
+int amdgpu_queue_mgr_fini(struct amdgpu_device *adev,
+			  struct amdgpu_queue_mgr *mgr);
+int amdgpu_queue_mgr_map(struct amdgpu_device *adev,
+			 struct amdgpu_queue_mgr *mgr,
+			 u32 hw_ip, u32 instance, u32 ring,
+			 struct amdgpu_ring **out_ring);
+
+/*
+ * context related structures
+ */
+
+struct amdgpu_ctx_ring {
+	uint64_t		sequence;
+	struct dma_fence	**fences;
+	struct drm_sched_entity	entity;
+};
+
+struct amdgpu_ctx {
+	struct kref		refcount;
+	struct amdgpu_device    *adev;
+	struct amdgpu_queue_mgr queue_mgr;
+	unsigned		reset_counter;
+	unsigned        reset_counter_query;
+	uint32_t		vram_lost_counter;
+	spinlock_t		ring_lock;
+	struct dma_fence	**fences;
+	struct amdgpu_ctx_ring	rings[AMDGPU_MAX_RINGS];
+	bool			preamble_presented;
+	enum drm_sched_priority init_priority;
+	enum drm_sched_priority override_priority;
+	struct rwlock            lock;
+	atomic_t	guilty;
+};
+
+struct amdgpu_ctx_mgr {
+	struct amdgpu_device	*adev;
+	struct rwlock		lock;
+	/* protected by lock */
+	struct idr		ctx_handles;
+};
+
+struct amdgpu_ctx *amdgpu_ctx_get(struct amdgpu_fpriv *fpriv, uint32_t id);
+int amdgpu_ctx_put(struct amdgpu_ctx *ctx);
+
+int amdgpu_ctx_add_fence(struct amdgpu_ctx *ctx, struct amdgpu_ring *ring,
+			      struct dma_fence *fence, uint64_t *seq);
+struct dma_fence *amdgpu_ctx_get_fence(struct amdgpu_ctx *ctx,
+				   struct amdgpu_ring *ring, uint64_t seq);
+void amdgpu_ctx_priority_override(struct amdgpu_ctx *ctx,
+				  enum drm_sched_priority priority);
+
+int amdgpu_ctx_ioctl(struct drm_device *dev, void *data,
+		     struct drm_file *filp);
+
+int amdgpu_ctx_wait_prev_fence(struct amdgpu_ctx *ctx, unsigned ring_id);
+
+void amdgpu_ctx_mgr_init(struct amdgpu_ctx_mgr *mgr);
+void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr);
+void amdgpu_ctx_mgr_entity_flush(struct amdgpu_ctx_mgr *mgr);
+void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr);
+
+
+/*
+ * file private structure
+ */
+
+struct amdgpu_fpriv {
+	struct amdgpu_vm	vm;
+	struct amdgpu_bo_va	*prt_va;
+	struct amdgpu_bo_va	*csa_va;
+	struct rwlock		bo_list_lock;
+	struct idr		bo_list_handles;
+	struct amdgpu_ctx_mgr	ctx_mgr;
+};
+
+/*
+ * GFX stuff
+ */
+#include "clearstate_defs.h"
+
+struct amdgpu_rlc_funcs {
+	void (*enter_safe_mode)(struct amdgpu_device *adev);
+	void (*exit_safe_mode)(struct amdgpu_device *adev);
+};
+
+struct amdgpu_rlc {
+	/* for power gating */
+	struct amdgpu_bo	*save_restore_obj;
+	uint64_t		save_restore_gpu_addr;
+	volatile uint32_t	*sr_ptr;
+	const u32               *reg_list;
+	u32                     reg_list_size;
+	/* for clear state */
+	struct amdgpu_bo	*clear_state_obj;
+	uint64_t		clear_state_gpu_addr;
+	volatile uint32_t	*cs_ptr;
+	const struct cs_section_def   *cs_data;
+	u32                     clear_state_size;
+	/* for cp tables */
+	struct amdgpu_bo	*cp_table_obj;
+	uint64_t		cp_table_gpu_addr;
+	volatile uint32_t	*cp_table_ptr;
+	u32                     cp_table_size;
+
+	/* safe mode for updating CG/PG state */
+	bool in_safe_mode;
+	const struct amdgpu_rlc_funcs *funcs;
+
+	/* for firmware data */
+	u32 save_and_restore_offset;
+	u32 clear_state_descriptor_offset;
+	u32 avail_scratch_ram_locations;
+	u32 reg_restore_list_size;
+	u32 reg_list_format_start;
+	u32 reg_list_format_separate_start;
+	u32 starting_offsets_start;
+	u32 reg_list_format_size_bytes;
+	u32 reg_list_size_bytes;
+	u32 reg_list_format_direct_reg_list_length;
+	u32 save_restore_list_cntl_size_bytes;
+	u32 save_restore_list_gpm_size_bytes;
+	u32 save_restore_list_srm_size_bytes;
+
+	u32 *register_list_format;
+	u32 *register_restore;
+	u8 *save_restore_list_cntl;
+	u8 *save_restore_list_gpm;
+	u8 *save_restore_list_srm;
+
+	bool is_rlc_v2_1;
+};
+
+#define AMDGPU_MAX_COMPUTE_QUEUES KGD_MAX_QUEUES
+
+struct amdgpu_mec {
+	struct amdgpu_bo	*hpd_eop_obj;
+	u64			hpd_eop_gpu_addr;
+	struct amdgpu_bo	*mec_fw_obj;
+	u64			mec_fw_gpu_addr;
+	u32 num_mec;
+	u32 num_pipe_per_mec;
+	u32 num_queue_per_pipe;
+	void			*mqd_backup[AMDGPU_MAX_COMPUTE_RINGS + 1];
+
+	/* These are the resources for which amdgpu takes ownership */
+	DECLARE_BITMAP(queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+};
+
+struct amdgpu_kiq {
+	u64			eop_gpu_addr;
+	struct amdgpu_bo	*eop_obj;
+	spinlock_t              ring_lock;
+	struct amdgpu_ring	ring;
+	struct amdgpu_irq_src	irq;
+};
+
+/*
+ * GPU scratch registers structures, functions & helpers
+ */
+struct amdgpu_scratch {
+	unsigned		num_reg;
+	uint32_t                reg_base;
+	uint32_t		free_mask;
+};
+
+/*
+ * GFX configurations
+ */
+#define AMDGPU_GFX_MAX_SE 4
+#define AMDGPU_GFX_MAX_SH_PER_SE 2
+
+struct amdgpu_rb_config {
+	uint32_t rb_backend_disable;
+	uint32_t user_rb_backend_disable;
+	uint32_t raster_config;
+	uint32_t raster_config_1;
+};
+
+struct gb_addr_config {
+	uint16_t pipe_interleave_size;
+	uint8_t num_pipes;
+	uint8_t max_compress_frags;
+	uint8_t num_banks;
+	uint8_t num_se;
+	uint8_t num_rb_per_se;
+};
+
+struct amdgpu_gfx_config {
+	unsigned max_shader_engines;
+	unsigned max_tile_pipes;
+	unsigned max_cu_per_sh;
+	unsigned max_sh_per_se;
+	unsigned max_backends_per_se;
+	unsigned max_texture_channel_caches;
+	unsigned max_gprs;
+	unsigned max_gs_threads;
+	unsigned max_hw_contexts;
+	unsigned sc_prim_fifo_size_frontend;
+	unsigned sc_prim_fifo_size_backend;
+	unsigned sc_hiz_tile_fifo_size;
+	unsigned sc_earlyz_tile_fifo_size;
+
+	unsigned num_tile_pipes;
+	unsigned backend_enable_mask;
+	unsigned mem_max_burst_length_bytes;
+	unsigned mem_row_size_in_kb;
+	unsigned shader_engine_tile_size;
+	unsigned num_gpus;
+	unsigned multi_gpu_tile_size;
+	unsigned mc_arb_ramcfg;
+	unsigned gb_addr_config;
+	unsigned num_rbs;
+	unsigned gs_vgt_table_depth;
+	unsigned gs_prim_buffer_depth;
+
+	uint32_t tile_mode_array[32];
+	uint32_t macrotile_mode_array[16];
+
+	struct gb_addr_config gb_addr_config_fields;
+	struct amdgpu_rb_config rb_config[AMDGPU_GFX_MAX_SE][AMDGPU_GFX_MAX_SH_PER_SE];
+
+	/* gfx configure feature */
+	uint32_t double_offchip_lds_buf;
+	/* cached value of DB_DEBUG2 */
+	uint32_t db_debug2;
+};
+
+struct amdgpu_cu_info {
+	uint32_t simd_per_cu;
+	uint32_t max_waves_per_simd;
+	uint32_t wave_front_size;
+	uint32_t max_scratch_slots_per_cu;
+	uint32_t lds_size;
+
+	/* total active CU number */
+	uint32_t number;
+	uint32_t ao_cu_mask;
+	uint32_t ao_cu_bitmap[4][4];
+	uint32_t bitmap[4][4];
+};
+
+struct amdgpu_gfx_funcs {
+	/* get the gpu clock counter */
+	uint64_t (*get_gpu_clock_counter)(struct amdgpu_device *adev);
+	void (*select_se_sh)(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 instance);
+	void (*read_wave_data)(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields);
+	void (*read_wave_vgprs)(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t thread, uint32_t start, uint32_t size, uint32_t *dst);
+	void (*read_wave_sgprs)(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t start, uint32_t size, uint32_t *dst);
+	void (*select_me_pipe_q)(struct amdgpu_device *adev, u32 me, u32 pipe, u32 queue);
+};
+
+struct amdgpu_ngg_buf {
+	struct amdgpu_bo	*bo;
+	uint64_t		gpu_addr;
+	uint32_t		size;
+	uint32_t		bo_size;
+};
+
+enum {
+	NGG_PRIM = 0,
+	NGG_POS,
+	NGG_CNTL,
+	NGG_PARAM,
+	NGG_BUF_MAX
+};
+
+struct amdgpu_ngg {
+	struct amdgpu_ngg_buf	buf[NGG_BUF_MAX];
+	uint32_t		gds_reserve_addr;
+	uint32_t		gds_reserve_size;
+	bool			init;
+};
+
+struct sq_work {
+	struct work_struct	work;
+	unsigned ih_data;
+};
+
+struct amdgpu_gfx {
+	struct rwlock			gpu_clock_mutex;
+	struct amdgpu_gfx_config	config;
+	struct amdgpu_rlc		rlc;
+	struct amdgpu_mec		mec;
+	struct amdgpu_kiq		kiq;
+	struct amdgpu_scratch		scratch;
+	const struct firmware		*me_fw;	/* ME firmware */
+	uint32_t			me_fw_version;
+	const struct firmware		*pfp_fw; /* PFP firmware */
+	uint32_t			pfp_fw_version;
+	const struct firmware		*ce_fw;	/* CE firmware */
+	uint32_t			ce_fw_version;
+	const struct firmware		*rlc_fw; /* RLC firmware */
+	uint32_t			rlc_fw_version;
+	const struct firmware		*mec_fw; /* MEC firmware */
+	uint32_t			mec_fw_version;
+	const struct firmware		*mec2_fw; /* MEC2 firmware */
+	uint32_t			mec2_fw_version;
+	uint32_t			me_feature_version;
+	uint32_t			ce_feature_version;
+	uint32_t			pfp_feature_version;
+	uint32_t			rlc_feature_version;
+	uint32_t			rlc_srlc_fw_version;
+	uint32_t			rlc_srlc_feature_version;
+	uint32_t			rlc_srlg_fw_version;
+	uint32_t			rlc_srlg_feature_version;
+	uint32_t			rlc_srls_fw_version;
+	uint32_t			rlc_srls_feature_version;
+	uint32_t			mec_feature_version;
+	uint32_t			mec2_feature_version;
+	struct amdgpu_ring		gfx_ring[AMDGPU_MAX_GFX_RINGS];
+	unsigned			num_gfx_rings;
+	struct amdgpu_ring		compute_ring[AMDGPU_MAX_COMPUTE_RINGS];
+	unsigned			num_compute_rings;
+	struct amdgpu_irq_src		eop_irq;
+	struct amdgpu_irq_src		priv_reg_irq;
+	struct amdgpu_irq_src		priv_inst_irq;
+	struct amdgpu_irq_src		cp_ecc_error_irq;
+	struct amdgpu_irq_src		sq_irq;
+	struct sq_work			sq_work;
+
+	/* gfx status */
+	uint32_t			gfx_current_status;
+	/* ce ram size*/
+	unsigned			ce_ram_size;
+	struct amdgpu_cu_info		cu_info;
+	const struct amdgpu_gfx_funcs	*funcs;
+
+	/* reset mask */
+	uint32_t                        grbm_soft_reset;
+	uint32_t                        srbm_soft_reset;
+	/* s3/s4 mask */
+	bool                            in_suspend;
+	/* NGG */
+	struct amdgpu_ngg		ngg;
+
+	/* pipe reservation */
+	struct rwlock			pipe_reserve_mutex;
+	DECLARE_BITMAP			(pipe_reserve_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+};
+
+int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		  unsigned size, struct amdgpu_ib *ib);
+void amdgpu_ib_free(struct amdgpu_device *adev, struct amdgpu_ib *ib,
+		    struct dma_fence *f);
+int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
+		       struct amdgpu_ib *ibs, struct amdgpu_job *job,
+		       struct dma_fence **f);
+int amdgpu_ib_pool_init(struct amdgpu_device *adev);
+void amdgpu_ib_pool_fini(struct amdgpu_device *adev);
+int amdgpu_ib_ring_tests(struct amdgpu_device *adev);
+
+/*
+ * CS.
+ */
+struct amdgpu_cs_chunk {
+	uint32_t		chunk_id;
+	uint32_t		length_dw;
+	void			*kdata;
+};
+
+struct amdgpu_cs_parser {
+	struct amdgpu_device	*adev;
+	struct drm_file		*filp;
+	struct amdgpu_ctx	*ctx;
+
+	/* chunks */
+	unsigned		nchunks;
+	struct amdgpu_cs_chunk	*chunks;
+
+	/* scheduler job object */
+	struct amdgpu_job	*job;
+	struct amdgpu_ring	*ring;
+
+	/* buffer objects */
+	struct ww_acquire_ctx		ticket;
+	struct amdgpu_bo_list		*bo_list;
+	struct amdgpu_mn		*mn;
+	struct amdgpu_bo_list_entry	vm_pd;
+	struct list_head		validated;
+	struct dma_fence		*fence;
+	uint64_t			bytes_moved_threshold;
+	uint64_t			bytes_moved_vis_threshold;
+	uint64_t			bytes_moved;
+	uint64_t			bytes_moved_vis;
+	struct amdgpu_bo_list_entry	*evictable;
+
+	/* user fence */
+	struct amdgpu_bo_list_entry	uf_entry;
+
+	unsigned num_post_dep_syncobjs;
+	struct drm_syncobj **post_dep_syncobjs;
+};
+
+static inline u32 amdgpu_get_ib_value(struct amdgpu_cs_parser *p,
+				      uint32_t ib_idx, int idx)
+{
+	return p->job->ibs[ib_idx].ptr[idx];
+}
+
+static inline void amdgpu_set_ib_value(struct amdgpu_cs_parser *p,
+				       uint32_t ib_idx, int idx,
+				       uint32_t value)
+{
+	p->job->ibs[ib_idx].ptr[idx] = value;
+}
+
+/*
+ * Writeback
+ */
+#define AMDGPU_MAX_WB 128	/* Reserve at most 128 WB slots for amdgpu-owned rings. */
+
+struct amdgpu_wb {
+	struct amdgpu_bo	*wb_obj;
+	volatile uint32_t	*wb;
+	uint64_t		gpu_addr;
+	u32			num_wb;	/* Number of wb slots actually reserved for amdgpu. */
+	unsigned long		used[DIV_ROUND_UP(AMDGPU_MAX_WB, BITS_PER_LONG)];
+};
+
+int amdgpu_device_wb_get(struct amdgpu_device *adev, u32 *wb);
+void amdgpu_device_wb_free(struct amdgpu_device *adev, u32 wb);
+
+/*
+ * SDMA
+ */
+struct amdgpu_sdma_instance {
+	/* SDMA firmware */
+	const struct firmware	*fw;
+	uint32_t		fw_version;
+	uint32_t		feature_version;
+
+	struct amdgpu_ring	ring;
+	bool			burst_nop;
+};
+
+struct amdgpu_sdma {
+	struct amdgpu_sdma_instance instance[AMDGPU_MAX_SDMA_INSTANCES];
+#ifdef CONFIG_DRM_AMDGPU_SI
+	//SI DMA has a difference trap irq number for the second engine
+	struct amdgpu_irq_src	trap_irq_1;
+#endif
+	struct amdgpu_irq_src	trap_irq;
+	struct amdgpu_irq_src	illegal_inst_irq;
+	int			num_instances;
+	uint32_t                    srbm_soft_reset;
+};
+
+/*
+ * Firmware
+ */
+enum amdgpu_firmware_load_type {
+	AMDGPU_FW_LOAD_DIRECT = 0,
+	AMDGPU_FW_LOAD_SMU,
+	AMDGPU_FW_LOAD_PSP,
+};
+
+struct amdgpu_firmware {
+	struct amdgpu_firmware_info ucode[AMDGPU_UCODE_ID_MAXIMUM];
+	enum amdgpu_firmware_load_type load_type;
+	struct amdgpu_bo *fw_buf;
+	unsigned int fw_size;
+	unsigned int max_ucodes;
+	/* firmwares are loaded by psp instead of smu from vega10 */
+	const struct amdgpu_psp_funcs *funcs;
+	struct amdgpu_bo *rbuf;
+	struct rwlock mutex;
+
+	/* gpu info firmware data pointer */
+	const struct firmware *gpu_info_fw;
+
+	void *fw_buf_ptr;
+	uint64_t fw_buf_mc;
+};
+
+/*
+ * Benchmarking
+ */
+void amdgpu_benchmark(struct amdgpu_device *adev, int test_number);
+
+
+/*
+ * Testing
+ */
+void amdgpu_test_moves(struct amdgpu_device *adev);
+
+
+/*
+ * amdgpu smumgr functions
+ */
+struct amdgpu_smumgr_funcs {
+	int (*check_fw_load_finish)(struct amdgpu_device *adev, uint32_t fwtype);
+	int (*request_smu_load_fw)(struct amdgpu_device *adev);
+	int (*request_smu_specific_fw)(struct amdgpu_device *adev, uint32_t fwtype);
+};
+
+/*
+ * amdgpu smumgr
+ */
+struct amdgpu_smumgr {
+	struct amdgpu_bo *toc_buf;
+	struct amdgpu_bo *smu_buf;
+	/* asic priv smu data */
+	void *priv;
+	spinlock_t smu_lock;
+	/* smumgr functions */
+	const struct amdgpu_smumgr_funcs *smumgr_funcs;
+	/* ucode loading complete flag */
+	uint32_t fw_flags;
+};
+
+/*
+ * ASIC specific register table accessible by UMD
+ */
+struct amdgpu_allowed_register_entry {
+	uint32_t reg_offset;
+	bool grbm_indexed;
+};
+
+/*
+ * ASIC specific functions.
+ */
+struct amdgpu_asic_funcs {
+	bool (*read_disabled_bios)(struct amdgpu_device *adev);
+	bool (*read_bios_from_rom)(struct amdgpu_device *adev,
+				   u8 *bios, u32 length_bytes);
+	int (*read_register)(struct amdgpu_device *adev, u32 se_num,
+			     u32 sh_num, u32 reg_offset, u32 *value);
+	void (*set_vga_state)(struct amdgpu_device *adev, bool state);
+	int (*reset)(struct amdgpu_device *adev);
+	/* get the reference clock */
+	u32 (*get_xclk)(struct amdgpu_device *adev);
+	/* MM block clocks */
+	int (*set_uvd_clocks)(struct amdgpu_device *adev, u32 vclk, u32 dclk);
+	int (*set_vce_clocks)(struct amdgpu_device *adev, u32 evclk, u32 ecclk);
+	/* static power management */
+	int (*get_pcie_lanes)(struct amdgpu_device *adev);
+	void (*set_pcie_lanes)(struct amdgpu_device *adev, int lanes);
+	/* get config memsize register */
+	u32 (*get_config_memsize)(struct amdgpu_device *adev);
+	/* flush hdp write queue */
+	void (*flush_hdp)(struct amdgpu_device *adev, struct amdgpu_ring *ring);
+	/* invalidate hdp read cache */
+	void (*invalidate_hdp)(struct amdgpu_device *adev,
+			       struct amdgpu_ring *ring);
+	/* check if the asic needs a full reset of if soft reset will work */
+	bool (*need_full_reset)(struct amdgpu_device *adev);
+};
+
+/*
+ * IOCTL.
+ */
+int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
+			    struct drm_file *filp);
+int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *filp);
+
+int amdgpu_gem_info_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp);
+int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
+			struct drm_file *filp);
+int amdgpu_gem_mmap_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp);
+int amdgpu_gem_wait_idle_ioctl(struct drm_device *dev, void *data,
+			      struct drm_file *filp);
+int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp);
+int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
+			struct drm_file *filp);
+int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp);
+int amdgpu_cs_fence_to_handle_ioctl(struct drm_device *dev, void *data,
+				    struct drm_file *filp);
+int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data, struct drm_file *filp);
+int amdgpu_cs_wait_fences_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *filp);
+
+int amdgpu_gem_metadata_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *filp);
+
+/* VRAM scratch page for HDP bug, default vram page */
+struct amdgpu_vram_scratch {
+	struct amdgpu_bo		*robj;
+	volatile uint32_t		*ptr;
+	u64				gpu_addr;
+};
+
+/*
+ * ACPI
+ */
+struct amdgpu_atcs_functions {
+	bool get_ext_state;
+	bool pcie_perf_req;
+	bool pcie_dev_rdy;
+	bool pcie_bus_width;
+};
+
+struct amdgpu_atcs {
+	struct amdgpu_atcs_functions functions;
+};
+
+/*
+ * Firmware VRAM reservation
+ */
+struct amdgpu_fw_vram_usage {
+	u64 start_offset;
+	u64 size;
+	struct amdgpu_bo *reserved_bo;
+	void *va;
+};
+
+/*
+ * CGS
+ */
+struct cgs_device *amdgpu_cgs_create_device(struct amdgpu_device *adev);
+void amdgpu_cgs_destroy_device(struct cgs_device *cgs_device);
+
+/*
+ * Core structure, functions and helpers.
+ */
+typedef uint32_t (*amdgpu_rreg_t)(struct amdgpu_device*, uint32_t);
+typedef void (*amdgpu_wreg_t)(struct amdgpu_device*, uint32_t, uint32_t);
+
+typedef uint32_t (*amdgpu_block_rreg_t)(struct amdgpu_device*, uint32_t, uint32_t);
+typedef void (*amdgpu_block_wreg_t)(struct amdgpu_device*, uint32_t, uint32_t, uint32_t);
+
+
+/*
+ * amdgpu nbio functions
+ *
+ */
+struct nbio_hdp_flush_reg {
+	u32 ref_and_mask_cp0;
+	u32 ref_and_mask_cp1;
+	u32 ref_and_mask_cp2;
+	u32 ref_and_mask_cp3;
+	u32 ref_and_mask_cp4;
+	u32 ref_and_mask_cp5;
+	u32 ref_and_mask_cp6;
+	u32 ref_and_mask_cp7;
+	u32 ref_and_mask_cp8;
+	u32 ref_and_mask_cp9;
+	u32 ref_and_mask_sdma0;
+	u32 ref_and_mask_sdma1;
+};
+
+struct amdgpu_nbio_funcs {
+	const struct nbio_hdp_flush_reg *hdp_flush_reg;
+	u32 (*get_hdp_flush_req_offset)(struct amdgpu_device *adev);
+	u32 (*get_hdp_flush_done_offset)(struct amdgpu_device *adev);
+	u32 (*get_pcie_index_offset)(struct amdgpu_device *adev);
+	u32 (*get_pcie_data_offset)(struct amdgpu_device *adev);
+	u32 (*get_rev_id)(struct amdgpu_device *adev);
+	void (*mc_access_enable)(struct amdgpu_device *adev, bool enable);
+	void (*hdp_flush)(struct amdgpu_device *adev, struct amdgpu_ring *ring);
+	u32 (*get_memsize)(struct amdgpu_device *adev);
+	void (*sdma_doorbell_range)(struct amdgpu_device *adev, int instance,
+				    bool use_doorbell, int doorbell_index);
+	void (*enable_doorbell_aperture)(struct amdgpu_device *adev,
+					 bool enable);
+	void (*enable_doorbell_selfring_aperture)(struct amdgpu_device *adev,
+						  bool enable);
+	void (*ih_doorbell_range)(struct amdgpu_device *adev,
+				  bool use_doorbell, int doorbell_index);
+	void (*update_medium_grain_clock_gating)(struct amdgpu_device *adev,
+						 bool enable);
+	void (*update_medium_grain_light_sleep)(struct amdgpu_device *adev,
+						bool enable);
+	void (*get_clockgating_state)(struct amdgpu_device *adev,
+				      u32 *flags);
+	void (*ih_control)(struct amdgpu_device *adev);
+	void (*init_registers)(struct amdgpu_device *adev);
+	void (*detect_hw_virt)(struct amdgpu_device *adev);
+};
+
+struct amdgpu_df_funcs {
+	void (*init)(struct amdgpu_device *adev);
+	void (*enable_broadcast_mode)(struct amdgpu_device *adev,
+				      bool enable);
+	u32 (*get_fb_channel_number)(struct amdgpu_device *adev);
+	u32 (*get_hbm_channel_number)(struct amdgpu_device *adev);
+	void (*update_medium_grain_clock_gating)(struct amdgpu_device *adev,
+						 bool enable);
+	void (*get_clockgating_state)(struct amdgpu_device *adev,
+				      u32 *flags);
+	void (*enable_ecc_force_par_wr_rmw)(struct amdgpu_device *adev,
+					    bool enable);
+};
+/* Define the HW IP blocks will be used in driver , add more if necessary */
+enum amd_hw_ip_block_type {
+	GC_HWIP = 1,
+	HDP_HWIP,
+	SDMA0_HWIP,
+	SDMA1_HWIP,
+	MMHUB_HWIP,
+	ATHUB_HWIP,
+	NBIO_HWIP,
+	MP0_HWIP,
+	MP1_HWIP,
+	UVD_HWIP,
+	VCN_HWIP = UVD_HWIP,
+	VCE_HWIP,
+	DF_HWIP,
+	DCE_HWIP,
+	OSSSYS_HWIP,
+	SMUIO_HWIP,
+	PWR_HWIP,
+	NBIF_HWIP,
+	THM_HWIP,
+	CLK_HWIP,
+	MAX_HWIP
+};
+
+#define HWIP_MAX_INSTANCE	6
+
+struct amd_powerplay {
+	void *pp_handle;
+	const struct amd_pm_funcs *pp_funcs;
+	uint32_t pp_feature;
+};
+
+#define AMDGPU_RESET_MAGIC_NUM 64
+struct amdgpu_device {
+	struct device			self;
+	struct device			*dev;
+	struct drm_device		*ddev;
+	struct pci_dev			*pdev;
+
+#ifdef CONFIG_DRM_AMD_ACP
+	struct amdgpu_acp		acp;
+#endif
+
+	pci_chipset_tag_t		pc;
+	pcitag_t			pa_tag;
+	pci_intr_handle_t		intrh;
+	bus_space_tag_t			iot;
+	bus_space_tag_t			memt;
+	bus_dma_tag_t			dmat;
+	void				*irqh;
+
+	void				(*switchcb)(void *, int, int);
+	void				*switchcbarg;
+	void				*switchcookie;
+	struct task			switchtask;
+	struct rasops_info		ro;
+	int				console;
+	int				primary;
+
+	struct task			burner_task;
+	int				burner_fblank;
+
+	unsigned long			fb_aper_offset;
+	unsigned long			fb_aper_size;
+
+	/* ASIC */
+	enum amd_asic_type		asic_type;
+	uint32_t			family;
+	uint32_t			rev_id;
+	uint32_t			external_rev_id;
+	unsigned long			flags;
+	int				usec_timeout;
+	const struct amdgpu_asic_funcs	*asic_funcs;
+	bool				shutdown;
+	bool				need_dma32;
+	bool				need_swiotlb;
+	bool				accel_working;
+	struct work_struct		reset_work;
+	struct notifier_block		acpi_nb;
+	struct amdgpu_i2c_chan		*i2c_bus[AMDGPU_MAX_I2C_BUS];
+	struct amdgpu_debugfs		debugfs[AMDGPU_DEBUGFS_MAX_COMPONENTS];
+	unsigned			debugfs_count;
+#if defined(CONFIG_DEBUG_FS)
+	struct dentry			*debugfs_regs[AMDGPU_DEBUGFS_MAX_COMPONENTS];
+#endif
+	struct amdgpu_atif		*atif;
+	struct amdgpu_atcs		atcs;
+	struct rwlock			srbm_mutex;
+	/* GRBM index mutex. Protects concurrent access to GRBM index */
+	struct rwlock                    grbm_idx_mutex;
+	struct dev_pm_domain		vga_pm_domain;
+	bool				have_disp_power_ref;
+
+	/* BIOS */
+	bool				is_atom_fw;
+	uint8_t				*bios;
+	uint32_t			bios_size;
+	struct amdgpu_bo		*stolen_vga_memory;
+	uint32_t			bios_scratch_reg_offset;
+	uint32_t			bios_scratch[AMDGPU_BIOS_NUM_SCRATCH];
+
+	/* Register/doorbell mmio */
+	resource_size_t			rmmio_base;
+	resource_size_t			rmmio_size;
+#ifdef __linux__
+	void __iomem			*rmmio;
+#endif
+	bus_space_tag_t			rmmio_bst;
+	bus_space_handle_t		rmmio_bsh;
+	/* protects concurrent MM_INDEX/DATA based register access */
+	spinlock_t mmio_idx_lock;
+	/* protects concurrent SMC based register access */
+	spinlock_t smc_idx_lock;
+	amdgpu_rreg_t			smc_rreg;
+	amdgpu_wreg_t			smc_wreg;
+	/* protects concurrent PCIE register access */
+	spinlock_t pcie_idx_lock;
+	amdgpu_rreg_t			pcie_rreg;
+	amdgpu_wreg_t			pcie_wreg;
+	amdgpu_rreg_t			pciep_rreg;
+	amdgpu_wreg_t			pciep_wreg;
+	/* protects concurrent UVD register access */
+	spinlock_t uvd_ctx_idx_lock;
+	amdgpu_rreg_t			uvd_ctx_rreg;
+	amdgpu_wreg_t			uvd_ctx_wreg;
+	/* protects concurrent DIDT register access */
+	spinlock_t didt_idx_lock;
+	amdgpu_rreg_t			didt_rreg;
+	amdgpu_wreg_t			didt_wreg;
+	/* protects concurrent gc_cac register access */
+	spinlock_t gc_cac_idx_lock;
+	amdgpu_rreg_t			gc_cac_rreg;
+	amdgpu_wreg_t			gc_cac_wreg;
+	/* protects concurrent se_cac register access */
+	spinlock_t se_cac_idx_lock;
+	amdgpu_rreg_t			se_cac_rreg;
+	amdgpu_wreg_t			se_cac_wreg;
+	/* protects concurrent ENDPOINT (audio) register access */
+	spinlock_t audio_endpt_idx_lock;
+	amdgpu_block_rreg_t		audio_endpt_rreg;
+	amdgpu_block_wreg_t		audio_endpt_wreg;
+#ifdef notyet
+	void __iomem			*rio_mem;
+#endif
+	bus_space_tag_t			rio_mem_bst;
+	bus_space_handle_t		rio_mem_bsh;
+	resource_size_t			rio_mem_size;
+	struct amdgpu_doorbell		doorbell;
+
+	/* clock/pll info */
+	struct amdgpu_clock            clock;
+
+	/* MC */
+	struct amdgpu_gmc		gmc;
+	struct amdgpu_gart		gart;
+	dma_addr_t			dummy_page_addr;
+	struct amdgpu_vm_manager	vm_manager;
+	struct amdgpu_vmhub             vmhub[AMDGPU_MAX_VMHUBS];
+
+	/* memory management */
+	struct amdgpu_mman		mman;
+	struct amdgpu_vram_scratch	vram_scratch;
+	struct amdgpu_wb		wb;
+	atomic64_t			num_bytes_moved;
+	atomic64_t			num_evictions;
+	atomic64_t			num_vram_cpu_page_faults;
+	atomic_t			gpu_reset_counter;
+	atomic_t			vram_lost_counter;
+
+	/* data for buffer migration throttling */
+	struct {
+		spinlock_t		lock;
+		s64			last_update_us;
+		s64			accum_us; /* accumulated microseconds */
+		s64			accum_us_vis; /* for visible VRAM */
+		u32			log2_max_MBps;
+	} mm_stats;
+
+	/* display */
+	bool				enable_virtual_display;
+	struct amdgpu_mode_info		mode_info;
+	/* For pre-DCE11. DCE11 and later are in "struct amdgpu_device->dm" */
+	struct work_struct		hotplug_work;
+	struct amdgpu_irq_src		crtc_irq;
+	struct amdgpu_irq_src		pageflip_irq;
+	struct amdgpu_irq_src		hpd_irq;
+
+	/* rings */
+	u64				fence_context;
+	unsigned			num_rings;
+	struct amdgpu_ring		*rings[AMDGPU_MAX_RINGS];
+	bool				ib_pool_ready;
+	struct amdgpu_sa_manager	ring_tmp_bo;
+
+	/* interrupts */
+	struct amdgpu_irq		irq;
+
+	/* powerplay */
+	struct amd_powerplay		powerplay;
+	bool				pp_force_state_enabled;
+
+	/* dpm */
+	struct amdgpu_pm		pm;
+	u32				cg_flags;
+	u32				pg_flags;
+
+	/* amdgpu smumgr */
+	struct amdgpu_smumgr smu;
+
+	/* gfx */
+	struct amdgpu_gfx		gfx;
+
+	/* sdma */
+	struct amdgpu_sdma		sdma;
+
+	/* uvd */
+	struct amdgpu_uvd		uvd;
+
+	/* vce */
+	struct amdgpu_vce		vce;
+
+	/* vcn */
+	struct amdgpu_vcn		vcn;
+
+	/* firmwares */
+	struct amdgpu_firmware		firmware;
+
+	/* PSP */
+	struct psp_context		psp;
+
+	/* GDS */
+	struct amdgpu_gds		gds;
+
+	/* display related functionality */
+	struct amdgpu_display_manager dm;
+
+	struct amdgpu_ip_block          ip_blocks[AMDGPU_MAX_IP_NUM];
+	int				num_ip_blocks;
+	struct rwlock	mn_lock;
+	DECLARE_HASHTABLE(mn_hash, 7);
+
+	/* tracking pinned memory */
+	atomic64_t vram_pin_size;
+	atomic64_t visible_pin_size;
+	atomic64_t gart_pin_size;
+
+	/* amdkfd interface */
+	struct kfd_dev          *kfd;
+
+	/* soc15 register offset based on ip, instance and  segment */
+	uint32_t 		*reg_offset[MAX_HWIP][HWIP_MAX_INSTANCE];
+
+	const struct amdgpu_nbio_funcs	*nbio_funcs;
+	const struct amdgpu_df_funcs	*df_funcs;
+
+	/* delayed work_func for deferring clockgating during resume */
+	struct delayed_work     late_init_work;
+
+	struct amdgpu_virt	virt;
+	/* firmware VRAM reservation */
+	struct amdgpu_fw_vram_usage fw_vram_usage;
+
+	/* link all shadow bo */
+	struct list_head                shadow_list;
+	struct rwlock                    shadow_list_lock;
+	/* keep an lru list of rings by HW IP */
+	struct list_head		ring_lru_list;
+	spinlock_t			ring_lru_list_lock;
+
+	/* record hw reset is performed */
+	bool has_hw_reset;
+	u8				reset_magic[AMDGPU_RESET_MAGIC_NUM];
+
+	/* record last mm index being written through WREG32*/
+	unsigned long last_mm_index;
+	bool                            in_gpu_reset;
+	struct rwlock  lock_reset;
+};
+
+static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_bo_device *bdev)
+{
+	return container_of(bdev, struct amdgpu_device, mman.bdev);
+}
+
+int amdgpu_device_init(struct amdgpu_device *adev,
+		       struct drm_device *ddev,
+		       struct pci_dev *pdev,
+		       uint32_t flags);
+void amdgpu_device_fini(struct amdgpu_device *adev);
+int amdgpu_gpu_wait_for_idle(struct amdgpu_device *adev);
+
+uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg,
+			uint32_t acc_flags);
+void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
+		    uint32_t acc_flags);
+void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value);
+uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset);
+
+u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg);
+void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v);
+
+u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index);
+void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v);
+u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index);
+void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v);
+
+bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type);
+bool amdgpu_device_has_dc_support(struct amdgpu_device *adev);
+
+int emu_soc_asic_init(struct amdgpu_device *adev);
+
+/*
+ * Registers read & write functions.
+ */
+
+#define AMDGPU_REGS_IDX       (1<<0)
+#define AMDGPU_REGS_NO_KIQ    (1<<1)
+
+#define RREG32_NO_KIQ(reg) amdgpu_mm_rreg(adev, (reg), AMDGPU_REGS_NO_KIQ)
+#define WREG32_NO_KIQ(reg, v) amdgpu_mm_wreg(adev, (reg), (v), AMDGPU_REGS_NO_KIQ)
+
+#define RREG8(reg) amdgpu_mm_rreg8(adev, (reg))
+#define WREG8(reg, v) amdgpu_mm_wreg8(adev, (reg), (v))
+
+#define RREG32(reg) amdgpu_mm_rreg(adev, (reg), 0)
+#define RREG32_IDX(reg) amdgpu_mm_rreg(adev, (reg), AMDGPU_REGS_IDX)
+#define DREG32(reg) printk(KERN_INFO "REGISTER: " #reg " : 0x%08X\n", amdgpu_mm_rreg(adev, (reg), 0))
+#define WREG32(reg, v) amdgpu_mm_wreg(adev, (reg), (v), 0)
+#define WREG32_IDX(reg, v) amdgpu_mm_wreg(adev, (reg), (v), AMDGPU_REGS_IDX)
+#define REG_SET(FIELD, v) (((v) << FIELD##_SHIFT) & FIELD##_MASK)
+#define REG_GET(FIELD, v) (((v) << FIELD##_SHIFT) & FIELD##_MASK)
+#define RREG32_PCIE(reg) adev->pcie_rreg(adev, (reg))
+#define WREG32_PCIE(reg, v) adev->pcie_wreg(adev, (reg), (v))
+#define RREG32_PCIE_PORT(reg) adev->pciep_rreg(adev, (reg))
+#define WREG32_PCIE_PORT(reg, v) adev->pciep_wreg(adev, (reg), (v))
+#define RREG32_SMC(reg) adev->smc_rreg(adev, (reg))
+#define WREG32_SMC(reg, v) adev->smc_wreg(adev, (reg), (v))
+#define RREG32_UVD_CTX(reg) adev->uvd_ctx_rreg(adev, (reg))
+#define WREG32_UVD_CTX(reg, v) adev->uvd_ctx_wreg(adev, (reg), (v))
+#define RREG32_DIDT(reg) adev->didt_rreg(adev, (reg))
+#define WREG32_DIDT(reg, v) adev->didt_wreg(adev, (reg), (v))
+#define RREG32_GC_CAC(reg) adev->gc_cac_rreg(adev, (reg))
+#define WREG32_GC_CAC(reg, v) adev->gc_cac_wreg(adev, (reg), (v))
+#define RREG32_SE_CAC(reg) adev->se_cac_rreg(adev, (reg))
+#define WREG32_SE_CAC(reg, v) adev->se_cac_wreg(adev, (reg), (v))
+#define RREG32_AUDIO_ENDPT(block, reg) adev->audio_endpt_rreg(adev, (block), (reg))
+#define WREG32_AUDIO_ENDPT(block, reg, v) adev->audio_endpt_wreg(adev, (block), (reg), (v))
+#define WREG32_P(reg, val, mask)				\
+	do {							\
+		uint32_t tmp_ = RREG32(reg);			\
+		tmp_ &= (mask);					\
+		tmp_ |= ((val) & ~(mask));			\
+		WREG32(reg, tmp_);				\
+	} while (0)
+#define WREG32_AND(reg, and) WREG32_P(reg, 0, and)
+#define WREG32_OR(reg, or) WREG32_P(reg, or, ~(or))
+#define WREG32_PLL_P(reg, val, mask)				\
+	do {							\
+		uint32_t tmp_ = RREG32_PLL(reg);		\
+		tmp_ &= (mask);					\
+		tmp_ |= ((val) & ~(mask));			\
+		WREG32_PLL(reg, tmp_);				\
+	} while (0)
+#define DREG32_SYS(sqf, adev, reg) seq_printf((sqf), #reg " : 0x%08X\n", amdgpu_mm_rreg((adev), (reg), false))
+#define RREG32_IO(reg) amdgpu_io_rreg(adev, (reg))
+#define WREG32_IO(reg, v) amdgpu_io_wreg(adev, (reg), (v))
+
+#define RDOORBELL32(index) amdgpu_mm_rdoorbell(adev, (index))
+#define WDOORBELL32(index, v) amdgpu_mm_wdoorbell(adev, (index), (v))
+#define RDOORBELL64(index) amdgpu_mm_rdoorbell64(adev, (index))
+#define WDOORBELL64(index, v) amdgpu_mm_wdoorbell64(adev, (index), (v))
+
+#define REG_FIELD_SHIFT(reg, field) reg##__##field##__SHIFT
+#define REG_FIELD_MASK(reg, field) reg##__##field##_MASK
+
+#define REG_SET_FIELD(orig_val, reg, field, field_val)			\
+	(((orig_val) & ~REG_FIELD_MASK(reg, field)) |			\
+	 (REG_FIELD_MASK(reg, field) & ((field_val) << REG_FIELD_SHIFT(reg, field))))
+
+#define REG_GET_FIELD(value, reg, field)				\
+	(((value) & REG_FIELD_MASK(reg, field)) >> REG_FIELD_SHIFT(reg, field))
+
+#define WREG32_FIELD(reg, field, val)	\
+	WREG32(mm##reg, (RREG32(mm##reg) & ~REG_FIELD_MASK(reg, field)) | (val) << REG_FIELD_SHIFT(reg, field))
+
+#define WREG32_FIELD_OFFSET(reg, offset, field, val)	\
+	WREG32(mm##reg + offset, (RREG32(mm##reg + offset) & ~REG_FIELD_MASK(reg, field)) | (val) << REG_FIELD_SHIFT(reg, field))
+
+/*
+ * BIOS helpers.
+ */
+#define RBIOS8(i) (adev->bios[i])
+#define RBIOS16(i) (RBIOS8(i) | (RBIOS8((i)+1) << 8))
+#define RBIOS32(i) ((RBIOS16(i)) | (RBIOS16((i)+2) << 16))
+
+static inline struct amdgpu_sdma_instance *
+amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		if (&adev->sdma.instance[i].ring == ring)
+			break;
+
+	if (i < AMDGPU_MAX_SDMA_INSTANCES)
+		return &adev->sdma.instance[i];
+	else
+		return NULL;
+}
+
+/*
+ * ASICs macro.
+ */
+#define amdgpu_asic_set_vga_state(adev, state) (adev)->asic_funcs->set_vga_state((adev), (state))
+#define amdgpu_asic_reset(adev) (adev)->asic_funcs->reset((adev))
+#define amdgpu_asic_get_xclk(adev) (adev)->asic_funcs->get_xclk((adev))
+#define amdgpu_asic_set_uvd_clocks(adev, v, d) (adev)->asic_funcs->set_uvd_clocks((adev), (v), (d))
+#define amdgpu_asic_set_vce_clocks(adev, ev, ec) (adev)->asic_funcs->set_vce_clocks((adev), (ev), (ec))
+#define amdgpu_get_pcie_lanes(adev) (adev)->asic_funcs->get_pcie_lanes((adev))
+#define amdgpu_set_pcie_lanes(adev, l) (adev)->asic_funcs->set_pcie_lanes((adev), (l))
+#define amdgpu_asic_get_gpu_clock_counter(adev) (adev)->asic_funcs->get_gpu_clock_counter((adev))
+#define amdgpu_asic_read_disabled_bios(adev) (adev)->asic_funcs->read_disabled_bios((adev))
+#define amdgpu_asic_read_bios_from_rom(adev, b, l) (adev)->asic_funcs->read_bios_from_rom((adev), (b), (l))
+#define amdgpu_asic_read_register(adev, se, sh, offset, v)((adev)->asic_funcs->read_register((adev), (se), (sh), (offset), (v)))
+#define amdgpu_asic_get_config_memsize(adev) (adev)->asic_funcs->get_config_memsize((adev))
+#define amdgpu_asic_flush_hdp(adev, r) (adev)->asic_funcs->flush_hdp((adev), (r))
+#define amdgpu_asic_invalidate_hdp(adev, r) (adev)->asic_funcs->invalidate_hdp((adev), (r))
+#define amdgpu_asic_need_full_reset(adev) (adev)->asic_funcs->need_full_reset((adev))
+#define amdgpu_gmc_flush_gpu_tlb(adev, vmid) (adev)->gmc.gmc_funcs->flush_gpu_tlb((adev), (vmid))
+#define amdgpu_gmc_emit_flush_gpu_tlb(r, vmid, addr) (r)->adev->gmc.gmc_funcs->emit_flush_gpu_tlb((r), (vmid), (addr))
+#define amdgpu_gmc_emit_pasid_mapping(r, vmid, pasid) (r)->adev->gmc.gmc_funcs->emit_pasid_mapping((r), (vmid), (pasid))
+#define amdgpu_gmc_set_pte_pde(adev, pt, idx, addr, flags) (adev)->gmc.gmc_funcs->set_pte_pde((adev), (pt), (idx), (addr), (flags))
+#define amdgpu_gmc_get_vm_pde(adev, level, dst, flags) (adev)->gmc.gmc_funcs->get_vm_pde((adev), (level), (dst), (flags))
+#define amdgpu_gmc_get_pte_flags(adev, flags) (adev)->gmc.gmc_funcs->get_vm_pte_flags((adev),(flags))
+#define amdgpu_vm_copy_pte(adev, ib, pe, src, count) ((adev)->vm_manager.vm_pte_funcs->copy_pte((ib), (pe), (src), (count)))
+#define amdgpu_vm_write_pte(adev, ib, pe, value, count, incr) ((adev)->vm_manager.vm_pte_funcs->write_pte((ib), (pe), (value), (count), (incr)))
+#define amdgpu_vm_set_pte_pde(adev, ib, pe, addr, count, incr, flags) ((adev)->vm_manager.vm_pte_funcs->set_pte_pde((ib), (pe), (addr), (count), (incr), (flags)))
+#define amdgpu_ring_parse_cs(r, p, ib) ((r)->funcs->parse_cs((p), (ib)))
+#define amdgpu_ring_patch_cs_in_place(r, p, ib) ((r)->funcs->patch_cs_in_place((p), (ib)))
+#define amdgpu_ring_test_ring(r) (r)->funcs->test_ring((r))
+#define amdgpu_ring_test_ib(r, t) (r)->funcs->test_ib((r), (t))
+#define amdgpu_ring_get_rptr(r) (r)->funcs->get_rptr((r))
+#define amdgpu_ring_get_wptr(r) (r)->funcs->get_wptr((r))
+#define amdgpu_ring_set_wptr(r) (r)->funcs->set_wptr((r))
+#define amdgpu_ring_emit_ib(r, ib, vmid, c) (r)->funcs->emit_ib((r), (ib), (vmid), (c))
+#define amdgpu_ring_emit_pipeline_sync(r) (r)->funcs->emit_pipeline_sync((r))
+#define amdgpu_ring_emit_vm_flush(r, vmid, addr) (r)->funcs->emit_vm_flush((r), (vmid), (addr))
+#define amdgpu_ring_emit_fence(r, addr, seq, flags) (r)->funcs->emit_fence((r), (addr), (seq), (flags))
+#define amdgpu_ring_emit_gds_switch(r, v, db, ds, wb, ws, ab, as) (r)->funcs->emit_gds_switch((r), (v), (db), (ds), (wb), (ws), (ab), (as))
+#define amdgpu_ring_emit_hdp_flush(r) (r)->funcs->emit_hdp_flush((r))
+#define amdgpu_ring_emit_switch_buffer(r) (r)->funcs->emit_switch_buffer((r))
+#define amdgpu_ring_emit_cntxcntl(r, d) (r)->funcs->emit_cntxcntl((r), (d))
+#define amdgpu_ring_emit_rreg(r, d) (r)->funcs->emit_rreg((r), (d))
+#define amdgpu_ring_emit_wreg(r, d, v) (r)->funcs->emit_wreg((r), (d), (v))
+#define amdgpu_ring_emit_reg_wait(r, d, v, m) (r)->funcs->emit_reg_wait((r), (d), (v), (m))
+#define amdgpu_ring_emit_reg_write_reg_wait(r, d0, d1, v, m) (r)->funcs->emit_reg_write_reg_wait((r), (d0), (d1), (v), (m))
+#define amdgpu_ring_emit_tmz(r, b) (r)->funcs->emit_tmz((r), (b))
+#define amdgpu_ring_pad_ib(r, ib) ((r)->funcs->pad_ib((r), (ib)))
+#define amdgpu_ring_init_cond_exec(r) (r)->funcs->init_cond_exec((r))
+#define amdgpu_ring_patch_cond_exec(r,o) (r)->funcs->patch_cond_exec((r),(o))
+#define amdgpu_ih_get_wptr(adev) (adev)->irq.ih_funcs->get_wptr((adev))
+#define amdgpu_ih_prescreen_iv(adev) (adev)->irq.ih_funcs->prescreen_iv((adev))
+#define amdgpu_ih_decode_iv(adev, iv) (adev)->irq.ih_funcs->decode_iv((adev), (iv))
+#define amdgpu_ih_set_rptr(adev) (adev)->irq.ih_funcs->set_rptr((adev))
+#define amdgpu_display_vblank_get_counter(adev, crtc) (adev)->mode_info.funcs->vblank_get_counter((adev), (crtc))
+#define amdgpu_display_backlight_set_level(adev, e, l) (adev)->mode_info.funcs->backlight_set_level((e), (l))
+#define amdgpu_display_backlight_get_level(adev, e) (adev)->mode_info.funcs->backlight_get_level((e))
+#define amdgpu_display_hpd_sense(adev, h) (adev)->mode_info.funcs->hpd_sense((adev), (h))
+#define amdgpu_display_hpd_set_polarity(adev, h) (adev)->mode_info.funcs->hpd_set_polarity((adev), (h))
+#define amdgpu_display_hpd_get_gpio_reg(adev) (adev)->mode_info.funcs->hpd_get_gpio_reg((adev))
+#define amdgpu_display_bandwidth_update(adev) (adev)->mode_info.funcs->bandwidth_update((adev))
+#define amdgpu_display_page_flip(adev, crtc, base, async) (adev)->mode_info.funcs->page_flip((adev), (crtc), (base), (async))
+#define amdgpu_display_page_flip_get_scanoutpos(adev, crtc, vbl, pos) (adev)->mode_info.funcs->page_flip_get_scanoutpos((adev), (crtc), (vbl), (pos))
+#define amdgpu_display_add_encoder(adev, e, s, c) (adev)->mode_info.funcs->add_encoder((adev), (e), (s), (c))
+#define amdgpu_display_add_connector(adev, ci, sd, ct, ib, coi, h, r) (adev)->mode_info.funcs->add_connector((adev), (ci), (sd), (ct), (ib), (coi), (h), (r))
+#define amdgpu_emit_copy_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_copy_buffer((ib),  (s), (d), (b))
+#define amdgpu_emit_fill_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_fill_buffer((ib), (s), (d), (b))
+#define amdgpu_gfx_get_gpu_clock_counter(adev) (adev)->gfx.funcs->get_gpu_clock_counter((adev))
+#define amdgpu_gfx_select_se_sh(adev, se, sh, instance) (adev)->gfx.funcs->select_se_sh((adev), (se), (sh), (instance))
+#define amdgpu_gds_switch(adev, r, v, d, w, a) (adev)->gds.funcs->patch_gds_switch((r), (v), (d), (w), (a))
+#define amdgpu_psp_check_fw_loading_status(adev, i) (adev)->firmware.funcs->check_fw_loading_status((adev), (i))
+#define amdgpu_gfx_select_me_pipe_q(adev, me, pipe, q) (adev)->gfx.funcs->select_me_pipe_q((adev), (me), (pipe), (q))
+
+/* Common functions */
+int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
+			      struct amdgpu_job* job, bool force);
+void amdgpu_device_pci_config_reset(struct amdgpu_device *adev);
+bool amdgpu_device_need_post(struct amdgpu_device *adev);
+void amdgpu_display_update_priority(struct amdgpu_device *adev);
+
+void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes,
+				  u64 num_vis_bytes);
+void amdgpu_device_vram_location(struct amdgpu_device *adev,
+				 struct amdgpu_gmc *mc, u64 base);
+void amdgpu_device_gart_location(struct amdgpu_device *adev,
+				 struct amdgpu_gmc *mc);
+int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev);
+void amdgpu_device_program_register_sequence(struct amdgpu_device *adev,
+					     const u32 *registers,
+					     const u32 array_size);
+
+bool amdgpu_device_is_px(struct drm_device *dev);
+/* atpx handler */
+#if defined(CONFIG_VGA_SWITCHEROO)
+void amdgpu_register_atpx_handler(void);
+void amdgpu_unregister_atpx_handler(void);
+bool amdgpu_has_atpx_dgpu_power_cntl(void);
+bool amdgpu_is_atpx_hybrid(void);
+bool amdgpu_atpx_dgpu_req_power_for_displays(void);
+bool amdgpu_has_atpx(void);
+#else
+static inline void amdgpu_register_atpx_handler(void) {}
+static inline void amdgpu_unregister_atpx_handler(void) {}
+static inline bool amdgpu_has_atpx_dgpu_power_cntl(void) { return false; }
+static inline bool amdgpu_is_atpx_hybrid(void) { return false; }
+static inline bool amdgpu_atpx_dgpu_req_power_for_displays(void) { return false; }
+static inline bool amdgpu_has_atpx(void) { return false; }
+#endif
+
+#if defined(CONFIG_VGA_SWITCHEROO) && defined(CONFIG_ACPI)
+void *amdgpu_atpx_get_dhandle(void);
+#else
+static inline void *amdgpu_atpx_get_dhandle(void) { return NULL; }
+#endif
+
+/*
+ * KMS
+ */
+extern const struct drm_ioctl_desc amdgpu_ioctls_kms[];
+extern const int amdgpu_max_kms_ioctl;
+
+int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags);
+void amdgpu_driver_unload_kms(struct drm_device *dev);
+void amdgpu_driver_lastclose_kms(struct drm_device *dev);
+int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv);
+void amdgpu_driver_postclose_kms(struct drm_device *dev,
+				 struct drm_file *file_priv);
+int amdgpu_device_ip_suspend(struct amdgpu_device *adev);
+int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon);
+int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon);
+u32 amdgpu_get_vblank_counter_kms(struct drm_device *dev, unsigned int pipe);
+int amdgpu_enable_vblank_kms(struct drm_device *dev, unsigned int pipe);
+void amdgpu_disable_vblank_kms(struct drm_device *dev, unsigned int pipe);
+long amdgpu_kms_compat_ioctl(struct file *filp, unsigned int cmd,
+			     unsigned long arg);
+
+/*
+ * functions used by amdgpu_encoder.c
+ */
+struct amdgpu_afmt_acr {
+	u32 clock;
+
+	int n_32khz;
+	int cts_32khz;
+
+	int n_44_1khz;
+	int cts_44_1khz;
+
+	int n_48khz;
+	int cts_48khz;
+
+};
+
+struct amdgpu_afmt_acr amdgpu_afmt_acr(uint32_t clock);
+
+/* amdgpu_acpi.c */
+#if defined(CONFIG_ACPI)
+int amdgpu_acpi_init(struct amdgpu_device *adev);
+void amdgpu_acpi_fini(struct amdgpu_device *adev);
+bool amdgpu_acpi_is_pcie_performance_request_supported(struct amdgpu_device *adev);
+int amdgpu_acpi_pcie_performance_request(struct amdgpu_device *adev,
+						u8 perf_req, bool advertise);
+int amdgpu_acpi_pcie_notify_device_ready(struct amdgpu_device *adev);
+#else
+static inline int amdgpu_acpi_init(struct amdgpu_device *adev) { return 0; }
+static inline void amdgpu_acpi_fini(struct amdgpu_device *adev) { }
+#endif
+
+int amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
+			   uint64_t addr, struct amdgpu_bo **bo,
+			   struct amdgpu_bo_va_mapping **mapping);
+
+#if defined(CONFIG_DRM_AMD_DC)
+int amdgpu_dm_display_resume(struct amdgpu_device *adev );
+#else
+static inline int amdgpu_dm_display_resume(struct amdgpu_device *adev) { return 0; }
+#endif
+
+#include "amdgpu_object.h"
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_acp.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_acp.c
new file mode 100644
index 00000000000..71efcf38f11
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_acp.c
@@ -0,0 +1,623 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#include <linux/irqdomain.h>
+#include <linux/pm_domain.h>
+#include <linux/platform_device.h>
+#include <sound/designware_i2s.h>
+#include <sound/pcm.h>
+
+#include "amdgpu.h"
+#include "atom.h"
+#include "amdgpu_acp.h"
+
+#include "acp_gfx_if.h"
+
+#define ACP_TILE_ON_MASK                	0x03
+#define ACP_TILE_OFF_MASK               	0x02
+#define ACP_TILE_ON_RETAIN_REG_MASK     	0x1f
+#define ACP_TILE_OFF_RETAIN_REG_MASK    	0x20
+
+#define ACP_TILE_P1_MASK                	0x3e
+#define ACP_TILE_P2_MASK                	0x3d
+#define ACP_TILE_DSP0_MASK              	0x3b
+#define ACP_TILE_DSP1_MASK              	0x37
+
+#define ACP_TILE_DSP2_MASK              	0x2f
+
+#define ACP_DMA_REGS_END			0x146c0
+#define ACP_I2S_PLAY_REGS_START			0x14840
+#define ACP_I2S_PLAY_REGS_END			0x148b4
+#define ACP_I2S_CAP_REGS_START			0x148b8
+#define ACP_I2S_CAP_REGS_END			0x1496c
+
+#define ACP_I2S_COMP1_CAP_REG_OFFSET		0xac
+#define ACP_I2S_COMP2_CAP_REG_OFFSET		0xa8
+#define ACP_I2S_COMP1_PLAY_REG_OFFSET		0x6c
+#define ACP_I2S_COMP2_PLAY_REG_OFFSET		0x68
+#define ACP_BT_PLAY_REGS_START			0x14970
+#define ACP_BT_PLAY_REGS_END			0x14a24
+#define ACP_BT_COMP1_REG_OFFSET			0xac
+#define ACP_BT_COMP2_REG_OFFSET			0xa8
+
+#define mmACP_PGFSM_RETAIN_REG			0x51c9
+#define mmACP_PGFSM_CONFIG_REG			0x51ca
+#define mmACP_PGFSM_READ_REG_0			0x51cc
+
+#define mmACP_MEM_SHUT_DOWN_REQ_LO		0x51f8
+#define mmACP_MEM_SHUT_DOWN_REQ_HI		0x51f9
+#define mmACP_MEM_SHUT_DOWN_STS_LO		0x51fa
+#define mmACP_MEM_SHUT_DOWN_STS_HI		0x51fb
+
+#define mmACP_CONTROL				0x5131
+#define mmACP_STATUS				0x5133
+#define mmACP_SOFT_RESET			0x5134
+#define ACP_CONTROL__ClkEn_MASK 		0x1
+#define ACP_SOFT_RESET__SoftResetAud_MASK 	0x100
+#define ACP_SOFT_RESET__SoftResetAudDone_MASK	0x1000000
+#define ACP_CLOCK_EN_TIME_OUT_VALUE		0x000000FF
+#define ACP_SOFT_RESET_DONE_TIME_OUT_VALUE	0x000000FF
+
+#define ACP_TIMEOUT_LOOP			0x000000FF
+#define ACP_DEVS				4
+#define ACP_SRC_ID				162
+
+enum {
+	ACP_TILE_P1 = 0,
+	ACP_TILE_P2,
+	ACP_TILE_DSP0,
+	ACP_TILE_DSP1,
+	ACP_TILE_DSP2,
+};
+
+static int acp_sw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->acp.parent = adev->dev;
+
+	adev->acp.cgs_device =
+		amdgpu_cgs_create_device(adev);
+	if (!adev->acp.cgs_device)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int acp_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->acp.cgs_device)
+		amdgpu_cgs_destroy_device(adev->acp.cgs_device);
+
+	return 0;
+}
+
+/* power off a tile/block within ACP */
+static int acp_suspend_tile(void *cgs_dev, int tile)
+{
+	u32 val = 0;
+	u32 count = 0;
+
+	if ((tile  < ACP_TILE_P1) || (tile > ACP_TILE_DSP2)) {
+		pr_err("Invalid ACP tile : %d to suspend\n", tile);
+		return -1;
+	}
+
+	val = cgs_read_register(cgs_dev, mmACP_PGFSM_READ_REG_0 + tile);
+	val &= ACP_TILE_ON_MASK;
+
+	if (val == 0x0) {
+		val = cgs_read_register(cgs_dev, mmACP_PGFSM_RETAIN_REG);
+		val = val | (1 << tile);
+		cgs_write_register(cgs_dev, mmACP_PGFSM_RETAIN_REG, val);
+		cgs_write_register(cgs_dev, mmACP_PGFSM_CONFIG_REG,
+					0x500 + tile);
+
+		count = ACP_TIMEOUT_LOOP;
+		while (true) {
+			val = cgs_read_register(cgs_dev, mmACP_PGFSM_READ_REG_0
+								+ tile);
+			val = val & ACP_TILE_ON_MASK;
+			if (val == ACP_TILE_OFF_MASK)
+				break;
+			if (--count == 0) {
+				pr_err("Timeout reading ACP PGFSM status\n");
+				return -ETIMEDOUT;
+			}
+			udelay(100);
+		}
+
+		val = cgs_read_register(cgs_dev, mmACP_PGFSM_RETAIN_REG);
+
+		val |= ACP_TILE_OFF_RETAIN_REG_MASK;
+		cgs_write_register(cgs_dev, mmACP_PGFSM_RETAIN_REG, val);
+	}
+	return 0;
+}
+
+/* power on a tile/block within ACP */
+static int acp_resume_tile(void *cgs_dev, int tile)
+{
+	u32 val = 0;
+	u32 count = 0;
+
+	if ((tile  < ACP_TILE_P1) || (tile > ACP_TILE_DSP2)) {
+		pr_err("Invalid ACP tile to resume\n");
+		return -1;
+	}
+
+	val = cgs_read_register(cgs_dev, mmACP_PGFSM_READ_REG_0 + tile);
+	val = val & ACP_TILE_ON_MASK;
+
+	if (val != 0x0) {
+		cgs_write_register(cgs_dev, mmACP_PGFSM_CONFIG_REG,
+					0x600 + tile);
+		count = ACP_TIMEOUT_LOOP;
+		while (true) {
+			val = cgs_read_register(cgs_dev, mmACP_PGFSM_READ_REG_0
+							+ tile);
+			val = val & ACP_TILE_ON_MASK;
+			if (val == 0x0)
+				break;
+			if (--count == 0) {
+				pr_err("Timeout reading ACP PGFSM status\n");
+				return -ETIMEDOUT;
+			}
+			udelay(100);
+		}
+		val = cgs_read_register(cgs_dev, mmACP_PGFSM_RETAIN_REG);
+		if (tile == ACP_TILE_P1)
+			val = val & (ACP_TILE_P1_MASK);
+		else if (tile == ACP_TILE_P2)
+			val = val & (ACP_TILE_P2_MASK);
+
+		cgs_write_register(cgs_dev, mmACP_PGFSM_RETAIN_REG, val);
+	}
+	return 0;
+}
+
+struct acp_pm_domain {
+	void *cgs_dev;
+	struct generic_pm_domain gpd;
+};
+
+static int acp_poweroff(struct generic_pm_domain *genpd)
+{
+	int i, ret;
+	struct acp_pm_domain *apd;
+
+	apd = container_of(genpd, struct acp_pm_domain, gpd);
+	if (apd != NULL) {
+		/* Donot return abruptly if any of power tile fails to suspend.
+		 * Log it and continue powering off other tile
+		 */
+		for (i = 4; i >= 0 ; i--) {
+			ret = acp_suspend_tile(apd->cgs_dev, ACP_TILE_P1 + i);
+			if (ret)
+				pr_err("ACP tile %d tile suspend failed\n", i);
+		}
+	}
+	return 0;
+}
+
+static int acp_poweron(struct generic_pm_domain *genpd)
+{
+	int i, ret;
+	struct acp_pm_domain *apd;
+
+	apd = container_of(genpd, struct acp_pm_domain, gpd);
+	if (apd != NULL) {
+		for (i = 0; i < 2; i++) {
+			ret = acp_resume_tile(apd->cgs_dev, ACP_TILE_P1 + i);
+			if (ret) {
+				pr_err("ACP tile %d resume failed\n", i);
+				break;
+			}
+		}
+
+		/* Disable DSPs which are not going to be used */
+		for (i = 0; i < 3; i++) {
+			ret = acp_suspend_tile(apd->cgs_dev, ACP_TILE_DSP0 + i);
+			/* Continue suspending other DSP, even if one fails */
+			if (ret)
+				pr_err("ACP DSP %d suspend failed\n", i);
+		}
+	}
+	return 0;
+}
+
+static struct device *get_mfd_cell_dev(const char *device_name, int r)
+{
+	char auto_dev_name[25];
+	struct device *dev;
+
+	snprintf(auto_dev_name, sizeof(auto_dev_name),
+		 "%s.%d.auto", device_name, r);
+	dev = bus_find_device_by_name(&platform_bus_type, NULL, auto_dev_name);
+	dev_info(dev, "device %s added to pm domain\n", auto_dev_name);
+
+	return dev;
+}
+
+/**
+ * acp_hw_init - start and test ACP block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ */
+static int acp_hw_init(void *handle)
+{
+	int r, i;
+	uint64_t acp_base;
+	u32 val = 0;
+	u32 count = 0;
+	struct device *dev;
+	struct i2s_platform_data *i2s_pdata;
+
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	const struct amdgpu_ip_block *ip_block =
+		amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_ACP);
+
+	if (!ip_block)
+		return -EINVAL;
+
+	r = amd_acp_hw_init(adev->acp.cgs_device,
+			    ip_block->version->major, ip_block->version->minor);
+	/* -ENODEV means board uses AZ rather than ACP */
+	if (r == -ENODEV)
+		return 0;
+	else if (r)
+		return r;
+
+	if (adev->rmmio_size == 0 || adev->rmmio_size < 0x5289)
+		return -EINVAL;
+
+	acp_base = adev->rmmio_base;
+
+	if (adev->asic_type != CHIP_STONEY) {
+		adev->acp.acp_genpd = kzalloc(sizeof(struct acp_pm_domain), GFP_KERNEL);
+		if (adev->acp.acp_genpd == NULL)
+			return -ENOMEM;
+
+		adev->acp.acp_genpd->gpd.name = "ACP_AUDIO";
+		adev->acp.acp_genpd->gpd.power_off = acp_poweroff;
+		adev->acp.acp_genpd->gpd.power_on = acp_poweron;
+
+
+		adev->acp.acp_genpd->cgs_dev = adev->acp.cgs_device;
+
+		pm_genpd_init(&adev->acp.acp_genpd->gpd, NULL, false);
+	}
+
+	adev->acp.acp_cell = kcalloc(ACP_DEVS, sizeof(struct mfd_cell),
+							GFP_KERNEL);
+
+	if (adev->acp.acp_cell == NULL)
+		return -ENOMEM;
+
+	adev->acp.acp_res = kcalloc(5, sizeof(struct resource), GFP_KERNEL);
+	if (adev->acp.acp_res == NULL) {
+		kfree(adev->acp.acp_cell);
+		return -ENOMEM;
+	}
+
+	i2s_pdata = kcalloc(3, sizeof(struct i2s_platform_data), GFP_KERNEL);
+	if (i2s_pdata == NULL) {
+		kfree(adev->acp.acp_res);
+		kfree(adev->acp.acp_cell);
+		return -ENOMEM;
+	}
+
+	switch (adev->asic_type) {
+	case CHIP_STONEY:
+		i2s_pdata[0].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
+			DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
+		break;
+	default:
+		i2s_pdata[0].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET;
+	}
+	i2s_pdata[0].cap = DWC_I2S_PLAY;
+	i2s_pdata[0].snd_rates = SNDRV_PCM_RATE_8000_96000;
+	i2s_pdata[0].i2s_reg_comp1 = ACP_I2S_COMP1_PLAY_REG_OFFSET;
+	i2s_pdata[0].i2s_reg_comp2 = ACP_I2S_COMP2_PLAY_REG_OFFSET;
+	switch (adev->asic_type) {
+	case CHIP_STONEY:
+		i2s_pdata[1].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
+			DW_I2S_QUIRK_COMP_PARAM1 |
+			DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
+		break;
+	default:
+		i2s_pdata[1].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
+			DW_I2S_QUIRK_COMP_PARAM1;
+	}
+
+	i2s_pdata[1].cap = DWC_I2S_RECORD;
+	i2s_pdata[1].snd_rates = SNDRV_PCM_RATE_8000_96000;
+	i2s_pdata[1].i2s_reg_comp1 = ACP_I2S_COMP1_CAP_REG_OFFSET;
+	i2s_pdata[1].i2s_reg_comp2 = ACP_I2S_COMP2_CAP_REG_OFFSET;
+
+	i2s_pdata[2].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET;
+	switch (adev->asic_type) {
+	case CHIP_STONEY:
+		i2s_pdata[2].quirks |= DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
+		break;
+	default:
+		break;
+	}
+
+	i2s_pdata[2].cap = DWC_I2S_PLAY | DWC_I2S_RECORD;
+	i2s_pdata[2].snd_rates = SNDRV_PCM_RATE_8000_96000;
+	i2s_pdata[2].i2s_reg_comp1 = ACP_BT_COMP1_REG_OFFSET;
+	i2s_pdata[2].i2s_reg_comp2 = ACP_BT_COMP2_REG_OFFSET;
+
+	adev->acp.acp_res[0].name = "acp2x_dma";
+	adev->acp.acp_res[0].flags = IORESOURCE_MEM;
+	adev->acp.acp_res[0].start = acp_base;
+	adev->acp.acp_res[0].end = acp_base + ACP_DMA_REGS_END;
+
+	adev->acp.acp_res[1].name = "acp2x_dw_i2s_play";
+	adev->acp.acp_res[1].flags = IORESOURCE_MEM;
+	adev->acp.acp_res[1].start = acp_base + ACP_I2S_PLAY_REGS_START;
+	adev->acp.acp_res[1].end = acp_base + ACP_I2S_PLAY_REGS_END;
+
+	adev->acp.acp_res[2].name = "acp2x_dw_i2s_cap";
+	adev->acp.acp_res[2].flags = IORESOURCE_MEM;
+	adev->acp.acp_res[2].start = acp_base + ACP_I2S_CAP_REGS_START;
+	adev->acp.acp_res[2].end = acp_base + ACP_I2S_CAP_REGS_END;
+
+	adev->acp.acp_res[3].name = "acp2x_dw_bt_i2s_play_cap";
+	adev->acp.acp_res[3].flags = IORESOURCE_MEM;
+	adev->acp.acp_res[3].start = acp_base + ACP_BT_PLAY_REGS_START;
+	adev->acp.acp_res[3].end = acp_base + ACP_BT_PLAY_REGS_END;
+
+	adev->acp.acp_res[4].name = "acp2x_dma_irq";
+	adev->acp.acp_res[4].flags = IORESOURCE_IRQ;
+	adev->acp.acp_res[4].start = amdgpu_irq_create_mapping(adev, 162);
+	adev->acp.acp_res[4].end = adev->acp.acp_res[4].start;
+
+	adev->acp.acp_cell[0].name = "acp_audio_dma";
+	adev->acp.acp_cell[0].num_resources = 5;
+	adev->acp.acp_cell[0].resources = &adev->acp.acp_res[0];
+	adev->acp.acp_cell[0].platform_data = &adev->asic_type;
+	adev->acp.acp_cell[0].pdata_size = sizeof(adev->asic_type);
+
+	adev->acp.acp_cell[1].name = "designware-i2s";
+	adev->acp.acp_cell[1].num_resources = 1;
+	adev->acp.acp_cell[1].resources = &adev->acp.acp_res[1];
+	adev->acp.acp_cell[1].platform_data = &i2s_pdata[0];
+	adev->acp.acp_cell[1].pdata_size = sizeof(struct i2s_platform_data);
+
+	adev->acp.acp_cell[2].name = "designware-i2s";
+	adev->acp.acp_cell[2].num_resources = 1;
+	adev->acp.acp_cell[2].resources = &adev->acp.acp_res[2];
+	adev->acp.acp_cell[2].platform_data = &i2s_pdata[1];
+	adev->acp.acp_cell[2].pdata_size = sizeof(struct i2s_platform_data);
+
+	adev->acp.acp_cell[3].name = "designware-i2s";
+	adev->acp.acp_cell[3].num_resources = 1;
+	adev->acp.acp_cell[3].resources = &adev->acp.acp_res[3];
+	adev->acp.acp_cell[3].platform_data = &i2s_pdata[2];
+	adev->acp.acp_cell[3].pdata_size = sizeof(struct i2s_platform_data);
+
+	r = mfd_add_hotplug_devices(adev->acp.parent, adev->acp.acp_cell,
+								ACP_DEVS);
+	if (r)
+		return r;
+
+	if (adev->asic_type != CHIP_STONEY) {
+		for (i = 0; i < ACP_DEVS ; i++) {
+			dev = get_mfd_cell_dev(adev->acp.acp_cell[i].name, i);
+			r = pm_genpd_add_device(&adev->acp.acp_genpd->gpd, dev);
+			if (r) {
+				dev_err(dev, "Failed to add dev to genpd\n");
+				return r;
+			}
+		}
+	}
+
+	/* Assert Soft reset of ACP */
+	val = cgs_read_register(adev->acp.cgs_device, mmACP_SOFT_RESET);
+
+	val |= ACP_SOFT_RESET__SoftResetAud_MASK;
+	cgs_write_register(adev->acp.cgs_device, mmACP_SOFT_RESET, val);
+
+	count = ACP_SOFT_RESET_DONE_TIME_OUT_VALUE;
+	while (true) {
+		val = cgs_read_register(adev->acp.cgs_device, mmACP_SOFT_RESET);
+		if (ACP_SOFT_RESET__SoftResetAudDone_MASK ==
+		    (val & ACP_SOFT_RESET__SoftResetAudDone_MASK))
+			break;
+		if (--count == 0) {
+			dev_err(&adev->pdev->dev, "Failed to reset ACP\n");
+			return -ETIMEDOUT;
+		}
+		udelay(100);
+	}
+	/* Enable clock to ACP and wait until the clock is enabled */
+	val = cgs_read_register(adev->acp.cgs_device, mmACP_CONTROL);
+	val = val | ACP_CONTROL__ClkEn_MASK;
+	cgs_write_register(adev->acp.cgs_device, mmACP_CONTROL, val);
+
+	count = ACP_CLOCK_EN_TIME_OUT_VALUE;
+
+	while (true) {
+		val = cgs_read_register(adev->acp.cgs_device, mmACP_STATUS);
+		if (val & (u32) 0x1)
+			break;
+		if (--count == 0) {
+			dev_err(&adev->pdev->dev, "Failed to reset ACP\n");
+			return -ETIMEDOUT;
+		}
+		udelay(100);
+	}
+	/* Deassert the SOFT RESET flags */
+	val = cgs_read_register(adev->acp.cgs_device, mmACP_SOFT_RESET);
+	val &= ~ACP_SOFT_RESET__SoftResetAud_MASK;
+	cgs_write_register(adev->acp.cgs_device, mmACP_SOFT_RESET, val);
+	return 0;
+}
+
+/**
+ * acp_hw_fini - stop the hardware block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ */
+static int acp_hw_fini(void *handle)
+{
+	int i, ret;
+	u32 val = 0;
+	u32 count = 0;
+	struct device *dev;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* return early if no ACP */
+	if (!adev->acp.acp_cell)
+		return 0;
+
+	/* Assert Soft reset of ACP */
+	val = cgs_read_register(adev->acp.cgs_device, mmACP_SOFT_RESET);
+
+	val |= ACP_SOFT_RESET__SoftResetAud_MASK;
+	cgs_write_register(adev->acp.cgs_device, mmACP_SOFT_RESET, val);
+
+	count = ACP_SOFT_RESET_DONE_TIME_OUT_VALUE;
+	while (true) {
+		val = cgs_read_register(adev->acp.cgs_device, mmACP_SOFT_RESET);
+		if (ACP_SOFT_RESET__SoftResetAudDone_MASK ==
+		    (val & ACP_SOFT_RESET__SoftResetAudDone_MASK))
+			break;
+		if (--count == 0) {
+			dev_err(&adev->pdev->dev, "Failed to reset ACP\n");
+			return -ETIMEDOUT;
+		}
+		udelay(100);
+	}
+	/* Disable ACP clock */
+	val = cgs_read_register(adev->acp.cgs_device, mmACP_CONTROL);
+	val &= ~ACP_CONTROL__ClkEn_MASK;
+	cgs_write_register(adev->acp.cgs_device, mmACP_CONTROL, val);
+
+	count = ACP_CLOCK_EN_TIME_OUT_VALUE;
+
+	while (true) {
+		val = cgs_read_register(adev->acp.cgs_device, mmACP_STATUS);
+		if (val & (u32) 0x1)
+			break;
+		if (--count == 0) {
+			dev_err(&adev->pdev->dev, "Failed to reset ACP\n");
+			return -ETIMEDOUT;
+		}
+		udelay(100);
+	}
+
+	if (adev->acp.acp_genpd) {
+		for (i = 0; i < ACP_DEVS ; i++) {
+			dev = get_mfd_cell_dev(adev->acp.acp_cell[i].name, i);
+			ret = pm_genpd_remove_device(dev);
+			/* If removal fails, dont giveup and try rest */
+			if (ret)
+				dev_err(dev, "remove dev from genpd failed\n");
+		}
+		kfree(adev->acp.acp_genpd);
+	}
+
+	mfd_remove_devices(adev->acp.parent);
+	kfree(adev->acp.acp_res);
+	kfree(adev->acp.acp_cell);
+
+	return 0;
+}
+
+static int acp_suspend(void *handle)
+{
+	return 0;
+}
+
+static int acp_resume(void *handle)
+{
+	return 0;
+}
+
+static int acp_early_init(void *handle)
+{
+	return 0;
+}
+
+static bool acp_is_idle(void *handle)
+{
+	return true;
+}
+
+static int acp_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static int acp_soft_reset(void *handle)
+{
+	return 0;
+}
+
+static int acp_set_clockgating_state(void *handle,
+				     enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int acp_set_powergating_state(void *handle,
+				     enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs acp_ip_funcs = {
+	.name = "acp_ip",
+	.early_init = acp_early_init,
+	.late_init = NULL,
+	.sw_init = acp_sw_init,
+	.sw_fini = acp_sw_fini,
+	.hw_init = acp_hw_init,
+	.hw_fini = acp_hw_fini,
+	.suspend = acp_suspend,
+	.resume = acp_resume,
+	.is_idle = acp_is_idle,
+	.wait_for_idle = acp_wait_for_idle,
+	.soft_reset = acp_soft_reset,
+	.set_clockgating_state = acp_set_clockgating_state,
+	.set_powergating_state = acp_set_powergating_state,
+};
+
+const struct amdgpu_ip_block_version acp_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_ACP,
+	.major = 2,
+	.minor = 2,
+	.rev = 0,
+	.funcs = &acp_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_acp.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_acp.h
new file mode 100644
index 00000000000..a288ce25c17
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_acp.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: AMD
+ *
+ */
+
+#ifndef __AMDGPU_ACP_H__
+#define __AMDGPU_ACP_H__
+
+#include <linux/mfd/core.h>
+
+struct amdgpu_acp {
+	struct device *parent;
+	struct cgs_device *cgs_device;
+	struct amd_acp_private *private;
+	struct mfd_cell *acp_cell;
+	struct resource *acp_res;
+	struct acp_pm_domain *acp_genpd;
+};
+
+extern const struct amdgpu_ip_block_version acp_ip_block;
+
+#endif /* __AMDGPU_ACP_H__ */
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_acpi.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_acpi.c
new file mode 100644
index 00000000000..f008804f0b9
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_acpi.c
@@ -0,0 +1,820 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/pci.h>
+#include <linux/acpi.h>
+#include <linux/slab.h>
+#include <linux/power_supply.h>
+#include <linux/pm_runtime.h>
+#include <acpi/video.h>
+#include <drm/drmP.h>
+#include <drm/drm_crtc_helper.h>
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amd_acpi.h"
+#include "atom.h"
+
+struct amdgpu_atif_notification_cfg {
+	bool enabled;
+	int command_code;
+};
+
+struct amdgpu_atif_notifications {
+	bool display_switch;
+	bool expansion_mode_change;
+	bool thermal_state;
+	bool forced_power_state;
+	bool system_power_state;
+	bool display_conf_change;
+	bool px_gfx_switch;
+	bool brightness_change;
+	bool dgpu_display_event;
+};
+
+struct amdgpu_atif_functions {
+	bool system_params;
+	bool sbios_requests;
+	bool select_active_disp;
+	bool lid_state;
+	bool get_tv_standard;
+	bool set_tv_standard;
+	bool get_panel_expansion_mode;
+	bool set_panel_expansion_mode;
+	bool temperature_change;
+	bool graphics_device_types;
+};
+
+struct amdgpu_atif {
+	acpi_handle handle;
+
+	struct amdgpu_atif_notifications notifications;
+	struct amdgpu_atif_functions functions;
+	struct amdgpu_atif_notification_cfg notification_cfg;
+	struct amdgpu_encoder *encoder_for_bl;
+};
+
+/* Call the ATIF method
+ */
+/**
+ * amdgpu_atif_call - call an ATIF method
+ *
+ * @handle: acpi handle
+ * @function: the ATIF function to execute
+ * @params: ATIF function params
+ *
+ * Executes the requested ATIF function (all asics).
+ * Returns a pointer to the acpi output buffer.
+ */
+static union acpi_object *amdgpu_atif_call(struct amdgpu_atif *atif,
+					   int function,
+					   struct acpi_buffer *params)
+{
+	acpi_status status;
+	union acpi_object atif_arg_elements[2];
+	struct acpi_object_list atif_arg;
+	struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
+
+	atif_arg.count = 2;
+	atif_arg.pointer = &atif_arg_elements[0];
+
+	atif_arg_elements[0].type = ACPI_TYPE_INTEGER;
+	atif_arg_elements[0].integer.value = function;
+
+	if (params) {
+		atif_arg_elements[1].type = ACPI_TYPE_BUFFER;
+		atif_arg_elements[1].buffer.length = params->length;
+		atif_arg_elements[1].buffer.pointer = params->pointer;
+	} else {
+		/* We need a second fake parameter */
+		atif_arg_elements[1].type = ACPI_TYPE_INTEGER;
+		atif_arg_elements[1].integer.value = 0;
+	}
+
+	status = acpi_evaluate_object(atif->handle, NULL, &atif_arg,
+				      &buffer);
+
+	/* Fail only if calling the method fails and ATIF is supported */
+	if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) {
+		DRM_DEBUG_DRIVER("failed to evaluate ATIF got %s\n",
+				 acpi_format_exception(status));
+		kfree(buffer.pointer);
+		return NULL;
+	}
+
+	return buffer.pointer;
+}
+
+/**
+ * amdgpu_atif_parse_notification - parse supported notifications
+ *
+ * @n: supported notifications struct
+ * @mask: supported notifications mask from ATIF
+ *
+ * Use the supported notifications mask from ATIF function
+ * ATIF_FUNCTION_VERIFY_INTERFACE to determine what notifications
+ * are supported (all asics).
+ */
+static void amdgpu_atif_parse_notification(struct amdgpu_atif_notifications *n, u32 mask)
+{
+	n->display_switch = mask & ATIF_DISPLAY_SWITCH_REQUEST_SUPPORTED;
+	n->expansion_mode_change = mask & ATIF_EXPANSION_MODE_CHANGE_REQUEST_SUPPORTED;
+	n->thermal_state = mask & ATIF_THERMAL_STATE_CHANGE_REQUEST_SUPPORTED;
+	n->forced_power_state = mask & ATIF_FORCED_POWER_STATE_CHANGE_REQUEST_SUPPORTED;
+	n->system_power_state = mask & ATIF_SYSTEM_POWER_SOURCE_CHANGE_REQUEST_SUPPORTED;
+	n->display_conf_change = mask & ATIF_DISPLAY_CONF_CHANGE_REQUEST_SUPPORTED;
+	n->px_gfx_switch = mask & ATIF_PX_GFX_SWITCH_REQUEST_SUPPORTED;
+	n->brightness_change = mask & ATIF_PANEL_BRIGHTNESS_CHANGE_REQUEST_SUPPORTED;
+	n->dgpu_display_event = mask & ATIF_DGPU_DISPLAY_EVENT_SUPPORTED;
+}
+
+/**
+ * amdgpu_atif_parse_functions - parse supported functions
+ *
+ * @f: supported functions struct
+ * @mask: supported functions mask from ATIF
+ *
+ * Use the supported functions mask from ATIF function
+ * ATIF_FUNCTION_VERIFY_INTERFACE to determine what functions
+ * are supported (all asics).
+ */
+static void amdgpu_atif_parse_functions(struct amdgpu_atif_functions *f, u32 mask)
+{
+	f->system_params = mask & ATIF_GET_SYSTEM_PARAMETERS_SUPPORTED;
+	f->sbios_requests = mask & ATIF_GET_SYSTEM_BIOS_REQUESTS_SUPPORTED;
+	f->select_active_disp = mask & ATIF_SELECT_ACTIVE_DISPLAYS_SUPPORTED;
+	f->lid_state = mask & ATIF_GET_LID_STATE_SUPPORTED;
+	f->get_tv_standard = mask & ATIF_GET_TV_STANDARD_FROM_CMOS_SUPPORTED;
+	f->set_tv_standard = mask & ATIF_SET_TV_STANDARD_IN_CMOS_SUPPORTED;
+	f->get_panel_expansion_mode = mask & ATIF_GET_PANEL_EXPANSION_MODE_FROM_CMOS_SUPPORTED;
+	f->set_panel_expansion_mode = mask & ATIF_SET_PANEL_EXPANSION_MODE_IN_CMOS_SUPPORTED;
+	f->temperature_change = mask & ATIF_TEMPERATURE_CHANGE_NOTIFICATION_SUPPORTED;
+	f->graphics_device_types = mask & ATIF_GET_GRAPHICS_DEVICE_TYPES_SUPPORTED;
+}
+
+/**
+ * amdgpu_atif_verify_interface - verify ATIF
+ *
+ * @handle: acpi handle
+ * @atif: amdgpu atif struct
+ *
+ * Execute the ATIF_FUNCTION_VERIFY_INTERFACE ATIF function
+ * to initialize ATIF and determine what features are supported
+ * (all asics).
+ * returns 0 on success, error on failure.
+ */
+static int amdgpu_atif_verify_interface(struct amdgpu_atif *atif)
+{
+	union acpi_object *info;
+	struct atif_verify_interface output;
+	size_t size;
+	int err = 0;
+
+	info = amdgpu_atif_call(atif, ATIF_FUNCTION_VERIFY_INTERFACE, NULL);
+	if (!info)
+		return -EIO;
+
+	memset(&output, 0, sizeof(output));
+
+	size = *(u16 *) info->buffer.pointer;
+	if (size < 12) {
+		DRM_INFO("ATIF buffer is too small: %zu\n", size);
+		err = -EINVAL;
+		goto out;
+	}
+	size = min(sizeof(output), size);
+
+	memcpy(&output, info->buffer.pointer, size);
+
+	/* TODO: check version? */
+	DRM_DEBUG_DRIVER("ATIF version %u\n", output.version);
+
+	amdgpu_atif_parse_notification(&atif->notifications, output.notification_mask);
+	amdgpu_atif_parse_functions(&atif->functions, output.function_bits);
+
+out:
+	kfree(info);
+	return err;
+}
+
+static acpi_handle amdgpu_atif_probe_handle(acpi_handle dhandle)
+{
+	acpi_handle handle = NULL;
+	char acpi_method_name[255] = { 0 };
+	struct acpi_buffer buffer = { sizeof(acpi_method_name), acpi_method_name };
+	acpi_status status;
+
+	/* For PX/HG systems, ATIF and ATPX are in the iGPU's namespace, on dGPU only
+	 * systems, ATIF is in the dGPU's namespace.
+	 */
+	status = acpi_get_handle(dhandle, "ATIF", &handle);
+	if (ACPI_SUCCESS(status))
+		goto out;
+
+	if (amdgpu_has_atpx()) {
+		status = acpi_get_handle(amdgpu_atpx_get_dhandle(), "ATIF",
+					 &handle);
+		if (ACPI_SUCCESS(status))
+			goto out;
+	}
+
+	DRM_DEBUG_DRIVER("No ATIF handle found\n");
+	return NULL;
+out:
+	acpi_get_name(handle, ACPI_FULL_PATHNAME, &buffer);
+	DRM_DEBUG_DRIVER("Found ATIF handle %s\n", acpi_method_name);
+	return handle;
+}
+
+/**
+ * amdgpu_atif_get_notification_params - determine notify configuration
+ *
+ * @handle: acpi handle
+ * @n: atif notification configuration struct
+ *
+ * Execute the ATIF_FUNCTION_GET_SYSTEM_PARAMETERS ATIF function
+ * to determine if a notifier is used and if so which one
+ * (all asics).  This is either Notify(VGA, 0x81) or Notify(VGA, n)
+ * where n is specified in the result if a notifier is used.
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atif_get_notification_params(struct amdgpu_atif *atif)
+{
+	union acpi_object *info;
+	struct amdgpu_atif_notification_cfg *n = &atif->notification_cfg;
+	struct atif_system_params params;
+	size_t size;
+	int err = 0;
+
+	info = amdgpu_atif_call(atif, ATIF_FUNCTION_GET_SYSTEM_PARAMETERS,
+				NULL);
+	if (!info) {
+		err = -EIO;
+		goto out;
+	}
+
+	size = *(u16 *) info->buffer.pointer;
+	if (size < 10) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	memset(&params, 0, sizeof(params));
+	size = min(sizeof(params), size);
+	memcpy(&params, info->buffer.pointer, size);
+
+	DRM_DEBUG_DRIVER("SYSTEM_PARAMS: mask = %#x, flags = %#x\n",
+			params.flags, params.valid_mask);
+	params.flags = params.flags & params.valid_mask;
+
+	if ((params.flags & ATIF_NOTIFY_MASK) == ATIF_NOTIFY_NONE) {
+		n->enabled = false;
+		n->command_code = 0;
+	} else if ((params.flags & ATIF_NOTIFY_MASK) == ATIF_NOTIFY_81) {
+		n->enabled = true;
+		n->command_code = 0x81;
+	} else {
+		if (size < 11) {
+			err = -EINVAL;
+			goto out;
+		}
+		n->enabled = true;
+		n->command_code = params.command_code;
+	}
+
+out:
+	DRM_DEBUG_DRIVER("Notification %s, command code = %#x\n",
+			(n->enabled ? "enabled" : "disabled"),
+			n->command_code);
+	kfree(info);
+	return err;
+}
+
+/**
+ * amdgpu_atif_get_sbios_requests - get requested sbios event
+ *
+ * @handle: acpi handle
+ * @req: atif sbios request struct
+ *
+ * Execute the ATIF_FUNCTION_GET_SYSTEM_BIOS_REQUESTS ATIF function
+ * to determine what requests the sbios is making to the driver
+ * (all asics).
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atif_get_sbios_requests(struct amdgpu_atif *atif,
+					  struct atif_sbios_requests *req)
+{
+	union acpi_object *info;
+	size_t size;
+	int count = 0;
+
+	info = amdgpu_atif_call(atif, ATIF_FUNCTION_GET_SYSTEM_BIOS_REQUESTS,
+				NULL);
+	if (!info)
+		return -EIO;
+
+	size = *(u16 *)info->buffer.pointer;
+	if (size < 0xd) {
+		count = -EINVAL;
+		goto out;
+	}
+	memset(req, 0, sizeof(*req));
+
+	size = min(sizeof(*req), size);
+	memcpy(req, info->buffer.pointer, size);
+	DRM_DEBUG_DRIVER("SBIOS pending requests: %#x\n", req->pending);
+
+	count = hweight32(req->pending);
+
+out:
+	kfree(info);
+	return count;
+}
+
+/**
+ * amdgpu_atif_handler - handle ATIF notify requests
+ *
+ * @adev: amdgpu_device pointer
+ * @event: atif sbios request struct
+ *
+ * Checks the acpi event and if it matches an atif event,
+ * handles it.
+ *
+ * Returns:
+ * NOTIFY_BAD or NOTIFY_DONE, depending on the event.
+ */
+static int amdgpu_atif_handler(struct amdgpu_device *adev,
+			       struct acpi_bus_event *event)
+{
+	struct amdgpu_atif *atif = adev->atif;
+	int count;
+
+	DRM_DEBUG_DRIVER("event, device_class = %s, type = %#x\n",
+			event->device_class, event->type);
+
+	if (strcmp(event->device_class, ACPI_VIDEO_CLASS) != 0)
+		return NOTIFY_DONE;
+
+	/* Is this actually our event? */
+	if (!atif ||
+	    !atif->notification_cfg.enabled ||
+	    event->type != atif->notification_cfg.command_code) {
+		/* These events will generate keypresses otherwise */
+		if (event->type == ACPI_VIDEO_NOTIFY_PROBE)
+			return NOTIFY_BAD;
+		else
+			return NOTIFY_DONE;
+	}
+
+	if (atif->functions.sbios_requests) {
+		struct atif_sbios_requests req;
+
+		/* Check pending SBIOS requests */
+		count = amdgpu_atif_get_sbios_requests(atif, &req);
+
+		if (count <= 0)
+			return NOTIFY_BAD;
+
+		DRM_DEBUG_DRIVER("ATIF: %d pending SBIOS requests\n", count);
+
+		/* todo: add DC handling */
+		if ((req.pending & ATIF_PANEL_BRIGHTNESS_CHANGE_REQUEST) &&
+		    !amdgpu_device_has_dc_support(adev)) {
+			struct amdgpu_encoder *enc = atif->encoder_for_bl;
+
+			if (enc) {
+				struct amdgpu_encoder_atom_dig *dig = enc->enc_priv;
+
+				DRM_DEBUG_DRIVER("Changing brightness to %d\n",
+						 req.backlight_level);
+
+				amdgpu_display_backlight_set_level(adev, enc, req.backlight_level);
+
+#if defined(CONFIG_BACKLIGHT_CLASS_DEVICE) || defined(CONFIG_BACKLIGHT_CLASS_DEVICE_MODULE)
+				backlight_force_update(dig->bl_dev,
+						       BACKLIGHT_UPDATE_HOTKEY);
+#endif
+			}
+		}
+		if (req.pending & ATIF_DGPU_DISPLAY_EVENT) {
+			if ((adev->flags & AMD_IS_PX) &&
+			    amdgpu_atpx_dgpu_req_power_for_displays()) {
+				pm_runtime_get_sync(adev->ddev->dev);
+				/* Just fire off a uevent and let userspace tell us what to do */
+				drm_helper_hpd_irq_event(adev->ddev);
+				pm_runtime_mark_last_busy(adev->ddev->dev);
+				pm_runtime_put_autosuspend(adev->ddev->dev);
+			}
+		}
+		/* TODO: check other events */
+	}
+
+	/* We've handled the event, stop the notifier chain. The ACPI interface
+	 * overloads ACPI_VIDEO_NOTIFY_PROBE, we don't want to send that to
+	 * userspace if the event was generated only to signal a SBIOS
+	 * request.
+	 */
+	return NOTIFY_BAD;
+}
+
+/* Call the ATCS method
+ */
+/**
+ * amdgpu_atcs_call - call an ATCS method
+ *
+ * @handle: acpi handle
+ * @function: the ATCS function to execute
+ * @params: ATCS function params
+ *
+ * Executes the requested ATCS function (all asics).
+ * Returns a pointer to the acpi output buffer.
+ */
+static union acpi_object *amdgpu_atcs_call(acpi_handle handle, int function,
+					   struct acpi_buffer *params)
+{
+	acpi_status status;
+	union acpi_object atcs_arg_elements[2];
+	struct acpi_object_list atcs_arg;
+	struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
+
+	atcs_arg.count = 2;
+	atcs_arg.pointer = &atcs_arg_elements[0];
+
+	atcs_arg_elements[0].type = ACPI_TYPE_INTEGER;
+	atcs_arg_elements[0].integer.value = function;
+
+	if (params) {
+		atcs_arg_elements[1].type = ACPI_TYPE_BUFFER;
+		atcs_arg_elements[1].buffer.length = params->length;
+		atcs_arg_elements[1].buffer.pointer = params->pointer;
+	} else {
+		/* We need a second fake parameter */
+		atcs_arg_elements[1].type = ACPI_TYPE_INTEGER;
+		atcs_arg_elements[1].integer.value = 0;
+	}
+
+	status = acpi_evaluate_object(handle, "ATCS", &atcs_arg, &buffer);
+
+	/* Fail only if calling the method fails and ATIF is supported */
+	if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) {
+		DRM_DEBUG_DRIVER("failed to evaluate ATCS got %s\n",
+				 acpi_format_exception(status));
+		kfree(buffer.pointer);
+		return NULL;
+	}
+
+	return buffer.pointer;
+}
+
+/**
+ * amdgpu_atcs_parse_functions - parse supported functions
+ *
+ * @f: supported functions struct
+ * @mask: supported functions mask from ATCS
+ *
+ * Use the supported functions mask from ATCS function
+ * ATCS_FUNCTION_VERIFY_INTERFACE to determine what functions
+ * are supported (all asics).
+ */
+static void amdgpu_atcs_parse_functions(struct amdgpu_atcs_functions *f, u32 mask)
+{
+	f->get_ext_state = mask & ATCS_GET_EXTERNAL_STATE_SUPPORTED;
+	f->pcie_perf_req = mask & ATCS_PCIE_PERFORMANCE_REQUEST_SUPPORTED;
+	f->pcie_dev_rdy = mask & ATCS_PCIE_DEVICE_READY_NOTIFICATION_SUPPORTED;
+	f->pcie_bus_width = mask & ATCS_SET_PCIE_BUS_WIDTH_SUPPORTED;
+}
+
+/**
+ * amdgpu_atcs_verify_interface - verify ATCS
+ *
+ * @handle: acpi handle
+ * @atcs: amdgpu atcs struct
+ *
+ * Execute the ATCS_FUNCTION_VERIFY_INTERFACE ATCS function
+ * to initialize ATCS and determine what features are supported
+ * (all asics).
+ * returns 0 on success, error on failure.
+ */
+static int amdgpu_atcs_verify_interface(acpi_handle handle,
+					struct amdgpu_atcs *atcs)
+{
+	union acpi_object *info;
+	struct atcs_verify_interface output;
+	size_t size;
+	int err = 0;
+
+	info = amdgpu_atcs_call(handle, ATCS_FUNCTION_VERIFY_INTERFACE, NULL);
+	if (!info)
+		return -EIO;
+
+	memset(&output, 0, sizeof(output));
+
+	size = *(u16 *) info->buffer.pointer;
+	if (size < 8) {
+		DRM_INFO("ATCS buffer is too small: %zu\n", size);
+		err = -EINVAL;
+		goto out;
+	}
+	size = min(sizeof(output), size);
+
+	memcpy(&output, info->buffer.pointer, size);
+
+	/* TODO: check version? */
+	DRM_DEBUG_DRIVER("ATCS version %u\n", output.version);
+
+	amdgpu_atcs_parse_functions(&atcs->functions, output.function_bits);
+
+out:
+	kfree(info);
+	return err;
+}
+
+/**
+ * amdgpu_acpi_is_pcie_performance_request_supported
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Check if the ATCS pcie_perf_req and pcie_dev_rdy methods
+ * are supported (all asics).
+ * returns true if supported, false if not.
+ */
+bool amdgpu_acpi_is_pcie_performance_request_supported(struct amdgpu_device *adev)
+{
+	struct amdgpu_atcs *atcs = &adev->atcs;
+
+	if (atcs->functions.pcie_perf_req && atcs->functions.pcie_dev_rdy)
+		return true;
+
+	return false;
+}
+
+/**
+ * amdgpu_acpi_pcie_notify_device_ready
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Executes the PCIE_DEVICE_READY_NOTIFICATION method
+ * (all asics).
+ * returns 0 on success, error on failure.
+ */
+int amdgpu_acpi_pcie_notify_device_ready(struct amdgpu_device *adev)
+{
+	acpi_handle handle;
+	union acpi_object *info;
+	struct amdgpu_atcs *atcs = &adev->atcs;
+
+	/* Get the device handle */
+	handle = ACPI_HANDLE(&adev->pdev->dev);
+	if (!handle)
+		return -EINVAL;
+
+	if (!atcs->functions.pcie_dev_rdy)
+		return -EINVAL;
+
+	info = amdgpu_atcs_call(handle, ATCS_FUNCTION_PCIE_DEVICE_READY_NOTIFICATION, NULL);
+	if (!info)
+		return -EIO;
+
+	kfree(info);
+
+	return 0;
+}
+
+/**
+ * amdgpu_acpi_pcie_performance_request
+ *
+ * @adev: amdgpu_device pointer
+ * @perf_req: requested perf level (pcie gen speed)
+ * @advertise: set advertise caps flag if set
+ *
+ * Executes the PCIE_PERFORMANCE_REQUEST method to
+ * change the pcie gen speed (all asics).
+ * returns 0 on success, error on failure.
+ */
+int amdgpu_acpi_pcie_performance_request(struct amdgpu_device *adev,
+					 u8 perf_req, bool advertise)
+{
+	acpi_handle handle;
+	union acpi_object *info;
+	struct amdgpu_atcs *atcs = &adev->atcs;
+	struct atcs_pref_req_input atcs_input;
+	struct atcs_pref_req_output atcs_output;
+	struct acpi_buffer params;
+	size_t size;
+	u32 retry = 3;
+
+	if (amdgpu_acpi_pcie_notify_device_ready(adev))
+		return -EINVAL;
+
+	/* Get the device handle */
+	handle = ACPI_HANDLE(&adev->pdev->dev);
+	if (!handle)
+		return -EINVAL;
+
+	if (!atcs->functions.pcie_perf_req)
+		return -EINVAL;
+
+	atcs_input.size = sizeof(struct atcs_pref_req_input);
+	/* client id (bit 2-0: func num, 7-3: dev num, 15-8: bus num) */
+	atcs_input.client_id = adev->pdev->devfn | (adev->pdev->bus->number << 8);
+	atcs_input.valid_flags_mask = ATCS_VALID_FLAGS_MASK;
+	atcs_input.flags = ATCS_WAIT_FOR_COMPLETION;
+	if (advertise)
+		atcs_input.flags |= ATCS_ADVERTISE_CAPS;
+	atcs_input.req_type = ATCS_PCIE_LINK_SPEED;
+	atcs_input.perf_req = perf_req;
+
+	params.length = sizeof(struct atcs_pref_req_input);
+	params.pointer = &atcs_input;
+
+	while (retry--) {
+		info = amdgpu_atcs_call(handle, ATCS_FUNCTION_PCIE_PERFORMANCE_REQUEST, &params);
+		if (!info)
+			return -EIO;
+
+		memset(&atcs_output, 0, sizeof(atcs_output));
+
+		size = *(u16 *) info->buffer.pointer;
+		if (size < 3) {
+			DRM_INFO("ATCS buffer is too small: %zu\n", size);
+			kfree(info);
+			return -EINVAL;
+		}
+		size = min(sizeof(atcs_output), size);
+
+		memcpy(&atcs_output, info->buffer.pointer, size);
+
+		kfree(info);
+
+		switch (atcs_output.ret_val) {
+		case ATCS_REQUEST_REFUSED:
+		default:
+			return -EINVAL;
+		case ATCS_REQUEST_COMPLETE:
+			return 0;
+		case ATCS_REQUEST_IN_PROGRESS:
+			udelay(10);
+			break;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_acpi_event - handle notify events
+ *
+ * @nb: notifier block
+ * @val: val
+ * @data: acpi event
+ *
+ * Calls relevant amdgpu functions in response to various
+ * acpi events.
+ * Returns NOTIFY code
+ */
+static int amdgpu_acpi_event(struct notifier_block *nb,
+			     unsigned long val,
+			     void *data)
+{
+	struct amdgpu_device *adev = container_of(nb, struct amdgpu_device, acpi_nb);
+	struct acpi_bus_event *entry = (struct acpi_bus_event *)data;
+
+	if (strcmp(entry->device_class, ACPI_AC_CLASS) == 0) {
+		if (power_supply_is_system_supplied() > 0)
+			DRM_DEBUG_DRIVER("pm: AC\n");
+		else
+			DRM_DEBUG_DRIVER("pm: DC\n");
+
+		amdgpu_pm_acpi_event_handler(adev);
+	}
+
+	/* Check for pending SBIOS requests */
+	return amdgpu_atif_handler(adev, entry);
+}
+
+/* Call all ACPI methods here */
+/**
+ * amdgpu_acpi_init - init driver acpi support
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Verifies the AMD ACPI interfaces and registers with the acpi
+ * notifier chain (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_acpi_init(struct amdgpu_device *adev)
+{
+	acpi_handle handle, atif_handle;
+	struct amdgpu_atif *atif;
+	struct amdgpu_atcs *atcs = &adev->atcs;
+	int ret;
+
+	/* Get the device handle */
+	handle = ACPI_HANDLE(&adev->pdev->dev);
+
+	if (!adev->bios || !handle)
+		return 0;
+
+	/* Call the ATCS method */
+	ret = amdgpu_atcs_verify_interface(handle, atcs);
+	if (ret) {
+		DRM_DEBUG_DRIVER("Call to ATCS verify_interface failed: %d\n", ret);
+	}
+
+	/* Probe for ATIF, and initialize it if found */
+	atif_handle = amdgpu_atif_probe_handle(handle);
+	if (!atif_handle)
+		goto out;
+
+	atif = kzalloc(sizeof(*atif), GFP_KERNEL);
+	if (!atif) {
+		DRM_WARN("Not enough memory to initialize ATIF\n");
+		goto out;
+	}
+	atif->handle = atif_handle;
+
+	/* Call the ATIF method */
+	ret = amdgpu_atif_verify_interface(atif);
+	if (ret) {
+		DRM_DEBUG_DRIVER("Call to ATIF verify_interface failed: %d\n", ret);
+		kfree(atif);
+		goto out;
+	}
+	adev->atif = atif;
+
+	if (atif->notifications.brightness_change) {
+		struct drm_encoder *tmp;
+
+		/* Find the encoder controlling the brightness */
+		list_for_each_entry(tmp, &adev->ddev->mode_config.encoder_list,
+				head) {
+			struct amdgpu_encoder *enc = to_amdgpu_encoder(tmp);
+
+			if ((enc->devices & (ATOM_DEVICE_LCD_SUPPORT)) &&
+			    enc->enc_priv) {
+				struct amdgpu_encoder_atom_dig *dig = enc->enc_priv;
+				if (dig->bl_dev) {
+					atif->encoder_for_bl = enc;
+					break;
+				}
+			}
+		}
+	}
+
+	if (atif->functions.sbios_requests && !atif->functions.system_params) {
+		/* XXX check this workraround, if sbios request function is
+		 * present we have to see how it's configured in the system
+		 * params
+		 */
+		atif->functions.system_params = true;
+	}
+
+	if (atif->functions.system_params) {
+		ret = amdgpu_atif_get_notification_params(atif);
+		if (ret) {
+			DRM_DEBUG_DRIVER("Call to GET_SYSTEM_PARAMS failed: %d\n",
+					ret);
+			/* Disable notification */
+			atif->notification_cfg.enabled = false;
+		}
+	}
+
+out:
+	adev->acpi_nb.notifier_call = amdgpu_acpi_event;
+	register_acpi_notifier(&adev->acpi_nb);
+
+	return ret;
+}
+
+/**
+ * amdgpu_acpi_fini - tear down driver acpi support
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Unregisters with the acpi notifier chain (all asics).
+ */
+void amdgpu_acpi_fini(struct amdgpu_device *adev)
+{
+	unregister_acpi_notifier(&adev->acpi_nb);
+	if (adev->atif)
+		kfree(adev->atif);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_afmt.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_afmt.c
new file mode 100644
index 00000000000..3889486f71f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_afmt.c
@@ -0,0 +1,105 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Christian König.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König
+ */
+#include <linux/hdmi.h>
+#include <linux/gcd.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+static const struct amdgpu_afmt_acr amdgpu_afmt_predefined_acr[] = {
+    /*	     32kHz	  44.1kHz	48kHz    */
+    /* Clock      N     CTS      N     CTS      N     CTS */
+    {  25175,  4096,  25175, 28224, 125875,  6144,  25175 }, /*  25,20/1.001 MHz */
+    {  25200,  4096,  25200,  6272,  28000,  6144,  25200 }, /*  25.20       MHz */
+    {  27000,  4096,  27000,  6272,  30000,  6144,  27000 }, /*  27.00       MHz */
+    {  27027,  4096,  27027,  6272,  30030,  6144,  27027 }, /*  27.00*1.001 MHz */
+    {  54000,  4096,  54000,  6272,  60000,  6144,  54000 }, /*  54.00       MHz */
+    {  54054,  4096,  54054,  6272,  60060,  6144,  54054 }, /*  54.00*1.001 MHz */
+    {  74176,  4096,  74176,  5733,  75335,  6144,  74176 }, /*  74.25/1.001 MHz */
+    {  74250,  4096,  74250,  6272,  82500,  6144,  74250 }, /*  74.25       MHz */
+    { 148352,  4096, 148352,  5733, 150670,  6144, 148352 }, /* 148.50/1.001 MHz */
+    { 148500,  4096, 148500,  6272, 165000,  6144, 148500 }, /* 148.50       MHz */
+};
+
+
+/*
+ * calculate CTS and N values if they are not found in the table
+ */
+static void amdgpu_afmt_calc_cts(uint32_t clock, int *CTS, int *N, int freq)
+{
+	int n, cts;
+	unsigned long div, mul;
+
+	/* Safe, but overly large values */
+	n = 128 * freq;
+	cts = clock * 1000;
+
+	/* Smallest valid fraction */
+	div = gcd(n, cts);
+
+	n /= div;
+	cts /= div;
+
+	/*
+	 * The optimal N is 128*freq/1000. Calculate the closest larger
+	 * value that doesn't truncate any bits.
+	 */
+	mul = ((128*freq/1000) + (n-1))/n;
+
+	n *= mul;
+	cts *= mul;
+
+	/* Check that we are in spec (not always possible) */
+	if (n < (128*freq/1500))
+		pr_warn("Calculated ACR N value is too small. You may experience audio problems.\n");
+	if (n > (128*freq/300))
+		pr_warn("Calculated ACR N value is too large. You may experience audio problems.\n");
+
+	*N = n;
+	*CTS = cts;
+
+	DRM_DEBUG("Calculated ACR timing N=%d CTS=%d for frequency %d\n",
+		  *N, *CTS, freq);
+}
+
+struct amdgpu_afmt_acr amdgpu_afmt_acr(uint32_t clock)
+{
+	struct amdgpu_afmt_acr res;
+	u8 i;
+
+	/* Precalculated values for common clocks */
+	for (i = 0; i < ARRAY_SIZE(amdgpu_afmt_predefined_acr); i++) {
+		if (amdgpu_afmt_predefined_acr[i].clock == clock)
+			return amdgpu_afmt_predefined_acr[i];
+	}
+
+	/* And odd clocks get manually calculated */
+	amdgpu_afmt_calc_cts(clock, &res.cts_32khz, &res.n_32khz, 32000);
+	amdgpu_afmt_calc_cts(clock, &res.cts_44_1khz, &res.n_44_1khz, 44100);
+	amdgpu_afmt_calc_cts(clock, &res.cts_48khz, &res.n_48khz, 48000);
+
+	return res;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd.c
new file mode 100644
index 00000000000..cc83b7b6540
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -0,0 +1,560 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "amdgpu_amdkfd.h"
+#include "amd_shared.h"
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include <linux/module.h>
+
+const struct kgd2kfd_calls *kgd2kfd;
+bool (*kgd2kfd_init_p)(unsigned int, const struct kgd2kfd_calls**);
+
+static const unsigned int compute_vmid_bitmap = 0xFF00;
+
+int amdgpu_amdkfd_init(void)
+{
+	int ret;
+
+#if defined(CONFIG_HSA_AMD_MODULE)
+	int (*kgd2kfd_init_p)(unsigned int, const struct kgd2kfd_calls**);
+
+	kgd2kfd_init_p = symbol_request(kgd2kfd_init);
+
+	if (kgd2kfd_init_p == NULL)
+		return -ENOENT;
+
+	ret = kgd2kfd_init_p(KFD_INTERFACE_VERSION, &kgd2kfd);
+	if (ret) {
+		symbol_put(kgd2kfd_init);
+		kgd2kfd = NULL;
+	}
+
+
+#elif defined(CONFIG_HSA_AMD)
+
+	ret = kgd2kfd_init(KFD_INTERFACE_VERSION, &kgd2kfd);
+	if (ret)
+		kgd2kfd = NULL;
+
+#else
+	kgd2kfd = NULL;
+	ret = -ENOENT;
+#endif
+
+#if defined(CONFIG_HSA_AMD_MODULE) || defined(CONFIG_HSA_AMD)
+	amdgpu_amdkfd_gpuvm_init_mem_limits();
+#endif
+
+	return ret;
+}
+
+void amdgpu_amdkfd_fini(void)
+{
+	if (kgd2kfd) {
+		kgd2kfd->exit();
+		symbol_put(kgd2kfd_init);
+	}
+}
+
+void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
+{
+	const struct kfd2kgd_calls *kfd2kgd;
+
+	if (!kgd2kfd)
+		return;
+
+	switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_KAVERI:
+	case CHIP_HAWAII:
+		kfd2kgd = amdgpu_amdkfd_gfx_7_get_functions();
+		break;
+#endif
+	case CHIP_CARRIZO:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+		kfd2kgd = amdgpu_amdkfd_gfx_8_0_get_functions();
+		break;
+	case CHIP_VEGA10:
+	case CHIP_RAVEN:
+		kfd2kgd = amdgpu_amdkfd_gfx_9_0_get_functions();
+		break;
+	default:
+		dev_info(adev->dev, "kfd not supported on this ASIC\n");
+		return;
+	}
+
+	adev->kfd = kgd2kfd->probe((struct kgd_dev *)adev,
+				   adev->pdev, kfd2kgd);
+}
+
+/**
+ * amdgpu_doorbell_get_kfd_info - Report doorbell configuration required to
+ *                                setup amdkfd
+ *
+ * @adev: amdgpu_device pointer
+ * @aperture_base: output returning doorbell aperture base physical address
+ * @aperture_size: output returning doorbell aperture size in bytes
+ * @start_offset: output returning # of doorbell bytes reserved for amdgpu.
+ *
+ * amdgpu and amdkfd share the doorbell aperture. amdgpu sets it up,
+ * takes doorbells required for its own rings and reports the setup to amdkfd.
+ * amdgpu reserved doorbells are at the start of the doorbell aperture.
+ */
+static void amdgpu_doorbell_get_kfd_info(struct amdgpu_device *adev,
+					 phys_addr_t *aperture_base,
+					 size_t *aperture_size,
+					 size_t *start_offset)
+{
+	/*
+	 * The first num_doorbells are used by amdgpu.
+	 * amdkfd takes whatever's left in the aperture.
+	 */
+	if (adev->doorbell.size > adev->doorbell.num_doorbells * sizeof(u32)) {
+		*aperture_base = adev->doorbell.base;
+		*aperture_size = adev->doorbell.size;
+		*start_offset = adev->doorbell.num_doorbells * sizeof(u32);
+	} else {
+		*aperture_base = 0;
+		*aperture_size = 0;
+		*start_offset = 0;
+	}
+}
+
+void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
+{
+	int i;
+	int last_valid_bit;
+	if (adev->kfd) {
+		struct kgd2kfd_shared_resources gpu_resources = {
+			.compute_vmid_bitmap = compute_vmid_bitmap,
+			.num_pipe_per_mec = adev->gfx.mec.num_pipe_per_mec,
+			.num_queue_per_pipe = adev->gfx.mec.num_queue_per_pipe,
+			.gpuvm_size = min(adev->vm_manager.max_pfn
+					  << AMDGPU_GPU_PAGE_SHIFT,
+					  AMDGPU_VA_HOLE_START),
+#ifdef notyet
+			.drm_render_minor = adev->ddev->render->index
+#endif
+		};
+
+		/* this is going to have a few of the MSBs set that we need to
+		 * clear */
+		bitmap_complement(gpu_resources.queue_bitmap,
+				  adev->gfx.mec.queue_bitmap,
+				  KGD_MAX_QUEUES);
+
+		/* remove the KIQ bit as well */
+		if (adev->gfx.kiq.ring.ready)
+			clear_bit(amdgpu_gfx_queue_to_bit(adev,
+							  adev->gfx.kiq.ring.me - 1,
+							  adev->gfx.kiq.ring.pipe,
+							  adev->gfx.kiq.ring.queue),
+				  gpu_resources.queue_bitmap);
+
+		/* According to linux/bitmap.h we shouldn't use bitmap_clear if
+		 * nbits is not compile time constant */
+		last_valid_bit = 1 /* only first MEC can have compute queues */
+				* adev->gfx.mec.num_pipe_per_mec
+				* adev->gfx.mec.num_queue_per_pipe;
+		for (i = last_valid_bit; i < KGD_MAX_QUEUES; ++i)
+			clear_bit(i, gpu_resources.queue_bitmap);
+
+		amdgpu_doorbell_get_kfd_info(adev,
+				&gpu_resources.doorbell_physical_address,
+				&gpu_resources.doorbell_aperture_size,
+				&gpu_resources.doorbell_start_offset);
+		if (adev->asic_type >= CHIP_VEGA10) {
+			/* On SOC15 the BIF is involved in routing
+			 * doorbells using the low 12 bits of the
+			 * address. Communicate the assignments to
+			 * KFD. KFD uses two doorbell pages per
+			 * process in case of 64-bit doorbells so we
+			 * can use each doorbell assignment twice.
+			 */
+			gpu_resources.sdma_doorbell[0][0] =
+				AMDGPU_DOORBELL64_sDMA_ENGINE0;
+			gpu_resources.sdma_doorbell[0][1] =
+				AMDGPU_DOORBELL64_sDMA_ENGINE0 + 0x200;
+			gpu_resources.sdma_doorbell[1][0] =
+				AMDGPU_DOORBELL64_sDMA_ENGINE1;
+			gpu_resources.sdma_doorbell[1][1] =
+				AMDGPU_DOORBELL64_sDMA_ENGINE1 + 0x200;
+			/* Doorbells 0x0f0-0ff and 0x2f0-2ff are reserved for
+			 * SDMA, IH and VCN. So don't use them for the CP.
+			 */
+			gpu_resources.reserved_doorbell_mask = 0x1f0;
+			gpu_resources.reserved_doorbell_val  = 0x0f0;
+		}
+
+		kgd2kfd->device_init(adev->kfd, &gpu_resources);
+	}
+}
+
+void amdgpu_amdkfd_device_fini(struct amdgpu_device *adev)
+{
+	if (adev->kfd) {
+		kgd2kfd->device_exit(adev->kfd);
+		adev->kfd = NULL;
+	}
+}
+
+void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev,
+		const void *ih_ring_entry)
+{
+	if (adev->kfd)
+		kgd2kfd->interrupt(adev->kfd, ih_ring_entry);
+}
+
+void amdgpu_amdkfd_suspend(struct amdgpu_device *adev)
+{
+	if (adev->kfd)
+		kgd2kfd->suspend(adev->kfd);
+}
+
+int amdgpu_amdkfd_resume(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (adev->kfd)
+		r = kgd2kfd->resume(adev->kfd);
+
+	return r;
+}
+
+int amdgpu_amdkfd_pre_reset(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (adev->kfd)
+		r = kgd2kfd->pre_reset(adev->kfd);
+
+	return r;
+}
+
+int amdgpu_amdkfd_post_reset(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (adev->kfd)
+		r = kgd2kfd->post_reset(adev->kfd);
+
+	return r;
+}
+
+void amdgpu_amdkfd_gpu_reset(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+
+	amdgpu_device_gpu_recover(adev, NULL, false);
+}
+
+int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
+			void **mem_obj, uint64_t *gpu_addr,
+			void **cpu_ptr, bool mqd_gfx9)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+	struct amdgpu_bo *bo = NULL;
+	struct amdgpu_bo_param bp;
+	int r;
+	void *cpu_ptr_tmp = NULL;
+
+	memset(&bp, 0, sizeof(bp));
+	bp.size = size;
+	bp.byte_align = PAGE_SIZE;
+	bp.domain = AMDGPU_GEM_DOMAIN_GTT;
+	bp.flags = AMDGPU_GEM_CREATE_CPU_GTT_USWC;
+	bp.type = ttm_bo_type_kernel;
+	bp.resv = NULL;
+
+	if (mqd_gfx9)
+		bp.flags |= AMDGPU_GEM_CREATE_MQD_GFX9;
+
+	r = amdgpu_bo_create(adev, &bp, &bo);
+	if (r) {
+		dev_err(adev->dev,
+			"failed to allocate BO for amdkfd (%d)\n", r);
+		return r;
+	}
+
+	/* map the buffer */
+	r = amdgpu_bo_reserve(bo, true);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to reserve bo for amdkfd\n", r);
+		goto allocate_mem_reserve_bo_failed;
+	}
+
+	r = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to pin bo for amdkfd\n", r);
+		goto allocate_mem_pin_bo_failed;
+	}
+
+	r = amdgpu_ttm_alloc_gart(&bo->tbo);
+	if (r) {
+		dev_err(adev->dev, "%p bind failed\n", bo);
+		goto allocate_mem_kmap_bo_failed;
+	}
+
+	r = amdgpu_bo_kmap(bo, &cpu_ptr_tmp);
+	if (r) {
+		dev_err(adev->dev,
+			"(%d) failed to map bo to kernel for amdkfd\n", r);
+		goto allocate_mem_kmap_bo_failed;
+	}
+
+	*mem_obj = bo;
+	*gpu_addr = amdgpu_bo_gpu_offset(bo);
+	*cpu_ptr = cpu_ptr_tmp;
+
+	amdgpu_bo_unreserve(bo);
+
+	return 0;
+
+allocate_mem_kmap_bo_failed:
+	amdgpu_bo_unpin(bo);
+allocate_mem_pin_bo_failed:
+	amdgpu_bo_unreserve(bo);
+allocate_mem_reserve_bo_failed:
+	amdgpu_bo_unref(&bo);
+
+	return r;
+}
+
+void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj)
+{
+	struct amdgpu_bo *bo = (struct amdgpu_bo *) mem_obj;
+
+	amdgpu_bo_reserve(bo, true);
+	amdgpu_bo_kunmap(bo);
+	amdgpu_bo_unpin(bo);
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&(bo));
+}
+
+void get_local_mem_info(struct kgd_dev *kgd,
+			struct kfd_local_mem_info *mem_info)
+{
+	STUB();
+#if 0
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+	uint64_t address_mask = adev->dev->dma_mask ? ~*adev->dev->dma_mask :
+					     ~((1ULL << 32) - 1);
+	resource_size_t aper_limit = adev->gmc.aper_base + adev->gmc.aper_size;
+
+	memset(mem_info, 0, sizeof(*mem_info));
+	if (!(adev->gmc.aper_base & address_mask || aper_limit & address_mask)) {
+		mem_info->local_mem_size_public = adev->gmc.visible_vram_size;
+		mem_info->local_mem_size_private = adev->gmc.real_vram_size -
+				adev->gmc.visible_vram_size;
+	} else {
+		mem_info->local_mem_size_public = 0;
+		mem_info->local_mem_size_private = adev->gmc.real_vram_size;
+	}
+	mem_info->vram_width = adev->gmc.vram_width;
+
+	pr_debug("Address base: %pap limit %pap public 0x%llx private 0x%llx\n",
+			&adev->gmc.aper_base, &aper_limit,
+			mem_info->local_mem_size_public,
+			mem_info->local_mem_size_private);
+
+	if (amdgpu_sriov_vf(adev))
+		mem_info->mem_clk_max = adev->clock.default_mclk / 100;
+	else if (adev->powerplay.pp_funcs)
+		mem_info->mem_clk_max = amdgpu_dpm_get_mclk(adev, false) / 100;
+	else
+		mem_info->mem_clk_max = 100;
+#endif
+}
+
+uint64_t get_gpu_clock_counter(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+
+	if (adev->gfx.funcs->get_gpu_clock_counter)
+		return adev->gfx.funcs->get_gpu_clock_counter(adev);
+	return 0;
+}
+
+uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+
+	/* the sclk is in quantas of 10kHz */
+	if (amdgpu_sriov_vf(adev))
+		return adev->clock.default_sclk / 100;
+	else if (adev->powerplay.pp_funcs)
+		return amdgpu_dpm_get_sclk(adev, false) / 100;
+	else
+		return 100;
+}
+
+void get_cu_info(struct kgd_dev *kgd, struct kfd_cu_info *cu_info)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+	struct amdgpu_cu_info acu_info = adev->gfx.cu_info;
+
+	memset(cu_info, 0, sizeof(*cu_info));
+	if (sizeof(cu_info->cu_bitmap) != sizeof(acu_info.bitmap))
+		return;
+
+	cu_info->cu_active_number = acu_info.number;
+	cu_info->cu_ao_mask = acu_info.ao_cu_mask;
+	memcpy(&cu_info->cu_bitmap[0], &acu_info.bitmap[0],
+	       sizeof(acu_info.bitmap));
+	cu_info->num_shader_engines = adev->gfx.config.max_shader_engines;
+	cu_info->num_shader_arrays_per_engine = adev->gfx.config.max_sh_per_se;
+	cu_info->num_cu_per_sh = adev->gfx.config.max_cu_per_sh;
+	cu_info->simd_per_cu = acu_info.simd_per_cu;
+	cu_info->max_waves_per_simd = acu_info.max_waves_per_simd;
+	cu_info->wave_front_size = acu_info.wave_front_size;
+	cu_info->max_scratch_slots_per_cu = acu_info.max_scratch_slots_per_cu;
+	cu_info->lds_size = acu_info.lds_size;
+}
+
+uint64_t amdgpu_amdkfd_get_vram_usage(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+
+	return amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+}
+
+int amdgpu_amdkfd_submit_ib(struct kgd_dev *kgd, enum kgd_engine_type engine,
+				uint32_t vmid, uint64_t gpu_addr,
+				uint32_t *ib_cmd, uint32_t ib_len)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct amdgpu_ring *ring;
+	struct dma_fence *f = NULL;
+	int ret;
+
+	switch (engine) {
+	case KGD_ENGINE_MEC1:
+		ring = &adev->gfx.compute_ring[0];
+		break;
+	case KGD_ENGINE_SDMA1:
+		ring = &adev->sdma.instance[0].ring;
+		break;
+	case KGD_ENGINE_SDMA2:
+		ring = &adev->sdma.instance[1].ring;
+		break;
+	default:
+		pr_err("Invalid engine in IB submission: %d\n", engine);
+		ret = -EINVAL;
+		goto err;
+	}
+
+	ret = amdgpu_job_alloc(adev, 1, &job, NULL);
+	if (ret)
+		goto err;
+
+	ib = &job->ibs[0];
+	memset(ib, 0, sizeof(struct amdgpu_ib));
+
+	ib->gpu_addr = gpu_addr;
+	ib->ptr = ib_cmd;
+	ib->length_dw = ib_len;
+	/* This works for NO_HWS. TODO: need to handle without knowing VMID */
+	job->vmid = vmid;
+
+	ret = amdgpu_ib_schedule(ring, 1, ib, job, &f);
+	if (ret) {
+		DRM_ERROR("amdgpu: failed to schedule IB.\n");
+		goto err_ib_sched;
+	}
+
+	ret = dma_fence_wait(f, false);
+
+err_ib_sched:
+	dma_fence_put(f);
+	amdgpu_job_free(job);
+err:
+	return ret;
+}
+
+void amdgpu_amdkfd_set_compute_idle(struct kgd_dev *kgd, bool idle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+
+	if (adev->powerplay.pp_funcs &&
+	    adev->powerplay.pp_funcs->switch_power_profile)
+		amdgpu_dpm_switch_power_profile(adev,
+						PP_SMC_POWER_PROFILE_COMPUTE,
+						!idle);
+}
+
+bool amdgpu_amdkfd_is_kfd_vmid(struct amdgpu_device *adev, u32 vmid)
+{
+	if (adev->kfd) {
+		if ((1 << vmid) & compute_vmid_bitmap)
+			return true;
+	}
+
+	return false;
+}
+
+#if !defined(CONFIG_HSA_AMD_MODULE) && !defined(CONFIG_HSA_AMD)
+bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm)
+{
+	return false;
+}
+
+void amdgpu_amdkfd_unreserve_system_memory_limit(struct amdgpu_bo *bo)
+{
+}
+
+void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev,
+					struct amdgpu_vm *vm)
+{
+}
+
+struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f)
+{
+	return NULL;
+}
+
+int amdgpu_amdkfd_evict_userptr(struct kgd_mem *mem, struct mm_struct *mm)
+{
+	return 0;
+}
+
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void)
+{
+	return NULL;
+}
+
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void)
+{
+	return NULL;
+}
+
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_9_0_get_functions(void)
+{
+	return NULL;
+}
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd.h
new file mode 100644
index 00000000000..caa4d6d95c6
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -0,0 +1,199 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/* amdgpu_amdkfd.h defines the private interface between amdgpu and amdkfd. */
+
+#ifndef AMDGPU_AMDKFD_H_INCLUDED
+#define AMDGPU_AMDKFD_H_INCLUDED
+
+#include <linux/types.h>
+#include <linux/mm.h>
+#include <linux/mmu_context.h>
+#include <linux/workqueue.h>
+#include <kgd_kfd_interface.h>
+#include <drm/ttm/ttm_execbuf_util.h>
+#include "amdgpu_sync.h"
+#include "amdgpu_vm.h"
+
+extern const struct kgd2kfd_calls *kgd2kfd;
+
+struct amdgpu_device;
+
+struct kfd_bo_va_list {
+	struct list_head bo_list;
+	struct amdgpu_bo_va *bo_va;
+	void *kgd_dev;
+	bool is_mapped;
+	uint64_t va;
+	uint64_t pte_flags;
+};
+
+struct kgd_mem {
+	struct rwlock lock;
+	struct amdgpu_bo *bo;
+	struct list_head bo_va_list;
+	/* protected by amdkfd_process_info.lock */
+	struct ttm_validate_buffer validate_list;
+	struct ttm_validate_buffer resv_list;
+	uint32_t domain;
+	unsigned int mapped_to_gpu_memory;
+	uint64_t va;
+
+	uint32_t mapping_flags;
+
+	atomic_t invalid;
+	struct amdkfd_process_info *process_info;
+	struct vm_page **user_pages;
+
+	struct amdgpu_sync sync;
+
+	bool aql_queue;
+};
+
+/* KFD Memory Eviction */
+struct amdgpu_amdkfd_fence {
+	struct dma_fence base;
+	struct mm_struct *mm;
+	spinlock_t lock;
+	char timeline_name[TASK_COMM_LEN];
+};
+
+struct amdgpu_amdkfd_fence *amdgpu_amdkfd_fence_create(u64 context,
+						       struct mm_struct *mm);
+bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm);
+struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f);
+
+struct amdkfd_process_info {
+	/* List head of all VMs that belong to a KFD process */
+	struct list_head vm_list_head;
+	/* List head for all KFD BOs that belong to a KFD process. */
+	struct list_head kfd_bo_list;
+	/* List of userptr BOs that are valid or invalid */
+	struct list_head userptr_valid_list;
+	struct list_head userptr_inval_list;
+	/* Lock to protect kfd_bo_list */
+	struct rwlock lock;
+
+	/* Number of VMs */
+	unsigned int n_vms;
+	/* Eviction Fence */
+	struct amdgpu_amdkfd_fence *eviction_fence;
+
+	/* MMU-notifier related fields */
+	atomic_t evicted_bos;
+	struct delayed_work restore_userptr_work;
+	struct pid *pid;
+};
+
+int amdgpu_amdkfd_init(void);
+void amdgpu_amdkfd_fini(void);
+
+void amdgpu_amdkfd_suspend(struct amdgpu_device *adev);
+int amdgpu_amdkfd_resume(struct amdgpu_device *adev);
+void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev,
+			const void *ih_ring_entry);
+void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev);
+void amdgpu_amdkfd_device_init(struct amdgpu_device *adev);
+void amdgpu_amdkfd_device_fini(struct amdgpu_device *adev);
+
+int amdgpu_amdkfd_evict_userptr(struct kgd_mem *mem, struct mm_struct *mm);
+int amdgpu_amdkfd_submit_ib(struct kgd_dev *kgd, enum kgd_engine_type engine,
+				uint32_t vmid, uint64_t gpu_addr,
+				uint32_t *ib_cmd, uint32_t ib_len);
+void amdgpu_amdkfd_set_compute_idle(struct kgd_dev *kgd, bool idle);
+
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void);
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void);
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_9_0_get_functions(void);
+
+bool amdgpu_amdkfd_is_kfd_vmid(struct amdgpu_device *adev, u32 vmid);
+
+int amdgpu_amdkfd_pre_reset(struct amdgpu_device *adev);
+
+int amdgpu_amdkfd_post_reset(struct amdgpu_device *adev);
+
+void amdgpu_amdkfd_gpu_reset(struct kgd_dev *kgd);
+
+/* Shared API */
+int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
+			void **mem_obj, uint64_t *gpu_addr,
+			void **cpu_ptr, bool mqd_gfx9);
+void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj);
+void get_local_mem_info(struct kgd_dev *kgd,
+			struct kfd_local_mem_info *mem_info);
+uint64_t get_gpu_clock_counter(struct kgd_dev *kgd);
+
+uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd);
+void get_cu_info(struct kgd_dev *kgd, struct kfd_cu_info *cu_info);
+uint64_t amdgpu_amdkfd_get_vram_usage(struct kgd_dev *kgd);
+
+#define read_user_wptr(mmptr, wptr, dst)				\
+	({								\
+		bool valid = false;					\
+		if ((mmptr) && (wptr)) {				\
+			if ((mmptr) == current->mm) {			\
+				valid = !get_user((dst), (wptr));	\
+			} else if (current->mm == NULL) {		\
+				use_mm(mmptr);				\
+				valid = !get_user((dst), (wptr));	\
+				unuse_mm(mmptr);			\
+			}						\
+		}							\
+		valid;							\
+	})
+
+/* GPUVM API */
+int amdgpu_amdkfd_gpuvm_create_process_vm(struct kgd_dev *kgd, void **vm,
+					void **process_info,
+					struct dma_fence **ef);
+int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct kgd_dev *kgd,
+					struct file *filp,
+					void **vm, void **process_info,
+					struct dma_fence **ef);
+void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm);
+void amdgpu_amdkfd_gpuvm_destroy_process_vm(struct kgd_dev *kgd, void *vm);
+uint32_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *vm);
+int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
+		struct kgd_dev *kgd, uint64_t va, uint64_t size,
+		void *vm, struct kgd_mem **mem,
+		uint64_t *offset, uint32_t flags);
+int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
+		struct kgd_dev *kgd, struct kgd_mem *mem);
+int amdgpu_amdkfd_gpuvm_map_memory_to_gpu(
+		struct kgd_dev *kgd, struct kgd_mem *mem, void *vm);
+int amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(
+		struct kgd_dev *kgd, struct kgd_mem *mem, void *vm);
+int amdgpu_amdkfd_gpuvm_sync_memory(
+		struct kgd_dev *kgd, struct kgd_mem *mem, bool intr);
+int amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(struct kgd_dev *kgd,
+		struct kgd_mem *mem, void **kptr, uint64_t *size);
+int amdgpu_amdkfd_gpuvm_restore_process_bos(void *process_info,
+					    struct dma_fence **ef);
+
+int amdgpu_amdkfd_gpuvm_get_vm_fault_info(struct kgd_dev *kgd,
+					      struct kfd_vm_fault_info *info);
+
+void amdgpu_amdkfd_gpuvm_init_mem_limits(void);
+void amdgpu_amdkfd_unreserve_system_memory_limit(struct amdgpu_bo *bo);
+
+#endif /* AMDGPU_AMDKFD_H_INCLUDED */
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_fence.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_fence.c
new file mode 100644
index 00000000000..b16935a5392
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_fence.c
@@ -0,0 +1,177 @@
+/*
+ * Copyright 2016-2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/dma-fence.h>
+#include <linux/spinlock.h>
+#include <linux/atomic.h>
+#include <linux/stacktrace.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/sched/mm.h>
+#include "amdgpu_amdkfd.h"
+
+static const struct dma_fence_ops amdkfd_fence_ops;
+static atomic_t fence_seq = ATOMIC_INIT(0);
+
+/* Eviction Fence
+ * Fence helper functions to deal with KFD memory eviction.
+ * Big Idea - Since KFD submissions are done by user queues, a BO cannot be
+ *  evicted unless all the user queues for that process are evicted.
+ *
+ * All the BOs in a process share an eviction fence. When process X wants
+ * to map VRAM memory but TTM can't find enough space, TTM will attempt to
+ * evict BOs from its LRU list. TTM checks if the BO is valuable to evict
+ * by calling ttm_bo_driver->eviction_valuable().
+ *
+ * ttm_bo_driver->eviction_valuable() - will return false if the BO belongs
+ *  to process X. Otherwise, it will return true to indicate BO can be
+ *  evicted by TTM.
+ *
+ * If ttm_bo_driver->eviction_valuable returns true, then TTM will continue
+ * the evcition process for that BO by calling ttm_bo_evict --> amdgpu_bo_move
+ * --> amdgpu_copy_buffer(). This sets up job in GPU scheduler.
+ *
+ * GPU Scheduler (amd_sched_main) - sets up a cb (fence_add_callback) to
+ *  nofity when the BO is free to move. fence_add_callback --> enable_signaling
+ *  --> amdgpu_amdkfd_fence.enable_signaling
+ *
+ * amdgpu_amdkfd_fence.enable_signaling - Start a work item that will quiesce
+ * user queues and signal fence. The work item will also start another delayed
+ * work item to restore BOs
+ */
+
+struct amdgpu_amdkfd_fence *amdgpu_amdkfd_fence_create(u64 context,
+						       struct mm_struct *mm)
+{
+	struct amdgpu_amdkfd_fence *fence;
+
+	fence = kzalloc(sizeof(*fence), GFP_KERNEL);
+	if (fence == NULL)
+		return NULL;
+
+	/* This reference gets released in amdkfd_fence_release */
+	mmgrab(mm);
+	fence->mm = mm;
+	get_task_comm(fence->timeline_name, current);
+	mtx_init(&fence->lock, IPL_TTY);
+
+	dma_fence_init(&fence->base, &amdkfd_fence_ops, &fence->lock,
+		   context, atomic_inc_return(&fence_seq));
+
+	return fence;
+}
+
+struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f)
+{
+	struct amdgpu_amdkfd_fence *fence;
+
+	if (!f)
+		return NULL;
+
+	fence = container_of(f, struct amdgpu_amdkfd_fence, base);
+	if (fence && f->ops == &amdkfd_fence_ops)
+		return fence;
+
+	return NULL;
+}
+
+static const char *amdkfd_fence_get_driver_name(struct dma_fence *f)
+{
+	return "amdgpu_amdkfd_fence";
+}
+
+static const char *amdkfd_fence_get_timeline_name(struct dma_fence *f)
+{
+	struct amdgpu_amdkfd_fence *fence = to_amdgpu_amdkfd_fence(f);
+
+	return fence->timeline_name;
+}
+
+/**
+ * amdkfd_fence_enable_signaling - This gets called when TTM wants to evict
+ *  a KFD BO and schedules a job to move the BO.
+ *  If fence is already signaled return true.
+ *  If fence is not signaled schedule a evict KFD process work item.
+ */
+static bool amdkfd_fence_enable_signaling(struct dma_fence *f)
+{
+	struct amdgpu_amdkfd_fence *fence = to_amdgpu_amdkfd_fence(f);
+
+	if (!fence)
+		return false;
+
+	if (dma_fence_is_signaled(f))
+		return true;
+
+	if (!kgd2kfd->schedule_evict_and_restore_process(fence->mm, f))
+		return true;
+
+	return false;
+}
+
+/**
+ * amdkfd_fence_release - callback that fence can be freed
+ *
+ * @fence: fence
+ *
+ * This function is called when the reference count becomes zero.
+ * Drops the mm_struct reference and RCU schedules freeing up the fence.
+ */
+static void amdkfd_fence_release(struct dma_fence *f)
+{
+	struct amdgpu_amdkfd_fence *fence = to_amdgpu_amdkfd_fence(f);
+
+	/* Unconditionally signal the fence. The process is getting
+	 * terminated.
+	 */
+	if (WARN_ON(!fence))
+		return; /* Not an amdgpu_amdkfd_fence */
+
+	mmdrop(fence->mm);
+	kfree_rcu(f, rcu);
+}
+
+/**
+ * amdkfd_fence_check_mm - Check if @mm is same as that of the fence @f
+ *  if same return TRUE else return FALSE.
+ *
+ * @f: [IN] fence
+ * @mm: [IN] mm that needs to be verified
+ */
+bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm)
+{
+	struct amdgpu_amdkfd_fence *fence = to_amdgpu_amdkfd_fence(f);
+
+	if (!fence)
+		return false;
+	else if (fence->mm == mm)
+		return true;
+
+	return false;
+}
+
+static const struct dma_fence_ops amdkfd_fence_ops = {
+	.get_driver_name = amdkfd_fence_get_driver_name,
+	.get_timeline_name = amdkfd_fence_get_timeline_name,
+	.enable_signaling = amdkfd_fence_enable_signaling,
+	.release = amdkfd_fence_release,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
new file mode 100644
index 00000000000..9803b91f3e7
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
@@ -0,0 +1,940 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/fdtable.h>
+#include <linux/uaccess.h>
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "cikd.h"
+#include "cik_sdma.h"
+#include "amdgpu_ucode.h"
+#include "gfx_v7_0.h"
+#include "gca/gfx_7_2_d.h"
+#include "gca/gfx_7_2_enum.h"
+#include "gca/gfx_7_2_sh_mask.h"
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+#include "cik_structs.h"
+
+enum hqd_dequeue_request_type {
+	NO_ACTION = 0,
+	DRAIN_PIPE,
+	RESET_WAVES
+};
+
+enum {
+	MAX_TRAPID = 8,		/* 3 bits in the bitfield. */
+	MAX_WATCH_ADDRESSES = 4
+};
+
+enum {
+	ADDRESS_WATCH_REG_ADDR_HI = 0,
+	ADDRESS_WATCH_REG_ADDR_LO,
+	ADDRESS_WATCH_REG_CNTL,
+	ADDRESS_WATCH_REG_MAX
+};
+
+/*  not defined in the CI/KV reg file  */
+enum {
+	ADDRESS_WATCH_REG_CNTL_ATC_BIT = 0x10000000UL,
+	ADDRESS_WATCH_REG_CNTL_DEFAULT_MASK = 0x00FFFFFF,
+	ADDRESS_WATCH_REG_ADDLOW_MASK_EXTENSION = 0x03000000,
+	/* extend the mask to 26 bits to match the low address field */
+	ADDRESS_WATCH_REG_ADDLOW_SHIFT = 6,
+	ADDRESS_WATCH_REG_ADDHIGH_MASK = 0xFFFF
+};
+
+static const uint32_t watchRegs[MAX_WATCH_ADDRESSES * ADDRESS_WATCH_REG_MAX] = {
+	mmTCP_WATCH0_ADDR_H, mmTCP_WATCH0_ADDR_L, mmTCP_WATCH0_CNTL,
+	mmTCP_WATCH1_ADDR_H, mmTCP_WATCH1_ADDR_L, mmTCP_WATCH1_CNTL,
+	mmTCP_WATCH2_ADDR_H, mmTCP_WATCH2_ADDR_L, mmTCP_WATCH2_CNTL,
+	mmTCP_WATCH3_ADDR_H, mmTCP_WATCH3_ADDR_L, mmTCP_WATCH3_CNTL
+};
+
+union TCP_WATCH_CNTL_BITS {
+	struct {
+		uint32_t mask:24;
+		uint32_t vmid:4;
+		uint32_t atc:1;
+		uint32_t mode:2;
+		uint32_t valid:1;
+	} bitfields, bits;
+	uint32_t u32All;
+	signed int i32All;
+	float f32All;
+};
+
+/*
+ * Register access functions
+ */
+
+static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t sh_mem_config,	uint32_t sh_mem_ape1_base,
+		uint32_t sh_mem_ape1_limit, uint32_t sh_mem_bases);
+
+static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
+					unsigned int vmid);
+
+static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id);
+static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
+			uint32_t queue_id, uint32_t __user *wptr,
+			uint32_t wptr_shift, uint32_t wptr_mask,
+			struct mm_struct *mm);
+static int kgd_hqd_dump(struct kgd_dev *kgd,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs);
+static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm);
+static int kgd_hqd_sdma_dump(struct kgd_dev *kgd,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs);
+static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
+				uint32_t pipe_id, uint32_t queue_id);
+
+static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
+				enum kfd_preempt_type reset_type,
+				unsigned int utimeout, uint32_t pipe_id,
+				uint32_t queue_id);
+static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd);
+static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
+				unsigned int utimeout);
+static int kgd_address_watch_disable(struct kgd_dev *kgd);
+static int kgd_address_watch_execute(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					uint32_t cntl_val,
+					uint32_t addr_hi,
+					uint32_t addr_lo);
+static int kgd_wave_control_execute(struct kgd_dev *kgd,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd);
+static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					unsigned int reg_offset);
+
+static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd, uint8_t vmid);
+static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
+							uint8_t vmid);
+
+static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
+static void set_scratch_backing_va(struct kgd_dev *kgd,
+					uint64_t va, uint32_t vmid);
+static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t page_table_base);
+static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid);
+static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid);
+static uint32_t read_vmid_from_vmfault_reg(struct kgd_dev *kgd);
+
+/* Because of REG_GET_FIELD() being used, we put this function in the
+ * asic specific file.
+ */
+static int get_tile_config(struct kgd_dev *kgd,
+		struct tile_config *config)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+
+	config->gb_addr_config = adev->gfx.config.gb_addr_config;
+	config->num_banks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
+				MC_ARB_RAMCFG, NOOFBANK);
+	config->num_ranks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
+				MC_ARB_RAMCFG, NOOFRANKS);
+
+	config->tile_config_ptr = adev->gfx.config.tile_mode_array;
+	config->num_tile_configs =
+			ARRAY_SIZE(adev->gfx.config.tile_mode_array);
+	config->macro_tile_config_ptr =
+			adev->gfx.config.macrotile_mode_array;
+	config->num_macro_tile_configs =
+			ARRAY_SIZE(adev->gfx.config.macrotile_mode_array);
+
+	return 0;
+}
+
+static const struct kfd2kgd_calls kfd2kgd = {
+	.init_gtt_mem_allocation = alloc_gtt_mem,
+	.free_gtt_mem = free_gtt_mem,
+	.get_local_mem_info = get_local_mem_info,
+	.get_gpu_clock_counter = get_gpu_clock_counter,
+	.get_max_engine_clock_in_mhz = get_max_engine_clock_in_mhz,
+	.alloc_pasid = amdgpu_pasid_alloc,
+	.free_pasid = amdgpu_pasid_free,
+	.program_sh_mem_settings = kgd_program_sh_mem_settings,
+	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
+	.init_interrupts = kgd_init_interrupts,
+	.hqd_load = kgd_hqd_load,
+	.hqd_sdma_load = kgd_hqd_sdma_load,
+	.hqd_dump = kgd_hqd_dump,
+	.hqd_sdma_dump = kgd_hqd_sdma_dump,
+	.hqd_is_occupied = kgd_hqd_is_occupied,
+	.hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied,
+	.hqd_destroy = kgd_hqd_destroy,
+	.hqd_sdma_destroy = kgd_hqd_sdma_destroy,
+	.address_watch_disable = kgd_address_watch_disable,
+	.address_watch_execute = kgd_address_watch_execute,
+	.wave_control_execute = kgd_wave_control_execute,
+	.address_watch_get_offset = kgd_address_watch_get_offset,
+	.get_atc_vmid_pasid_mapping_pasid = get_atc_vmid_pasid_mapping_pasid,
+	.get_atc_vmid_pasid_mapping_valid = get_atc_vmid_pasid_mapping_valid,
+	.get_fw_version = get_fw_version,
+	.set_scratch_backing_va = set_scratch_backing_va,
+	.get_tile_config = get_tile_config,
+	.get_cu_info = get_cu_info,
+	.get_vram_usage = amdgpu_amdkfd_get_vram_usage,
+	.create_process_vm = amdgpu_amdkfd_gpuvm_create_process_vm,
+	.acquire_process_vm = amdgpu_amdkfd_gpuvm_acquire_process_vm,
+	.destroy_process_vm = amdgpu_amdkfd_gpuvm_destroy_process_vm,
+	.get_process_page_dir = amdgpu_amdkfd_gpuvm_get_process_page_dir,
+	.set_vm_context_page_table_base = set_vm_context_page_table_base,
+	.alloc_memory_of_gpu = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu,
+	.free_memory_of_gpu = amdgpu_amdkfd_gpuvm_free_memory_of_gpu,
+	.map_memory_to_gpu = amdgpu_amdkfd_gpuvm_map_memory_to_gpu,
+	.unmap_memory_to_gpu = amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu,
+	.sync_memory = amdgpu_amdkfd_gpuvm_sync_memory,
+	.map_gtt_bo_to_kernel = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel,
+	.restore_process_bos = amdgpu_amdkfd_gpuvm_restore_process_bos,
+	.invalidate_tlbs = invalidate_tlbs,
+	.invalidate_tlbs_vmid = invalidate_tlbs_vmid,
+	.submit_ib = amdgpu_amdkfd_submit_ib,
+	.get_vm_fault_info = amdgpu_amdkfd_gpuvm_get_vm_fault_info,
+	.read_vmid_from_vmfault_reg = read_vmid_from_vmfault_reg,
+	.gpu_recover = amdgpu_amdkfd_gpu_reset,
+	.set_compute_idle = amdgpu_amdkfd_set_compute_idle
+};
+
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void)
+{
+	return (struct kfd2kgd_calls *)&kfd2kgd;
+}
+
+static inline struct amdgpu_device *get_amdgpu_device(struct kgd_dev *kgd)
+{
+	return (struct amdgpu_device *)kgd;
+}
+
+static void lock_srbm(struct kgd_dev *kgd, uint32_t mec, uint32_t pipe,
+			uint32_t queue, uint32_t vmid)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t value = PIPEID(pipe) | MEID(mec) | VMID(vmid) | QUEUEID(queue);
+
+	mutex_lock(&adev->srbm_mutex);
+	WREG32(mmSRBM_GFX_CNTL, value);
+}
+
+static void unlock_srbm(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	WREG32(mmSRBM_GFX_CNTL, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void acquire_queue(struct kgd_dev *kgd, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(kgd, mec, pipe, queue_id, 0);
+}
+
+static void release_queue(struct kgd_dev *kgd)
+{
+	unlock_srbm(kgd);
+}
+
+static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
+					uint32_t sh_mem_config,
+					uint32_t sh_mem_ape1_base,
+					uint32_t sh_mem_ape1_limit,
+					uint32_t sh_mem_bases)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	lock_srbm(kgd, 0, 0, 0, vmid);
+
+	WREG32(mmSH_MEM_CONFIG, sh_mem_config);
+	WREG32(mmSH_MEM_APE1_BASE, sh_mem_ape1_base);
+	WREG32(mmSH_MEM_APE1_LIMIT, sh_mem_ape1_limit);
+	WREG32(mmSH_MEM_BASES, sh_mem_bases);
+
+	unlock_srbm(kgd);
+}
+
+static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
+					unsigned int vmid)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	/*
+	 * We have to assume that there is no outstanding mapping.
+	 * The ATC_VMID_PASID_MAPPING_UPDATE_STATUS bit could be 0 because
+	 * a mapping is in progress or because a mapping finished and the
+	 * SW cleared it. So the protocol is to always wait & clear.
+	 */
+	uint32_t pasid_mapping = (pasid == 0) ? 0 : (uint32_t)pasid |
+			ATC_VMID0_PASID_MAPPING__VALID_MASK;
+
+	WREG32(mmATC_VMID0_PASID_MAPPING + vmid, pasid_mapping);
+
+	while (!(RREG32(mmATC_VMID_PASID_MAPPING_UPDATE_STATUS) & (1U << vmid)))
+		cpu_relax();
+	WREG32(mmATC_VMID_PASID_MAPPING_UPDATE_STATUS, 1U << vmid);
+
+	/* Mapping vmid to pasid also for IH block */
+	WREG32(mmIH_VMID_0_LUT + vmid, pasid_mapping);
+
+	return 0;
+}
+
+static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t mec;
+	uint32_t pipe;
+
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(kgd, mec, pipe, 0, 0);
+
+	WREG32(mmCPC_INT_CNTL, CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
+			CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
+
+	unlock_srbm(kgd);
+
+	return 0;
+}
+
+static inline uint32_t get_sdma_base_addr(struct cik_sdma_rlc_registers *m)
+{
+	uint32_t retval;
+
+	retval = m->sdma_engine_id * SDMA1_REGISTER_OFFSET +
+			m->sdma_queue_id * KFD_CIK_SDMA_QUEUE_OFFSET;
+
+	pr_debug("kfd: sdma base address: 0x%x\n", retval);
+
+	return retval;
+}
+
+static inline struct cik_mqd *get_mqd(void *mqd)
+{
+	return (struct cik_mqd *)mqd;
+}
+
+static inline struct cik_sdma_rlc_registers *get_sdma_mqd(void *mqd)
+{
+	return (struct cik_sdma_rlc_registers *)mqd;
+}
+
+static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
+			uint32_t queue_id, uint32_t __user *wptr,
+			uint32_t wptr_shift, uint32_t wptr_mask,
+			struct mm_struct *mm)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct cik_mqd *m;
+	uint32_t *mqd_hqd;
+	uint32_t reg, wptr_val, data;
+	bool valid_wptr = false;
+
+	m = get_mqd(mqd);
+
+	acquire_queue(kgd, pipe_id, queue_id);
+
+	/* HQD registers extend from CP_MQD_BASE_ADDR to CP_MQD_CONTROL. */
+	mqd_hqd = &m->cp_mqd_base_addr_lo;
+
+	for (reg = mmCP_MQD_BASE_ADDR; reg <= mmCP_MQD_CONTROL; reg++)
+		WREG32(reg, mqd_hqd[reg - mmCP_MQD_BASE_ADDR]);
+
+	/* Copy userspace write pointer value to register.
+	 * Activate doorbell logic to monitor subsequent changes.
+	 */
+	data = REG_SET_FIELD(m->cp_hqd_pq_doorbell_control,
+			     CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
+	WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, data);
+
+	/* read_user_ptr may take the mm->mmap_sem.
+	 * release srbm_mutex to avoid circular dependency between
+	 * srbm_mutex->mm_sem->reservation_ww_class_mutex->srbm_mutex.
+	 */
+	release_queue(kgd);
+	valid_wptr = read_user_wptr(mm, wptr, wptr_val);
+	acquire_queue(kgd, pipe_id, queue_id);
+	if (valid_wptr)
+		WREG32(mmCP_HQD_PQ_WPTR, (wptr_val << wptr_shift) & wptr_mask);
+
+	data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
+	WREG32(mmCP_HQD_ACTIVE, data);
+
+	release_queue(kgd);
+
+	return 0;
+}
+
+static int kgd_hqd_dump(struct kgd_dev *kgd,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t i = 0, reg;
+#define HQD_N_REGS (35+4)
+#define DUMP_REG(addr) do {				\
+		if (WARN_ON_ONCE(i >= HQD_N_REGS))	\
+			break;				\
+		(*dump)[i][0] = (addr) << 2;		\
+		(*dump)[i++][1] = RREG32(addr);		\
+	} while (0)
+
+	*dump = kmalloc_array(HQD_N_REGS * 2, sizeof(uint32_t), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE0);
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE1);
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE2);
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE3);
+
+	for (reg = mmCP_MQD_BASE_ADDR; reg <= mmCP_MQD_CONTROL; reg++)
+		DUMP_REG(reg);
+
+	release_queue(kgd);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct cik_sdma_rlc_registers *m;
+	unsigned long end_jiffies;
+	uint32_t sdma_base_addr;
+	uint32_t data;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(m);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL,
+		m->sdma_rlc_rb_cntl & (~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK));
+
+	end_jiffies = msecs_to_jiffies(2000) + jiffies;
+	while (true) {
+		data = RREG32(sdma_base_addr + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (data & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies))
+			return -ETIME;
+		usleep_range(500, 1000);
+	}
+	if (m->sdma_engine_id) {
+		data = RREG32(mmSDMA1_GFX_CONTEXT_CNTL);
+		data = REG_SET_FIELD(data, SDMA1_GFX_CONTEXT_CNTL,
+				RESUME_CTX, 0);
+		WREG32(mmSDMA1_GFX_CONTEXT_CNTL, data);
+	} else {
+		data = RREG32(mmSDMA0_GFX_CONTEXT_CNTL);
+		data = REG_SET_FIELD(data, SDMA0_GFX_CONTEXT_CNTL,
+				RESUME_CTX, 0);
+		WREG32(mmSDMA0_GFX_CONTEXT_CNTL, data);
+	}
+
+	data = REG_SET_FIELD(m->sdma_rlc_doorbell, SDMA0_RLC0_DOORBELL,
+			     ENABLE, 1);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL, data);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR, m->sdma_rlc_rb_rptr);
+
+	if (read_user_wptr(mm, wptr, data))
+		WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR, data);
+	else
+		WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR,
+		       m->sdma_rlc_rb_rptr);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_VIRTUAL_ADDR,
+				m->sdma_rlc_virtual_addr);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE, m->sdma_rlc_rb_base);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE_HI,
+			m->sdma_rlc_rb_base_hi);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_ADDR_LO,
+			m->sdma_rlc_rb_rptr_addr_lo);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_ADDR_HI,
+			m->sdma_rlc_rb_rptr_addr_hi);
+
+	data = REG_SET_FIELD(m->sdma_rlc_rb_cntl, SDMA0_RLC0_RB_CNTL,
+			     RB_ENABLE, 1);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL, data);
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_dump(struct kgd_dev *kgd,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t sdma_offset = engine_id * SDMA1_REGISTER_OFFSET +
+		queue_id * KFD_CIK_SDMA_QUEUE_OFFSET;
+	uint32_t i = 0, reg;
+#undef HQD_N_REGS
+#define HQD_N_REGS (19+4)
+
+	*dump = kmalloc_array(HQD_N_REGS * 2, sizeof(uint32_t), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	for (reg = mmSDMA0_RLC0_RB_CNTL; reg <= mmSDMA0_RLC0_DOORBELL; reg++)
+		DUMP_REG(sdma_offset + reg);
+	for (reg = mmSDMA0_RLC0_VIRTUAL_ADDR; reg <= mmSDMA0_RLC0_WATERMARK;
+	     reg++)
+		DUMP_REG(sdma_offset + reg);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
+				uint32_t pipe_id, uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t act;
+	bool retval = false;
+	uint32_t low, high;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+	act = RREG32(mmCP_HQD_ACTIVE);
+	if (act) {
+		low = lower_32_bits(queue_address >> 8);
+		high = upper_32_bits(queue_address >> 8);
+
+		if (low == RREG32(mmCP_HQD_PQ_BASE) &&
+				high == RREG32(mmCP_HQD_PQ_BASE_HI))
+			retval = true;
+	}
+	release_queue(kgd);
+	return retval;
+}
+
+static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct cik_sdma_rlc_registers *m;
+	uint32_t sdma_base_addr;
+	uint32_t sdma_rlc_rb_cntl;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(m);
+
+	sdma_rlc_rb_cntl = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+
+	if (sdma_rlc_rb_cntl & SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK)
+		return true;
+
+	return false;
+}
+
+static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
+				enum kfd_preempt_type reset_type,
+				unsigned int utimeout, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t temp;
+	enum hqd_dequeue_request_type type;
+	unsigned long flags, end_jiffies;
+	int retry;
+
+	if (adev->in_gpu_reset)
+		return -EIO;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+	WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, 0);
+
+	switch (reset_type) {
+	case KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN:
+		type = DRAIN_PIPE;
+		break;
+	case KFD_PREEMPT_TYPE_WAVEFRONT_RESET:
+		type = RESET_WAVES;
+		break;
+	default:
+		type = DRAIN_PIPE;
+		break;
+	}
+
+	/* Workaround: If IQ timer is active and the wait time is close to or
+	 * equal to 0, dequeueing is not safe. Wait until either the wait time
+	 * is larger or timer is cleared. Also, ensure that IQ_REQ_PEND is
+	 * cleared before continuing. Also, ensure wait times are set to at
+	 * least 0x3.
+	 */
+	local_irq_save(flags);
+	preempt_disable();
+	retry = 5000; /* wait for 500 usecs at maximum */
+	while (true) {
+		temp = RREG32(mmCP_HQD_IQ_TIMER);
+		if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, PROCESSING_IQ)) {
+			pr_debug("HW is processing IQ\n");
+			goto loop;
+		}
+		if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, ACTIVE)) {
+			if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, RETRY_TYPE)
+					== 3) /* SEM-rearm is safe */
+				break;
+			/* Wait time 3 is safe for CP, but our MMIO read/write
+			 * time is close to 1 microsecond, so check for 10 to
+			 * leave more buffer room
+			 */
+			if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, WAIT_TIME)
+					>= 10)
+				break;
+			pr_debug("IQ timer is active\n");
+		} else
+			break;
+loop:
+		if (!retry) {
+			pr_err("CP HQD IQ timer status time out\n");
+			break;
+		}
+		ndelay(100);
+		--retry;
+	}
+	retry = 1000;
+	while (true) {
+		temp = RREG32(mmCP_HQD_DEQUEUE_REQUEST);
+		if (!(temp & CP_HQD_DEQUEUE_REQUEST__IQ_REQ_PEND_MASK))
+			break;
+		pr_debug("Dequeue request is pending\n");
+
+		if (!retry) {
+			pr_err("CP HQD dequeue request time out\n");
+			break;
+		}
+		ndelay(100);
+		--retry;
+	}
+	local_irq_restore(flags);
+	preempt_enable();
+
+	WREG32(mmCP_HQD_DEQUEUE_REQUEST, type);
+
+	end_jiffies = (utimeout * HZ / 1000) + jiffies;
+	while (true) {
+		temp = RREG32(mmCP_HQD_ACTIVE);
+		if (!(temp & CP_HQD_ACTIVE__ACTIVE_MASK))
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("cp queue preemption time out\n");
+			release_queue(kgd);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	release_queue(kgd);
+	return 0;
+}
+
+static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
+				unsigned int utimeout)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct cik_sdma_rlc_registers *m;
+	uint32_t sdma_base_addr;
+	uint32_t temp;
+	unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(m);
+
+	temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+	temp = temp & ~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL, temp);
+
+	while (true) {
+		temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (temp & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies))
+			return -ETIME;
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL, 0);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL,
+		RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL) |
+		SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK);
+
+	m->sdma_rlc_rb_rptr = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR);
+
+	return 0;
+}
+
+static int kgd_address_watch_disable(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	union TCP_WATCH_CNTL_BITS cntl;
+	unsigned int i;
+
+	cntl.u32All = 0;
+
+	cntl.bitfields.valid = 0;
+	cntl.bitfields.mask = ADDRESS_WATCH_REG_CNTL_DEFAULT_MASK;
+	cntl.bitfields.atc = 1;
+
+	/* Turning off this address until we set all the registers */
+	for (i = 0; i < MAX_WATCH_ADDRESSES; i++)
+		WREG32(watchRegs[i * ADDRESS_WATCH_REG_MAX +
+			ADDRESS_WATCH_REG_CNTL], cntl.u32All);
+
+	return 0;
+}
+
+static int kgd_address_watch_execute(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					uint32_t cntl_val,
+					uint32_t addr_hi,
+					uint32_t addr_lo)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	union TCP_WATCH_CNTL_BITS cntl;
+
+	cntl.u32All = cntl_val;
+
+	/* Turning off this watch point until we set all the registers */
+	cntl.bitfields.valid = 0;
+	WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX +
+		ADDRESS_WATCH_REG_CNTL], cntl.u32All);
+
+	WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX +
+		ADDRESS_WATCH_REG_ADDR_HI], addr_hi);
+
+	WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX +
+		ADDRESS_WATCH_REG_ADDR_LO], addr_lo);
+
+	/* Enable the watch point */
+	cntl.bitfields.valid = 1;
+
+	WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX +
+		ADDRESS_WATCH_REG_CNTL], cntl.u32All);
+
+	return 0;
+}
+
+static int kgd_wave_control_execute(struct kgd_dev *kgd,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t data;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	WREG32(mmGRBM_GFX_INDEX, gfx_index_val);
+	WREG32(mmSQ_CMD, sq_cmd);
+
+	/*  Restore the GRBM_GFX_INDEX register  */
+
+	data = GRBM_GFX_INDEX__INSTANCE_BROADCAST_WRITES_MASK |
+		GRBM_GFX_INDEX__SH_BROADCAST_WRITES_MASK |
+		GRBM_GFX_INDEX__SE_BROADCAST_WRITES_MASK;
+
+	WREG32(mmGRBM_GFX_INDEX, data);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					unsigned int reg_offset)
+{
+	return watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX + reg_offset];
+}
+
+static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
+							uint8_t vmid)
+{
+	uint32_t reg;
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
+}
+
+static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
+								uint8_t vmid)
+{
+	uint32_t reg;
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+	return reg & ATC_VMID0_PASID_MAPPING__PASID_MASK;
+}
+
+static void set_scratch_backing_va(struct kgd_dev *kgd,
+					uint64_t va, uint32_t vmid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	lock_srbm(kgd, 0, 0, 0, vmid);
+	WREG32(mmSH_HIDDEN_PRIVATE_BASE_VMID, va);
+	unlock_srbm(kgd);
+}
+
+static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	const union amdgpu_firmware_header *hdr;
+
+	switch (type) {
+	case KGD_ENGINE_PFP:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->gfx.pfp_fw->data;
+		break;
+
+	case KGD_ENGINE_ME:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->gfx.me_fw->data;
+		break;
+
+	case KGD_ENGINE_CE:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->gfx.ce_fw->data;
+		break;
+
+	case KGD_ENGINE_MEC1:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->gfx.mec_fw->data;
+		break;
+
+	case KGD_ENGINE_MEC2:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->gfx.mec2_fw->data;
+		break;
+
+	case KGD_ENGINE_RLC:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->gfx.rlc_fw->data;
+		break;
+
+	case KGD_ENGINE_SDMA1:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->sdma.instance[0].fw->data;
+		break;
+
+	case KGD_ENGINE_SDMA2:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->sdma.instance[1].fw->data;
+		break;
+
+	default:
+		return 0;
+	}
+
+	if (hdr == NULL)
+		return 0;
+
+	/* Only 12 bit in use*/
+	return hdr->common.ucode_version;
+}
+
+static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
+			uint32_t page_table_base)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("trying to set page table base for wrong VMID\n");
+		return;
+	}
+	WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8, page_table_base);
+}
+
+static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	int vmid;
+	unsigned int tmp;
+
+	if (adev->in_gpu_reset)
+		return -EIO;
+
+	for (vmid = 0; vmid < 16; vmid++) {
+		if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid))
+			continue;
+
+		tmp = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+		if ((tmp & ATC_VMID0_PASID_MAPPING__VALID_MASK) &&
+			(tmp & ATC_VMID0_PASID_MAPPING__PASID_MASK) == pasid) {
+			WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+			RREG32(mmVM_INVALIDATE_RESPONSE);
+			break;
+		}
+	}
+
+	return 0;
+}
+
+static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("non kfd vmid\n");
+		return 0;
+	}
+
+	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+	RREG32(mmVM_INVALIDATE_RESPONSE);
+	return 0;
+}
+
+ /**
+  * read_vmid_from_vmfault_reg - read vmid from register
+  *
+  * adev: amdgpu_device pointer
+  * @vmid: vmid pointer
+  * read vmid from register (CIK).
+  */
+static uint32_t read_vmid_from_vmfault_reg(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	uint32_t status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
+
+	return REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
new file mode 100644
index 00000000000..f6e53e9352b
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
@@ -0,0 +1,884 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/module.h>
+#include <linux/fdtable.h>
+#include <linux/uaccess.h>
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_ucode.h"
+#include "gfx_v8_0.h"
+#include "gca/gfx_8_0_sh_mask.h"
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_enum.h"
+#include "oss/oss_3_0_sh_mask.h"
+#include "oss/oss_3_0_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+#include "gmc/gmc_8_1_d.h"
+#include "vi_structs.h"
+#include "vid.h"
+
+enum hqd_dequeue_request_type {
+	NO_ACTION = 0,
+	DRAIN_PIPE,
+	RESET_WAVES
+};
+
+struct vi_sdma_mqd;
+
+/*
+ * Register access functions
+ */
+
+static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t sh_mem_config,
+		uint32_t sh_mem_ape1_base, uint32_t sh_mem_ape1_limit,
+		uint32_t sh_mem_bases);
+static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
+		unsigned int vmid);
+static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id);
+static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
+			uint32_t queue_id, uint32_t __user *wptr,
+			uint32_t wptr_shift, uint32_t wptr_mask,
+			struct mm_struct *mm);
+static int kgd_hqd_dump(struct kgd_dev *kgd,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs);
+static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm);
+static int kgd_hqd_sdma_dump(struct kgd_dev *kgd,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs);
+static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
+		uint32_t pipe_id, uint32_t queue_id);
+static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd);
+static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
+				enum kfd_preempt_type reset_type,
+				unsigned int utimeout, uint32_t pipe_id,
+				uint32_t queue_id);
+static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
+				unsigned int utimeout);
+static int kgd_address_watch_disable(struct kgd_dev *kgd);
+static int kgd_address_watch_execute(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					uint32_t cntl_val,
+					uint32_t addr_hi,
+					uint32_t addr_lo);
+static int kgd_wave_control_execute(struct kgd_dev *kgd,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd);
+static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					unsigned int reg_offset);
+
+static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
+		uint8_t vmid);
+static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
+		uint8_t vmid);
+static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
+static void set_scratch_backing_va(struct kgd_dev *kgd,
+					uint64_t va, uint32_t vmid);
+static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t page_table_base);
+static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid);
+static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid);
+
+/* Because of REG_GET_FIELD() being used, we put this function in the
+ * asic specific file.
+ */
+static int get_tile_config(struct kgd_dev *kgd,
+		struct tile_config *config)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+
+	config->gb_addr_config = adev->gfx.config.gb_addr_config;
+	config->num_banks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
+				MC_ARB_RAMCFG, NOOFBANK);
+	config->num_ranks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
+				MC_ARB_RAMCFG, NOOFRANKS);
+
+	config->tile_config_ptr = adev->gfx.config.tile_mode_array;
+	config->num_tile_configs =
+			ARRAY_SIZE(adev->gfx.config.tile_mode_array);
+	config->macro_tile_config_ptr =
+			adev->gfx.config.macrotile_mode_array;
+	config->num_macro_tile_configs =
+			ARRAY_SIZE(adev->gfx.config.macrotile_mode_array);
+
+	return 0;
+}
+
+static const struct kfd2kgd_calls kfd2kgd = {
+	.init_gtt_mem_allocation = alloc_gtt_mem,
+	.free_gtt_mem = free_gtt_mem,
+	.get_local_mem_info = get_local_mem_info,
+	.get_gpu_clock_counter = get_gpu_clock_counter,
+	.get_max_engine_clock_in_mhz = get_max_engine_clock_in_mhz,
+	.alloc_pasid = amdgpu_pasid_alloc,
+	.free_pasid = amdgpu_pasid_free,
+	.program_sh_mem_settings = kgd_program_sh_mem_settings,
+	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
+	.init_interrupts = kgd_init_interrupts,
+	.hqd_load = kgd_hqd_load,
+	.hqd_sdma_load = kgd_hqd_sdma_load,
+	.hqd_dump = kgd_hqd_dump,
+	.hqd_sdma_dump = kgd_hqd_sdma_dump,
+	.hqd_is_occupied = kgd_hqd_is_occupied,
+	.hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied,
+	.hqd_destroy = kgd_hqd_destroy,
+	.hqd_sdma_destroy = kgd_hqd_sdma_destroy,
+	.address_watch_disable = kgd_address_watch_disable,
+	.address_watch_execute = kgd_address_watch_execute,
+	.wave_control_execute = kgd_wave_control_execute,
+	.address_watch_get_offset = kgd_address_watch_get_offset,
+	.get_atc_vmid_pasid_mapping_pasid =
+			get_atc_vmid_pasid_mapping_pasid,
+	.get_atc_vmid_pasid_mapping_valid =
+			get_atc_vmid_pasid_mapping_valid,
+	.get_fw_version = get_fw_version,
+	.set_scratch_backing_va = set_scratch_backing_va,
+	.get_tile_config = get_tile_config,
+	.get_cu_info = get_cu_info,
+	.get_vram_usage = amdgpu_amdkfd_get_vram_usage,
+	.create_process_vm = amdgpu_amdkfd_gpuvm_create_process_vm,
+	.acquire_process_vm = amdgpu_amdkfd_gpuvm_acquire_process_vm,
+	.destroy_process_vm = amdgpu_amdkfd_gpuvm_destroy_process_vm,
+	.get_process_page_dir = amdgpu_amdkfd_gpuvm_get_process_page_dir,
+	.set_vm_context_page_table_base = set_vm_context_page_table_base,
+	.alloc_memory_of_gpu = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu,
+	.free_memory_of_gpu = amdgpu_amdkfd_gpuvm_free_memory_of_gpu,
+	.map_memory_to_gpu = amdgpu_amdkfd_gpuvm_map_memory_to_gpu,
+	.unmap_memory_to_gpu = amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu,
+	.sync_memory = amdgpu_amdkfd_gpuvm_sync_memory,
+	.map_gtt_bo_to_kernel = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel,
+	.restore_process_bos = amdgpu_amdkfd_gpuvm_restore_process_bos,
+	.invalidate_tlbs = invalidate_tlbs,
+	.invalidate_tlbs_vmid = invalidate_tlbs_vmid,
+	.submit_ib = amdgpu_amdkfd_submit_ib,
+	.get_vm_fault_info = amdgpu_amdkfd_gpuvm_get_vm_fault_info,
+	.gpu_recover = amdgpu_amdkfd_gpu_reset,
+	.set_compute_idle = amdgpu_amdkfd_set_compute_idle
+};
+
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void)
+{
+	return (struct kfd2kgd_calls *)&kfd2kgd;
+}
+
+static inline struct amdgpu_device *get_amdgpu_device(struct kgd_dev *kgd)
+{
+	return (struct amdgpu_device *)kgd;
+}
+
+static void lock_srbm(struct kgd_dev *kgd, uint32_t mec, uint32_t pipe,
+			uint32_t queue, uint32_t vmid)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t value = PIPEID(pipe) | MEID(mec) | VMID(vmid) | QUEUEID(queue);
+
+	mutex_lock(&adev->srbm_mutex);
+	WREG32(mmSRBM_GFX_CNTL, value);
+}
+
+static void unlock_srbm(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	WREG32(mmSRBM_GFX_CNTL, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void acquire_queue(struct kgd_dev *kgd, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(kgd, mec, pipe, queue_id, 0);
+}
+
+static void release_queue(struct kgd_dev *kgd)
+{
+	unlock_srbm(kgd);
+}
+
+static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
+					uint32_t sh_mem_config,
+					uint32_t sh_mem_ape1_base,
+					uint32_t sh_mem_ape1_limit,
+					uint32_t sh_mem_bases)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	lock_srbm(kgd, 0, 0, 0, vmid);
+
+	WREG32(mmSH_MEM_CONFIG, sh_mem_config);
+	WREG32(mmSH_MEM_APE1_BASE, sh_mem_ape1_base);
+	WREG32(mmSH_MEM_APE1_LIMIT, sh_mem_ape1_limit);
+	WREG32(mmSH_MEM_BASES, sh_mem_bases);
+
+	unlock_srbm(kgd);
+}
+
+static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
+					unsigned int vmid)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	/*
+	 * We have to assume that there is no outstanding mapping.
+	 * The ATC_VMID_PASID_MAPPING_UPDATE_STATUS bit could be 0 because
+	 * a mapping is in progress or because a mapping finished
+	 * and the SW cleared it.
+	 * So the protocol is to always wait & clear.
+	 */
+	uint32_t pasid_mapping = (pasid == 0) ? 0 : (uint32_t)pasid |
+			ATC_VMID0_PASID_MAPPING__VALID_MASK;
+
+	WREG32(mmATC_VMID0_PASID_MAPPING + vmid, pasid_mapping);
+
+	while (!(RREG32(mmATC_VMID_PASID_MAPPING_UPDATE_STATUS) & (1U << vmid)))
+		cpu_relax();
+	WREG32(mmATC_VMID_PASID_MAPPING_UPDATE_STATUS, 1U << vmid);
+
+	/* Mapping vmid to pasid also for IH block */
+	WREG32(mmIH_VMID_0_LUT + vmid, pasid_mapping);
+
+	return 0;
+}
+
+static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t mec;
+	uint32_t pipe;
+
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(kgd, mec, pipe, 0, 0);
+
+	WREG32(mmCPC_INT_CNTL, CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK);
+
+	unlock_srbm(kgd);
+
+	return 0;
+}
+
+static inline uint32_t get_sdma_base_addr(struct vi_sdma_mqd *m)
+{
+	uint32_t retval;
+
+	retval = m->sdma_engine_id * SDMA1_REGISTER_OFFSET +
+		m->sdma_queue_id * KFD_VI_SDMA_QUEUE_OFFSET;
+	pr_debug("kfd: sdma base address: 0x%x\n", retval);
+
+	return retval;
+}
+
+static inline struct vi_mqd *get_mqd(void *mqd)
+{
+	return (struct vi_mqd *)mqd;
+}
+
+static inline struct vi_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct vi_sdma_mqd *)mqd;
+}
+
+static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
+			uint32_t queue_id, uint32_t __user *wptr,
+			uint32_t wptr_shift, uint32_t wptr_mask,
+			struct mm_struct *mm)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct vi_mqd *m;
+	uint32_t *mqd_hqd;
+	uint32_t reg, wptr_val, data;
+	bool valid_wptr = false;
+
+	m = get_mqd(mqd);
+
+	acquire_queue(kgd, pipe_id, queue_id);
+
+	/* HIQ is set during driver init period with vmid set to 0*/
+	if (m->cp_hqd_vmid == 0) {
+		uint32_t value, mec, pipe;
+
+		mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+		pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+		pr_debug("kfd: set HIQ, mec:%d, pipe:%d, queue:%d.\n",
+			mec, pipe, queue_id);
+		value = RREG32(mmRLC_CP_SCHEDULERS);
+		value = REG_SET_FIELD(value, RLC_CP_SCHEDULERS, scheduler1,
+			((mec << 5) | (pipe << 3) | queue_id | 0x80));
+		WREG32(mmRLC_CP_SCHEDULERS, value);
+	}
+
+	/* HQD registers extend from CP_MQD_BASE_ADDR to CP_HQD_EOP_WPTR_MEM. */
+	mqd_hqd = &m->cp_mqd_base_addr_lo;
+
+	for (reg = mmCP_MQD_BASE_ADDR; reg <= mmCP_HQD_EOP_CONTROL; reg++)
+		WREG32(reg, mqd_hqd[reg - mmCP_MQD_BASE_ADDR]);
+
+	/* Tonga errata: EOP RPTR/WPTR should be left unmodified.
+	 * This is safe since EOP RPTR==WPTR for any inactive HQD
+	 * on ASICs that do not support context-save.
+	 * EOP writes/reads can start anywhere in the ring.
+	 */
+	if (get_amdgpu_device(kgd)->asic_type != CHIP_TONGA) {
+		WREG32(mmCP_HQD_EOP_RPTR, m->cp_hqd_eop_rptr);
+		WREG32(mmCP_HQD_EOP_WPTR, m->cp_hqd_eop_wptr);
+		WREG32(mmCP_HQD_EOP_WPTR_MEM, m->cp_hqd_eop_wptr_mem);
+	}
+
+	for (reg = mmCP_HQD_EOP_EVENTS; reg <= mmCP_HQD_ERROR; reg++)
+		WREG32(reg, mqd_hqd[reg - mmCP_MQD_BASE_ADDR]);
+
+	/* Copy userspace write pointer value to register.
+	 * Activate doorbell logic to monitor subsequent changes.
+	 */
+	data = REG_SET_FIELD(m->cp_hqd_pq_doorbell_control,
+			     CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
+	WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, data);
+
+	/* read_user_ptr may take the mm->mmap_sem.
+	 * release srbm_mutex to avoid circular dependency between
+	 * srbm_mutex->mm_sem->reservation_ww_class_mutex->srbm_mutex.
+	 */
+	release_queue(kgd);
+	valid_wptr = read_user_wptr(mm, wptr, wptr_val);
+	acquire_queue(kgd, pipe_id, queue_id);
+	if (valid_wptr)
+		WREG32(mmCP_HQD_PQ_WPTR, (wptr_val << wptr_shift) & wptr_mask);
+
+	data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
+	WREG32(mmCP_HQD_ACTIVE, data);
+
+	release_queue(kgd);
+
+	return 0;
+}
+
+static int kgd_hqd_dump(struct kgd_dev *kgd,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t i = 0, reg;
+#define HQD_N_REGS (54+4)
+#define DUMP_REG(addr) do {				\
+		if (WARN_ON_ONCE(i >= HQD_N_REGS))	\
+			break;				\
+		(*dump)[i][0] = (addr) << 2;		\
+		(*dump)[i++][1] = RREG32(addr);		\
+	} while (0)
+
+	*dump = kmalloc_array(HQD_N_REGS * 2, sizeof(uint32_t), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE0);
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE1);
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE2);
+	DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE3);
+
+	for (reg = mmCP_MQD_BASE_ADDR; reg <= mmCP_HQD_EOP_DONES; reg++)
+		DUMP_REG(reg);
+
+	release_queue(kgd);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct vi_sdma_mqd *m;
+	unsigned long end_jiffies;
+	uint32_t sdma_base_addr;
+	uint32_t data;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(m);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL,
+		m->sdmax_rlcx_rb_cntl & (~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK));
+
+	end_jiffies = msecs_to_jiffies(2000) + jiffies;
+	while (true) {
+		data = RREG32(sdma_base_addr + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (data & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies))
+			return -ETIME;
+		usleep_range(500, 1000);
+	}
+	if (m->sdma_engine_id) {
+		data = RREG32(mmSDMA1_GFX_CONTEXT_CNTL);
+		data = REG_SET_FIELD(data, SDMA1_GFX_CONTEXT_CNTL,
+				RESUME_CTX, 0);
+		WREG32(mmSDMA1_GFX_CONTEXT_CNTL, data);
+	} else {
+		data = RREG32(mmSDMA0_GFX_CONTEXT_CNTL);
+		data = REG_SET_FIELD(data, SDMA0_GFX_CONTEXT_CNTL,
+				RESUME_CTX, 0);
+		WREG32(mmSDMA0_GFX_CONTEXT_CNTL, data);
+	}
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_doorbell, SDMA0_RLC0_DOORBELL,
+			     ENABLE, 1);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL, data);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR, m->sdmax_rlcx_rb_rptr);
+
+	if (read_user_wptr(mm, wptr, data))
+		WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR, data);
+	else
+		WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR,
+		       m->sdmax_rlcx_rb_rptr);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_VIRTUAL_ADDR,
+				m->sdmax_rlcx_virtual_addr);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE, m->sdmax_rlcx_rb_base);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE_HI,
+			m->sdmax_rlcx_rb_base_hi);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_ADDR_LO,
+			m->sdmax_rlcx_rb_rptr_addr_lo);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_ADDR_HI,
+			m->sdmax_rlcx_rb_rptr_addr_hi);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_rb_cntl, SDMA0_RLC0_RB_CNTL,
+			     RB_ENABLE, 1);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL, data);
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_dump(struct kgd_dev *kgd,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t sdma_offset = engine_id * SDMA1_REGISTER_OFFSET +
+		queue_id * KFD_VI_SDMA_QUEUE_OFFSET;
+	uint32_t i = 0, reg;
+#undef HQD_N_REGS
+#define HQD_N_REGS (19+4+2+3+7)
+
+	*dump = kmalloc_array(HQD_N_REGS * 2, sizeof(uint32_t), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	for (reg = mmSDMA0_RLC0_RB_CNTL; reg <= mmSDMA0_RLC0_DOORBELL; reg++)
+		DUMP_REG(sdma_offset + reg);
+	for (reg = mmSDMA0_RLC0_VIRTUAL_ADDR; reg <= mmSDMA0_RLC0_WATERMARK;
+	     reg++)
+		DUMP_REG(sdma_offset + reg);
+	for (reg = mmSDMA0_RLC0_CSA_ADDR_LO; reg <= mmSDMA0_RLC0_CSA_ADDR_HI;
+	     reg++)
+		DUMP_REG(sdma_offset + reg);
+	for (reg = mmSDMA0_RLC0_IB_SUB_REMAIN; reg <= mmSDMA0_RLC0_DUMMY_REG;
+	     reg++)
+		DUMP_REG(sdma_offset + reg);
+	for (reg = mmSDMA0_RLC0_MIDCMD_DATA0; reg <= mmSDMA0_RLC0_MIDCMD_CNTL;
+	     reg++)
+		DUMP_REG(sdma_offset + reg);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
+				uint32_t pipe_id, uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t act;
+	bool retval = false;
+	uint32_t low, high;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+	act = RREG32(mmCP_HQD_ACTIVE);
+	if (act) {
+		low = lower_32_bits(queue_address >> 8);
+		high = upper_32_bits(queue_address >> 8);
+
+		if (low == RREG32(mmCP_HQD_PQ_BASE) &&
+				high == RREG32(mmCP_HQD_PQ_BASE_HI))
+			retval = true;
+	}
+	release_queue(kgd);
+	return retval;
+}
+
+static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct vi_sdma_mqd *m;
+	uint32_t sdma_base_addr;
+	uint32_t sdma_rlc_rb_cntl;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(m);
+
+	sdma_rlc_rb_cntl = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+
+	if (sdma_rlc_rb_cntl & SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK)
+		return true;
+
+	return false;
+}
+
+static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
+				enum kfd_preempt_type reset_type,
+				unsigned int utimeout, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t temp;
+	enum hqd_dequeue_request_type type;
+	unsigned long flags, end_jiffies;
+	int retry;
+	struct vi_mqd *m = get_mqd(mqd);
+
+	if (adev->in_gpu_reset)
+		return -EIO;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+
+	if (m->cp_hqd_vmid == 0)
+		WREG32_FIELD(RLC_CP_SCHEDULERS, scheduler1, 0);
+
+	switch (reset_type) {
+	case KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN:
+		type = DRAIN_PIPE;
+		break;
+	case KFD_PREEMPT_TYPE_WAVEFRONT_RESET:
+		type = RESET_WAVES;
+		break;
+	default:
+		type = DRAIN_PIPE;
+		break;
+	}
+
+	/* Workaround: If IQ timer is active and the wait time is close to or
+	 * equal to 0, dequeueing is not safe. Wait until either the wait time
+	 * is larger or timer is cleared. Also, ensure that IQ_REQ_PEND is
+	 * cleared before continuing. Also, ensure wait times are set to at
+	 * least 0x3.
+	 */
+	local_irq_save(flags);
+	preempt_disable();
+	retry = 5000; /* wait for 500 usecs at maximum */
+	while (true) {
+		temp = RREG32(mmCP_HQD_IQ_TIMER);
+		if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, PROCESSING_IQ)) {
+			pr_debug("HW is processing IQ\n");
+			goto loop;
+		}
+		if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, ACTIVE)) {
+			if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, RETRY_TYPE)
+					== 3) /* SEM-rearm is safe */
+				break;
+			/* Wait time 3 is safe for CP, but our MMIO read/write
+			 * time is close to 1 microsecond, so check for 10 to
+			 * leave more buffer room
+			 */
+			if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, WAIT_TIME)
+					>= 10)
+				break;
+			pr_debug("IQ timer is active\n");
+		} else
+			break;
+loop:
+		if (!retry) {
+			pr_err("CP HQD IQ timer status time out\n");
+			break;
+		}
+		ndelay(100);
+		--retry;
+	}
+	retry = 1000;
+	while (true) {
+		temp = RREG32(mmCP_HQD_DEQUEUE_REQUEST);
+		if (!(temp & CP_HQD_DEQUEUE_REQUEST__IQ_REQ_PEND_MASK))
+			break;
+		pr_debug("Dequeue request is pending\n");
+
+		if (!retry) {
+			pr_err("CP HQD dequeue request time out\n");
+			break;
+		}
+		ndelay(100);
+		--retry;
+	}
+	local_irq_restore(flags);
+	preempt_enable();
+
+	WREG32(mmCP_HQD_DEQUEUE_REQUEST, type);
+
+	end_jiffies = (utimeout * HZ / 1000) + jiffies;
+	while (true) {
+		temp = RREG32(mmCP_HQD_ACTIVE);
+		if (!(temp & CP_HQD_ACTIVE__ACTIVE_MASK))
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("cp queue preemption time out.\n");
+			release_queue(kgd);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	release_queue(kgd);
+	return 0;
+}
+
+static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
+				unsigned int utimeout)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct vi_sdma_mqd *m;
+	uint32_t sdma_base_addr;
+	uint32_t temp;
+	unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(m);
+
+	temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+	temp = temp & ~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL, temp);
+
+	while (true) {
+		temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (temp & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies))
+			return -ETIME;
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL, 0);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL,
+		RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL) |
+		SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK);
+
+	m->sdmax_rlcx_rb_rptr = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR);
+
+	return 0;
+}
+
+static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
+							uint8_t vmid)
+{
+	uint32_t reg;
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
+}
+
+static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
+								uint8_t vmid)
+{
+	uint32_t reg;
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+	return reg & ATC_VMID0_PASID_MAPPING__PASID_MASK;
+}
+
+static int kgd_address_watch_disable(struct kgd_dev *kgd)
+{
+	return 0;
+}
+
+static int kgd_address_watch_execute(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					uint32_t cntl_val,
+					uint32_t addr_hi,
+					uint32_t addr_lo)
+{
+	return 0;
+}
+
+static int kgd_wave_control_execute(struct kgd_dev *kgd,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t data = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	WREG32(mmGRBM_GFX_INDEX, gfx_index_val);
+	WREG32(mmSQ_CMD, sq_cmd);
+
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		INSTANCE_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SH_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SE_BROADCAST_WRITES, 1);
+
+	WREG32(mmGRBM_GFX_INDEX, data);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					unsigned int reg_offset)
+{
+	return 0;
+}
+
+static void set_scratch_backing_va(struct kgd_dev *kgd,
+					uint64_t va, uint32_t vmid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	lock_srbm(kgd, 0, 0, 0, vmid);
+	WREG32(mmSH_HIDDEN_PRIVATE_BASE_VMID, va);
+	unlock_srbm(kgd);
+}
+
+static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	const union amdgpu_firmware_header *hdr;
+
+	switch (type) {
+	case KGD_ENGINE_PFP:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->gfx.pfp_fw->data;
+		break;
+
+	case KGD_ENGINE_ME:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->gfx.me_fw->data;
+		break;
+
+	case KGD_ENGINE_CE:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->gfx.ce_fw->data;
+		break;
+
+	case KGD_ENGINE_MEC1:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->gfx.mec_fw->data;
+		break;
+
+	case KGD_ENGINE_MEC2:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->gfx.mec2_fw->data;
+		break;
+
+	case KGD_ENGINE_RLC:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->gfx.rlc_fw->data;
+		break;
+
+	case KGD_ENGINE_SDMA1:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->sdma.instance[0].fw->data;
+		break;
+
+	case KGD_ENGINE_SDMA2:
+		hdr = (const union amdgpu_firmware_header *)
+						adev->sdma.instance[1].fw->data;
+		break;
+
+	default:
+		return 0;
+	}
+
+	if (hdr == NULL)
+		return 0;
+
+	/* Only 12 bit in use*/
+	return hdr->common.ucode_version;
+}
+
+static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t page_table_base)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("trying to set page table base for wrong VMID\n");
+		return;
+	}
+	WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8, page_table_base);
+}
+
+static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	int vmid;
+	unsigned int tmp;
+
+	if (adev->in_gpu_reset)
+		return -EIO;
+
+	for (vmid = 0; vmid < 16; vmid++) {
+		if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid))
+			continue;
+
+		tmp = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+		if ((tmp & ATC_VMID0_PASID_MAPPING__VALID_MASK) &&
+			(tmp & ATC_VMID0_PASID_MAPPING__PASID_MASK) == pasid) {
+			WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+			RREG32(mmVM_INVALIDATE_RESPONSE);
+			break;
+		}
+	}
+
+	return 0;
+}
+
+static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("non kfd vmid %d\n", vmid);
+		return -EINVAL;
+	}
+
+	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+	RREG32(mmVM_INVALIDATE_RESPONSE);
+	return 0;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
new file mode 100644
index 00000000000..8efedfcb9df
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
@@ -0,0 +1,1051 @@
+/*
+ * Copyright 2014-2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#define pr_fmt(fmt) "kfd2kgd: " fmt
+
+#include <linux/module.h>
+#include <linux/fdtable.h>
+#include <linux/uaccess.h>
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_ucode.h"
+#include "soc15_hw_ip.h"
+#include "gc/gc_9_0_offset.h"
+#include "gc/gc_9_0_sh_mask.h"
+#include "vega10_enum.h"
+#include "sdma0/sdma0_4_0_offset.h"
+#include "sdma0/sdma0_4_0_sh_mask.h"
+#include "sdma1/sdma1_4_0_offset.h"
+#include "sdma1/sdma1_4_0_sh_mask.h"
+#include "athub/athub_1_0_offset.h"
+#include "athub/athub_1_0_sh_mask.h"
+#include "oss/osssys_4_0_offset.h"
+#include "oss/osssys_4_0_sh_mask.h"
+#include "soc15_common.h"
+#include "v9_structs.h"
+#include "soc15.h"
+#include "soc15d.h"
+
+/* HACK: MMHUB and GC both have VM-related register with the same
+ * names but different offsets. Define the MMHUB register we need here
+ * with a prefix. A proper solution would be to move the functions
+ * programming these registers into gfx_v9_0.c and mmhub_v1_0.c
+ * respectively.
+ */
+#define mmMMHUB_VM_INVALIDATE_ENG16_REQ				0x06f3
+#define mmMMHUB_VM_INVALIDATE_ENG16_REQ_BASE_IDX		0
+
+#define mmMMHUB_VM_INVALIDATE_ENG16_ACK				0x0705
+#define mmMMHUB_VM_INVALIDATE_ENG16_ACK_BASE_IDX		0
+
+#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32		0x072b
+#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32_BASE_IDX	0
+#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32		0x072c
+#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32_BASE_IDX	0
+
+#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32		0x074b
+#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32_BASE_IDX	0
+#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32		0x074c
+#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32_BASE_IDX	0
+
+#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32		0x076b
+#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32_BASE_IDX	0
+#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32		0x076c
+#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32_BASE_IDX	0
+
+#define mmMMHUB_VM_INVALIDATE_ENG16_ADDR_RANGE_LO32		0x0727
+#define mmMMHUB_VM_INVALIDATE_ENG16_ADDR_RANGE_LO32_BASE_IDX	0
+#define mmMMHUB_VM_INVALIDATE_ENG16_ADDR_RANGE_HI32		0x0728
+#define mmMMHUB_VM_INVALIDATE_ENG16_ADDR_RANGE_HI32_BASE_IDX	0
+
+#define V9_PIPE_PER_MEC		(4)
+#define V9_QUEUES_PER_PIPE_MEC	(8)
+
+enum hqd_dequeue_request_type {
+	NO_ACTION = 0,
+	DRAIN_PIPE,
+	RESET_WAVES
+};
+
+/*
+ * Register access functions
+ */
+
+static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t sh_mem_config,
+		uint32_t sh_mem_ape1_base, uint32_t sh_mem_ape1_limit,
+		uint32_t sh_mem_bases);
+static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
+		unsigned int vmid);
+static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id);
+static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
+			uint32_t queue_id, uint32_t __user *wptr,
+			uint32_t wptr_shift, uint32_t wptr_mask,
+			struct mm_struct *mm);
+static int kgd_hqd_dump(struct kgd_dev *kgd,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs);
+static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm);
+static int kgd_hqd_sdma_dump(struct kgd_dev *kgd,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs);
+static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
+		uint32_t pipe_id, uint32_t queue_id);
+static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd);
+static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
+				enum kfd_preempt_type reset_type,
+				unsigned int utimeout, uint32_t pipe_id,
+				uint32_t queue_id);
+static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
+				unsigned int utimeout);
+static int kgd_address_watch_disable(struct kgd_dev *kgd);
+static int kgd_address_watch_execute(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					uint32_t cntl_val,
+					uint32_t addr_hi,
+					uint32_t addr_lo);
+static int kgd_wave_control_execute(struct kgd_dev *kgd,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd);
+static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					unsigned int reg_offset);
+
+static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
+		uint8_t vmid);
+static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
+		uint8_t vmid);
+static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t page_table_base);
+static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
+static void set_scratch_backing_va(struct kgd_dev *kgd,
+					uint64_t va, uint32_t vmid);
+static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid);
+static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid);
+
+/* Because of REG_GET_FIELD() being used, we put this function in the
+ * asic specific file.
+ */
+static int amdgpu_amdkfd_get_tile_config(struct kgd_dev *kgd,
+		struct tile_config *config)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+
+	config->gb_addr_config = adev->gfx.config.gb_addr_config;
+
+	config->tile_config_ptr = adev->gfx.config.tile_mode_array;
+	config->num_tile_configs =
+			ARRAY_SIZE(adev->gfx.config.tile_mode_array);
+	config->macro_tile_config_ptr =
+			adev->gfx.config.macrotile_mode_array;
+	config->num_macro_tile_configs =
+			ARRAY_SIZE(adev->gfx.config.macrotile_mode_array);
+
+	return 0;
+}
+
+static const struct kfd2kgd_calls kfd2kgd = {
+	.init_gtt_mem_allocation = alloc_gtt_mem,
+	.free_gtt_mem = free_gtt_mem,
+	.get_local_mem_info = get_local_mem_info,
+	.get_gpu_clock_counter = get_gpu_clock_counter,
+	.get_max_engine_clock_in_mhz = get_max_engine_clock_in_mhz,
+	.alloc_pasid = amdgpu_pasid_alloc,
+	.free_pasid = amdgpu_pasid_free,
+	.program_sh_mem_settings = kgd_program_sh_mem_settings,
+	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
+	.init_interrupts = kgd_init_interrupts,
+	.hqd_load = kgd_hqd_load,
+	.hqd_sdma_load = kgd_hqd_sdma_load,
+	.hqd_dump = kgd_hqd_dump,
+	.hqd_sdma_dump = kgd_hqd_sdma_dump,
+	.hqd_is_occupied = kgd_hqd_is_occupied,
+	.hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied,
+	.hqd_destroy = kgd_hqd_destroy,
+	.hqd_sdma_destroy = kgd_hqd_sdma_destroy,
+	.address_watch_disable = kgd_address_watch_disable,
+	.address_watch_execute = kgd_address_watch_execute,
+	.wave_control_execute = kgd_wave_control_execute,
+	.address_watch_get_offset = kgd_address_watch_get_offset,
+	.get_atc_vmid_pasid_mapping_pasid =
+			get_atc_vmid_pasid_mapping_pasid,
+	.get_atc_vmid_pasid_mapping_valid =
+			get_atc_vmid_pasid_mapping_valid,
+	.get_fw_version = get_fw_version,
+	.set_scratch_backing_va = set_scratch_backing_va,
+	.get_tile_config = amdgpu_amdkfd_get_tile_config,
+	.get_cu_info = get_cu_info,
+	.get_vram_usage = amdgpu_amdkfd_get_vram_usage,
+	.create_process_vm = amdgpu_amdkfd_gpuvm_create_process_vm,
+	.acquire_process_vm = amdgpu_amdkfd_gpuvm_acquire_process_vm,
+	.destroy_process_vm = amdgpu_amdkfd_gpuvm_destroy_process_vm,
+	.get_process_page_dir = amdgpu_amdkfd_gpuvm_get_process_page_dir,
+	.set_vm_context_page_table_base = set_vm_context_page_table_base,
+	.alloc_memory_of_gpu = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu,
+	.free_memory_of_gpu = amdgpu_amdkfd_gpuvm_free_memory_of_gpu,
+	.map_memory_to_gpu = amdgpu_amdkfd_gpuvm_map_memory_to_gpu,
+	.unmap_memory_to_gpu = amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu,
+	.sync_memory = amdgpu_amdkfd_gpuvm_sync_memory,
+	.map_gtt_bo_to_kernel = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel,
+	.restore_process_bos = amdgpu_amdkfd_gpuvm_restore_process_bos,
+	.invalidate_tlbs = invalidate_tlbs,
+	.invalidate_tlbs_vmid = invalidate_tlbs_vmid,
+	.submit_ib = amdgpu_amdkfd_submit_ib,
+	.gpu_recover = amdgpu_amdkfd_gpu_reset,
+	.set_compute_idle = amdgpu_amdkfd_set_compute_idle
+};
+
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_9_0_get_functions(void)
+{
+	return (struct kfd2kgd_calls *)&kfd2kgd;
+}
+
+static inline struct amdgpu_device *get_amdgpu_device(struct kgd_dev *kgd)
+{
+	return (struct amdgpu_device *)kgd;
+}
+
+static void lock_srbm(struct kgd_dev *kgd, uint32_t mec, uint32_t pipe,
+			uint32_t queue, uint32_t vmid)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	mutex_lock(&adev->srbm_mutex);
+	soc15_grbm_select(adev, mec, pipe, queue, vmid);
+}
+
+static void unlock_srbm(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	soc15_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void acquire_queue(struct kgd_dev *kgd, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(kgd, mec, pipe, queue_id, 0);
+}
+
+static uint32_t get_queue_mask(struct amdgpu_device *adev,
+			       uint32_t pipe_id, uint32_t queue_id)
+{
+	unsigned int bit = (pipe_id * adev->gfx.mec.num_queue_per_pipe +
+			    queue_id) & 31;
+
+	return ((uint32_t)1) << bit;
+}
+
+static void release_queue(struct kgd_dev *kgd)
+{
+	unlock_srbm(kgd);
+}
+
+static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
+					uint32_t sh_mem_config,
+					uint32_t sh_mem_ape1_base,
+					uint32_t sh_mem_ape1_limit,
+					uint32_t sh_mem_bases)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	lock_srbm(kgd, 0, 0, 0, vmid);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSH_MEM_CONFIG), sh_mem_config);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSH_MEM_BASES), sh_mem_bases);
+	/* APE1 no longer exists on GFX9 */
+
+	unlock_srbm(kgd);
+}
+
+static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
+					unsigned int vmid)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	/*
+	 * We have to assume that there is no outstanding mapping.
+	 * The ATC_VMID_PASID_MAPPING_UPDATE_STATUS bit could be 0 because
+	 * a mapping is in progress or because a mapping finished
+	 * and the SW cleared it.
+	 * So the protocol is to always wait & clear.
+	 */
+	uint32_t pasid_mapping = (pasid == 0) ? 0 : (uint32_t)pasid |
+			ATC_VMID0_PASID_MAPPING__VALID_MASK;
+
+	/*
+	 * need to do this twice, once for gfx and once for mmhub
+	 * for ATC add 16 to VMID for mmhub, for IH different registers.
+	 * ATC_VMID0..15 registers are separate from ATC_VMID16..31.
+	 */
+
+	WREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING) + vmid,
+	       pasid_mapping);
+
+	while (!(RREG32(SOC15_REG_OFFSET(
+				ATHUB, 0,
+				mmATC_VMID_PASID_MAPPING_UPDATE_STATUS)) &
+		 (1U << vmid)))
+		cpu_relax();
+
+	WREG32(SOC15_REG_OFFSET(ATHUB, 0,
+				mmATC_VMID_PASID_MAPPING_UPDATE_STATUS),
+	       1U << vmid);
+
+	/* Mapping vmid to pasid also for IH block */
+	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT) + vmid,
+	       pasid_mapping);
+
+	WREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID16_PASID_MAPPING) + vmid,
+	       pasid_mapping);
+
+	while (!(RREG32(SOC15_REG_OFFSET(
+				ATHUB, 0,
+				mmATC_VMID_PASID_MAPPING_UPDATE_STATUS)) &
+		 (1U << (vmid + 16))))
+		cpu_relax();
+
+	WREG32(SOC15_REG_OFFSET(ATHUB, 0,
+				mmATC_VMID_PASID_MAPPING_UPDATE_STATUS),
+	       1U << (vmid + 16));
+
+	/* Mapping vmid to pasid also for IH block */
+	WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT_MM) + vmid,
+	       pasid_mapping);
+	return 0;
+}
+
+/* TODO - RING0 form of field is obsolete, seems to date back to SI
+ * but still works
+ */
+
+static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t mec;
+	uint32_t pipe;
+
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+	lock_srbm(kgd, mec, pipe, 0, 0);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmCPC_INT_CNTL),
+		CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
+		CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
+
+	unlock_srbm(kgd);
+
+	return 0;
+}
+
+static uint32_t get_sdma_base_addr(struct amdgpu_device *adev,
+				unsigned int engine_id,
+				unsigned int queue_id)
+{
+	uint32_t base[2] = {
+		SOC15_REG_OFFSET(SDMA0, 0,
+				 mmSDMA0_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL,
+		SOC15_REG_OFFSET(SDMA1, 0,
+				 mmSDMA1_RLC0_RB_CNTL) - mmSDMA1_RLC0_RB_CNTL
+	};
+	uint32_t retval;
+
+	retval = base[engine_id] + queue_id * (mmSDMA0_RLC1_RB_CNTL -
+					       mmSDMA0_RLC0_RB_CNTL);
+
+	pr_debug("sdma base address: 0x%x\n", retval);
+
+	return retval;
+}
+
+static inline struct v9_mqd *get_mqd(void *mqd)
+{
+	return (struct v9_mqd *)mqd;
+}
+
+static inline struct v9_sdma_mqd *get_sdma_mqd(void *mqd)
+{
+	return (struct v9_sdma_mqd *)mqd;
+}
+
+static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
+			uint32_t queue_id, uint32_t __user *wptr,
+			uint32_t wptr_shift, uint32_t wptr_mask,
+			struct mm_struct *mm)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct v9_mqd *m;
+	uint32_t *mqd_hqd;
+	uint32_t reg, hqd_base, data;
+
+	m = get_mqd(mqd);
+
+	acquire_queue(kgd, pipe_id, queue_id);
+
+	/* HIQ is set during driver init period with vmid set to 0*/
+	if (m->cp_hqd_vmid == 0) {
+		uint32_t value, mec, pipe;
+
+		mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+		pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+		pr_debug("kfd: set HIQ, mec:%d, pipe:%d, queue:%d.\n",
+			mec, pipe, queue_id);
+		value = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_CP_SCHEDULERS));
+		value = REG_SET_FIELD(value, RLC_CP_SCHEDULERS, scheduler1,
+			((mec << 5) | (pipe << 3) | queue_id | 0x80));
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_CP_SCHEDULERS), value);
+	}
+
+	/* HQD registers extend from CP_MQD_BASE_ADDR to CP_HQD_EOP_WPTR_MEM. */
+	mqd_hqd = &m->cp_mqd_base_addr_lo;
+	hqd_base = SOC15_REG_OFFSET(GC, 0, mmCP_MQD_BASE_ADDR);
+
+	for (reg = hqd_base;
+	     reg <= SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_HI); reg++)
+		WREG32(reg, mqd_hqd[reg - hqd_base]);
+
+
+	/* Activate doorbell logic before triggering WPTR poll. */
+	data = REG_SET_FIELD(m->cp_hqd_pq_doorbell_control,
+			     CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL), data);
+
+	if (wptr) {
+		/* Don't read wptr with get_user because the user
+		 * context may not be accessible (if this function
+		 * runs in a work queue). Instead trigger a one-shot
+		 * polling read from memory in the CP. This assumes
+		 * that wptr is GPU-accessible in the queue's VMID via
+		 * ATC or SVM. WPTR==RPTR before starting the poll so
+		 * the CP starts fetching new commands from the right
+		 * place.
+		 *
+		 * Guessing a 64-bit WPTR from a 32-bit RPTR is a bit
+		 * tricky. Assume that the queue didn't overflow. The
+		 * number of valid bits in the 32-bit RPTR depends on
+		 * the queue size. The remaining bits are taken from
+		 * the saved 64-bit WPTR. If the WPTR wrapped, add the
+		 * queue size.
+		 */
+		uint32_t queue_size =
+			2 << REG_GET_FIELD(m->cp_hqd_pq_control,
+					   CP_HQD_PQ_CONTROL, QUEUE_SIZE);
+		uint64_t guessed_wptr = m->cp_hqd_pq_rptr & (queue_size - 1);
+
+		if ((m->cp_hqd_pq_wptr_lo & (queue_size - 1)) < guessed_wptr)
+			guessed_wptr += queue_size;
+		guessed_wptr += m->cp_hqd_pq_wptr_lo & ~(queue_size - 1);
+		guessed_wptr += (uint64_t)m->cp_hqd_pq_wptr_hi << 32;
+
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_LO),
+		       lower_32_bits(guessed_wptr));
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_HI),
+		       upper_32_bits(guessed_wptr));
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR),
+		       lower_32_bits((uintptr_t)wptr));
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR_HI),
+		       upper_32_bits((uintptr_t)wptr));
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_PQ_WPTR_POLL_CNTL1),
+		       get_queue_mask(adev, pipe_id, queue_id));
+	}
+
+	/* Start the EOP fetcher */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_EOP_RPTR),
+	       REG_SET_FIELD(m->cp_hqd_eop_rptr,
+			     CP_HQD_EOP_RPTR, INIT_FETCHER, 1));
+
+	data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_ACTIVE), data);
+
+	release_queue(kgd);
+
+	return 0;
+}
+
+static int kgd_hqd_dump(struct kgd_dev *kgd,
+			uint32_t pipe_id, uint32_t queue_id,
+			uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t i = 0, reg;
+#define HQD_N_REGS 56
+#define DUMP_REG(addr) do {				\
+		if (WARN_ON_ONCE(i >= HQD_N_REGS))	\
+			break;				\
+		(*dump)[i][0] = (addr) << 2;		\
+		(*dump)[i++][1] = RREG32(addr);		\
+	} while (0)
+
+	*dump = kmalloc_array(HQD_N_REGS * 2, sizeof(uint32_t), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+
+	for (reg = SOC15_REG_OFFSET(GC, 0, mmCP_MQD_BASE_ADDR);
+	     reg <= SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_HI); reg++)
+		DUMP_REG(reg);
+
+	release_queue(kgd);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd,
+			     uint32_t __user *wptr, struct mm_struct *mm)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct v9_sdma_mqd *m;
+	uint32_t sdma_base_addr, sdmax_gfx_context_cntl;
+	unsigned long end_jiffies;
+	uint32_t data;
+	uint64_t data64;
+	uint64_t __user *wptr64 = (uint64_t __user *)wptr;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+	sdmax_gfx_context_cntl = m->sdma_engine_id ?
+		SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_GFX_CONTEXT_CNTL) :
+		SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_GFX_CONTEXT_CNTL);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL,
+		m->sdmax_rlcx_rb_cntl & (~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK));
+
+	end_jiffies = msecs_to_jiffies(2000) + jiffies;
+	while (true) {
+		data = RREG32(sdma_base_addr + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (data & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies))
+			return -ETIME;
+		usleep_range(500, 1000);
+	}
+	data = RREG32(sdmax_gfx_context_cntl);
+	data = REG_SET_FIELD(data, SDMA0_GFX_CONTEXT_CNTL,
+			     RESUME_CTX, 0);
+	WREG32(sdmax_gfx_context_cntl, data);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL_OFFSET,
+	       m->sdmax_rlcx_doorbell_offset);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_doorbell, SDMA0_RLC0_DOORBELL,
+			     ENABLE, 1);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL, data);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR, m->sdmax_rlcx_rb_rptr);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_HI,
+				m->sdmax_rlcx_rb_rptr_hi);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_MINOR_PTR_UPDATE, 1);
+	if (read_user_wptr(mm, wptr64, data64)) {
+		WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR,
+		       lower_32_bits(data64));
+		WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR_HI,
+		       upper_32_bits(data64));
+	} else {
+		WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR,
+		       m->sdmax_rlcx_rb_rptr);
+		WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR_HI,
+		       m->sdmax_rlcx_rb_rptr_hi);
+	}
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_MINOR_PTR_UPDATE, 0);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE, m->sdmax_rlcx_rb_base);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE_HI,
+			m->sdmax_rlcx_rb_base_hi);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_ADDR_LO,
+			m->sdmax_rlcx_rb_rptr_addr_lo);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_ADDR_HI,
+			m->sdmax_rlcx_rb_rptr_addr_hi);
+
+	data = REG_SET_FIELD(m->sdmax_rlcx_rb_cntl, SDMA0_RLC0_RB_CNTL,
+			     RB_ENABLE, 1);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL, data);
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_dump(struct kgd_dev *kgd,
+			     uint32_t engine_id, uint32_t queue_id,
+			     uint32_t (**dump)[2], uint32_t *n_regs)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t sdma_base_addr = get_sdma_base_addr(adev, engine_id, queue_id);
+	uint32_t i = 0, reg;
+#undef HQD_N_REGS
+#define HQD_N_REGS (19+6+7+10)
+
+	*dump = kmalloc_array(HQD_N_REGS * 2, sizeof(uint32_t), GFP_KERNEL);
+	if (*dump == NULL)
+		return -ENOMEM;
+
+	for (reg = mmSDMA0_RLC0_RB_CNTL; reg <= mmSDMA0_RLC0_DOORBELL; reg++)
+		DUMP_REG(sdma_base_addr + reg);
+	for (reg = mmSDMA0_RLC0_STATUS; reg <= mmSDMA0_RLC0_CSA_ADDR_HI; reg++)
+		DUMP_REG(sdma_base_addr + reg);
+	for (reg = mmSDMA0_RLC0_IB_SUB_REMAIN;
+	     reg <= mmSDMA0_RLC0_MINOR_PTR_UPDATE; reg++)
+		DUMP_REG(sdma_base_addr + reg);
+	for (reg = mmSDMA0_RLC0_MIDCMD_DATA0;
+	     reg <= mmSDMA0_RLC0_MIDCMD_CNTL; reg++)
+		DUMP_REG(sdma_base_addr + reg);
+
+	WARN_ON_ONCE(i != HQD_N_REGS);
+	*n_regs = i;
+
+	return 0;
+}
+
+static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
+				uint32_t pipe_id, uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t act;
+	bool retval = false;
+	uint32_t low, high;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+	act = RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_ACTIVE));
+	if (act) {
+		low = lower_32_bits(queue_address >> 8);
+		high = upper_32_bits(queue_address >> 8);
+
+		if (low == RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_BASE)) &&
+		   high == RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_BASE_HI)))
+			retval = true;
+	}
+	release_queue(kgd);
+	return retval;
+}
+
+static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct v9_sdma_mqd *m;
+	uint32_t sdma_base_addr;
+	uint32_t sdma_rlc_rb_cntl;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	sdma_rlc_rb_cntl = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+
+	if (sdma_rlc_rb_cntl & SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK)
+		return true;
+
+	return false;
+}
+
+static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
+				enum kfd_preempt_type reset_type,
+				unsigned int utimeout, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	enum hqd_dequeue_request_type type;
+	unsigned long end_jiffies;
+	uint32_t temp;
+	struct v9_mqd *m = get_mqd(mqd);
+
+	if (adev->in_gpu_reset)
+		return -EIO;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+
+	if (m->cp_hqd_vmid == 0)
+		WREG32_FIELD15(GC, 0, RLC_CP_SCHEDULERS, scheduler1, 0);
+
+	switch (reset_type) {
+	case KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN:
+		type = DRAIN_PIPE;
+		break;
+	case KFD_PREEMPT_TYPE_WAVEFRONT_RESET:
+		type = RESET_WAVES;
+		break;
+	default:
+		type = DRAIN_PIPE;
+		break;
+	}
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_DEQUEUE_REQUEST), type);
+
+	end_jiffies = (utimeout * HZ / 1000) + jiffies;
+	while (true) {
+		temp = RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_ACTIVE));
+		if (!(temp & CP_HQD_ACTIVE__ACTIVE_MASK))
+			break;
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("cp queue preemption time out.\n");
+			release_queue(kgd);
+			return -ETIME;
+		}
+		usleep_range(500, 1000);
+	}
+
+	release_queue(kgd);
+	return 0;
+}
+
+static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
+				unsigned int utimeout)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct v9_sdma_mqd *m;
+	uint32_t sdma_base_addr;
+	uint32_t temp;
+	unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(adev, m->sdma_engine_id,
+					    m->sdma_queue_id);
+
+	temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+	temp = temp & ~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL, temp);
+
+	while (true) {
+		temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (temp & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
+			break;
+		if (time_after(jiffies, end_jiffies))
+			return -ETIME;
+		usleep_range(500, 1000);
+	}
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL, 0);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL,
+		RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL) |
+		SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK);
+
+	m->sdmax_rlcx_rb_rptr = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR);
+	m->sdmax_rlcx_rb_rptr_hi =
+		RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_HI);
+
+	return 0;
+}
+
+static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
+							uint8_t vmid)
+{
+	uint32_t reg;
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	reg = RREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING)
+		     + vmid);
+	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
+}
+
+static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
+								uint8_t vmid)
+{
+	uint32_t reg;
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	reg = RREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING)
+		     + vmid);
+	return reg & ATC_VMID0_PASID_MAPPING__PASID_MASK;
+}
+
+static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	uint32_t req = (1 << vmid) |
+		(0 << VM_INVALIDATE_ENG16_REQ__FLUSH_TYPE__SHIFT) | /* legacy */
+		VM_INVALIDATE_ENG16_REQ__INVALIDATE_L2_PTES_MASK |
+		VM_INVALIDATE_ENG16_REQ__INVALIDATE_L2_PDE0_MASK |
+		VM_INVALIDATE_ENG16_REQ__INVALIDATE_L2_PDE1_MASK |
+		VM_INVALIDATE_ENG16_REQ__INVALIDATE_L2_PDE2_MASK |
+		VM_INVALIDATE_ENG16_REQ__INVALIDATE_L1_PTES_MASK;
+
+	mutex_lock(&adev->srbm_mutex);
+
+	/* Use legacy mode tlb invalidation.
+	 *
+	 * Currently on Raven the code below is broken for anything but
+	 * legacy mode due to a MMHUB power gating problem. A workaround
+	 * is for MMHUB to wait until the condition PER_VMID_INVALIDATE_REQ
+	 * == PER_VMID_INVALIDATE_ACK instead of simply waiting for the ack
+	 * bit.
+	 *
+	 * TODO 1: agree on the right set of invalidation registers for
+	 * KFD use. Use the last one for now. Invalidate both GC and
+	 * MMHUB.
+	 *
+	 * TODO 2: support range-based invalidation, requires kfg2kgd
+	 * interface change
+	 */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG16_ADDR_RANGE_LO32),
+				0xffffffff);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG16_ADDR_RANGE_HI32),
+				0x0000001f);
+
+	WREG32(SOC15_REG_OFFSET(MMHUB, 0,
+				mmMMHUB_VM_INVALIDATE_ENG16_ADDR_RANGE_LO32),
+				0xffffffff);
+	WREG32(SOC15_REG_OFFSET(MMHUB, 0,
+				mmMMHUB_VM_INVALIDATE_ENG16_ADDR_RANGE_HI32),
+				0x0000001f);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG16_REQ), req);
+
+	WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_INVALIDATE_ENG16_REQ),
+				req);
+
+	while (!(RREG32(SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG16_ACK)) &
+					(1 << vmid)))
+		cpu_relax();
+
+	while (!(RREG32(SOC15_REG_OFFSET(MMHUB, 0,
+					mmMMHUB_VM_INVALIDATE_ENG16_ACK)) &
+					(1 << vmid)))
+		cpu_relax();
+
+	mutex_unlock(&adev->srbm_mutex);
+
+}
+
+static int invalidate_tlbs_with_kiq(struct amdgpu_device *adev, uint16_t pasid)
+{
+	signed long r;
+	uint32_t seq;
+	struct amdgpu_ring *ring = &adev->gfx.kiq.ring;
+
+	spin_lock(&adev->gfx.kiq.ring_lock);
+	amdgpu_ring_alloc(ring, 12); /* fence + invalidate_tlbs package*/
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INVALIDATE_TLBS, 0));
+	amdgpu_ring_write(ring,
+			PACKET3_INVALIDATE_TLBS_DST_SEL(1) |
+			PACKET3_INVALIDATE_TLBS_ALL_HUB(1) |
+			PACKET3_INVALIDATE_TLBS_PASID(pasid) |
+			PACKET3_INVALIDATE_TLBS_FLUSH_TYPE(0)); /* legacy */
+	amdgpu_fence_emit_polling(ring, &seq);
+	amdgpu_ring_commit(ring);
+	spin_unlock(&adev->gfx.kiq.ring_lock);
+
+	r = amdgpu_fence_wait_polling(ring, seq, adev->usec_timeout);
+	if (r < 1) {
+		DRM_ERROR("wait for kiq fence error: %ld.\n", r);
+		return -ETIME;
+	}
+
+	return 0;
+}
+
+static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	int vmid;
+	struct amdgpu_ring *ring = &adev->gfx.kiq.ring;
+
+	if (adev->in_gpu_reset)
+		return -EIO;
+
+	if (ring->ready)
+		return invalidate_tlbs_with_kiq(adev, pasid);
+
+	for (vmid = 0; vmid < 16; vmid++) {
+		if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid))
+			continue;
+		if (get_atc_vmid_pasid_mapping_valid(kgd, vmid)) {
+			if (get_atc_vmid_pasid_mapping_pasid(kgd, vmid)
+				== pasid) {
+				write_vmid_invalidate_request(kgd, vmid);
+				break;
+			}
+		}
+	}
+
+	return 0;
+}
+
+static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("non kfd vmid %d\n", vmid);
+		return 0;
+	}
+
+	write_vmid_invalidate_request(kgd, vmid);
+	return 0;
+}
+
+static int kgd_address_watch_disable(struct kgd_dev *kgd)
+{
+	return 0;
+}
+
+static int kgd_address_watch_execute(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					uint32_t cntl_val,
+					uint32_t addr_hi,
+					uint32_t addr_lo)
+{
+	return 0;
+}
+
+static int kgd_wave_control_execute(struct kgd_dev *kgd,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t data = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_INDEX), gfx_index_val);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_CMD), sq_cmd);
+
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		INSTANCE_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SH_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SE_BROADCAST_WRITES, 1);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_INDEX), data);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					unsigned int reg_offset)
+{
+	return 0;
+}
+
+static void set_scratch_backing_va(struct kgd_dev *kgd,
+					uint64_t va, uint32_t vmid)
+{
+	/* No longer needed on GFXv9. The scratch base address is
+	 * passed to the shader by the CP. It's the user mode driver's
+	 * responsibility.
+	 */
+}
+
+/* FIXME: Does this need to be ASIC-specific code? */
+static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	const union amdgpu_firmware_header *hdr;
+
+	switch (type) {
+	case KGD_ENGINE_PFP:
+		hdr = (const union amdgpu_firmware_header *)adev->gfx.pfp_fw->data;
+		break;
+
+	case KGD_ENGINE_ME:
+		hdr = (const union amdgpu_firmware_header *)adev->gfx.me_fw->data;
+		break;
+
+	case KGD_ENGINE_CE:
+		hdr = (const union amdgpu_firmware_header *)adev->gfx.ce_fw->data;
+		break;
+
+	case KGD_ENGINE_MEC1:
+		hdr = (const union amdgpu_firmware_header *)adev->gfx.mec_fw->data;
+		break;
+
+	case KGD_ENGINE_MEC2:
+		hdr = (const union amdgpu_firmware_header *)adev->gfx.mec2_fw->data;
+		break;
+
+	case KGD_ENGINE_RLC:
+		hdr = (const union amdgpu_firmware_header *)adev->gfx.rlc_fw->data;
+		break;
+
+	case KGD_ENGINE_SDMA1:
+		hdr = (const union amdgpu_firmware_header *)adev->sdma.instance[0].fw->data;
+		break;
+
+	case KGD_ENGINE_SDMA2:
+		hdr = (const union amdgpu_firmware_header *)adev->sdma.instance[1].fw->data;
+		break;
+
+	default:
+		return 0;
+	}
+
+	if (hdr == NULL)
+		return 0;
+
+	/* Only 12 bit in use*/
+	return hdr->common.ucode_version;
+}
+
+static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t page_table_base)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint64_t base = (uint64_t)page_table_base << PAGE_SHIFT |
+		AMDGPU_PTE_VALID;
+
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("trying to set page table base for wrong VMID %u\n",
+		       vmid);
+		return;
+	}
+
+	/* TODO: take advantage of per-process address space size. For
+	 * now, all processes share the same address space size, like
+	 * on GFX8 and older.
+	 */
+	WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32) + (vmid*2), 0);
+	WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32) + (vmid*2), 0);
+
+	WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32) + (vmid*2),
+			lower_32_bits(adev->vm_manager.max_pfn - 1));
+	WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32) + (vmid*2),
+			upper_32_bits(adev->vm_manager.max_pfn - 1));
+
+	WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32) + (vmid*2), lower_32_bits(base));
+	WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32) + (vmid*2), upper_32_bits(base));
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32) + (vmid*2), 0);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32) + (vmid*2), 0);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32) + (vmid*2),
+			lower_32_bits(adev->vm_manager.max_pfn - 1));
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32) + (vmid*2),
+			upper_32_bits(adev->vm_manager.max_pfn - 1));
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32) + (vmid*2), lower_32_bits(base));
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32) + (vmid*2), upper_32_bits(base));
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
new file mode 100644
index 00000000000..a5905a9b0d2
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -0,0 +1,2114 @@
+/*
+ * Copyright 2014-2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#define pr_fmt(fmt) "kfd2kgd: " fmt
+
+#include <linux/list.h>
+#include <linux/pagemap.h>
+#include <linux/sched/mm.h>
+#include <drm/drmP.h>
+#include "amdgpu_object.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_amdkfd.h"
+
+/* Special VM and GART address alignment needed for VI pre-Fiji due to
+ * a HW bug.
+ */
+#define VI_BO_SIZE_ALIGN (0x8000)
+
+/* BO flag to indicate a KFD userptr BO */
+#define AMDGPU_AMDKFD_USERPTR_BO (1ULL << 63)
+
+/* Userptr restore delay, just long enough to allow consecutive VM
+ * changes to accumulate
+ */
+#define AMDGPU_USERPTR_RESTORE_DELAY_MS 1
+
+/* Impose limit on how much memory KFD can use */
+static struct {
+	uint64_t max_system_mem_limit;
+	uint64_t max_userptr_mem_limit;
+	int64_t system_mem_used;
+	int64_t userptr_mem_used;
+	spinlock_t mem_limit_lock;
+} kfd_mem_limit;
+
+/* Struct used for amdgpu_amdkfd_bo_validate */
+struct amdgpu_vm_parser {
+	uint32_t        domain;
+	bool            wait;
+};
+
+static const char * const domain_bit_to_string[] = {
+		"CPU",
+		"GTT",
+		"VRAM",
+		"GDS",
+		"GWS",
+		"OA"
+};
+
+#define domain_string(domain) domain_bit_to_string[ffs(domain)-1]
+
+static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work);
+
+
+static inline struct amdgpu_device *get_amdgpu_device(struct kgd_dev *kgd)
+{
+	return (struct amdgpu_device *)kgd;
+}
+
+static bool check_if_add_bo_to_vm(struct amdgpu_vm *avm,
+		struct kgd_mem *mem)
+{
+	struct kfd_bo_va_list *entry;
+
+	list_for_each_entry(entry, &mem->bo_va_list, bo_list)
+		if (entry->bo_va->base.vm == avm)
+			return false;
+
+	return true;
+}
+
+/* Set memory usage limits. Current, limits are
+ *  System (kernel) memory - 3/8th System RAM
+ *  Userptr memory - 3/4th System RAM
+ */
+void amdgpu_amdkfd_gpuvm_init_mem_limits(void)
+{
+	struct sysinfo si;
+	uint64_t mem;
+
+	si_meminfo(&si);
+	mem = si.totalram - si.totalhigh;
+	mem *= si.mem_unit;
+
+	mtx_init(&kfd_mem_limit.mem_limit_lock, IPL_TTY);
+	kfd_mem_limit.max_system_mem_limit = (mem >> 1) - (mem >> 3);
+	kfd_mem_limit.max_userptr_mem_limit = mem - (mem >> 2);
+	pr_debug("Kernel memory limit %lluM, userptr limit %lluM\n",
+		(kfd_mem_limit.max_system_mem_limit >> 20),
+		(kfd_mem_limit.max_userptr_mem_limit >> 20));
+}
+
+static int amdgpu_amdkfd_reserve_system_mem_limit(struct amdgpu_device *adev,
+					      uint64_t size, u32 domain)
+{
+	size_t acc_size;
+	int ret = 0;
+
+	acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size,
+				       sizeof(struct amdgpu_bo));
+
+	spin_lock(&kfd_mem_limit.mem_limit_lock);
+	if (domain == AMDGPU_GEM_DOMAIN_GTT) {
+		if (kfd_mem_limit.system_mem_used + (acc_size + size) >
+			kfd_mem_limit.max_system_mem_limit) {
+			ret = -ENOMEM;
+			goto err_no_mem;
+		}
+		kfd_mem_limit.system_mem_used += (acc_size + size);
+	} else if (domain == AMDGPU_GEM_DOMAIN_CPU) {
+		if ((kfd_mem_limit.system_mem_used + acc_size >
+			kfd_mem_limit.max_system_mem_limit) ||
+			(kfd_mem_limit.userptr_mem_used + (size + acc_size) >
+			kfd_mem_limit.max_userptr_mem_limit)) {
+			ret = -ENOMEM;
+			goto err_no_mem;
+		}
+		kfd_mem_limit.system_mem_used += acc_size;
+		kfd_mem_limit.userptr_mem_used += size;
+	}
+err_no_mem:
+	spin_unlock(&kfd_mem_limit.mem_limit_lock);
+	return ret;
+}
+
+static void unreserve_system_mem_limit(struct amdgpu_device *adev,
+				       uint64_t size, u32 domain)
+{
+	size_t acc_size;
+
+	acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size,
+				       sizeof(struct amdgpu_bo));
+
+	spin_lock(&kfd_mem_limit.mem_limit_lock);
+	if (domain == AMDGPU_GEM_DOMAIN_GTT) {
+		kfd_mem_limit.system_mem_used -= (acc_size + size);
+	} else if (domain == AMDGPU_GEM_DOMAIN_CPU) {
+		kfd_mem_limit.system_mem_used -= acc_size;
+		kfd_mem_limit.userptr_mem_used -= size;
+	}
+	WARN_ONCE(kfd_mem_limit.system_mem_used < 0,
+		  "kfd system memory accounting unbalanced");
+	WARN_ONCE(kfd_mem_limit.userptr_mem_used < 0,
+		  "kfd userptr memory accounting unbalanced");
+
+	spin_unlock(&kfd_mem_limit.mem_limit_lock);
+}
+
+void amdgpu_amdkfd_unreserve_system_memory_limit(struct amdgpu_bo *bo)
+{
+	spin_lock(&kfd_mem_limit.mem_limit_lock);
+
+	if (bo->flags & AMDGPU_AMDKFD_USERPTR_BO) {
+		kfd_mem_limit.system_mem_used -= bo->tbo.acc_size;
+		kfd_mem_limit.userptr_mem_used -= amdgpu_bo_size(bo);
+	} else if (bo->preferred_domains == AMDGPU_GEM_DOMAIN_GTT) {
+		kfd_mem_limit.system_mem_used -=
+			(bo->tbo.acc_size + amdgpu_bo_size(bo));
+	}
+	WARN_ONCE(kfd_mem_limit.system_mem_used < 0,
+		  "kfd system memory accounting unbalanced");
+	WARN_ONCE(kfd_mem_limit.userptr_mem_used < 0,
+		  "kfd userptr memory accounting unbalanced");
+
+	spin_unlock(&kfd_mem_limit.mem_limit_lock);
+}
+
+
+/* amdgpu_amdkfd_remove_eviction_fence - Removes eviction fence(s) from BO's
+ *  reservation object.
+ *
+ * @bo: [IN] Remove eviction fence(s) from this BO
+ * @ef: [IN] If ef is specified, then this eviction fence is removed if it
+ *  is present in the shared list.
+ * @ef_list: [OUT] Returns list of eviction fences. These fences are removed
+ *  from BO's reservation object shared list.
+ * @ef_count: [OUT] Number of fences in ef_list.
+ *
+ * NOTE: If called with ef_list, then amdgpu_amdkfd_add_eviction_fence must be
+ *  called to restore the eviction fences and to avoid memory leak. This is
+ *  useful for shared BOs.
+ * NOTE: Must be called with BO reserved i.e. bo->tbo.resv->lock held.
+ */
+static int amdgpu_amdkfd_remove_eviction_fence(struct amdgpu_bo *bo,
+					struct amdgpu_amdkfd_fence *ef,
+					struct amdgpu_amdkfd_fence ***ef_list,
+					unsigned int *ef_count)
+{
+	struct reservation_object *resv = bo->tbo.resv;
+	struct reservation_object_list *old, *new;
+	unsigned int i, j, k;
+
+	if (!ef && !ef_list)
+		return -EINVAL;
+
+	if (ef_list) {
+		*ef_list = NULL;
+		*ef_count = 0;
+	}
+
+	old = reservation_object_get_list(resv);
+	if (!old)
+		return 0;
+
+	new = kmalloc(offsetof(typeof(*new), shared[old->shared_max]),
+		      GFP_KERNEL);
+	if (!new)
+		return -ENOMEM;
+
+	/* Go through all the shared fences in the resevation object and sort
+	 * the interesting ones to the end of the list.
+	 */
+	for (i = 0, j = old->shared_count, k = 0; i < old->shared_count; ++i) {
+		struct dma_fence *f;
+
+		f = rcu_dereference_protected(old->shared[i],
+					      reservation_object_held(resv));
+
+		if ((ef && f->context == ef->base.context) ||
+		    (!ef && to_amdgpu_amdkfd_fence(f)))
+			RCU_INIT_POINTER(new->shared[--j], f);
+		else
+			RCU_INIT_POINTER(new->shared[k++], f);
+	}
+	new->shared_max = old->shared_max;
+	new->shared_count = k;
+
+	if (!ef) {
+		unsigned int count = old->shared_count - j;
+
+		/* Alloc memory for count number of eviction fence pointers.
+		 * Fill the ef_list array and ef_count
+		 */
+		*ef_list = kcalloc(count, sizeof(**ef_list), GFP_KERNEL);
+		*ef_count = count;
+
+		if (!*ef_list) {
+			kfree(new);
+			return -ENOMEM;
+		}
+	}
+
+	/* Install the new fence list, seqcount provides the barriers */
+	preempt_disable();
+	write_seqcount_begin(&resv->seq);
+	RCU_INIT_POINTER(resv->fence, new);
+	write_seqcount_end(&resv->seq);
+	preempt_enable();
+
+	/* Drop the references to the removed fences or move them to ef_list */
+	for (i = j, k = 0; i < old->shared_count; ++i) {
+		struct dma_fence *f;
+
+		f = rcu_dereference_protected(new->shared[i],
+					      reservation_object_held(resv));
+		if (!ef)
+			(*ef_list)[k++] = to_amdgpu_amdkfd_fence(f);
+		else
+			dma_fence_put(f);
+	}
+	kfree_rcu(old, rcu);
+
+	return 0;
+}
+
+/* amdgpu_amdkfd_add_eviction_fence - Adds eviction fence(s) back into BO's
+ *  reservation object.
+ *
+ * @bo: [IN] Add eviction fences to this BO
+ * @ef_list: [IN] List of eviction fences to be added
+ * @ef_count: [IN] Number of fences in ef_list.
+ *
+ * NOTE: Must call amdgpu_amdkfd_remove_eviction_fence before calling this
+ *  function.
+ */
+static void amdgpu_amdkfd_add_eviction_fence(struct amdgpu_bo *bo,
+				struct amdgpu_amdkfd_fence **ef_list,
+				unsigned int ef_count)
+{
+	int i;
+
+	if (!ef_list || !ef_count)
+		return;
+
+	for (i = 0; i < ef_count; i++) {
+		amdgpu_bo_fence(bo, &ef_list[i]->base, true);
+		/* Re-adding the fence takes an additional reference. Drop that
+		 * reference.
+		 */
+		dma_fence_put(&ef_list[i]->base);
+	}
+
+	kfree(ef_list);
+}
+
+static int amdgpu_amdkfd_bo_validate(struct amdgpu_bo *bo, uint32_t domain,
+				     bool wait)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+	int ret;
+
+	if (WARN(amdgpu_ttm_tt_get_usermm(bo->tbo.ttm),
+		 "Called with userptr BO"))
+		return -EINVAL;
+
+	amdgpu_bo_placement_from_domain(bo, domain);
+
+	ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (ret)
+		goto validate_fail;
+	if (wait) {
+		struct amdgpu_amdkfd_fence **ef_list;
+		unsigned int ef_count;
+
+		ret = amdgpu_amdkfd_remove_eviction_fence(bo, NULL, &ef_list,
+							  &ef_count);
+		if (ret)
+			goto validate_fail;
+
+		ttm_bo_wait(&bo->tbo, false, false);
+		amdgpu_amdkfd_add_eviction_fence(bo, ef_list, ef_count);
+	}
+
+validate_fail:
+	return ret;
+}
+
+static int amdgpu_amdkfd_validate(void *param, struct amdgpu_bo *bo)
+{
+	struct amdgpu_vm_parser *p = param;
+
+	return amdgpu_amdkfd_bo_validate(bo, p->domain, p->wait);
+}
+
+/* vm_validate_pt_pd_bos - Validate page table and directory BOs
+ *
+ * Page directories are not updated here because huge page handling
+ * during page table updates can invalidate page directory entries
+ * again. Page directories are only updated after updating page
+ * tables.
+ */
+static int vm_validate_pt_pd_bos(struct amdgpu_vm *vm)
+{
+	struct amdgpu_bo *pd = vm->root.base.bo;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(pd->tbo.bdev);
+	struct amdgpu_vm_parser param;
+	uint64_t addr, flags = AMDGPU_PTE_VALID;
+	int ret;
+
+	param.domain = AMDGPU_GEM_DOMAIN_VRAM;
+	param.wait = false;
+
+	ret = amdgpu_vm_validate_pt_bos(adev, vm, amdgpu_amdkfd_validate,
+					&param);
+	if (ret) {
+		pr_err("amdgpu: failed to validate PT BOs\n");
+		return ret;
+	}
+
+	ret = amdgpu_amdkfd_validate(&param, pd);
+	if (ret) {
+		pr_err("amdgpu: failed to validate PD\n");
+		return ret;
+	}
+
+	addr = amdgpu_bo_gpu_offset(vm->root.base.bo);
+	amdgpu_gmc_get_vm_pde(adev, -1, &addr, &flags);
+	vm->pd_phys_addr = addr;
+
+	if (vm->use_cpu_for_update) {
+		ret = amdgpu_bo_kmap(pd, NULL);
+		if (ret) {
+			pr_err("amdgpu: failed to kmap PD, ret=%d\n", ret);
+			return ret;
+		}
+	}
+
+	return 0;
+}
+
+static int sync_vm_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync,
+			 struct dma_fence *f)
+{
+	int ret = amdgpu_sync_fence(adev, sync, f, false);
+
+	/* Sync objects can't handle multiple GPUs (contexts) updating
+	 * sync->last_vm_update. Fortunately we don't need it for
+	 * KFD's purposes, so we can just drop that fence.
+	 */
+	if (sync->last_vm_update) {
+		dma_fence_put(sync->last_vm_update);
+		sync->last_vm_update = NULL;
+	}
+
+	return ret;
+}
+
+static int vm_update_pds(struct amdgpu_vm *vm, struct amdgpu_sync *sync)
+{
+	struct amdgpu_bo *pd = vm->root.base.bo;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(pd->tbo.bdev);
+	int ret;
+
+	ret = amdgpu_vm_update_directories(adev, vm);
+	if (ret)
+		return ret;
+
+	return sync_vm_fence(adev, sync, vm->last_update);
+}
+
+/* add_bo_to_vm - Add a BO to a VM
+ *
+ * Everything that needs to bo done only once when a BO is first added
+ * to a VM. It can later be mapped and unmapped many times without
+ * repeating these steps.
+ *
+ * 1. Allocate and initialize BO VA entry data structure
+ * 2. Add BO to the VM
+ * 3. Determine ASIC-specific PTE flags
+ * 4. Alloc page tables and directories if needed
+ * 4a.  Validate new page tables and directories
+ */
+static int add_bo_to_vm(struct amdgpu_device *adev, struct kgd_mem *mem,
+		struct amdgpu_vm *vm, bool is_aql,
+		struct kfd_bo_va_list **p_bo_va_entry)
+{
+	int ret;
+	struct kfd_bo_va_list *bo_va_entry;
+	struct amdgpu_bo *pd = vm->root.base.bo;
+	struct amdgpu_bo *bo = mem->bo;
+	uint64_t va = mem->va;
+	struct list_head *list_bo_va = &mem->bo_va_list;
+	unsigned long bo_size = bo->tbo.mem.size;
+
+	if (!va) {
+		pr_err("Invalid VA when adding BO to VM\n");
+		return -EINVAL;
+	}
+
+	if (is_aql)
+		va += bo_size;
+
+	bo_va_entry = kzalloc(sizeof(*bo_va_entry), GFP_KERNEL);
+	if (!bo_va_entry)
+		return -ENOMEM;
+
+	pr_debug("\t add VA 0x%llx - 0x%llx to vm %p\n", va,
+			va + bo_size, vm);
+
+	/* Add BO to VM internal data structures*/
+	bo_va_entry->bo_va = amdgpu_vm_bo_add(adev, vm, bo);
+	if (!bo_va_entry->bo_va) {
+		ret = -EINVAL;
+		pr_err("Failed to add BO object to VM. ret == %d\n",
+				ret);
+		goto err_vmadd;
+	}
+
+	bo_va_entry->va = va;
+	bo_va_entry->pte_flags = amdgpu_gmc_get_pte_flags(adev,
+							 mem->mapping_flags);
+	bo_va_entry->kgd_dev = (void *)adev;
+	list_add(&bo_va_entry->bo_list, list_bo_va);
+
+	if (p_bo_va_entry)
+		*p_bo_va_entry = bo_va_entry;
+
+	/* Allocate new page tables if needed and validate
+	 * them. Clearing of new page tables and validate need to wait
+	 * on move fences. We don't want that to trigger the eviction
+	 * fence, so remove it temporarily.
+	 */
+	amdgpu_amdkfd_remove_eviction_fence(pd,
+					vm->process_info->eviction_fence,
+					NULL, NULL);
+
+	ret = amdgpu_vm_alloc_pts(adev, vm, va, amdgpu_bo_size(bo));
+	if (ret) {
+		pr_err("Failed to allocate pts, err=%d\n", ret);
+		goto err_alloc_pts;
+	}
+
+	ret = vm_validate_pt_pd_bos(vm);
+	if (ret) {
+		pr_err("validate_pt_pd_bos() failed\n");
+		goto err_alloc_pts;
+	}
+
+	/* Add the eviction fence back */
+	amdgpu_bo_fence(pd, &vm->process_info->eviction_fence->base, true);
+
+	return 0;
+
+err_alloc_pts:
+	amdgpu_bo_fence(pd, &vm->process_info->eviction_fence->base, true);
+	amdgpu_vm_bo_rmv(adev, bo_va_entry->bo_va);
+	list_del(&bo_va_entry->bo_list);
+err_vmadd:
+	kfree(bo_va_entry);
+	return ret;
+}
+
+static void remove_bo_from_vm(struct amdgpu_device *adev,
+		struct kfd_bo_va_list *entry, unsigned long size)
+{
+	pr_debug("\t remove VA 0x%llx - 0x%llx in entry %p\n",
+			entry->va,
+			entry->va + size, entry);
+	amdgpu_vm_bo_rmv(adev, entry->bo_va);
+	list_del(&entry->bo_list);
+	kfree(entry);
+}
+
+static void add_kgd_mem_to_kfd_bo_list(struct kgd_mem *mem,
+				struct amdkfd_process_info *process_info,
+				bool userptr)
+{
+	struct ttm_validate_buffer *entry = &mem->validate_list;
+	struct amdgpu_bo *bo = mem->bo;
+
+	INIT_LIST_HEAD(&entry->head);
+	entry->shared = true;
+	entry->bo = &bo->tbo;
+	mutex_lock(&process_info->lock);
+	if (userptr)
+		list_add_tail(&entry->head, &process_info->userptr_valid_list);
+	else
+		list_add_tail(&entry->head, &process_info->kfd_bo_list);
+	mutex_unlock(&process_info->lock);
+}
+
+/* Initializes user pages. It registers the MMU notifier and validates
+ * the userptr BO in the GTT domain.
+ *
+ * The BO must already be on the userptr_valid_list. Otherwise an
+ * eviction and restore may happen that leaves the new BO unmapped
+ * with the user mode queues running.
+ *
+ * Takes the process_info->lock to protect against concurrent restore
+ * workers.
+ *
+ * Returns 0 for success, negative errno for errors.
+ */
+static int init_user_pages(struct kgd_mem *mem, struct mm_struct *mm,
+			   uint64_t user_addr)
+{
+	struct amdkfd_process_info *process_info = mem->process_info;
+	struct amdgpu_bo *bo = mem->bo;
+	struct ttm_operation_ctx ctx = { true, false };
+	int ret = 0;
+
+	mutex_lock(&process_info->lock);
+
+	ret = amdgpu_ttm_tt_set_userptr(bo->tbo.ttm, user_addr, 0);
+	if (ret) {
+		pr_err("%s: Failed to set userptr: %d\n", __func__, ret);
+		goto out;
+	}
+
+	ret = amdgpu_mn_register(bo, user_addr);
+	if (ret) {
+		pr_err("%s: Failed to register MMU notifier: %d\n",
+		       __func__, ret);
+		goto out;
+	}
+
+	/* If no restore worker is running concurrently, user_pages
+	 * should not be allocated
+	 */
+	WARN(mem->user_pages, "Leaking user_pages array");
+
+	mem->user_pages = kvmalloc_array(bo->tbo.ttm->num_pages,
+					   sizeof(struct vm_page *),
+					   GFP_KERNEL | __GFP_ZERO);
+	if (!mem->user_pages) {
+		pr_err("%s: Failed to allocate pages array\n", __func__);
+		ret = -ENOMEM;
+		goto unregister_out;
+	}
+
+	ret = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm, mem->user_pages);
+	if (ret) {
+		pr_err("%s: Failed to get user pages: %d\n", __func__, ret);
+		goto free_out;
+	}
+
+	amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm, mem->user_pages);
+
+	ret = amdgpu_bo_reserve(bo, true);
+	if (ret) {
+		pr_err("%s: Failed to reserve BO\n", __func__);
+		goto release_out;
+	}
+	amdgpu_bo_placement_from_domain(bo, mem->domain);
+	ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (ret)
+		pr_err("%s: failed to validate BO\n", __func__);
+	amdgpu_bo_unreserve(bo);
+
+release_out:
+	if (ret)
+		release_pages(mem->user_pages, bo->tbo.ttm->num_pages);
+free_out:
+	kvfree(mem->user_pages);
+	mem->user_pages = NULL;
+unregister_out:
+	if (ret)
+		amdgpu_mn_unregister(bo);
+out:
+	mutex_unlock(&process_info->lock);
+	return ret;
+}
+
+/* Reserving a BO and its page table BOs must happen atomically to
+ * avoid deadlocks. Some operations update multiple VMs at once. Track
+ * all the reservation info in a context structure. Optionally a sync
+ * object can track VM updates.
+ */
+struct bo_vm_reservation_context {
+	struct amdgpu_bo_list_entry kfd_bo; /* BO list entry for the KFD BO */
+	unsigned int n_vms;		    /* Number of VMs reserved	    */
+	struct amdgpu_bo_list_entry *vm_pd; /* Array of VM BO list entries  */
+	struct ww_acquire_ctx ticket;	    /* Reservation ticket	    */
+	struct list_head list, duplicates;  /* BO lists			    */
+	struct amdgpu_sync *sync;	    /* Pointer to sync object	    */
+	bool reserved;			    /* Whether BOs are reserved	    */
+};
+
+enum bo_vm_match {
+	BO_VM_NOT_MAPPED = 0,	/* Match VMs where a BO is not mapped */
+	BO_VM_MAPPED,		/* Match VMs where a BO is mapped     */
+	BO_VM_ALL,		/* Match all VMs a BO was added to    */
+};
+
+/**
+ * reserve_bo_and_vm - reserve a BO and a VM unconditionally.
+ * @mem: KFD BO structure.
+ * @vm: the VM to reserve.
+ * @ctx: the struct that will be used in unreserve_bo_and_vms().
+ */
+static int reserve_bo_and_vm(struct kgd_mem *mem,
+			      struct amdgpu_vm *vm,
+			      struct bo_vm_reservation_context *ctx)
+{
+	struct amdgpu_bo *bo = mem->bo;
+	int ret;
+
+	WARN_ON(!vm);
+
+	ctx->reserved = false;
+	ctx->n_vms = 1;
+	ctx->sync = &mem->sync;
+
+	INIT_LIST_HEAD(&ctx->list);
+	INIT_LIST_HEAD(&ctx->duplicates);
+
+	ctx->vm_pd = kcalloc(ctx->n_vms, sizeof(*ctx->vm_pd), GFP_KERNEL);
+	if (!ctx->vm_pd)
+		return -ENOMEM;
+
+	ctx->kfd_bo.robj = bo;
+	ctx->kfd_bo.priority = 0;
+	ctx->kfd_bo.tv.bo = &bo->tbo;
+	ctx->kfd_bo.tv.shared = true;
+	ctx->kfd_bo.user_pages = NULL;
+	list_add(&ctx->kfd_bo.tv.head, &ctx->list);
+
+	amdgpu_vm_get_pd_bo(vm, &ctx->list, &ctx->vm_pd[0]);
+
+	ret = ttm_eu_reserve_buffers(&ctx->ticket, &ctx->list,
+				     false, &ctx->duplicates);
+	if (!ret)
+		ctx->reserved = true;
+	else {
+		pr_err("Failed to reserve buffers in ttm\n");
+		kfree(ctx->vm_pd);
+		ctx->vm_pd = NULL;
+	}
+
+	return ret;
+}
+
+/**
+ * reserve_bo_and_cond_vms - reserve a BO and some VMs conditionally
+ * @mem: KFD BO structure.
+ * @vm: the VM to reserve. If NULL, then all VMs associated with the BO
+ * is used. Otherwise, a single VM associated with the BO.
+ * @map_type: the mapping status that will be used to filter the VMs.
+ * @ctx: the struct that will be used in unreserve_bo_and_vms().
+ *
+ * Returns 0 for success, negative for failure.
+ */
+static int reserve_bo_and_cond_vms(struct kgd_mem *mem,
+				struct amdgpu_vm *vm, enum bo_vm_match map_type,
+				struct bo_vm_reservation_context *ctx)
+{
+	struct amdgpu_bo *bo = mem->bo;
+	struct kfd_bo_va_list *entry;
+	unsigned int i;
+	int ret;
+
+	ctx->reserved = false;
+	ctx->n_vms = 0;
+	ctx->vm_pd = NULL;
+	ctx->sync = &mem->sync;
+
+	INIT_LIST_HEAD(&ctx->list);
+	INIT_LIST_HEAD(&ctx->duplicates);
+
+	list_for_each_entry(entry, &mem->bo_va_list, bo_list) {
+		if ((vm && vm != entry->bo_va->base.vm) ||
+			(entry->is_mapped != map_type
+			&& map_type != BO_VM_ALL))
+			continue;
+
+		ctx->n_vms++;
+	}
+
+	if (ctx->n_vms != 0) {
+		ctx->vm_pd = kcalloc(ctx->n_vms, sizeof(*ctx->vm_pd),
+				     GFP_KERNEL);
+		if (!ctx->vm_pd)
+			return -ENOMEM;
+	}
+
+	ctx->kfd_bo.robj = bo;
+	ctx->kfd_bo.priority = 0;
+	ctx->kfd_bo.tv.bo = &bo->tbo;
+	ctx->kfd_bo.tv.shared = true;
+	ctx->kfd_bo.user_pages = NULL;
+	list_add(&ctx->kfd_bo.tv.head, &ctx->list);
+
+	i = 0;
+	list_for_each_entry(entry, &mem->bo_va_list, bo_list) {
+		if ((vm && vm != entry->bo_va->base.vm) ||
+			(entry->is_mapped != map_type
+			&& map_type != BO_VM_ALL))
+			continue;
+
+		amdgpu_vm_get_pd_bo(entry->bo_va->base.vm, &ctx->list,
+				&ctx->vm_pd[i]);
+		i++;
+	}
+
+	ret = ttm_eu_reserve_buffers(&ctx->ticket, &ctx->list,
+				     false, &ctx->duplicates);
+	if (!ret)
+		ctx->reserved = true;
+	else
+		pr_err("Failed to reserve buffers in ttm.\n");
+
+	if (ret) {
+		kfree(ctx->vm_pd);
+		ctx->vm_pd = NULL;
+	}
+
+	return ret;
+}
+
+/**
+ * unreserve_bo_and_vms - Unreserve BO and VMs from a reservation context
+ * @ctx: Reservation context to unreserve
+ * @wait: Optionally wait for a sync object representing pending VM updates
+ * @intr: Whether the wait is interruptible
+ *
+ * Also frees any resources allocated in
+ * reserve_bo_and_(cond_)vm(s). Returns the status from
+ * amdgpu_sync_wait.
+ */
+static int unreserve_bo_and_vms(struct bo_vm_reservation_context *ctx,
+				 bool wait, bool intr)
+{
+	int ret = 0;
+
+	if (wait)
+		ret = amdgpu_sync_wait(ctx->sync, intr);
+
+	if (ctx->reserved)
+		ttm_eu_backoff_reservation(&ctx->ticket, &ctx->list);
+	kfree(ctx->vm_pd);
+
+	ctx->sync = NULL;
+
+	ctx->reserved = false;
+	ctx->vm_pd = NULL;
+
+	return ret;
+}
+
+static int unmap_bo_from_gpuvm(struct amdgpu_device *adev,
+				struct kfd_bo_va_list *entry,
+				struct amdgpu_sync *sync)
+{
+	struct amdgpu_bo_va *bo_va = entry->bo_va;
+	struct amdgpu_vm *vm = bo_va->base.vm;
+	struct amdgpu_bo *pd = vm->root.base.bo;
+
+	/* Remove eviction fence from PD (and thereby from PTs too as
+	 * they share the resv. object). Otherwise during PT update
+	 * job (see amdgpu_vm_bo_update_mapping), eviction fence would
+	 * get added to job->sync object and job execution would
+	 * trigger the eviction fence.
+	 */
+	amdgpu_amdkfd_remove_eviction_fence(pd,
+					    vm->process_info->eviction_fence,
+					    NULL, NULL);
+	amdgpu_vm_bo_unmap(adev, bo_va, entry->va);
+
+	amdgpu_vm_clear_freed(adev, vm, &bo_va->last_pt_update);
+
+	/* Add the eviction fence back */
+	amdgpu_bo_fence(pd, &vm->process_info->eviction_fence->base, true);
+
+	sync_vm_fence(adev, sync, bo_va->last_pt_update);
+
+	return 0;
+}
+
+static int update_gpuvm_pte(struct amdgpu_device *adev,
+		struct kfd_bo_va_list *entry,
+		struct amdgpu_sync *sync)
+{
+	int ret;
+	struct amdgpu_vm *vm;
+	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_bo *bo;
+
+	bo_va = entry->bo_va;
+	vm = bo_va->base.vm;
+	bo = bo_va->base.bo;
+
+	/* Update the page tables  */
+	ret = amdgpu_vm_bo_update(adev, bo_va, false);
+	if (ret) {
+		pr_err("amdgpu_vm_bo_update failed\n");
+		return ret;
+	}
+
+	return sync_vm_fence(adev, sync, bo_va->last_pt_update);
+}
+
+static int map_bo_to_gpuvm(struct amdgpu_device *adev,
+		struct kfd_bo_va_list *entry, struct amdgpu_sync *sync,
+		bool no_update_pte)
+{
+	int ret;
+
+	/* Set virtual address for the allocation */
+	ret = amdgpu_vm_bo_map(adev, entry->bo_va, entry->va, 0,
+			       amdgpu_bo_size(entry->bo_va->base.bo),
+			       entry->pte_flags);
+	if (ret) {
+		pr_err("Failed to map VA 0x%llx in vm. ret %d\n",
+				entry->va, ret);
+		return ret;
+	}
+
+	if (no_update_pte)
+		return 0;
+
+	ret = update_gpuvm_pte(adev, entry, sync);
+	if (ret) {
+		pr_err("update_gpuvm_pte() failed\n");
+		goto update_gpuvm_pte_failed;
+	}
+
+	return 0;
+
+update_gpuvm_pte_failed:
+	unmap_bo_from_gpuvm(adev, entry, sync);
+	return ret;
+}
+
+static int process_validate_vms(struct amdkfd_process_info *process_info)
+{
+	struct amdgpu_vm *peer_vm;
+	int ret;
+
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node) {
+		ret = vm_validate_pt_pd_bos(peer_vm);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int process_update_pds(struct amdkfd_process_info *process_info,
+			      struct amdgpu_sync *sync)
+{
+	struct amdgpu_vm *peer_vm;
+	int ret;
+
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node) {
+		ret = vm_update_pds(peer_vm, sync);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int init_kfd_vm(struct amdgpu_vm *vm, void **process_info,
+		       struct dma_fence **ef)
+{
+	struct amdkfd_process_info *info = NULL;
+	int ret;
+
+	if (!*process_info) {
+		info = kzalloc(sizeof(*info), GFP_KERNEL);
+		if (!info)
+			return -ENOMEM;
+
+		rw_init(&info->lock, "aginfo");
+		INIT_LIST_HEAD(&info->vm_list_head);
+		INIT_LIST_HEAD(&info->kfd_bo_list);
+		INIT_LIST_HEAD(&info->userptr_valid_list);
+		INIT_LIST_HEAD(&info->userptr_inval_list);
+
+		info->eviction_fence =
+			amdgpu_amdkfd_fence_create(dma_fence_context_alloc(1),
+						   current->mm);
+		if (!info->eviction_fence) {
+			pr_err("Failed to create eviction fence\n");
+			ret = -ENOMEM;
+			goto create_evict_fence_fail;
+		}
+
+		info->pid = get_task_pid(current->group_leader, PIDTYPE_PID);
+		atomic_set(&info->evicted_bos, 0);
+		INIT_DELAYED_WORK(&info->restore_userptr_work,
+				  amdgpu_amdkfd_restore_userptr_worker);
+
+		*process_info = info;
+		*ef = dma_fence_get(&info->eviction_fence->base);
+	}
+
+	vm->process_info = *process_info;
+
+	/* Validate page directory and attach eviction fence */
+	ret = amdgpu_bo_reserve(vm->root.base.bo, true);
+	if (ret)
+		goto reserve_pd_fail;
+	ret = vm_validate_pt_pd_bos(vm);
+	if (ret) {
+		pr_err("validate_pt_pd_bos() failed\n");
+		goto validate_pd_fail;
+	}
+	ret = ttm_bo_wait(&vm->root.base.bo->tbo, false, false);
+	if (ret)
+		goto wait_pd_fail;
+	amdgpu_bo_fence(vm->root.base.bo,
+			&vm->process_info->eviction_fence->base, true);
+	amdgpu_bo_unreserve(vm->root.base.bo);
+
+	/* Update process info */
+	mutex_lock(&vm->process_info->lock);
+	list_add_tail(&vm->vm_list_node,
+			&(vm->process_info->vm_list_head));
+	vm->process_info->n_vms++;
+	mutex_unlock(&vm->process_info->lock);
+
+	return 0;
+
+wait_pd_fail:
+validate_pd_fail:
+	amdgpu_bo_unreserve(vm->root.base.bo);
+reserve_pd_fail:
+	vm->process_info = NULL;
+	if (info) {
+		/* Two fence references: one in info and one in *ef */
+		dma_fence_put(&info->eviction_fence->base);
+		dma_fence_put(*ef);
+		*ef = NULL;
+		*process_info = NULL;
+		put_pid(info->pid);
+create_evict_fence_fail:
+		mutex_destroy(&info->lock);
+		kfree(info);
+	}
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_create_process_vm(struct kgd_dev *kgd, void **vm,
+					  void **process_info,
+					  struct dma_fence **ef)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct amdgpu_vm *new_vm;
+	int ret;
+
+	new_vm = kzalloc(sizeof(*new_vm), GFP_KERNEL);
+	if (!new_vm)
+		return -ENOMEM;
+
+	/* Initialize AMDGPU part of the VM */
+	ret = amdgpu_vm_init(adev, new_vm, AMDGPU_VM_CONTEXT_COMPUTE, 0);
+	if (ret) {
+		pr_err("Failed init vm ret %d\n", ret);
+		goto amdgpu_vm_init_fail;
+	}
+
+	/* Initialize KFD part of the VM and process info */
+	ret = init_kfd_vm(new_vm, process_info, ef);
+	if (ret)
+		goto init_kfd_vm_fail;
+
+	*vm = (void *) new_vm;
+
+	return 0;
+
+init_kfd_vm_fail:
+	amdgpu_vm_fini(adev, new_vm);
+amdgpu_vm_init_fail:
+	kfree(new_vm);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct kgd_dev *kgd,
+					   struct file *filp,
+					   void **vm, void **process_info,
+					   struct dma_fence **ef)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct drm_file *drm_priv = filp->private_data;
+	struct amdgpu_fpriv *drv_priv = drm_priv->driver_priv;
+	struct amdgpu_vm *avm = &drv_priv->vm;
+	int ret;
+
+	/* Already a compute VM? */
+	if (avm->process_info)
+		return -EINVAL;
+
+	/* Convert VM into a compute VM */
+	ret = amdgpu_vm_make_compute(adev, avm);
+	if (ret)
+		return ret;
+
+	/* Initialize KFD part of the VM and process info */
+	ret = init_kfd_vm(avm, process_info, ef);
+	if (ret)
+		return ret;
+
+	*vm = (void *)avm;
+
+	return 0;
+}
+
+void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev,
+				    struct amdgpu_vm *vm)
+{
+	struct amdkfd_process_info *process_info = vm->process_info;
+	struct amdgpu_bo *pd = vm->root.base.bo;
+
+	if (!process_info)
+		return;
+
+	/* Release eviction fence from PD */
+	amdgpu_bo_reserve(pd, false);
+	amdgpu_bo_fence(pd, NULL, false);
+	amdgpu_bo_unreserve(pd);
+
+	/* Update process info */
+	mutex_lock(&process_info->lock);
+	process_info->n_vms--;
+	list_del(&vm->vm_list_node);
+	mutex_unlock(&process_info->lock);
+
+	/* Release per-process resources when last compute VM is destroyed */
+	if (!process_info->n_vms) {
+		WARN_ON(!list_empty(&process_info->kfd_bo_list));
+		WARN_ON(!list_empty(&process_info->userptr_valid_list));
+		WARN_ON(!list_empty(&process_info->userptr_inval_list));
+
+		dma_fence_put(&process_info->eviction_fence->base);
+		cancel_delayed_work_sync(&process_info->restore_userptr_work);
+		put_pid(process_info->pid);
+		mutex_destroy(&process_info->lock);
+		kfree(process_info);
+	}
+}
+
+void amdgpu_amdkfd_gpuvm_destroy_process_vm(struct kgd_dev *kgd, void *vm)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct amdgpu_vm *avm = (struct amdgpu_vm *)vm;
+
+	if (WARN_ON(!kgd || !vm))
+		return;
+
+	pr_debug("Destroying process vm %p\n", vm);
+
+	/* Release the VM context */
+	amdgpu_vm_fini(adev, avm);
+	kfree(vm);
+}
+
+uint32_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *vm)
+{
+	struct amdgpu_vm *avm = (struct amdgpu_vm *)vm;
+
+	return avm->pd_phys_addr >> AMDGPU_GPU_PAGE_SHIFT;
+}
+
+int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
+		struct kgd_dev *kgd, uint64_t va, uint64_t size,
+		void *vm, struct kgd_mem **mem,
+		uint64_t *offset, uint32_t flags)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct amdgpu_vm *avm = (struct amdgpu_vm *)vm;
+	uint64_t user_addr = 0;
+	struct amdgpu_bo *bo;
+	struct amdgpu_bo_param bp;
+	int byte_align;
+	u32 domain, alloc_domain;
+	u64 alloc_flags;
+	uint32_t mapping_flags;
+	int ret;
+
+	/*
+	 * Check on which domain to allocate BO
+	 */
+	if (flags & ALLOC_MEM_FLAGS_VRAM) {
+		domain = alloc_domain = AMDGPU_GEM_DOMAIN_VRAM;
+		alloc_flags = AMDGPU_GEM_CREATE_VRAM_CLEARED;
+		alloc_flags |= (flags & ALLOC_MEM_FLAGS_PUBLIC) ?
+			AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED :
+			AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
+	} else if (flags & ALLOC_MEM_FLAGS_GTT) {
+		domain = alloc_domain = AMDGPU_GEM_DOMAIN_GTT;
+		alloc_flags = 0;
+	} else if (flags & ALLOC_MEM_FLAGS_USERPTR) {
+		domain = AMDGPU_GEM_DOMAIN_GTT;
+		alloc_domain = AMDGPU_GEM_DOMAIN_CPU;
+		alloc_flags = 0;
+		if (!offset || !*offset)
+			return -EINVAL;
+		user_addr = *offset;
+	} else {
+		return -EINVAL;
+	}
+
+	*mem = kzalloc(sizeof(struct kgd_mem), GFP_KERNEL);
+	if (!*mem)
+		return -ENOMEM;
+	INIT_LIST_HEAD(&(*mem)->bo_va_list);
+	rw_init(&(*mem)->lock, "gpuvm");
+	(*mem)->aql_queue = !!(flags & ALLOC_MEM_FLAGS_AQL_QUEUE_MEM);
+
+	/* Workaround for AQL queue wraparound bug. Map the same
+	 * memory twice. That means we only actually allocate half
+	 * the memory.
+	 */
+	if ((*mem)->aql_queue)
+		size = size >> 1;
+
+	/* Workaround for TLB bug on older VI chips */
+	byte_align = (adev->family == AMDGPU_FAMILY_VI &&
+			adev->asic_type != CHIP_FIJI &&
+			adev->asic_type != CHIP_POLARIS10 &&
+			adev->asic_type != CHIP_POLARIS11) ?
+			VI_BO_SIZE_ALIGN : 1;
+
+	mapping_flags = AMDGPU_VM_PAGE_READABLE;
+	if (flags & ALLOC_MEM_FLAGS_WRITABLE)
+		mapping_flags |= AMDGPU_VM_PAGE_WRITEABLE;
+	if (flags & ALLOC_MEM_FLAGS_EXECUTABLE)
+		mapping_flags |= AMDGPU_VM_PAGE_EXECUTABLE;
+	if (flags & ALLOC_MEM_FLAGS_COHERENT)
+		mapping_flags |= AMDGPU_VM_MTYPE_UC;
+	else
+		mapping_flags |= AMDGPU_VM_MTYPE_NC;
+	(*mem)->mapping_flags = mapping_flags;
+
+	amdgpu_sync_create(&(*mem)->sync);
+
+	ret = amdgpu_amdkfd_reserve_system_mem_limit(adev, size, alloc_domain);
+	if (ret) {
+		pr_debug("Insufficient system memory\n");
+		goto err_reserve_system_mem;
+	}
+
+	pr_debug("\tcreate BO VA 0x%llx size 0x%llx domain %s\n",
+			va, size, domain_string(alloc_domain));
+
+	memset(&bp, 0, sizeof(bp));
+	bp.size = size;
+	bp.byte_align = byte_align;
+	bp.domain = alloc_domain;
+	bp.flags = alloc_flags;
+	bp.type = ttm_bo_type_device;
+	bp.resv = NULL;
+	ret = amdgpu_bo_create(adev, &bp, &bo);
+	if (ret) {
+		pr_debug("Failed to create BO on domain %s. ret %d\n",
+				domain_string(alloc_domain), ret);
+		goto err_bo_create;
+	}
+	bo->kfd_bo = *mem;
+	(*mem)->bo = bo;
+	if (user_addr)
+		bo->flags |= AMDGPU_AMDKFD_USERPTR_BO;
+
+	(*mem)->va = va;
+	(*mem)->domain = domain;
+	(*mem)->mapped_to_gpu_memory = 0;
+	(*mem)->process_info = avm->process_info;
+	add_kgd_mem_to_kfd_bo_list(*mem, avm->process_info, user_addr);
+
+	if (user_addr) {
+		ret = init_user_pages(*mem, current->mm, user_addr);
+		if (ret) {
+			mutex_lock(&avm->process_info->lock);
+			list_del(&(*mem)->validate_list.head);
+			mutex_unlock(&avm->process_info->lock);
+			goto allocate_init_user_pages_failed;
+		}
+	}
+
+	if (offset)
+		*offset = amdgpu_bo_mmap_offset(bo);
+
+	return 0;
+
+allocate_init_user_pages_failed:
+	amdgpu_bo_unref(&bo);
+	/* Don't unreserve system mem limit twice */
+	goto err_reserve_system_mem;
+err_bo_create:
+	unreserve_system_mem_limit(adev, size, alloc_domain);
+err_reserve_system_mem:
+	mutex_destroy(&(*mem)->lock);
+	kfree(*mem);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
+		struct kgd_dev *kgd, struct kgd_mem *mem)
+{
+	struct amdkfd_process_info *process_info = mem->process_info;
+	unsigned long bo_size = mem->bo->tbo.mem.size;
+	struct kfd_bo_va_list *entry, *tmp;
+	struct bo_vm_reservation_context ctx;
+	struct ttm_validate_buffer *bo_list_entry;
+	int ret;
+
+	mutex_lock(&mem->lock);
+
+	if (mem->mapped_to_gpu_memory > 0) {
+		pr_debug("BO VA 0x%llx size 0x%lx is still mapped.\n",
+				mem->va, bo_size);
+		mutex_unlock(&mem->lock);
+		return -EBUSY;
+	}
+
+	mutex_unlock(&mem->lock);
+	/* lock is not needed after this, since mem is unused and will
+	 * be freed anyway
+	 */
+
+	/* No more MMU notifiers */
+	amdgpu_mn_unregister(mem->bo);
+
+	/* Make sure restore workers don't access the BO any more */
+	bo_list_entry = &mem->validate_list;
+	mutex_lock(&process_info->lock);
+	list_del(&bo_list_entry->head);
+	mutex_unlock(&process_info->lock);
+
+	/* Free user pages if necessary */
+	if (mem->user_pages) {
+		pr_debug("%s: Freeing user_pages array\n", __func__);
+		if (mem->user_pages[0])
+			release_pages(mem->user_pages,
+					mem->bo->tbo.ttm->num_pages);
+		kvfree(mem->user_pages);
+	}
+
+	ret = reserve_bo_and_cond_vms(mem, NULL, BO_VM_ALL, &ctx);
+	if (unlikely(ret))
+		return ret;
+
+	/* The eviction fence should be removed by the last unmap.
+	 * TODO: Log an error condition if the bo still has the eviction fence
+	 * attached
+	 */
+	amdgpu_amdkfd_remove_eviction_fence(mem->bo,
+					process_info->eviction_fence,
+					NULL, NULL);
+	pr_debug("Release VA 0x%llx - 0x%llx\n", mem->va,
+		mem->va + bo_size * (1 + mem->aql_queue));
+
+	/* Remove from VM internal data structures */
+	list_for_each_entry_safe(entry, tmp, &mem->bo_va_list, bo_list)
+		remove_bo_from_vm((struct amdgpu_device *)entry->kgd_dev,
+				entry, bo_size);
+
+	ret = unreserve_bo_and_vms(&ctx, false, false);
+
+	/* Free the sync object */
+	amdgpu_sync_free(&mem->sync);
+
+	/* Free the BO*/
+	amdgpu_bo_unref(&mem->bo);
+	mutex_destroy(&mem->lock);
+	kfree(mem);
+
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_map_memory_to_gpu(
+		struct kgd_dev *kgd, struct kgd_mem *mem, void *vm)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct amdgpu_vm *avm = (struct amdgpu_vm *)vm;
+	int ret;
+	struct amdgpu_bo *bo;
+	uint32_t domain;
+	struct kfd_bo_va_list *entry;
+	struct bo_vm_reservation_context ctx;
+	struct kfd_bo_va_list *bo_va_entry = NULL;
+	struct kfd_bo_va_list *bo_va_entry_aql = NULL;
+	unsigned long bo_size;
+	bool is_invalid_userptr = false;
+
+	bo = mem->bo;
+	if (!bo) {
+		pr_err("Invalid BO when mapping memory to GPU\n");
+		return -EINVAL;
+	}
+
+	/* Make sure restore is not running concurrently. Since we
+	 * don't map invalid userptr BOs, we rely on the next restore
+	 * worker to do the mapping
+	 */
+	mutex_lock(&mem->process_info->lock);
+
+	/* Lock mmap-sem. If we find an invalid userptr BO, we can be
+	 * sure that the MMU notifier is no longer running
+	 * concurrently and the queues are actually stopped
+	 */
+	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm)) {
+		down_write(&current->mm->mmap_sem);
+		is_invalid_userptr = atomic_read(&mem->invalid);
+		up_write(&current->mm->mmap_sem);
+	}
+
+	mutex_lock(&mem->lock);
+
+	domain = mem->domain;
+	bo_size = bo->tbo.mem.size;
+
+	pr_debug("Map VA 0x%llx - 0x%llx to vm %p domain %s\n",
+			mem->va,
+			mem->va + bo_size * (1 + mem->aql_queue),
+			vm, domain_string(domain));
+
+	ret = reserve_bo_and_vm(mem, vm, &ctx);
+	if (unlikely(ret))
+		goto out;
+
+	/* Userptr can be marked as "not invalid", but not actually be
+	 * validated yet (still in the system domain). In that case
+	 * the queues are still stopped and we can leave mapping for
+	 * the next restore worker
+	 */
+	if (bo->tbo.mem.mem_type == TTM_PL_SYSTEM)
+		is_invalid_userptr = true;
+
+	if (check_if_add_bo_to_vm(avm, mem)) {
+		ret = add_bo_to_vm(adev, mem, avm, false,
+				&bo_va_entry);
+		if (ret)
+			goto add_bo_to_vm_failed;
+		if (mem->aql_queue) {
+			ret = add_bo_to_vm(adev, mem, avm,
+					true, &bo_va_entry_aql);
+			if (ret)
+				goto add_bo_to_vm_failed_aql;
+		}
+	} else {
+		ret = vm_validate_pt_pd_bos(avm);
+		if (unlikely(ret))
+			goto add_bo_to_vm_failed;
+	}
+
+	if (mem->mapped_to_gpu_memory == 0 &&
+	    !amdgpu_ttm_tt_get_usermm(bo->tbo.ttm)) {
+		/* Validate BO only once. The eviction fence gets added to BO
+		 * the first time it is mapped. Validate will wait for all
+		 * background evictions to complete.
+		 */
+		ret = amdgpu_amdkfd_bo_validate(bo, domain, true);
+		if (ret) {
+			pr_debug("Validate failed\n");
+			goto map_bo_to_gpuvm_failed;
+		}
+	}
+
+	list_for_each_entry(entry, &mem->bo_va_list, bo_list) {
+		if (entry->bo_va->base.vm == vm && !entry->is_mapped) {
+			pr_debug("\t map VA 0x%llx - 0x%llx in entry %p\n",
+					entry->va, entry->va + bo_size,
+					entry);
+
+			ret = map_bo_to_gpuvm(adev, entry, ctx.sync,
+					      is_invalid_userptr);
+			if (ret) {
+				pr_err("Failed to map radeon bo to gpuvm\n");
+				goto map_bo_to_gpuvm_failed;
+			}
+
+			ret = vm_update_pds(vm, ctx.sync);
+			if (ret) {
+				pr_err("Failed to update page directories\n");
+				goto map_bo_to_gpuvm_failed;
+			}
+
+			entry->is_mapped = true;
+			mem->mapped_to_gpu_memory++;
+			pr_debug("\t INC mapping count %d\n",
+					mem->mapped_to_gpu_memory);
+		}
+	}
+
+	if (!amdgpu_ttm_tt_get_usermm(bo->tbo.ttm) && !bo->pin_count)
+		amdgpu_bo_fence(bo,
+				&avm->process_info->eviction_fence->base,
+				true);
+	ret = unreserve_bo_and_vms(&ctx, false, false);
+
+	goto out;
+
+map_bo_to_gpuvm_failed:
+	if (bo_va_entry_aql)
+		remove_bo_from_vm(adev, bo_va_entry_aql, bo_size);
+add_bo_to_vm_failed_aql:
+	if (bo_va_entry)
+		remove_bo_from_vm(adev, bo_va_entry, bo_size);
+add_bo_to_vm_failed:
+	unreserve_bo_and_vms(&ctx, false, false);
+out:
+	mutex_unlock(&mem->process_info->lock);
+	mutex_unlock(&mem->lock);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(
+		struct kgd_dev *kgd, struct kgd_mem *mem, void *vm)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct amdkfd_process_info *process_info =
+		((struct amdgpu_vm *)vm)->process_info;
+	unsigned long bo_size = mem->bo->tbo.mem.size;
+	struct kfd_bo_va_list *entry;
+	struct bo_vm_reservation_context ctx;
+	int ret;
+
+	mutex_lock(&mem->lock);
+
+	ret = reserve_bo_and_cond_vms(mem, vm, BO_VM_MAPPED, &ctx);
+	if (unlikely(ret))
+		goto out;
+	/* If no VMs were reserved, it means the BO wasn't actually mapped */
+	if (ctx.n_vms == 0) {
+		ret = -EINVAL;
+		goto unreserve_out;
+	}
+
+	ret = vm_validate_pt_pd_bos((struct amdgpu_vm *)vm);
+	if (unlikely(ret))
+		goto unreserve_out;
+
+	pr_debug("Unmap VA 0x%llx - 0x%llx from vm %p\n",
+		mem->va,
+		mem->va + bo_size * (1 + mem->aql_queue),
+		vm);
+
+	list_for_each_entry(entry, &mem->bo_va_list, bo_list) {
+		if (entry->bo_va->base.vm == vm && entry->is_mapped) {
+			pr_debug("\t unmap VA 0x%llx - 0x%llx from entry %p\n",
+					entry->va,
+					entry->va + bo_size,
+					entry);
+
+			ret = unmap_bo_from_gpuvm(adev, entry, ctx.sync);
+			if (ret == 0) {
+				entry->is_mapped = false;
+			} else {
+				pr_err("failed to unmap VA 0x%llx\n",
+						mem->va);
+				goto unreserve_out;
+			}
+
+			mem->mapped_to_gpu_memory--;
+			pr_debug("\t DEC mapping count %d\n",
+					mem->mapped_to_gpu_memory);
+		}
+	}
+
+	/* If BO is unmapped from all VMs, unfence it. It can be evicted if
+	 * required.
+	 */
+	if (mem->mapped_to_gpu_memory == 0 &&
+	    !amdgpu_ttm_tt_get_usermm(mem->bo->tbo.ttm) && !mem->bo->pin_count)
+		amdgpu_amdkfd_remove_eviction_fence(mem->bo,
+						process_info->eviction_fence,
+						    NULL, NULL);
+
+unreserve_out:
+	unreserve_bo_and_vms(&ctx, false, false);
+out:
+	mutex_unlock(&mem->lock);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_sync_memory(
+		struct kgd_dev *kgd, struct kgd_mem *mem, bool intr)
+{
+	struct amdgpu_sync sync;
+	int ret;
+
+	amdgpu_sync_create(&sync);
+
+	mutex_lock(&mem->lock);
+	amdgpu_sync_clone(&mem->sync, &sync);
+	mutex_unlock(&mem->lock);
+
+	ret = amdgpu_sync_wait(&sync, intr);
+	amdgpu_sync_free(&sync);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(struct kgd_dev *kgd,
+		struct kgd_mem *mem, void **kptr, uint64_t *size)
+{
+	int ret;
+	struct amdgpu_bo *bo = mem->bo;
+
+	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm)) {
+		pr_err("userptr can't be mapped to kernel\n");
+		return -EINVAL;
+	}
+
+	/* delete kgd_mem from kfd_bo_list to avoid re-validating
+	 * this BO in BO's restoring after eviction.
+	 */
+	mutex_lock(&mem->process_info->lock);
+
+	ret = amdgpu_bo_reserve(bo, true);
+	if (ret) {
+		pr_err("Failed to reserve bo. ret %d\n", ret);
+		goto bo_reserve_failed;
+	}
+
+	ret = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT);
+	if (ret) {
+		pr_err("Failed to pin bo. ret %d\n", ret);
+		goto pin_failed;
+	}
+
+	ret = amdgpu_bo_kmap(bo, kptr);
+	if (ret) {
+		pr_err("Failed to map bo to kernel. ret %d\n", ret);
+		goto kmap_failed;
+	}
+
+	amdgpu_amdkfd_remove_eviction_fence(
+		bo, mem->process_info->eviction_fence, NULL, NULL);
+	list_del_init(&mem->validate_list.head);
+
+	if (size)
+		*size = amdgpu_bo_size(bo);
+
+	amdgpu_bo_unreserve(bo);
+
+	mutex_unlock(&mem->process_info->lock);
+	return 0;
+
+kmap_failed:
+	amdgpu_bo_unpin(bo);
+pin_failed:
+	amdgpu_bo_unreserve(bo);
+bo_reserve_failed:
+	mutex_unlock(&mem->process_info->lock);
+
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_get_vm_fault_info(struct kgd_dev *kgd,
+					      struct kfd_vm_fault_info *mem)
+{
+	struct amdgpu_device *adev;
+
+	adev = (struct amdgpu_device *)kgd;
+	if (atomic_read(&adev->gmc.vm_fault_info_updated) == 1) {
+		*mem = *adev->gmc.vm_fault_info;
+		mb();
+		atomic_set(&adev->gmc.vm_fault_info_updated, 0);
+	}
+	return 0;
+}
+
+/* Evict a userptr BO by stopping the queues if necessary
+ *
+ * Runs in MMU notifier, may be in RECLAIM_FS context. This means it
+ * cannot do any memory allocations, and cannot take any locks that
+ * are held elsewhere while allocating memory. Therefore this is as
+ * simple as possible, using atomic counters.
+ *
+ * It doesn't do anything to the BO itself. The real work happens in
+ * restore, where we get updated page addresses. This function only
+ * ensures that GPU access to the BO is stopped.
+ */
+int amdgpu_amdkfd_evict_userptr(struct kgd_mem *mem,
+				struct mm_struct *mm)
+{
+	struct amdkfd_process_info *process_info = mem->process_info;
+	int invalid, evicted_bos;
+	int r = 0;
+
+	invalid = atomic_inc_return(&mem->invalid);
+	evicted_bos = atomic_inc_return(&process_info->evicted_bos);
+	if (evicted_bos == 1) {
+		/* First eviction, stop the queues */
+		r = kgd2kfd->quiesce_mm(mm);
+		if (r)
+			pr_err("Failed to quiesce KFD\n");
+		schedule_delayed_work(&process_info->restore_userptr_work,
+			msecs_to_jiffies(AMDGPU_USERPTR_RESTORE_DELAY_MS));
+	}
+
+	return r;
+}
+
+/* Update invalid userptr BOs
+ *
+ * Moves invalidated (evicted) userptr BOs from userptr_valid_list to
+ * userptr_inval_list and updates user pages for all BOs that have
+ * been invalidated since their last update.
+ */
+static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
+				     struct mm_struct *mm)
+{
+	struct kgd_mem *mem, *tmp_mem;
+	struct amdgpu_bo *bo;
+	struct ttm_operation_ctx ctx = { false, false };
+	int invalid, ret;
+
+	/* Move all invalidated BOs to the userptr_inval_list and
+	 * release their user pages by migration to the CPU domain
+	 */
+	list_for_each_entry_safe(mem, tmp_mem,
+				 &process_info->userptr_valid_list,
+				 validate_list.head) {
+		if (!atomic_read(&mem->invalid))
+			continue; /* BO is still valid */
+
+		bo = mem->bo;
+
+		if (amdgpu_bo_reserve(bo, true))
+			return -EAGAIN;
+		amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_CPU);
+		ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+		amdgpu_bo_unreserve(bo);
+		if (ret) {
+			pr_err("%s: Failed to invalidate userptr BO\n",
+			       __func__);
+			return -EAGAIN;
+		}
+
+		list_move_tail(&mem->validate_list.head,
+			       &process_info->userptr_inval_list);
+	}
+
+	if (list_empty(&process_info->userptr_inval_list))
+		return 0; /* All evicted userptr BOs were freed */
+
+	/* Go through userptr_inval_list and update any invalid user_pages */
+	list_for_each_entry(mem, &process_info->userptr_inval_list,
+			    validate_list.head) {
+		invalid = atomic_read(&mem->invalid);
+		if (!invalid)
+			/* BO hasn't been invalidated since the last
+			 * revalidation attempt. Keep its BO list.
+			 */
+			continue;
+
+		bo = mem->bo;
+
+		if (!mem->user_pages) {
+			mem->user_pages =
+				kvmalloc_array(bo->tbo.ttm->num_pages,
+						 sizeof(struct vm_page *),
+						 GFP_KERNEL | __GFP_ZERO);
+			if (!mem->user_pages) {
+				pr_err("%s: Failed to allocate pages array\n",
+				       __func__);
+				return -ENOMEM;
+			}
+		} else if (mem->user_pages[0]) {
+			release_pages(mem->user_pages, bo->tbo.ttm->num_pages);
+		}
+
+		/* Get updated user pages */
+		ret = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm,
+						   mem->user_pages);
+		if (ret) {
+			mem->user_pages[0] = NULL;
+			pr_info("%s: Failed to get user pages: %d\n",
+				__func__, ret);
+			/* Pretend it succeeded. It will fail later
+			 * with a VM fault if the GPU tries to access
+			 * it. Better than hanging indefinitely with
+			 * stalled user mode queues.
+			 */
+		}
+
+		/* Mark the BO as valid unless it was invalidated
+		 * again concurrently
+		 */
+		if (atomic_cmpxchg(&mem->invalid, invalid, 0) != invalid)
+			return -EAGAIN;
+	}
+
+	return 0;
+}
+
+/* Validate invalid userptr BOs
+ *
+ * Validates BOs on the userptr_inval_list, and moves them back to the
+ * userptr_valid_list. Also updates GPUVM page tables with new page
+ * addresses and waits for the page table updates to complete.
+ */
+static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
+{
+	struct amdgpu_bo_list_entry *pd_bo_list_entries;
+	struct list_head resv_list, duplicates;
+	struct ww_acquire_ctx ticket;
+	struct amdgpu_sync sync;
+
+	struct amdgpu_vm *peer_vm;
+	struct kgd_mem *mem, *tmp_mem;
+	struct amdgpu_bo *bo;
+	struct ttm_operation_ctx ctx = { false, false };
+	int i, ret;
+
+	pd_bo_list_entries = kcalloc(process_info->n_vms,
+				     sizeof(struct amdgpu_bo_list_entry),
+				     GFP_KERNEL);
+	if (!pd_bo_list_entries) {
+		pr_err("%s: Failed to allocate PD BO list entries\n", __func__);
+		return -ENOMEM;
+	}
+
+	INIT_LIST_HEAD(&resv_list);
+	INIT_LIST_HEAD(&duplicates);
+
+	/* Get all the page directory BOs that need to be reserved */
+	i = 0;
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node)
+		amdgpu_vm_get_pd_bo(peer_vm, &resv_list,
+				    &pd_bo_list_entries[i++]);
+	/* Add the userptr_inval_list entries to resv_list */
+	list_for_each_entry(mem, &process_info->userptr_inval_list,
+			    validate_list.head) {
+		list_add_tail(&mem->resv_list.head, &resv_list);
+		mem->resv_list.bo = mem->validate_list.bo;
+		mem->resv_list.shared = mem->validate_list.shared;
+	}
+
+	/* Reserve all BOs and page tables for validation */
+	ret = ttm_eu_reserve_buffers(&ticket, &resv_list, false, &duplicates);
+	WARN(!list_empty(&duplicates), "Duplicates should be empty");
+	if (ret)
+		goto out;
+
+	amdgpu_sync_create(&sync);
+
+	/* Avoid triggering eviction fences when unmapping invalid
+	 * userptr BOs (waits for all fences, doesn't use
+	 * FENCE_OWNER_VM)
+	 */
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node)
+		amdgpu_amdkfd_remove_eviction_fence(peer_vm->root.base.bo,
+						process_info->eviction_fence,
+						NULL, NULL);
+
+	ret = process_validate_vms(process_info);
+	if (ret)
+		goto unreserve_out;
+
+	/* Validate BOs and update GPUVM page tables */
+	list_for_each_entry_safe(mem, tmp_mem,
+				 &process_info->userptr_inval_list,
+				 validate_list.head) {
+		struct kfd_bo_va_list *bo_va_entry;
+
+		bo = mem->bo;
+
+		/* Copy pages array and validate the BO if we got user pages */
+		if (mem->user_pages[0]) {
+			amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm,
+						     mem->user_pages);
+			amdgpu_bo_placement_from_domain(bo, mem->domain);
+			ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+			if (ret) {
+				pr_err("%s: failed to validate BO\n", __func__);
+				goto unreserve_out;
+			}
+		}
+
+		/* Validate succeeded, now the BO owns the pages, free
+		 * our copy of the pointer array. Put this BO back on
+		 * the userptr_valid_list. If we need to revalidate
+		 * it, we need to start from scratch.
+		 */
+		kvfree(mem->user_pages);
+		mem->user_pages = NULL;
+		list_move_tail(&mem->validate_list.head,
+			       &process_info->userptr_valid_list);
+
+		/* Update mapping. If the BO was not validated
+		 * (because we couldn't get user pages), this will
+		 * clear the page table entries, which will result in
+		 * VM faults if the GPU tries to access the invalid
+		 * memory.
+		 */
+		list_for_each_entry(bo_va_entry, &mem->bo_va_list, bo_list) {
+			if (!bo_va_entry->is_mapped)
+				continue;
+
+			ret = update_gpuvm_pte((struct amdgpu_device *)
+					       bo_va_entry->kgd_dev,
+					       bo_va_entry, &sync);
+			if (ret) {
+				pr_err("%s: update PTE failed\n", __func__);
+				/* make sure this gets validated again */
+				atomic_inc(&mem->invalid);
+				goto unreserve_out;
+			}
+		}
+	}
+
+	/* Update page directories */
+	ret = process_update_pds(process_info, &sync);
+
+unreserve_out:
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node)
+		amdgpu_bo_fence(peer_vm->root.base.bo,
+				&process_info->eviction_fence->base, true);
+	ttm_eu_backoff_reservation(&ticket, &resv_list);
+	amdgpu_sync_wait(&sync, false);
+	amdgpu_sync_free(&sync);
+out:
+	kfree(pd_bo_list_entries);
+
+	return ret;
+}
+
+/* Worker callback to restore evicted userptr BOs
+ *
+ * Tries to update and validate all userptr BOs. If successful and no
+ * concurrent evictions happened, the queues are restarted. Otherwise,
+ * reschedule for another attempt later.
+ */
+static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
+{
+	struct delayed_work *dwork = to_delayed_work(work);
+	struct amdkfd_process_info *process_info =
+		container_of(dwork, struct amdkfd_process_info,
+			     restore_userptr_work);
+	struct task_struct *usertask;
+	struct mm_struct *mm;
+	int evicted_bos;
+
+	evicted_bos = atomic_read(&process_info->evicted_bos);
+	if (!evicted_bos)
+		return;
+
+	/* Reference task and mm in case of concurrent process termination */
+	usertask = get_pid_task(process_info->pid, PIDTYPE_PID);
+	if (!usertask)
+		return;
+	mm = get_task_mm(usertask);
+	if (!mm) {
+		put_task_struct(usertask);
+		return;
+	}
+
+	mutex_lock(&process_info->lock);
+
+	if (update_invalid_user_pages(process_info, mm))
+		goto unlock_out;
+	/* userptr_inval_list can be empty if all evicted userptr BOs
+	 * have been freed. In that case there is nothing to validate
+	 * and we can just restart the queues.
+	 */
+	if (!list_empty(&process_info->userptr_inval_list)) {
+		if (atomic_read(&process_info->evicted_bos) != evicted_bos)
+			goto unlock_out; /* Concurrent eviction, try again */
+
+		if (validate_invalid_user_pages(process_info))
+			goto unlock_out;
+	}
+	/* Final check for concurrent evicton and atomic update. If
+	 * another eviction happens after successful update, it will
+	 * be a first eviction that calls quiesce_mm. The eviction
+	 * reference counting inside KFD will handle this case.
+	 */
+	if (atomic_cmpxchg(&process_info->evicted_bos, evicted_bos, 0) !=
+	    evicted_bos)
+		goto unlock_out;
+	evicted_bos = 0;
+	if (kgd2kfd->resume_mm(mm)) {
+		pr_err("%s: Failed to resume KFD\n", __func__);
+		/* No recovery from this failure. Probably the CP is
+		 * hanging. No point trying again.
+		 */
+	}
+unlock_out:
+	mutex_unlock(&process_info->lock);
+	mmput(mm);
+	put_task_struct(usertask);
+
+	/* If validation failed, reschedule another attempt */
+	if (evicted_bos)
+		schedule_delayed_work(&process_info->restore_userptr_work,
+			msecs_to_jiffies(AMDGPU_USERPTR_RESTORE_DELAY_MS));
+}
+
+/** amdgpu_amdkfd_gpuvm_restore_process_bos - Restore all BOs for the given
+ *   KFD process identified by process_info
+ *
+ * @process_info: amdkfd_process_info of the KFD process
+ *
+ * After memory eviction, restore thread calls this function. The function
+ * should be called when the Process is still valid. BO restore involves -
+ *
+ * 1.  Release old eviction fence and create new one
+ * 2.  Get two copies of PD BO list from all the VMs. Keep one copy as pd_list.
+ * 3   Use the second PD list and kfd_bo_list to create a list (ctx.list) of
+ *     BOs that need to be reserved.
+ * 4.  Reserve all the BOs
+ * 5.  Validate of PD and PT BOs.
+ * 6.  Validate all KFD BOs using kfd_bo_list and Map them and add new fence
+ * 7.  Add fence to all PD and PT BOs.
+ * 8.  Unreserve all BOs
+ */
+int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence **ef)
+{
+	struct amdgpu_bo_list_entry *pd_bo_list;
+	struct amdkfd_process_info *process_info = info;
+	struct amdgpu_vm *peer_vm;
+	struct kgd_mem *mem;
+	struct bo_vm_reservation_context ctx;
+	struct amdgpu_amdkfd_fence *new_fence;
+	int ret = 0, i;
+	struct list_head duplicate_save;
+	struct amdgpu_sync sync_obj;
+
+	INIT_LIST_HEAD(&duplicate_save);
+	INIT_LIST_HEAD(&ctx.list);
+	INIT_LIST_HEAD(&ctx.duplicates);
+
+	pd_bo_list = kcalloc(process_info->n_vms,
+			     sizeof(struct amdgpu_bo_list_entry),
+			     GFP_KERNEL);
+	if (!pd_bo_list)
+		return -ENOMEM;
+
+	i = 0;
+	mutex_lock(&process_info->lock);
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			vm_list_node)
+		amdgpu_vm_get_pd_bo(peer_vm, &ctx.list, &pd_bo_list[i++]);
+
+	/* Reserve all BOs and page tables/directory. Add all BOs from
+	 * kfd_bo_list to ctx.list
+	 */
+	list_for_each_entry(mem, &process_info->kfd_bo_list,
+			    validate_list.head) {
+
+		list_add_tail(&mem->resv_list.head, &ctx.list);
+		mem->resv_list.bo = mem->validate_list.bo;
+		mem->resv_list.shared = mem->validate_list.shared;
+	}
+
+	ret = ttm_eu_reserve_buffers(&ctx.ticket, &ctx.list,
+				     false, &duplicate_save);
+	if (ret) {
+		pr_debug("Memory eviction: TTM Reserve Failed. Try again\n");
+		goto ttm_reserve_fail;
+	}
+
+	amdgpu_sync_create(&sync_obj);
+
+	/* Validate PDs and PTs */
+	ret = process_validate_vms(process_info);
+	if (ret)
+		goto validate_map_fail;
+
+	/* Wait for PD/PTs validate to finish */
+	/* FIXME: I think this isn't needed */
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node) {
+		struct amdgpu_bo *bo = peer_vm->root.base.bo;
+
+		ttm_bo_wait(&bo->tbo, false, false);
+	}
+
+	/* Validate BOs and map them to GPUVM (update VM page tables). */
+	list_for_each_entry(mem, &process_info->kfd_bo_list,
+			    validate_list.head) {
+
+		struct amdgpu_bo *bo = mem->bo;
+		uint32_t domain = mem->domain;
+		struct kfd_bo_va_list *bo_va_entry;
+
+		ret = amdgpu_amdkfd_bo_validate(bo, domain, false);
+		if (ret) {
+			pr_debug("Memory eviction: Validate BOs failed. Try again\n");
+			goto validate_map_fail;
+		}
+
+		list_for_each_entry(bo_va_entry, &mem->bo_va_list,
+				    bo_list) {
+			ret = update_gpuvm_pte((struct amdgpu_device *)
+					      bo_va_entry->kgd_dev,
+					      bo_va_entry,
+					      &sync_obj);
+			if (ret) {
+				pr_debug("Memory eviction: update PTE failed. Try again\n");
+				goto validate_map_fail;
+			}
+		}
+	}
+
+	/* Update page directories */
+	ret = process_update_pds(process_info, &sync_obj);
+	if (ret) {
+		pr_debug("Memory eviction: update PDs failed. Try again\n");
+		goto validate_map_fail;
+	}
+
+	amdgpu_sync_wait(&sync_obj, false);
+
+	/* Release old eviction fence and create new one, because fence only
+	 * goes from unsignaled to signaled, fence cannot be reused.
+	 * Use context and mm from the old fence.
+	 */
+	new_fence = amdgpu_amdkfd_fence_create(
+				process_info->eviction_fence->base.context,
+				process_info->eviction_fence->mm);
+	if (!new_fence) {
+		pr_err("Failed to create eviction fence\n");
+		ret = -ENOMEM;
+		goto validate_map_fail;
+	}
+	dma_fence_put(&process_info->eviction_fence->base);
+	process_info->eviction_fence = new_fence;
+	*ef = dma_fence_get(&new_fence->base);
+
+	/* Wait for validate to finish and attach new eviction fence */
+	list_for_each_entry(mem, &process_info->kfd_bo_list,
+		validate_list.head)
+		ttm_bo_wait(&mem->bo->tbo, false, false);
+	list_for_each_entry(mem, &process_info->kfd_bo_list,
+		validate_list.head)
+		amdgpu_bo_fence(mem->bo,
+			&process_info->eviction_fence->base, true);
+
+	/* Attach eviction fence to PD / PT BOs */
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node) {
+		struct amdgpu_bo *bo = peer_vm->root.base.bo;
+
+		amdgpu_bo_fence(bo, &process_info->eviction_fence->base, true);
+	}
+
+validate_map_fail:
+	ttm_eu_backoff_reservation(&ctx.ticket, &ctx.list);
+	amdgpu_sync_free(&sync_obj);
+ttm_reserve_fail:
+	mutex_unlock(&process_info->lock);
+	kfree(pd_bo_list);
+	return ret;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_atom.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atom.c
new file mode 100644
index 00000000000..bcbc16a963f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atom.c
@@ -0,0 +1,1422 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: Stanislaw Skowronek
+ */
+
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <asm/unaligned.h>
+
+#define ATOM_DEBUG
+
+#include "atom.h"
+#include "atom-names.h"
+#include "atom-bits.h"
+#include "amdgpu.h"
+
+#define ATOM_COND_ABOVE		0
+#define ATOM_COND_ABOVEOREQUAL	1
+#define ATOM_COND_ALWAYS	2
+#define ATOM_COND_BELOW		3
+#define ATOM_COND_BELOWOREQUAL	4
+#define ATOM_COND_EQUAL		5
+#define ATOM_COND_NOTEQUAL	6
+
+#define ATOM_PORT_ATI	0
+#define ATOM_PORT_PCI	1
+#define ATOM_PORT_SYSIO	2
+
+#define ATOM_UNIT_MICROSEC	0
+#define ATOM_UNIT_MILLISEC	1
+
+#define PLL_INDEX	2
+#define PLL_DATA	3
+
+typedef struct {
+	struct atom_context *ctx;
+	uint32_t *ps, *ws;
+	int ps_shift;
+	uint16_t start;
+	unsigned last_jump;
+	unsigned long last_jump_jiffies;
+	bool abort;
+} atom_exec_context;
+
+int amdgpu_atom_debug = 0;
+static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t * params);
+int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t * params);
+
+static uint32_t atom_arg_mask[8] =
+    { 0xFFFFFFFF, 0xFFFF, 0xFFFF00, 0xFFFF0000, 0xFF, 0xFF00, 0xFF0000,
+0xFF000000 };
+static int atom_arg_shift[8] = { 0, 0, 8, 16, 0, 8, 16, 24 };
+
+static int atom_dst_to_src[8][4] = {
+	/* translate destination alignment field to the source alignment encoding */
+	{0, 0, 0, 0},
+	{1, 2, 3, 0},
+	{1, 2, 3, 0},
+	{1, 2, 3, 0},
+	{4, 5, 6, 7},
+	{4, 5, 6, 7},
+	{4, 5, 6, 7},
+	{4, 5, 6, 7},
+};
+static int atom_def_dst[8] = { 0, 0, 1, 2, 0, 1, 2, 3 };
+
+static int debug_depth = 0;
+#ifdef ATOM_DEBUG
+static void debug_print_spaces(int n)
+{
+	while (n--)
+		printk("   ");
+}
+
+#define DEBUG(...) do if (amdgpu_atom_debug) { printk(KERN_DEBUG __VA_ARGS__); } while (0)
+#define SDEBUG(...) do if (amdgpu_atom_debug) { printk(KERN_DEBUG); debug_print_spaces(debug_depth); printk(__VA_ARGS__); } while (0)
+#else
+#define DEBUG(...) do { } while (0)
+#define SDEBUG(...) do { } while (0)
+#endif
+
+static uint32_t atom_iio_execute(struct atom_context *ctx, int base,
+				 uint32_t index, uint32_t data)
+{
+	uint32_t temp = 0xCDCDCDCD;
+
+	while (1)
+		switch (CU8(base)) {
+		case ATOM_IIO_NOP:
+			base++;
+			break;
+		case ATOM_IIO_READ:
+			temp = ctx->card->ioreg_read(ctx->card, CU16(base + 1));
+			base += 3;
+			break;
+		case ATOM_IIO_WRITE:
+			ctx->card->ioreg_write(ctx->card, CU16(base + 1), temp);
+			base += 3;
+			break;
+		case ATOM_IIO_CLEAR:
+			temp &=
+			    ~((0xFFFFFFFF >> (32 - CU8(base + 1))) <<
+			      CU8(base + 2));
+			base += 3;
+			break;
+		case ATOM_IIO_SET:
+			temp |=
+			    (0xFFFFFFFF >> (32 - CU8(base + 1))) << CU8(base +
+									2);
+			base += 3;
+			break;
+		case ATOM_IIO_MOVE_INDEX:
+			temp &=
+			    ~((0xFFFFFFFF >> (32 - CU8(base + 1))) <<
+			      CU8(base + 3));
+			temp |=
+			    ((index >> CU8(base + 2)) &
+			     (0xFFFFFFFF >> (32 - CU8(base + 1)))) << CU8(base +
+									  3);
+			base += 4;
+			break;
+		case ATOM_IIO_MOVE_DATA:
+			temp &=
+			    ~((0xFFFFFFFF >> (32 - CU8(base + 1))) <<
+			      CU8(base + 3));
+			temp |=
+			    ((data >> CU8(base + 2)) &
+			     (0xFFFFFFFF >> (32 - CU8(base + 1)))) << CU8(base +
+									  3);
+			base += 4;
+			break;
+		case ATOM_IIO_MOVE_ATTR:
+			temp &=
+			    ~((0xFFFFFFFF >> (32 - CU8(base + 1))) <<
+			      CU8(base + 3));
+			temp |=
+			    ((ctx->
+			      io_attr >> CU8(base + 2)) & (0xFFFFFFFF >> (32 -
+									  CU8
+									  (base
+									   +
+									   1))))
+			    << CU8(base + 3);
+			base += 4;
+			break;
+		case ATOM_IIO_END:
+			return temp;
+		default:
+			pr_info("Unknown IIO opcode\n");
+			return 0;
+		}
+}
+
+static uint32_t atom_get_src_int(atom_exec_context *ctx, uint8_t attr,
+				 int *ptr, uint32_t *saved, int print)
+{
+	uint32_t idx, val = 0xCDCDCDCD, align, arg;
+	struct atom_context *gctx = ctx->ctx;
+	arg = attr & 7;
+	align = (attr >> 3) & 7;
+	switch (arg) {
+	case ATOM_ARG_REG:
+		idx = U16(*ptr);
+		(*ptr) += 2;
+		if (print)
+			DEBUG("REG[0x%04X]", idx);
+		idx += gctx->reg_block;
+		switch (gctx->io_mode) {
+		case ATOM_IO_MM:
+			val = gctx->card->reg_read(gctx->card, idx);
+			break;
+		case ATOM_IO_PCI:
+			pr_info("PCI registers are not implemented\n");
+			return 0;
+		case ATOM_IO_SYSIO:
+			pr_info("SYSIO registers are not implemented\n");
+			return 0;
+		default:
+			if (!(gctx->io_mode & 0x80)) {
+				pr_info("Bad IO mode\n");
+				return 0;
+			}
+			if (!gctx->iio[gctx->io_mode & 0x7F]) {
+				pr_info("Undefined indirect IO read method %d\n",
+					gctx->io_mode & 0x7F);
+				return 0;
+			}
+			val =
+			    atom_iio_execute(gctx,
+					     gctx->iio[gctx->io_mode & 0x7F],
+					     idx, 0);
+		}
+		break;
+	case ATOM_ARG_PS:
+		idx = U8(*ptr);
+		(*ptr)++;
+		/* get_unaligned_le32 avoids unaligned accesses from atombios
+		 * tables, noticed on a DEC Alpha. */
+		val = get_unaligned_le32((u32 *)&ctx->ps[idx]);
+		if (print)
+			DEBUG("PS[0x%02X,0x%04X]", idx, val);
+		break;
+	case ATOM_ARG_WS:
+		idx = U8(*ptr);
+		(*ptr)++;
+		if (print)
+			DEBUG("WS[0x%02X]", idx);
+		switch (idx) {
+		case ATOM_WS_QUOTIENT:
+			val = gctx->divmul[0];
+			break;
+		case ATOM_WS_REMAINDER:
+			val = gctx->divmul[1];
+			break;
+		case ATOM_WS_DATAPTR:
+			val = gctx->data_block;
+			break;
+		case ATOM_WS_SHIFT:
+			val = gctx->shift;
+			break;
+		case ATOM_WS_OR_MASK:
+			val = 1 << gctx->shift;
+			break;
+		case ATOM_WS_AND_MASK:
+			val = ~(1 << gctx->shift);
+			break;
+		case ATOM_WS_FB_WINDOW:
+			val = gctx->fb_base;
+			break;
+		case ATOM_WS_ATTRIBUTES:
+			val = gctx->io_attr;
+			break;
+		case ATOM_WS_REGPTR:
+			val = gctx->reg_block;
+			break;
+		default:
+			val = ctx->ws[idx];
+		}
+		break;
+	case ATOM_ARG_ID:
+		idx = U16(*ptr);
+		(*ptr) += 2;
+		if (print) {
+			if (gctx->data_block)
+				DEBUG("ID[0x%04X+%04X]", idx, gctx->data_block);
+			else
+				DEBUG("ID[0x%04X]", idx);
+		}
+		val = U32(idx + gctx->data_block);
+		break;
+	case ATOM_ARG_FB:
+		idx = U8(*ptr);
+		(*ptr)++;
+		if ((gctx->fb_base + (idx * 4)) > gctx->scratch_size_bytes) {
+			DRM_ERROR("ATOM: fb read beyond scratch region: %d vs. %d\n",
+				  gctx->fb_base + (idx * 4), gctx->scratch_size_bytes);
+			val = 0;
+		} else
+			val = gctx->scratch[(gctx->fb_base / 4) + idx];
+		if (print)
+			DEBUG("FB[0x%02X]", idx);
+		break;
+	case ATOM_ARG_IMM:
+		switch (align) {
+		case ATOM_SRC_DWORD:
+			val = U32(*ptr);
+			(*ptr) += 4;
+			if (print)
+				DEBUG("IMM 0x%08X\n", val);
+			return val;
+		case ATOM_SRC_WORD0:
+		case ATOM_SRC_WORD8:
+		case ATOM_SRC_WORD16:
+			val = U16(*ptr);
+			(*ptr) += 2;
+			if (print)
+				DEBUG("IMM 0x%04X\n", val);
+			return val;
+		case ATOM_SRC_BYTE0:
+		case ATOM_SRC_BYTE8:
+		case ATOM_SRC_BYTE16:
+		case ATOM_SRC_BYTE24:
+			val = U8(*ptr);
+			(*ptr)++;
+			if (print)
+				DEBUG("IMM 0x%02X\n", val);
+			return val;
+		}
+		return 0;
+	case ATOM_ARG_PLL:
+		idx = U8(*ptr);
+		(*ptr)++;
+		if (print)
+			DEBUG("PLL[0x%02X]", idx);
+		val = gctx->card->pll_read(gctx->card, idx);
+		break;
+	case ATOM_ARG_MC:
+		idx = U8(*ptr);
+		(*ptr)++;
+		if (print)
+			DEBUG("MC[0x%02X]", idx);
+		val = gctx->card->mc_read(gctx->card, idx);
+		break;
+	}
+	if (saved)
+		*saved = val;
+	val &= atom_arg_mask[align];
+	val >>= atom_arg_shift[align];
+	if (print)
+		switch (align) {
+		case ATOM_SRC_DWORD:
+			DEBUG(".[31:0] -> 0x%08X\n", val);
+			break;
+		case ATOM_SRC_WORD0:
+			DEBUG(".[15:0] -> 0x%04X\n", val);
+			break;
+		case ATOM_SRC_WORD8:
+			DEBUG(".[23:8] -> 0x%04X\n", val);
+			break;
+		case ATOM_SRC_WORD16:
+			DEBUG(".[31:16] -> 0x%04X\n", val);
+			break;
+		case ATOM_SRC_BYTE0:
+			DEBUG(".[7:0] -> 0x%02X\n", val);
+			break;
+		case ATOM_SRC_BYTE8:
+			DEBUG(".[15:8] -> 0x%02X\n", val);
+			break;
+		case ATOM_SRC_BYTE16:
+			DEBUG(".[23:16] -> 0x%02X\n", val);
+			break;
+		case ATOM_SRC_BYTE24:
+			DEBUG(".[31:24] -> 0x%02X\n", val);
+			break;
+		}
+	return val;
+}
+
+static void atom_skip_src_int(atom_exec_context *ctx, uint8_t attr, int *ptr)
+{
+	uint32_t align = (attr >> 3) & 7, arg = attr & 7;
+	switch (arg) {
+	case ATOM_ARG_REG:
+	case ATOM_ARG_ID:
+		(*ptr) += 2;
+		break;
+	case ATOM_ARG_PLL:
+	case ATOM_ARG_MC:
+	case ATOM_ARG_PS:
+	case ATOM_ARG_WS:
+	case ATOM_ARG_FB:
+		(*ptr)++;
+		break;
+	case ATOM_ARG_IMM:
+		switch (align) {
+		case ATOM_SRC_DWORD:
+			(*ptr) += 4;
+			return;
+		case ATOM_SRC_WORD0:
+		case ATOM_SRC_WORD8:
+		case ATOM_SRC_WORD16:
+			(*ptr) += 2;
+			return;
+		case ATOM_SRC_BYTE0:
+		case ATOM_SRC_BYTE8:
+		case ATOM_SRC_BYTE16:
+		case ATOM_SRC_BYTE24:
+			(*ptr)++;
+			return;
+		}
+		return;
+	}
+}
+
+static uint32_t atom_get_src(atom_exec_context *ctx, uint8_t attr, int *ptr)
+{
+	return atom_get_src_int(ctx, attr, ptr, NULL, 1);
+}
+
+static uint32_t atom_get_src_direct(atom_exec_context *ctx, uint8_t align, int *ptr)
+{
+	uint32_t val = 0xCDCDCDCD;
+
+	switch (align) {
+	case ATOM_SRC_DWORD:
+		val = U32(*ptr);
+		(*ptr) += 4;
+		break;
+	case ATOM_SRC_WORD0:
+	case ATOM_SRC_WORD8:
+	case ATOM_SRC_WORD16:
+		val = U16(*ptr);
+		(*ptr) += 2;
+		break;
+	case ATOM_SRC_BYTE0:
+	case ATOM_SRC_BYTE8:
+	case ATOM_SRC_BYTE16:
+	case ATOM_SRC_BYTE24:
+		val = U8(*ptr);
+		(*ptr)++;
+		break;
+	}
+	return val;
+}
+
+static uint32_t atom_get_dst(atom_exec_context *ctx, int arg, uint8_t attr,
+			     int *ptr, uint32_t *saved, int print)
+{
+	return atom_get_src_int(ctx,
+				arg | atom_dst_to_src[(attr >> 3) &
+						      7][(attr >> 6) & 3] << 3,
+				ptr, saved, print);
+}
+
+static void atom_skip_dst(atom_exec_context *ctx, int arg, uint8_t attr, int *ptr)
+{
+	atom_skip_src_int(ctx,
+			  arg | atom_dst_to_src[(attr >> 3) & 7][(attr >> 6) &
+								 3] << 3, ptr);
+}
+
+static void atom_put_dst(atom_exec_context *ctx, int arg, uint8_t attr,
+			 int *ptr, uint32_t val, uint32_t saved)
+{
+	uint32_t align =
+	    atom_dst_to_src[(attr >> 3) & 7][(attr >> 6) & 3], old_val =
+	    val, idx;
+	struct atom_context *gctx = ctx->ctx;
+	old_val &= atom_arg_mask[align] >> atom_arg_shift[align];
+	val <<= atom_arg_shift[align];
+	val &= atom_arg_mask[align];
+	saved &= ~atom_arg_mask[align];
+	val |= saved;
+	switch (arg) {
+	case ATOM_ARG_REG:
+		idx = U16(*ptr);
+		(*ptr) += 2;
+		DEBUG("REG[0x%04X]", idx);
+		idx += gctx->reg_block;
+		switch (gctx->io_mode) {
+		case ATOM_IO_MM:
+			if (idx == 0)
+				gctx->card->reg_write(gctx->card, idx,
+						      val << 2);
+			else
+				gctx->card->reg_write(gctx->card, idx, val);
+			break;
+		case ATOM_IO_PCI:
+			pr_info("PCI registers are not implemented\n");
+			return;
+		case ATOM_IO_SYSIO:
+			pr_info("SYSIO registers are not implemented\n");
+			return;
+		default:
+			if (!(gctx->io_mode & 0x80)) {
+				pr_info("Bad IO mode\n");
+				return;
+			}
+			if (!gctx->iio[gctx->io_mode & 0xFF]) {
+				pr_info("Undefined indirect IO write method %d\n",
+					gctx->io_mode & 0x7F);
+				return;
+			}
+			atom_iio_execute(gctx, gctx->iio[gctx->io_mode & 0xFF],
+					 idx, val);
+		}
+		break;
+	case ATOM_ARG_PS:
+		idx = U8(*ptr);
+		(*ptr)++;
+		DEBUG("PS[0x%02X]", idx);
+		ctx->ps[idx] = cpu_to_le32(val);
+		break;
+	case ATOM_ARG_WS:
+		idx = U8(*ptr);
+		(*ptr)++;
+		DEBUG("WS[0x%02X]", idx);
+		switch (idx) {
+		case ATOM_WS_QUOTIENT:
+			gctx->divmul[0] = val;
+			break;
+		case ATOM_WS_REMAINDER:
+			gctx->divmul[1] = val;
+			break;
+		case ATOM_WS_DATAPTR:
+			gctx->data_block = val;
+			break;
+		case ATOM_WS_SHIFT:
+			gctx->shift = val;
+			break;
+		case ATOM_WS_OR_MASK:
+		case ATOM_WS_AND_MASK:
+			break;
+		case ATOM_WS_FB_WINDOW:
+			gctx->fb_base = val;
+			break;
+		case ATOM_WS_ATTRIBUTES:
+			gctx->io_attr = val;
+			break;
+		case ATOM_WS_REGPTR:
+			gctx->reg_block = val;
+			break;
+		default:
+			ctx->ws[idx] = val;
+		}
+		break;
+	case ATOM_ARG_FB:
+		idx = U8(*ptr);
+		(*ptr)++;
+		if ((gctx->fb_base + (idx * 4)) > gctx->scratch_size_bytes) {
+			DRM_ERROR("ATOM: fb write beyond scratch region: %d vs. %d\n",
+				  gctx->fb_base + (idx * 4), gctx->scratch_size_bytes);
+		} else
+			gctx->scratch[(gctx->fb_base / 4) + idx] = val;
+		DEBUG("FB[0x%02X]", idx);
+		break;
+	case ATOM_ARG_PLL:
+		idx = U8(*ptr);
+		(*ptr)++;
+		DEBUG("PLL[0x%02X]", idx);
+		gctx->card->pll_write(gctx->card, idx, val);
+		break;
+	case ATOM_ARG_MC:
+		idx = U8(*ptr);
+		(*ptr)++;
+		DEBUG("MC[0x%02X]", idx);
+		gctx->card->mc_write(gctx->card, idx, val);
+		return;
+	}
+	switch (align) {
+	case ATOM_SRC_DWORD:
+		DEBUG(".[31:0] <- 0x%08X\n", old_val);
+		break;
+	case ATOM_SRC_WORD0:
+		DEBUG(".[15:0] <- 0x%04X\n", old_val);
+		break;
+	case ATOM_SRC_WORD8:
+		DEBUG(".[23:8] <- 0x%04X\n", old_val);
+		break;
+	case ATOM_SRC_WORD16:
+		DEBUG(".[31:16] <- 0x%04X\n", old_val);
+		break;
+	case ATOM_SRC_BYTE0:
+		DEBUG(".[7:0] <- 0x%02X\n", old_val);
+		break;
+	case ATOM_SRC_BYTE8:
+		DEBUG(".[15:8] <- 0x%02X\n", old_val);
+		break;
+	case ATOM_SRC_BYTE16:
+		DEBUG(".[23:16] <- 0x%02X\n", old_val);
+		break;
+	case ATOM_SRC_BYTE24:
+		DEBUG(".[31:24] <- 0x%02X\n", old_val);
+		break;
+	}
+}
+
+static void atom_op_add(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src, saved;
+	int dptr = *ptr;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	dst += src;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_and(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src, saved;
+	int dptr = *ptr;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	dst &= src;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_beep(atom_exec_context *ctx, int *ptr, int arg)
+{
+	printk("ATOM BIOS beeped!\n");
+}
+
+static void atom_op_calltable(atom_exec_context *ctx, int *ptr, int arg)
+{
+	int idx = U8((*ptr)++);
+	int r = 0;
+
+	if (idx < ATOM_TABLE_NAMES_CNT)
+		SDEBUG("   table: %d (%s)\n", idx, atom_table_names[idx]);
+	else
+		SDEBUG("   table: %d\n", idx);
+	if (U16(ctx->ctx->cmd_table + 4 + 2 * idx))
+		r = amdgpu_atom_execute_table_locked(ctx->ctx, idx, ctx->ps + ctx->ps_shift);
+	if (r) {
+		ctx->abort = true;
+	}
+}
+
+static void atom_op_clear(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t saved;
+	int dptr = *ptr;
+	attr &= 0x38;
+	attr |= atom_def_dst[attr >> 3] << 6;
+	atom_get_dst(ctx, arg, attr, ptr, &saved, 0);
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, 0, saved);
+}
+
+static void atom_op_compare(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src;
+	SDEBUG("   src1: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1);
+	SDEBUG("   src2: ");
+	src = atom_get_src(ctx, attr, ptr);
+	ctx->ctx->cs_equal = (dst == src);
+	ctx->ctx->cs_above = (dst > src);
+	SDEBUG("   result: %s %s\n", ctx->ctx->cs_equal ? "EQ" : "NE",
+	       ctx->ctx->cs_above ? "GT" : "LE");
+}
+
+static void atom_op_delay(atom_exec_context *ctx, int *ptr, int arg)
+{
+	unsigned count = U8((*ptr)++);
+	SDEBUG("   count: %d\n", count);
+	if (arg == ATOM_UNIT_MICROSEC)
+		udelay(count);
+	else if (!drm_can_sleep())
+		mdelay(count);
+	else
+		drm_msleep(count);
+}
+
+static void atom_op_div(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src;
+	SDEBUG("   src1: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1);
+	SDEBUG("   src2: ");
+	src = atom_get_src(ctx, attr, ptr);
+	if (src != 0) {
+		ctx->ctx->divmul[0] = dst / src;
+		ctx->ctx->divmul[1] = dst % src;
+	} else {
+		ctx->ctx->divmul[0] = 0;
+		ctx->ctx->divmul[1] = 0;
+	}
+}
+
+static void atom_op_div32(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint64_t val64;
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src;
+	SDEBUG("   src1: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1);
+	SDEBUG("   src2: ");
+	src = atom_get_src(ctx, attr, ptr);
+	if (src != 0) {
+		val64 = dst;
+		val64 |= ((uint64_t)ctx->ctx->divmul[1]) << 32;
+		do_div(val64, src);
+		ctx->ctx->divmul[0] = lower_32_bits(val64);
+		ctx->ctx->divmul[1] = upper_32_bits(val64);
+	} else {
+		ctx->ctx->divmul[0] = 0;
+		ctx->ctx->divmul[1] = 0;
+	}
+}
+
+static void atom_op_eot(atom_exec_context *ctx, int *ptr, int arg)
+{
+	/* functionally, a nop */
+}
+
+static void atom_op_jump(atom_exec_context *ctx, int *ptr, int arg)
+{
+	int execute = 0, target = U16(*ptr);
+	unsigned long cjiffies;
+
+	(*ptr) += 2;
+	switch (arg) {
+	case ATOM_COND_ABOVE:
+		execute = ctx->ctx->cs_above;
+		break;
+	case ATOM_COND_ABOVEOREQUAL:
+		execute = ctx->ctx->cs_above || ctx->ctx->cs_equal;
+		break;
+	case ATOM_COND_ALWAYS:
+		execute = 1;
+		break;
+	case ATOM_COND_BELOW:
+		execute = !(ctx->ctx->cs_above || ctx->ctx->cs_equal);
+		break;
+	case ATOM_COND_BELOWOREQUAL:
+		execute = !ctx->ctx->cs_above;
+		break;
+	case ATOM_COND_EQUAL:
+		execute = ctx->ctx->cs_equal;
+		break;
+	case ATOM_COND_NOTEQUAL:
+		execute = !ctx->ctx->cs_equal;
+		break;
+	}
+	if (arg != ATOM_COND_ALWAYS)
+		SDEBUG("   taken: %s\n", execute ? "yes" : "no");
+	SDEBUG("   target: 0x%04X\n", target);
+	if (execute) {
+		if (ctx->last_jump == (ctx->start + target)) {
+			cjiffies = jiffies;
+			if (time_after(cjiffies, ctx->last_jump_jiffies)) {
+				cjiffies -= ctx->last_jump_jiffies;
+				if ((jiffies_to_msecs(cjiffies) > 5000)) {
+					DRM_ERROR("atombios stuck in loop for more than 5secs aborting\n");
+					ctx->abort = true;
+				}
+			} else {
+				/* jiffies wrap around we will just wait a little longer */
+				ctx->last_jump_jiffies = jiffies;
+			}
+		} else {
+			ctx->last_jump = ctx->start + target;
+			ctx->last_jump_jiffies = jiffies;
+		}
+		*ptr = ctx->start + target;
+	}
+}
+
+static void atom_op_mask(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, mask, src, saved;
+	int dptr = *ptr;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	mask = atom_get_src_direct(ctx, ((attr >> 3) & 7), ptr);
+	SDEBUG("   mask: 0x%08x", mask);
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	dst &= mask;
+	dst |= src;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_move(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t src, saved;
+	int dptr = *ptr;
+	if (((attr >> 3) & 7) != ATOM_SRC_DWORD)
+		atom_get_dst(ctx, arg, attr, ptr, &saved, 0);
+	else {
+		atom_skip_dst(ctx, arg, attr, ptr);
+		saved = 0xCDCDCDCD;
+	}
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, src, saved);
+}
+
+static void atom_op_mul(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src;
+	SDEBUG("   src1: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1);
+	SDEBUG("   src2: ");
+	src = atom_get_src(ctx, attr, ptr);
+	ctx->ctx->divmul[0] = dst * src;
+}
+
+static void atom_op_mul32(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint64_t val64;
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src;
+	SDEBUG("   src1: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1);
+	SDEBUG("   src2: ");
+	src = atom_get_src(ctx, attr, ptr);
+	val64 = (uint64_t)dst * (uint64_t)src;
+	ctx->ctx->divmul[0] = lower_32_bits(val64);
+	ctx->ctx->divmul[1] = upper_32_bits(val64);
+}
+
+static void atom_op_nop(atom_exec_context *ctx, int *ptr, int arg)
+{
+	/* nothing */
+}
+
+static void atom_op_or(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src, saved;
+	int dptr = *ptr;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	dst |= src;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_postcard(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t val = U8((*ptr)++);
+	SDEBUG("POST card output: 0x%02X\n", val);
+}
+
+static void atom_op_repeat(atom_exec_context *ctx, int *ptr, int arg)
+{
+	pr_info("unimplemented!\n");
+}
+
+static void atom_op_restorereg(atom_exec_context *ctx, int *ptr, int arg)
+{
+	pr_info("unimplemented!\n");
+}
+
+static void atom_op_savereg(atom_exec_context *ctx, int *ptr, int arg)
+{
+	pr_info("unimplemented!\n");
+}
+
+static void atom_op_setdatablock(atom_exec_context *ctx, int *ptr, int arg)
+{
+	int idx = U8(*ptr);
+	(*ptr)++;
+	SDEBUG("   block: %d\n", idx);
+	if (!idx)
+		ctx->ctx->data_block = 0;
+	else if (idx == 255)
+		ctx->ctx->data_block = ctx->start;
+	else
+		ctx->ctx->data_block = U16(ctx->ctx->data_table + 4 + 2 * idx);
+	SDEBUG("   base: 0x%04X\n", ctx->ctx->data_block);
+}
+
+static void atom_op_setfbbase(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	SDEBUG("   fb_base: ");
+	ctx->ctx->fb_base = atom_get_src(ctx, attr, ptr);
+}
+
+static void atom_op_setport(atom_exec_context *ctx, int *ptr, int arg)
+{
+	int port;
+	switch (arg) {
+	case ATOM_PORT_ATI:
+		port = U16(*ptr);
+		if (port < ATOM_IO_NAMES_CNT)
+			SDEBUG("   port: %d (%s)\n", port, atom_io_names[port]);
+		else
+			SDEBUG("   port: %d\n", port);
+		if (!port)
+			ctx->ctx->io_mode = ATOM_IO_MM;
+		else
+			ctx->ctx->io_mode = ATOM_IO_IIO | port;
+		(*ptr) += 2;
+		break;
+	case ATOM_PORT_PCI:
+		ctx->ctx->io_mode = ATOM_IO_PCI;
+		(*ptr)++;
+		break;
+	case ATOM_PORT_SYSIO:
+		ctx->ctx->io_mode = ATOM_IO_SYSIO;
+		(*ptr)++;
+		break;
+	}
+}
+
+static void atom_op_setregblock(atom_exec_context *ctx, int *ptr, int arg)
+{
+	ctx->ctx->reg_block = U16(*ptr);
+	(*ptr) += 2;
+	SDEBUG("   base: 0x%04X\n", ctx->ctx->reg_block);
+}
+
+static void atom_op_shift_left(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++), shift;
+	uint32_t saved, dst;
+	int dptr = *ptr;
+	attr &= 0x38;
+	attr |= atom_def_dst[attr >> 3] << 6;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	shift = atom_get_src_direct(ctx, ATOM_SRC_BYTE0, ptr);
+	SDEBUG("   shift: %d\n", shift);
+	dst <<= shift;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_shift_right(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++), shift;
+	uint32_t saved, dst;
+	int dptr = *ptr;
+	attr &= 0x38;
+	attr |= atom_def_dst[attr >> 3] << 6;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	shift = atom_get_src_direct(ctx, ATOM_SRC_BYTE0, ptr);
+	SDEBUG("   shift: %d\n", shift);
+	dst >>= shift;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_shl(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++), shift;
+	uint32_t saved, dst;
+	int dptr = *ptr;
+	uint32_t dst_align = atom_dst_to_src[(attr >> 3) & 7][(attr >> 6) & 3];
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	/* op needs to full dst value */
+	dst = saved;
+	shift = atom_get_src(ctx, attr, ptr);
+	SDEBUG("   shift: %d\n", shift);
+	dst <<= shift;
+	dst &= atom_arg_mask[dst_align];
+	dst >>= atom_arg_shift[dst_align];
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_shr(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++), shift;
+	uint32_t saved, dst;
+	int dptr = *ptr;
+	uint32_t dst_align = atom_dst_to_src[(attr >> 3) & 7][(attr >> 6) & 3];
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	/* op needs to full dst value */
+	dst = saved;
+	shift = atom_get_src(ctx, attr, ptr);
+	SDEBUG("   shift: %d\n", shift);
+	dst >>= shift;
+	dst &= atom_arg_mask[dst_align];
+	dst >>= atom_arg_shift[dst_align];
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_sub(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src, saved;
+	int dptr = *ptr;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	dst -= src;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_switch(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t src, val, target;
+	SDEBUG("   switch: ");
+	src = atom_get_src(ctx, attr, ptr);
+	while (U16(*ptr) != ATOM_CASE_END)
+		if (U8(*ptr) == ATOM_CASE_MAGIC) {
+			(*ptr)++;
+			SDEBUG("   case: ");
+			val =
+			    atom_get_src(ctx, (attr & 0x38) | ATOM_ARG_IMM,
+					 ptr);
+			target = U16(*ptr);
+			if (val == src) {
+				SDEBUG("   target: %04X\n", target);
+				*ptr = ctx->start + target;
+				return;
+			}
+			(*ptr) += 2;
+		} else {
+			pr_info("Bad case\n");
+			return;
+		}
+	(*ptr) += 2;
+}
+
+static void atom_op_test(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src;
+	SDEBUG("   src1: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1);
+	SDEBUG("   src2: ");
+	src = atom_get_src(ctx, attr, ptr);
+	ctx->ctx->cs_equal = ((dst & src) == 0);
+	SDEBUG("   result: %s\n", ctx->ctx->cs_equal ? "EQ" : "NE");
+}
+
+static void atom_op_xor(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src, saved;
+	int dptr = *ptr;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	dst ^= src;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_debug(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t val = U8((*ptr)++);
+	SDEBUG("DEBUG output: 0x%02X\n", val);
+}
+
+static void atom_op_processds(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint16_t val = U16(*ptr);
+	(*ptr) += val + 2;
+	SDEBUG("PROCESSDS output: 0x%02X\n", val);
+}
+
+static struct {
+	void (*func) (atom_exec_context *, int *, int);
+	int arg;
+} opcode_table[ATOM_OP_CNT] = {
+	{
+	NULL, 0}, {
+	atom_op_move, ATOM_ARG_REG}, {
+	atom_op_move, ATOM_ARG_PS}, {
+	atom_op_move, ATOM_ARG_WS}, {
+	atom_op_move, ATOM_ARG_FB}, {
+	atom_op_move, ATOM_ARG_PLL}, {
+	atom_op_move, ATOM_ARG_MC}, {
+	atom_op_and, ATOM_ARG_REG}, {
+	atom_op_and, ATOM_ARG_PS}, {
+	atom_op_and, ATOM_ARG_WS}, {
+	atom_op_and, ATOM_ARG_FB}, {
+	atom_op_and, ATOM_ARG_PLL}, {
+	atom_op_and, ATOM_ARG_MC}, {
+	atom_op_or, ATOM_ARG_REG}, {
+	atom_op_or, ATOM_ARG_PS}, {
+	atom_op_or, ATOM_ARG_WS}, {
+	atom_op_or, ATOM_ARG_FB}, {
+	atom_op_or, ATOM_ARG_PLL}, {
+	atom_op_or, ATOM_ARG_MC}, {
+	atom_op_shift_left, ATOM_ARG_REG}, {
+	atom_op_shift_left, ATOM_ARG_PS}, {
+	atom_op_shift_left, ATOM_ARG_WS}, {
+	atom_op_shift_left, ATOM_ARG_FB}, {
+	atom_op_shift_left, ATOM_ARG_PLL}, {
+	atom_op_shift_left, ATOM_ARG_MC}, {
+	atom_op_shift_right, ATOM_ARG_REG}, {
+	atom_op_shift_right, ATOM_ARG_PS}, {
+	atom_op_shift_right, ATOM_ARG_WS}, {
+	atom_op_shift_right, ATOM_ARG_FB}, {
+	atom_op_shift_right, ATOM_ARG_PLL}, {
+	atom_op_shift_right, ATOM_ARG_MC}, {
+	atom_op_mul, ATOM_ARG_REG}, {
+	atom_op_mul, ATOM_ARG_PS}, {
+	atom_op_mul, ATOM_ARG_WS}, {
+	atom_op_mul, ATOM_ARG_FB}, {
+	atom_op_mul, ATOM_ARG_PLL}, {
+	atom_op_mul, ATOM_ARG_MC}, {
+	atom_op_div, ATOM_ARG_REG}, {
+	atom_op_div, ATOM_ARG_PS}, {
+	atom_op_div, ATOM_ARG_WS}, {
+	atom_op_div, ATOM_ARG_FB}, {
+	atom_op_div, ATOM_ARG_PLL}, {
+	atom_op_div, ATOM_ARG_MC}, {
+	atom_op_add, ATOM_ARG_REG}, {
+	atom_op_add, ATOM_ARG_PS}, {
+	atom_op_add, ATOM_ARG_WS}, {
+	atom_op_add, ATOM_ARG_FB}, {
+	atom_op_add, ATOM_ARG_PLL}, {
+	atom_op_add, ATOM_ARG_MC}, {
+	atom_op_sub, ATOM_ARG_REG}, {
+	atom_op_sub, ATOM_ARG_PS}, {
+	atom_op_sub, ATOM_ARG_WS}, {
+	atom_op_sub, ATOM_ARG_FB}, {
+	atom_op_sub, ATOM_ARG_PLL}, {
+	atom_op_sub, ATOM_ARG_MC}, {
+	atom_op_setport, ATOM_PORT_ATI}, {
+	atom_op_setport, ATOM_PORT_PCI}, {
+	atom_op_setport, ATOM_PORT_SYSIO}, {
+	atom_op_setregblock, 0}, {
+	atom_op_setfbbase, 0}, {
+	atom_op_compare, ATOM_ARG_REG}, {
+	atom_op_compare, ATOM_ARG_PS}, {
+	atom_op_compare, ATOM_ARG_WS}, {
+	atom_op_compare, ATOM_ARG_FB}, {
+	atom_op_compare, ATOM_ARG_PLL}, {
+	atom_op_compare, ATOM_ARG_MC}, {
+	atom_op_switch, 0}, {
+	atom_op_jump, ATOM_COND_ALWAYS}, {
+	atom_op_jump, ATOM_COND_EQUAL}, {
+	atom_op_jump, ATOM_COND_BELOW}, {
+	atom_op_jump, ATOM_COND_ABOVE}, {
+	atom_op_jump, ATOM_COND_BELOWOREQUAL}, {
+	atom_op_jump, ATOM_COND_ABOVEOREQUAL}, {
+	atom_op_jump, ATOM_COND_NOTEQUAL}, {
+	atom_op_test, ATOM_ARG_REG}, {
+	atom_op_test, ATOM_ARG_PS}, {
+	atom_op_test, ATOM_ARG_WS}, {
+	atom_op_test, ATOM_ARG_FB}, {
+	atom_op_test, ATOM_ARG_PLL}, {
+	atom_op_test, ATOM_ARG_MC}, {
+	atom_op_delay, ATOM_UNIT_MILLISEC}, {
+	atom_op_delay, ATOM_UNIT_MICROSEC}, {
+	atom_op_calltable, 0}, {
+	atom_op_repeat, 0}, {
+	atom_op_clear, ATOM_ARG_REG}, {
+	atom_op_clear, ATOM_ARG_PS}, {
+	atom_op_clear, ATOM_ARG_WS}, {
+	atom_op_clear, ATOM_ARG_FB}, {
+	atom_op_clear, ATOM_ARG_PLL}, {
+	atom_op_clear, ATOM_ARG_MC}, {
+	atom_op_nop, 0}, {
+	atom_op_eot, 0}, {
+	atom_op_mask, ATOM_ARG_REG}, {
+	atom_op_mask, ATOM_ARG_PS}, {
+	atom_op_mask, ATOM_ARG_WS}, {
+	atom_op_mask, ATOM_ARG_FB}, {
+	atom_op_mask, ATOM_ARG_PLL}, {
+	atom_op_mask, ATOM_ARG_MC}, {
+	atom_op_postcard, 0}, {
+	atom_op_beep, 0}, {
+	atom_op_savereg, 0}, {
+	atom_op_restorereg, 0}, {
+	atom_op_setdatablock, 0}, {
+	atom_op_xor, ATOM_ARG_REG}, {
+	atom_op_xor, ATOM_ARG_PS}, {
+	atom_op_xor, ATOM_ARG_WS}, {
+	atom_op_xor, ATOM_ARG_FB}, {
+	atom_op_xor, ATOM_ARG_PLL}, {
+	atom_op_xor, ATOM_ARG_MC}, {
+	atom_op_shl, ATOM_ARG_REG}, {
+	atom_op_shl, ATOM_ARG_PS}, {
+	atom_op_shl, ATOM_ARG_WS}, {
+	atom_op_shl, ATOM_ARG_FB}, {
+	atom_op_shl, ATOM_ARG_PLL}, {
+	atom_op_shl, ATOM_ARG_MC}, {
+	atom_op_shr, ATOM_ARG_REG}, {
+	atom_op_shr, ATOM_ARG_PS}, {
+	atom_op_shr, ATOM_ARG_WS}, {
+	atom_op_shr, ATOM_ARG_FB}, {
+	atom_op_shr, ATOM_ARG_PLL}, {
+	atom_op_shr, ATOM_ARG_MC}, {
+	atom_op_debug, 0}, {
+	atom_op_processds, 0}, {
+	atom_op_mul32, ATOM_ARG_PS}, {
+	atom_op_mul32, ATOM_ARG_WS}, {
+	atom_op_div32, ATOM_ARG_PS}, {
+	atom_op_div32, ATOM_ARG_WS},
+};
+
+static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t * params)
+{
+	int base = CU16(ctx->cmd_table + 4 + 2 * index);
+	int len, ws, ps, ptr;
+	unsigned char op;
+	atom_exec_context ectx;
+	int ret = 0;
+
+	if (!base)
+		return -EINVAL;
+
+	len = CU16(base + ATOM_CT_SIZE_PTR);
+	ws = CU8(base + ATOM_CT_WS_PTR);
+	ps = CU8(base + ATOM_CT_PS_PTR) & ATOM_CT_PS_MASK;
+	ptr = base + ATOM_CT_CODE_PTR;
+
+	SDEBUG(">> execute %04X (len %d, WS %d, PS %d)\n", base, len, ws, ps);
+
+	ectx.ctx = ctx;
+	ectx.ps_shift = ps / 4;
+	ectx.start = base;
+	ectx.ps = params;
+	ectx.abort = false;
+	ectx.last_jump = 0;
+	if (ws)
+		ectx.ws = kcalloc(4, ws, GFP_KERNEL);
+	else
+		ectx.ws = NULL;
+
+	debug_depth++;
+	while (1) {
+		op = CU8(ptr++);
+		if (op < ATOM_OP_NAMES_CNT)
+			SDEBUG("%s @ 0x%04X\n", atom_op_names[op], ptr - 1);
+		else
+			SDEBUG("[%d] @ 0x%04X\n", op, ptr - 1);
+		if (ectx.abort) {
+			DRM_ERROR("atombios stuck executing %04X (len %d, WS %d, PS %d) @ 0x%04X\n",
+				base, len, ws, ps, ptr - 1);
+			ret = -EINVAL;
+			goto free;
+		}
+
+		if (op < ATOM_OP_CNT && op > 0)
+			opcode_table[op].func(&ectx, &ptr,
+					      opcode_table[op].arg);
+		else
+			break;
+
+		if (op == ATOM_OP_EOT)
+			break;
+	}
+	debug_depth--;
+	SDEBUG("<<\n");
+
+free:
+	if (ws)
+		kfree(ectx.ws);
+	return ret;
+}
+
+int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t * params)
+{
+	int r;
+
+	mutex_lock(&ctx->mutex);
+	/* reset data block */
+	ctx->data_block = 0;
+	/* reset reg block */
+	ctx->reg_block = 0;
+	/* reset fb window */
+	ctx->fb_base = 0;
+	/* reset io mode */
+	ctx->io_mode = ATOM_IO_MM;
+	/* reset divmul */
+	ctx->divmul[0] = 0;
+	ctx->divmul[1] = 0;
+	r = amdgpu_atom_execute_table_locked(ctx, index, params);
+	mutex_unlock(&ctx->mutex);
+	return r;
+}
+
+static int atom_iio_len[] = { 1, 2, 3, 3, 3, 3, 4, 4, 4, 3 };
+
+static void atom_index_iio(struct atom_context *ctx, int base)
+{
+	ctx->iio = kzalloc(2 * 256, GFP_KERNEL);
+	if (!ctx->iio)
+		return;
+	while (CU8(base) == ATOM_IIO_START) {
+		ctx->iio[CU8(base + 1)] = base + 2;
+		base += 2;
+		while (CU8(base) != ATOM_IIO_END)
+			base += atom_iio_len[CU8(base)];
+		base += 3;
+	}
+}
+
+struct atom_context *amdgpu_atom_parse(struct card_info *card, void *bios)
+{
+	int base;
+	struct atom_context *ctx =
+	    kzalloc(sizeof(struct atom_context), GFP_KERNEL);
+	char *str;
+	u16 idx;
+
+	if (!ctx)
+		return NULL;
+
+	ctx->card = card;
+	ctx->bios = bios;
+
+	if (CU16(0) != ATOM_BIOS_MAGIC) {
+		pr_info("Invalid BIOS magic\n");
+		kfree(ctx);
+		return NULL;
+	}
+	if (strncmp
+	    (CSTR(ATOM_ATI_MAGIC_PTR), ATOM_ATI_MAGIC,
+	     strlen(ATOM_ATI_MAGIC))) {
+		pr_info("Invalid ATI magic\n");
+		kfree(ctx);
+		return NULL;
+	}
+
+	base = CU16(ATOM_ROM_TABLE_PTR);
+	if (strncmp
+	    (CSTR(base + ATOM_ROM_MAGIC_PTR), ATOM_ROM_MAGIC,
+	     strlen(ATOM_ROM_MAGIC))) {
+		pr_info("Invalid ATOM magic\n");
+		kfree(ctx);
+		return NULL;
+	}
+
+	ctx->cmd_table = CU16(base + ATOM_ROM_CMD_PTR);
+	ctx->data_table = CU16(base + ATOM_ROM_DATA_PTR);
+	atom_index_iio(ctx, CU16(ctx->data_table + ATOM_DATA_IIO_PTR) + 4);
+	if (!ctx->iio) {
+		amdgpu_atom_destroy(ctx);
+		return NULL;
+	}
+
+	idx = CU16(ATOM_ROM_PART_NUMBER_PTR);
+	if (idx == 0)
+		idx = 0x80;
+
+	str = CSTR(idx);
+	if (*str != '\0') {
+		pr_info("ATOM BIOS: %s\n", str);
+		strlcpy(ctx->vbios_version, str, sizeof(ctx->vbios_version));
+	}
+
+
+	return ctx;
+}
+
+int amdgpu_atom_asic_init(struct atom_context *ctx)
+{
+	int hwi = CU16(ctx->data_table + ATOM_DATA_FWI_PTR);
+	uint32_t ps[16];
+	int ret;
+
+	memset(ps, 0, 64);
+
+	ps[0] = cpu_to_le32(CU32(hwi + ATOM_FWI_DEFSCLK_PTR));
+	ps[1] = cpu_to_le32(CU32(hwi + ATOM_FWI_DEFMCLK_PTR));
+	if (!ps[0] || !ps[1])
+		return 1;
+
+	if (!CU16(ctx->cmd_table + 4 + 2 * ATOM_CMD_INIT))
+		return 1;
+	ret = amdgpu_atom_execute_table(ctx, ATOM_CMD_INIT, ps);
+	if (ret)
+		return ret;
+
+	memset(ps, 0, 64);
+
+	return ret;
+}
+
+void amdgpu_atom_destroy(struct atom_context *ctx)
+{
+	kfree(ctx->iio);
+	kfree(ctx);
+}
+
+bool amdgpu_atom_parse_data_header(struct atom_context *ctx, int index,
+			    uint16_t * size, uint8_t * frev, uint8_t * crev,
+			    uint16_t * data_start)
+{
+	int offset = index * 2 + 4;
+	int idx = CU16(ctx->data_table + offset);
+	u16 *mdt = (u16 *)(ctx->bios + ctx->data_table + 4);
+
+	if (!mdt[index])
+		return false;
+
+	if (size)
+		*size = CU16(idx);
+	if (frev)
+		*frev = CU8(idx + 2);
+	if (crev)
+		*crev = CU8(idx + 3);
+	*data_start = idx;
+	return true;
+}
+
+bool amdgpu_atom_parse_cmd_header(struct atom_context *ctx, int index, uint8_t * frev,
+			   uint8_t * crev)
+{
+	int offset = index * 2 + 4;
+	int idx = CU16(ctx->cmd_table + offset);
+	u16 *mct = (u16 *)(ctx->bios + ctx->cmd_table + 4);
+
+	if (!mct[index])
+		return false;
+
+	if (frev)
+		*frev = CU8(idx + 2);
+	if (crev)
+		*crev = CU8(idx + 3);
+	return true;
+}
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios.c
new file mode 100644
index 00000000000..9a60b8441ea
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios.c
@@ -0,0 +1,2053 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_atomfirmware.h"
+#include "amdgpu_i2c.h"
+
+#include "atom.h"
+#include "atom-bits.h"
+#include "atombios_encoders.h"
+#include "bif/bif_4_1_d.h"
+
+static void amdgpu_atombios_lookup_i2c_gpio_quirks(struct amdgpu_device *adev,
+					  ATOM_GPIO_I2C_ASSIGMENT *gpio,
+					  u8 index)
+{
+
+}
+
+static struct amdgpu_i2c_bus_rec amdgpu_atombios_get_bus_rec_for_i2c_gpio(ATOM_GPIO_I2C_ASSIGMENT *gpio)
+{
+	struct amdgpu_i2c_bus_rec i2c;
+
+	memset(&i2c, 0, sizeof(struct amdgpu_i2c_bus_rec));
+
+	i2c.mask_clk_reg = le16_to_cpu(gpio->usClkMaskRegisterIndex);
+	i2c.mask_data_reg = le16_to_cpu(gpio->usDataMaskRegisterIndex);
+	i2c.en_clk_reg = le16_to_cpu(gpio->usClkEnRegisterIndex);
+	i2c.en_data_reg = le16_to_cpu(gpio->usDataEnRegisterIndex);
+	i2c.y_clk_reg = le16_to_cpu(gpio->usClkY_RegisterIndex);
+	i2c.y_data_reg = le16_to_cpu(gpio->usDataY_RegisterIndex);
+	i2c.a_clk_reg = le16_to_cpu(gpio->usClkA_RegisterIndex);
+	i2c.a_data_reg = le16_to_cpu(gpio->usDataA_RegisterIndex);
+	i2c.mask_clk_mask = (1 << gpio->ucClkMaskShift);
+	i2c.mask_data_mask = (1 << gpio->ucDataMaskShift);
+	i2c.en_clk_mask = (1 << gpio->ucClkEnShift);
+	i2c.en_data_mask = (1 << gpio->ucDataEnShift);
+	i2c.y_clk_mask = (1 << gpio->ucClkY_Shift);
+	i2c.y_data_mask = (1 << gpio->ucDataY_Shift);
+	i2c.a_clk_mask = (1 << gpio->ucClkA_Shift);
+	i2c.a_data_mask = (1 << gpio->ucDataA_Shift);
+
+	if (gpio->sucI2cId.sbfAccess.bfHW_Capable)
+		i2c.hw_capable = true;
+	else
+		i2c.hw_capable = false;
+
+	if (gpio->sucI2cId.ucAccess == 0xa0)
+		i2c.mm_i2c = true;
+	else
+		i2c.mm_i2c = false;
+
+	i2c.i2c_id = gpio->sucI2cId.ucAccess;
+
+	if (i2c.mask_clk_reg)
+		i2c.valid = true;
+	else
+		i2c.valid = false;
+
+	return i2c;
+}
+
+struct amdgpu_i2c_bus_rec amdgpu_atombios_lookup_i2c_gpio(struct amdgpu_device *adev,
+							  uint8_t id)
+{
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	ATOM_GPIO_I2C_ASSIGMENT *gpio;
+	struct amdgpu_i2c_bus_rec i2c;
+	int index = GetIndexIntoMasterTable(DATA, GPIO_I2C_Info);
+	struct _ATOM_GPIO_I2C_INFO *i2c_info;
+	uint16_t data_offset, size;
+	int i, num_indices;
+
+	memset(&i2c, 0, sizeof(struct amdgpu_i2c_bus_rec));
+	i2c.valid = false;
+
+	if (amdgpu_atom_parse_data_header(ctx, index, &size, NULL, NULL, &data_offset)) {
+		i2c_info = (struct _ATOM_GPIO_I2C_INFO *)(ctx->bios + data_offset);
+
+		num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+			sizeof(ATOM_GPIO_I2C_ASSIGMENT);
+
+		gpio = &i2c_info->asGPIO_Info[0];
+		for (i = 0; i < num_indices; i++) {
+
+			amdgpu_atombios_lookup_i2c_gpio_quirks(adev, gpio, i);
+
+			if (gpio->sucI2cId.ucAccess == id) {
+				i2c = amdgpu_atombios_get_bus_rec_for_i2c_gpio(gpio);
+				break;
+			}
+			gpio = (ATOM_GPIO_I2C_ASSIGMENT *)
+				((u8 *)gpio + sizeof(ATOM_GPIO_I2C_ASSIGMENT));
+		}
+	}
+
+	return i2c;
+}
+
+void amdgpu_atombios_i2c_init(struct amdgpu_device *adev)
+{
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	ATOM_GPIO_I2C_ASSIGMENT *gpio;
+	struct amdgpu_i2c_bus_rec i2c;
+	int index = GetIndexIntoMasterTable(DATA, GPIO_I2C_Info);
+	struct _ATOM_GPIO_I2C_INFO *i2c_info;
+	uint16_t data_offset, size;
+	int i, num_indices;
+	char stmp[32];
+
+	if (amdgpu_atom_parse_data_header(ctx, index, &size, NULL, NULL, &data_offset)) {
+		i2c_info = (struct _ATOM_GPIO_I2C_INFO *)(ctx->bios + data_offset);
+
+		num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+			sizeof(ATOM_GPIO_I2C_ASSIGMENT);
+
+		gpio = &i2c_info->asGPIO_Info[0];
+		for (i = 0; i < num_indices; i++) {
+			amdgpu_atombios_lookup_i2c_gpio_quirks(adev, gpio, i);
+
+			i2c = amdgpu_atombios_get_bus_rec_for_i2c_gpio(gpio);
+
+			if (i2c.valid) {
+				snprintf(stmp, sizeof(stmp), "0x%x", i2c.i2c_id);
+				adev->i2c_bus[i] = amdgpu_i2c_create(adev->ddev, &i2c, stmp);
+			}
+			gpio = (ATOM_GPIO_I2C_ASSIGMENT *)
+				((u8 *)gpio + sizeof(ATOM_GPIO_I2C_ASSIGMENT));
+		}
+	}
+}
+
+struct amdgpu_gpio_rec
+amdgpu_atombios_lookup_gpio(struct amdgpu_device *adev,
+			    u8 id)
+{
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	struct amdgpu_gpio_rec gpio;
+	int index = GetIndexIntoMasterTable(DATA, GPIO_Pin_LUT);
+	struct _ATOM_GPIO_PIN_LUT *gpio_info;
+	ATOM_GPIO_PIN_ASSIGNMENT *pin;
+	u16 data_offset, size;
+	int i, num_indices;
+
+	memset(&gpio, 0, sizeof(struct amdgpu_gpio_rec));
+	gpio.valid = false;
+
+	if (amdgpu_atom_parse_data_header(ctx, index, &size, NULL, NULL, &data_offset)) {
+		gpio_info = (struct _ATOM_GPIO_PIN_LUT *)(ctx->bios + data_offset);
+
+		num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+			sizeof(ATOM_GPIO_PIN_ASSIGNMENT);
+
+		pin = gpio_info->asGPIO_Pin;
+		for (i = 0; i < num_indices; i++) {
+			if (id == pin->ucGPIO_ID) {
+				gpio.id = pin->ucGPIO_ID;
+				gpio.reg = le16_to_cpu(pin->usGpioPin_AIndex);
+				gpio.shift = pin->ucGpioPinBitShift;
+				gpio.mask = (1 << pin->ucGpioPinBitShift);
+				gpio.valid = true;
+				break;
+			}
+			pin = (ATOM_GPIO_PIN_ASSIGNMENT *)
+				((u8 *)pin + sizeof(ATOM_GPIO_PIN_ASSIGNMENT));
+		}
+	}
+
+	return gpio;
+}
+
+static struct amdgpu_hpd
+amdgpu_atombios_get_hpd_info_from_gpio(struct amdgpu_device *adev,
+				       struct amdgpu_gpio_rec *gpio)
+{
+	struct amdgpu_hpd hpd;
+	u32 reg;
+
+	memset(&hpd, 0, sizeof(struct amdgpu_hpd));
+
+	reg = amdgpu_display_hpd_get_gpio_reg(adev);
+
+	hpd.gpio = *gpio;
+	if (gpio->reg == reg) {
+		switch(gpio->mask) {
+		case (1 << 0):
+			hpd.hpd = AMDGPU_HPD_1;
+			break;
+		case (1 << 8):
+			hpd.hpd = AMDGPU_HPD_2;
+			break;
+		case (1 << 16):
+			hpd.hpd = AMDGPU_HPD_3;
+			break;
+		case (1 << 24):
+			hpd.hpd = AMDGPU_HPD_4;
+			break;
+		case (1 << 26):
+			hpd.hpd = AMDGPU_HPD_5;
+			break;
+		case (1 << 28):
+			hpd.hpd = AMDGPU_HPD_6;
+			break;
+		default:
+			hpd.hpd = AMDGPU_HPD_NONE;
+			break;
+		}
+	} else
+		hpd.hpd = AMDGPU_HPD_NONE;
+	return hpd;
+}
+
+static const int object_connector_convert[] = {
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_DVII,
+	DRM_MODE_CONNECTOR_DVII,
+	DRM_MODE_CONNECTOR_DVID,
+	DRM_MODE_CONNECTOR_DVID,
+	DRM_MODE_CONNECTOR_VGA,
+	DRM_MODE_CONNECTOR_Composite,
+	DRM_MODE_CONNECTOR_SVIDEO,
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_9PinDIN,
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_HDMIA,
+	DRM_MODE_CONNECTOR_HDMIB,
+	DRM_MODE_CONNECTOR_LVDS,
+	DRM_MODE_CONNECTOR_9PinDIN,
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_DisplayPort,
+	DRM_MODE_CONNECTOR_eDP,
+	DRM_MODE_CONNECTOR_Unknown
+};
+
+bool amdgpu_atombios_has_dce_engine_info(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	struct atom_context *ctx = mode_info->atom_context;
+	int index = GetIndexIntoMasterTable(DATA, Object_Header);
+	u16 size, data_offset;
+	u8 frev, crev;
+	ATOM_DISPLAY_OBJECT_PATH_TABLE *path_obj;
+	ATOM_OBJECT_HEADER *obj_header;
+
+	if (!amdgpu_atom_parse_data_header(ctx, index, &size, &frev, &crev, &data_offset))
+		return false;
+
+	if (crev < 2)
+		return false;
+
+	obj_header = (ATOM_OBJECT_HEADER *) (ctx->bios + data_offset);
+	path_obj = (ATOM_DISPLAY_OBJECT_PATH_TABLE *)
+	    (ctx->bios + data_offset +
+	     le16_to_cpu(obj_header->usDisplayPathTableOffset));
+
+	if (path_obj->ucNumOfDispPath)
+		return true;
+	else
+		return false;
+}
+
+bool amdgpu_atombios_get_connector_info_from_object_table(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	struct atom_context *ctx = mode_info->atom_context;
+	int index = GetIndexIntoMasterTable(DATA, Object_Header);
+	u16 size, data_offset;
+	u8 frev, crev;
+	ATOM_CONNECTOR_OBJECT_TABLE *con_obj;
+	ATOM_ENCODER_OBJECT_TABLE *enc_obj;
+	ATOM_OBJECT_TABLE *router_obj;
+	ATOM_DISPLAY_OBJECT_PATH_TABLE *path_obj;
+	ATOM_OBJECT_HEADER *obj_header;
+	int i, j, k, path_size, device_support;
+	int connector_type;
+	u16 conn_id, connector_object_id;
+	struct amdgpu_i2c_bus_rec ddc_bus;
+	struct amdgpu_router router;
+	struct amdgpu_gpio_rec gpio;
+	struct amdgpu_hpd hpd;
+
+	if (!amdgpu_atom_parse_data_header(ctx, index, &size, &frev, &crev, &data_offset))
+		return false;
+
+	if (crev < 2)
+		return false;
+
+	obj_header = (ATOM_OBJECT_HEADER *) (ctx->bios + data_offset);
+	path_obj = (ATOM_DISPLAY_OBJECT_PATH_TABLE *)
+	    (ctx->bios + data_offset +
+	     le16_to_cpu(obj_header->usDisplayPathTableOffset));
+	con_obj = (ATOM_CONNECTOR_OBJECT_TABLE *)
+	    (ctx->bios + data_offset +
+	     le16_to_cpu(obj_header->usConnectorObjectTableOffset));
+	enc_obj = (ATOM_ENCODER_OBJECT_TABLE *)
+	    (ctx->bios + data_offset +
+	     le16_to_cpu(obj_header->usEncoderObjectTableOffset));
+	router_obj = (ATOM_OBJECT_TABLE *)
+		(ctx->bios + data_offset +
+		 le16_to_cpu(obj_header->usRouterObjectTableOffset));
+	device_support = le16_to_cpu(obj_header->usDeviceSupport);
+
+	path_size = 0;
+	for (i = 0; i < path_obj->ucNumOfDispPath; i++) {
+		uint8_t *addr = (uint8_t *) path_obj->asDispPath;
+		ATOM_DISPLAY_OBJECT_PATH *path;
+		addr += path_size;
+		path = (ATOM_DISPLAY_OBJECT_PATH *) addr;
+		path_size += le16_to_cpu(path->usSize);
+
+		if (device_support & le16_to_cpu(path->usDeviceTag)) {
+			uint8_t con_obj_id, con_obj_num, con_obj_type;
+
+			con_obj_id =
+			    (le16_to_cpu(path->usConnObjectId) & OBJECT_ID_MASK)
+			    >> OBJECT_ID_SHIFT;
+			con_obj_num =
+			    (le16_to_cpu(path->usConnObjectId) & ENUM_ID_MASK)
+			    >> ENUM_ID_SHIFT;
+			con_obj_type =
+			    (le16_to_cpu(path->usConnObjectId) &
+			     OBJECT_TYPE_MASK) >> OBJECT_TYPE_SHIFT;
+
+			/* Skip TV/CV support */
+			if ((le16_to_cpu(path->usDeviceTag) ==
+			     ATOM_DEVICE_TV1_SUPPORT) ||
+			    (le16_to_cpu(path->usDeviceTag) ==
+			     ATOM_DEVICE_CV_SUPPORT))
+				continue;
+
+			if (con_obj_id >= ARRAY_SIZE(object_connector_convert)) {
+				DRM_ERROR("invalid con_obj_id %d for device tag 0x%04x\n",
+					  con_obj_id, le16_to_cpu(path->usDeviceTag));
+				continue;
+			}
+
+			connector_type =
+				object_connector_convert[con_obj_id];
+			connector_object_id = con_obj_id;
+
+			if (connector_type == DRM_MODE_CONNECTOR_Unknown)
+				continue;
+
+			router.ddc_valid = false;
+			router.cd_valid = false;
+			for (j = 0; j < ((le16_to_cpu(path->usSize) - 8) / 2); j++) {
+				uint8_t grph_obj_id, grph_obj_num, grph_obj_type;
+
+				grph_obj_id =
+				    (le16_to_cpu(path->usGraphicObjIds[j]) &
+				     OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+				grph_obj_num =
+				    (le16_to_cpu(path->usGraphicObjIds[j]) &
+				     ENUM_ID_MASK) >> ENUM_ID_SHIFT;
+				grph_obj_type =
+				    (le16_to_cpu(path->usGraphicObjIds[j]) &
+				     OBJECT_TYPE_MASK) >> OBJECT_TYPE_SHIFT;
+
+				if (grph_obj_type == GRAPH_OBJECT_TYPE_ENCODER) {
+					for (k = 0; k < enc_obj->ucNumberOfObjects; k++) {
+						u16 encoder_obj = le16_to_cpu(enc_obj->asObjects[k].usObjectID);
+						if (le16_to_cpu(path->usGraphicObjIds[j]) == encoder_obj) {
+							ATOM_COMMON_RECORD_HEADER *record = (ATOM_COMMON_RECORD_HEADER *)
+								(ctx->bios + data_offset +
+								 le16_to_cpu(enc_obj->asObjects[k].usRecordOffset));
+							ATOM_ENCODER_CAP_RECORD *cap_record;
+							u16 caps = 0;
+
+							while (record->ucRecordSize > 0 &&
+							       record->ucRecordType > 0 &&
+							       record->ucRecordType <= ATOM_MAX_OBJECT_RECORD_NUMBER) {
+								switch (record->ucRecordType) {
+								case ATOM_ENCODER_CAP_RECORD_TYPE:
+									cap_record =(ATOM_ENCODER_CAP_RECORD *)
+										record;
+									caps = le16_to_cpu(cap_record->usEncoderCap);
+									break;
+								}
+								record = (ATOM_COMMON_RECORD_HEADER *)
+									((char *)record + record->ucRecordSize);
+							}
+							amdgpu_display_add_encoder(adev, encoder_obj,
+										    le16_to_cpu(path->usDeviceTag),
+										    caps);
+						}
+					}
+				} else if (grph_obj_type == GRAPH_OBJECT_TYPE_ROUTER) {
+					for (k = 0; k < router_obj->ucNumberOfObjects; k++) {
+						u16 router_obj_id = le16_to_cpu(router_obj->asObjects[k].usObjectID);
+						if (le16_to_cpu(path->usGraphicObjIds[j]) == router_obj_id) {
+							ATOM_COMMON_RECORD_HEADER *record = (ATOM_COMMON_RECORD_HEADER *)
+								(ctx->bios + data_offset +
+								 le16_to_cpu(router_obj->asObjects[k].usRecordOffset));
+							ATOM_I2C_RECORD *i2c_record;
+							ATOM_I2C_ID_CONFIG_ACCESS *i2c_config;
+							ATOM_ROUTER_DDC_PATH_SELECT_RECORD *ddc_path;
+							ATOM_ROUTER_DATA_CLOCK_PATH_SELECT_RECORD *cd_path;
+							ATOM_SRC_DST_TABLE_FOR_ONE_OBJECT *router_src_dst_table =
+								(ATOM_SRC_DST_TABLE_FOR_ONE_OBJECT *)
+								(ctx->bios + data_offset +
+								 le16_to_cpu(router_obj->asObjects[k].usSrcDstTableOffset));
+							u8 *num_dst_objs = (u8 *)
+								((u8 *)router_src_dst_table + 1 +
+								 (router_src_dst_table->ucNumberOfSrc * 2));
+							u16 *dst_objs = (u16 *)(num_dst_objs + 1);
+							int enum_id;
+
+							router.router_id = router_obj_id;
+							for (enum_id = 0; enum_id < (*num_dst_objs); enum_id++) {
+								if (le16_to_cpu(path->usConnObjectId) ==
+								    le16_to_cpu(dst_objs[enum_id]))
+									break;
+							}
+
+							while (record->ucRecordSize > 0 &&
+							       record->ucRecordType > 0 &&
+							       record->ucRecordType <= ATOM_MAX_OBJECT_RECORD_NUMBER) {
+								switch (record->ucRecordType) {
+								case ATOM_I2C_RECORD_TYPE:
+									i2c_record =
+										(ATOM_I2C_RECORD *)
+										record;
+									i2c_config =
+										(ATOM_I2C_ID_CONFIG_ACCESS *)
+										&i2c_record->sucI2cId;
+									router.i2c_info =
+										amdgpu_atombios_lookup_i2c_gpio(adev,
+												       i2c_config->
+												       ucAccess);
+									router.i2c_addr = i2c_record->ucI2CAddr >> 1;
+									break;
+								case ATOM_ROUTER_DDC_PATH_SELECT_RECORD_TYPE:
+									ddc_path = (ATOM_ROUTER_DDC_PATH_SELECT_RECORD *)
+										record;
+									router.ddc_valid = true;
+									router.ddc_mux_type = ddc_path->ucMuxType;
+									router.ddc_mux_control_pin = ddc_path->ucMuxControlPin;
+									router.ddc_mux_state = ddc_path->ucMuxState[enum_id];
+									break;
+								case ATOM_ROUTER_DATA_CLOCK_PATH_SELECT_RECORD_TYPE:
+									cd_path = (ATOM_ROUTER_DATA_CLOCK_PATH_SELECT_RECORD *)
+										record;
+									router.cd_valid = true;
+									router.cd_mux_type = cd_path->ucMuxType;
+									router.cd_mux_control_pin = cd_path->ucMuxControlPin;
+									router.cd_mux_state = cd_path->ucMuxState[enum_id];
+									break;
+								}
+								record = (ATOM_COMMON_RECORD_HEADER *)
+									((char *)record + record->ucRecordSize);
+							}
+						}
+					}
+				}
+			}
+
+			/* look up gpio for ddc, hpd */
+			ddc_bus.valid = false;
+			hpd.hpd = AMDGPU_HPD_NONE;
+			if ((le16_to_cpu(path->usDeviceTag) &
+			     (ATOM_DEVICE_TV_SUPPORT | ATOM_DEVICE_CV_SUPPORT)) == 0) {
+				for (j = 0; j < con_obj->ucNumberOfObjects; j++) {
+					if (le16_to_cpu(path->usConnObjectId) ==
+					    le16_to_cpu(con_obj->asObjects[j].
+							usObjectID)) {
+						ATOM_COMMON_RECORD_HEADER
+						    *record =
+						    (ATOM_COMMON_RECORD_HEADER
+						     *)
+						    (ctx->bios + data_offset +
+						     le16_to_cpu(con_obj->
+								 asObjects[j].
+								 usRecordOffset));
+						ATOM_I2C_RECORD *i2c_record;
+						ATOM_HPD_INT_RECORD *hpd_record;
+						ATOM_I2C_ID_CONFIG_ACCESS *i2c_config;
+
+						while (record->ucRecordSize > 0 &&
+						       record->ucRecordType > 0 &&
+						       record->ucRecordType <= ATOM_MAX_OBJECT_RECORD_NUMBER) {
+							switch (record->ucRecordType) {
+							case ATOM_I2C_RECORD_TYPE:
+								i2c_record =
+								    (ATOM_I2C_RECORD *)
+									record;
+								i2c_config =
+									(ATOM_I2C_ID_CONFIG_ACCESS *)
+									&i2c_record->sucI2cId;
+								ddc_bus = amdgpu_atombios_lookup_i2c_gpio(adev,
+												 i2c_config->
+												 ucAccess);
+								break;
+							case ATOM_HPD_INT_RECORD_TYPE:
+								hpd_record =
+									(ATOM_HPD_INT_RECORD *)
+									record;
+								gpio = amdgpu_atombios_lookup_gpio(adev,
+											  hpd_record->ucHPDIntGPIOID);
+								hpd = amdgpu_atombios_get_hpd_info_from_gpio(adev, &gpio);
+								hpd.plugged_state = hpd_record->ucPlugged_PinState;
+								break;
+							}
+							record =
+							    (ATOM_COMMON_RECORD_HEADER
+							     *) ((char *)record
+								 +
+								 record->
+								 ucRecordSize);
+						}
+						break;
+					}
+				}
+			}
+
+			/* needed for aux chan transactions */
+			ddc_bus.hpd = hpd.hpd;
+
+			conn_id = le16_to_cpu(path->usConnObjectId);
+
+			amdgpu_display_add_connector(adev,
+						      conn_id,
+						      le16_to_cpu(path->usDeviceTag),
+						      connector_type, &ddc_bus,
+						      connector_object_id,
+						      &hpd,
+						      &router);
+
+		}
+	}
+
+	amdgpu_link_encoder_connector(adev->ddev);
+
+	return true;
+}
+
+union firmware_info {
+	ATOM_FIRMWARE_INFO info;
+	ATOM_FIRMWARE_INFO_V1_2 info_12;
+	ATOM_FIRMWARE_INFO_V1_3 info_13;
+	ATOM_FIRMWARE_INFO_V1_4 info_14;
+	ATOM_FIRMWARE_INFO_V2_1 info_21;
+	ATOM_FIRMWARE_INFO_V2_2 info_22;
+};
+
+int amdgpu_atombios_get_clock_info(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, FirmwareInfo);
+	uint8_t frev, crev;
+	uint16_t data_offset;
+	int ret = -EINVAL;
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		int i;
+		struct amdgpu_pll *ppll = &adev->clock.ppll[0];
+		struct amdgpu_pll *spll = &adev->clock.spll;
+		struct amdgpu_pll *mpll = &adev->clock.mpll;
+		union firmware_info *firmware_info =
+			(union firmware_info *)(mode_info->atom_context->bios +
+						data_offset);
+		/* pixel clocks */
+		ppll->reference_freq =
+		    le16_to_cpu(firmware_info->info.usReferenceClock);
+		ppll->reference_div = 0;
+
+		ppll->pll_out_min =
+			le32_to_cpu(firmware_info->info_12.ulMinPixelClockPLL_Output);
+		ppll->pll_out_max =
+		    le32_to_cpu(firmware_info->info.ulMaxPixelClockPLL_Output);
+
+		ppll->lcd_pll_out_min =
+			le16_to_cpu(firmware_info->info_14.usLcdMinPixelClockPLL_Output) * 100;
+		if (ppll->lcd_pll_out_min == 0)
+			ppll->lcd_pll_out_min = ppll->pll_out_min;
+		ppll->lcd_pll_out_max =
+			le16_to_cpu(firmware_info->info_14.usLcdMaxPixelClockPLL_Output) * 100;
+		if (ppll->lcd_pll_out_max == 0)
+			ppll->lcd_pll_out_max = ppll->pll_out_max;
+
+		if (ppll->pll_out_min == 0)
+			ppll->pll_out_min = 64800;
+
+		ppll->pll_in_min =
+		    le16_to_cpu(firmware_info->info.usMinPixelClockPLL_Input);
+		ppll->pll_in_max =
+		    le16_to_cpu(firmware_info->info.usMaxPixelClockPLL_Input);
+
+		ppll->min_post_div = 2;
+		ppll->max_post_div = 0x7f;
+		ppll->min_frac_feedback_div = 0;
+		ppll->max_frac_feedback_div = 9;
+		ppll->min_ref_div = 2;
+		ppll->max_ref_div = 0x3ff;
+		ppll->min_feedback_div = 4;
+		ppll->max_feedback_div = 0xfff;
+		ppll->best_vco = 0;
+
+		for (i = 1; i < AMDGPU_MAX_PPLL; i++)
+			adev->clock.ppll[i] = *ppll;
+
+		/* system clock */
+		spll->reference_freq =
+			le16_to_cpu(firmware_info->info_21.usCoreReferenceClock);
+		spll->reference_div = 0;
+
+		spll->pll_out_min =
+		    le16_to_cpu(firmware_info->info.usMinEngineClockPLL_Output);
+		spll->pll_out_max =
+		    le32_to_cpu(firmware_info->info.ulMaxEngineClockPLL_Output);
+
+		/* ??? */
+		if (spll->pll_out_min == 0)
+			spll->pll_out_min = 64800;
+
+		spll->pll_in_min =
+		    le16_to_cpu(firmware_info->info.usMinEngineClockPLL_Input);
+		spll->pll_in_max =
+		    le16_to_cpu(firmware_info->info.usMaxEngineClockPLL_Input);
+
+		spll->min_post_div = 1;
+		spll->max_post_div = 1;
+		spll->min_ref_div = 2;
+		spll->max_ref_div = 0xff;
+		spll->min_feedback_div = 4;
+		spll->max_feedback_div = 0xff;
+		spll->best_vco = 0;
+
+		/* memory clock */
+		mpll->reference_freq =
+			le16_to_cpu(firmware_info->info_21.usMemoryReferenceClock);
+		mpll->reference_div = 0;
+
+		mpll->pll_out_min =
+		    le16_to_cpu(firmware_info->info.usMinMemoryClockPLL_Output);
+		mpll->pll_out_max =
+		    le32_to_cpu(firmware_info->info.ulMaxMemoryClockPLL_Output);
+
+		/* ??? */
+		if (mpll->pll_out_min == 0)
+			mpll->pll_out_min = 64800;
+
+		mpll->pll_in_min =
+		    le16_to_cpu(firmware_info->info.usMinMemoryClockPLL_Input);
+		mpll->pll_in_max =
+		    le16_to_cpu(firmware_info->info.usMaxMemoryClockPLL_Input);
+
+		adev->clock.default_sclk =
+		    le32_to_cpu(firmware_info->info.ulDefaultEngineClock);
+		adev->clock.default_mclk =
+		    le32_to_cpu(firmware_info->info.ulDefaultMemoryClock);
+
+		mpll->min_post_div = 1;
+		mpll->max_post_div = 1;
+		mpll->min_ref_div = 2;
+		mpll->max_ref_div = 0xff;
+		mpll->min_feedback_div = 4;
+		mpll->max_feedback_div = 0xff;
+		mpll->best_vco = 0;
+
+		/* disp clock */
+		adev->clock.default_dispclk =
+			le32_to_cpu(firmware_info->info_21.ulDefaultDispEngineClkFreq);
+		/* set a reasonable default for DP */
+		if (adev->clock.default_dispclk < 53900) {
+			DRM_DEBUG("Changing default dispclk from %dMhz to 600Mhz\n",
+				  adev->clock.default_dispclk / 100);
+			adev->clock.default_dispclk = 60000;
+		} else if (adev->clock.default_dispclk <= 60000) {
+			DRM_DEBUG("Changing default dispclk from %dMhz to 625Mhz\n",
+				  adev->clock.default_dispclk / 100);
+			adev->clock.default_dispclk = 62500;
+		}
+		adev->clock.dp_extclk =
+			le16_to_cpu(firmware_info->info_21.usUniphyDPModeExtClkFreq);
+		adev->clock.current_dispclk = adev->clock.default_dispclk;
+
+		adev->clock.max_pixel_clock = le16_to_cpu(firmware_info->info.usMaxPixelClock);
+		if (adev->clock.max_pixel_clock == 0)
+			adev->clock.max_pixel_clock = 40000;
+
+		/* not technically a clock, but... */
+		adev->mode_info.firmware_flags =
+			le16_to_cpu(firmware_info->info.usFirmwareCapability.susAccess);
+
+		ret = 0;
+	}
+
+	adev->pm.current_sclk = adev->clock.default_sclk;
+	adev->pm.current_mclk = adev->clock.default_mclk;
+
+	return ret;
+}
+
+union gfx_info {
+	ATOM_GFX_INFO_V2_1 info;
+};
+
+int amdgpu_atombios_get_gfx_info(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, GFX_Info);
+	uint8_t frev, crev;
+	uint16_t data_offset;
+	int ret = -EINVAL;
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		union gfx_info *gfx_info = (union gfx_info *)
+			(mode_info->atom_context->bios + data_offset);
+
+		adev->gfx.config.max_shader_engines = gfx_info->info.max_shader_engines;
+		adev->gfx.config.max_tile_pipes = gfx_info->info.max_tile_pipes;
+		adev->gfx.config.max_cu_per_sh = gfx_info->info.max_cu_per_sh;
+		adev->gfx.config.max_sh_per_se = gfx_info->info.max_sh_per_se;
+		adev->gfx.config.max_backends_per_se = gfx_info->info.max_backends_per_se;
+		adev->gfx.config.max_texture_channel_caches =
+			gfx_info->info.max_texture_channel_caches;
+
+		ret = 0;
+	}
+	return ret;
+}
+
+union igp_info {
+	struct _ATOM_INTEGRATED_SYSTEM_INFO info;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V2 info_2;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V6 info_6;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_7 info_7;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_8 info_8;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_9 info_9;
+};
+
+/*
+ * Return vram width from integrated system info table, if available,
+ * or 0 if not.
+ */
+int amdgpu_atombios_get_vram_width(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, IntegratedSystemInfo);
+	u16 data_offset, size;
+	union igp_info *igp_info;
+	u8 frev, crev;
+
+	/* get any igp specific overrides */
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		igp_info = (union igp_info *)
+			(mode_info->atom_context->bios + data_offset);
+		switch (crev) {
+		case 8:
+		case 9:
+			return igp_info->info_8.ucUMAChannelNumber * 64;
+		default:
+			return 0;
+		}
+	}
+
+	return 0;
+}
+
+static void amdgpu_atombios_get_igp_ss_overrides(struct amdgpu_device *adev,
+						 struct amdgpu_atom_ss *ss,
+						 int id)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, IntegratedSystemInfo);
+	u16 data_offset, size;
+	union igp_info *igp_info;
+	u8 frev, crev;
+	u16 percentage = 0, rate = 0;
+
+	/* get any igp specific overrides */
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		igp_info = (union igp_info *)
+			(mode_info->atom_context->bios + data_offset);
+		switch (crev) {
+		case 6:
+			switch (id) {
+			case ASIC_INTERNAL_SS_ON_TMDS:
+				percentage = le16_to_cpu(igp_info->info_6.usDVISSPercentage);
+				rate = le16_to_cpu(igp_info->info_6.usDVISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_HDMI:
+				percentage = le16_to_cpu(igp_info->info_6.usHDMISSPercentage);
+				rate = le16_to_cpu(igp_info->info_6.usHDMISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_LVDS:
+				percentage = le16_to_cpu(igp_info->info_6.usLvdsSSPercentage);
+				rate = le16_to_cpu(igp_info->info_6.usLvdsSSpreadRateIn10Hz);
+				break;
+			}
+			break;
+		case 7:
+			switch (id) {
+			case ASIC_INTERNAL_SS_ON_TMDS:
+				percentage = le16_to_cpu(igp_info->info_7.usDVISSPercentage);
+				rate = le16_to_cpu(igp_info->info_7.usDVISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_HDMI:
+				percentage = le16_to_cpu(igp_info->info_7.usHDMISSPercentage);
+				rate = le16_to_cpu(igp_info->info_7.usHDMISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_LVDS:
+				percentage = le16_to_cpu(igp_info->info_7.usLvdsSSPercentage);
+				rate = le16_to_cpu(igp_info->info_7.usLvdsSSpreadRateIn10Hz);
+				break;
+			}
+			break;
+		case 8:
+			switch (id) {
+			case ASIC_INTERNAL_SS_ON_TMDS:
+				percentage = le16_to_cpu(igp_info->info_8.usDVISSPercentage);
+				rate = le16_to_cpu(igp_info->info_8.usDVISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_HDMI:
+				percentage = le16_to_cpu(igp_info->info_8.usHDMISSPercentage);
+				rate = le16_to_cpu(igp_info->info_8.usHDMISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_LVDS:
+				percentage = le16_to_cpu(igp_info->info_8.usLvdsSSPercentage);
+				rate = le16_to_cpu(igp_info->info_8.usLvdsSSpreadRateIn10Hz);
+				break;
+			}
+			break;
+		case 9:
+			switch (id) {
+			case ASIC_INTERNAL_SS_ON_TMDS:
+				percentage = le16_to_cpu(igp_info->info_9.usDVISSPercentage);
+				rate = le16_to_cpu(igp_info->info_9.usDVISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_HDMI:
+				percentage = le16_to_cpu(igp_info->info_9.usHDMISSPercentage);
+				rate = le16_to_cpu(igp_info->info_9.usHDMISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_LVDS:
+				percentage = le16_to_cpu(igp_info->info_9.usLvdsSSPercentage);
+				rate = le16_to_cpu(igp_info->info_9.usLvdsSSpreadRateIn10Hz);
+				break;
+			}
+			break;
+		default:
+			DRM_ERROR("Unsupported IGP table: %d %d\n", frev, crev);
+			break;
+		}
+		if (percentage)
+			ss->percentage = percentage;
+		if (rate)
+			ss->rate = rate;
+	}
+}
+
+union asic_ss_info {
+	struct _ATOM_ASIC_INTERNAL_SS_INFO info;
+	struct _ATOM_ASIC_INTERNAL_SS_INFO_V2 info_2;
+	struct _ATOM_ASIC_INTERNAL_SS_INFO_V3 info_3;
+};
+
+union asic_ss_assignment {
+	struct _ATOM_ASIC_SS_ASSIGNMENT v1;
+	struct _ATOM_ASIC_SS_ASSIGNMENT_V2 v2;
+	struct _ATOM_ASIC_SS_ASSIGNMENT_V3 v3;
+};
+
+bool amdgpu_atombios_get_asic_ss_info(struct amdgpu_device *adev,
+				      struct amdgpu_atom_ss *ss,
+				      int id, u32 clock)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, ASIC_InternalSS_Info);
+	uint16_t data_offset, size;
+	union asic_ss_info *ss_info;
+	union asic_ss_assignment *ss_assign;
+	uint8_t frev, crev;
+	int i, num_indices;
+
+	if (id == ASIC_INTERNAL_MEMORY_SS) {
+		if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_MEMORY_CLOCK_SS_SUPPORT))
+			return false;
+	}
+	if (id == ASIC_INTERNAL_ENGINE_SS) {
+		if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_ENGINE_CLOCK_SS_SUPPORT))
+			return false;
+	}
+
+	memset(ss, 0, sizeof(struct amdgpu_atom_ss));
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+
+		ss_info =
+			(union asic_ss_info *)(mode_info->atom_context->bios + data_offset);
+
+		switch (frev) {
+		case 1:
+			num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+				sizeof(ATOM_ASIC_SS_ASSIGNMENT);
+
+			ss_assign = (union asic_ss_assignment *)((u8 *)&ss_info->info.asSpreadSpectrum[0]);
+			for (i = 0; i < num_indices; i++) {
+				if ((ss_assign->v1.ucClockIndication == id) &&
+				    (clock <= le32_to_cpu(ss_assign->v1.ulTargetClockRange))) {
+					ss->percentage =
+						le16_to_cpu(ss_assign->v1.usSpreadSpectrumPercentage);
+					ss->type = ss_assign->v1.ucSpreadSpectrumMode;
+					ss->rate = le16_to_cpu(ss_assign->v1.usSpreadRateInKhz);
+					ss->percentage_divider = 100;
+					return true;
+				}
+				ss_assign = (union asic_ss_assignment *)
+					((u8 *)ss_assign + sizeof(ATOM_ASIC_SS_ASSIGNMENT));
+			}
+			break;
+		case 2:
+			num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+				sizeof(ATOM_ASIC_SS_ASSIGNMENT_V2);
+			ss_assign = (union asic_ss_assignment *)((u8 *)&ss_info->info_2.asSpreadSpectrum[0]);
+			for (i = 0; i < num_indices; i++) {
+				if ((ss_assign->v2.ucClockIndication == id) &&
+				    (clock <= le32_to_cpu(ss_assign->v2.ulTargetClockRange))) {
+					ss->percentage =
+						le16_to_cpu(ss_assign->v2.usSpreadSpectrumPercentage);
+					ss->type = ss_assign->v2.ucSpreadSpectrumMode;
+					ss->rate = le16_to_cpu(ss_assign->v2.usSpreadRateIn10Hz);
+					ss->percentage_divider = 100;
+					if ((crev == 2) &&
+					    ((id == ASIC_INTERNAL_ENGINE_SS) ||
+					     (id == ASIC_INTERNAL_MEMORY_SS)))
+						ss->rate /= 100;
+					return true;
+				}
+				ss_assign = (union asic_ss_assignment *)
+					((u8 *)ss_assign + sizeof(ATOM_ASIC_SS_ASSIGNMENT_V2));
+			}
+			break;
+		case 3:
+			num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+				sizeof(ATOM_ASIC_SS_ASSIGNMENT_V3);
+			ss_assign = (union asic_ss_assignment *)((u8 *)&ss_info->info_3.asSpreadSpectrum[0]);
+			for (i = 0; i < num_indices; i++) {
+				if ((ss_assign->v3.ucClockIndication == id) &&
+				    (clock <= le32_to_cpu(ss_assign->v3.ulTargetClockRange))) {
+					ss->percentage =
+						le16_to_cpu(ss_assign->v3.usSpreadSpectrumPercentage);
+					ss->type = ss_assign->v3.ucSpreadSpectrumMode;
+					ss->rate = le16_to_cpu(ss_assign->v3.usSpreadRateIn10Hz);
+					if (ss_assign->v3.ucSpreadSpectrumMode &
+					    SS_MODE_V3_PERCENTAGE_DIV_BY_1000_MASK)
+						ss->percentage_divider = 1000;
+					else
+						ss->percentage_divider = 100;
+					if ((id == ASIC_INTERNAL_ENGINE_SS) ||
+					    (id == ASIC_INTERNAL_MEMORY_SS))
+						ss->rate /= 100;
+					if (adev->flags & AMD_IS_APU)
+						amdgpu_atombios_get_igp_ss_overrides(adev, ss, id);
+					return true;
+				}
+				ss_assign = (union asic_ss_assignment *)
+					((u8 *)ss_assign + sizeof(ATOM_ASIC_SS_ASSIGNMENT_V3));
+			}
+			break;
+		default:
+			DRM_ERROR("Unsupported ASIC_InternalSS_Info table: %d %d\n", frev, crev);
+			break;
+		}
+
+	}
+	return false;
+}
+
+union get_clock_dividers {
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS v1;
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V2 v2;
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3 v3;
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V4 v4;
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5 v5;
+	struct _COMPUTE_GPU_CLOCK_INPUT_PARAMETERS_V1_6 v6_in;
+	struct _COMPUTE_GPU_CLOCK_OUTPUT_PARAMETERS_V1_6 v6_out;
+};
+
+int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
+				       u8 clock_type,
+				       u32 clock,
+				       bool strobe_mode,
+				       struct atom_clock_dividers *dividers)
+{
+	union get_clock_dividers args;
+	int index = GetIndexIntoMasterTable(COMMAND, ComputeMemoryEnginePLL);
+	u8 frev, crev;
+
+	memset(&args, 0, sizeof(args));
+	memset(dividers, 0, sizeof(struct atom_clock_dividers));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return -EINVAL;
+
+	switch (crev) {
+	case 2:
+	case 3:
+	case 5:
+		/* r6xx, r7xx, evergreen, ni, si.
+		 * TODO: add support for asic_type <= CHIP_RV770*/
+		if (clock_type == COMPUTE_ENGINE_PLL_PARAM) {
+			args.v3.ulClockParams = cpu_to_le32((clock_type << 24) | clock);
+
+			amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+			dividers->post_div = args.v3.ucPostDiv;
+			dividers->enable_post_div = (args.v3.ucCntlFlag &
+						     ATOM_PLL_CNTL_FLAG_PLL_POST_DIV_EN) ? true : false;
+			dividers->enable_dithen = (args.v3.ucCntlFlag &
+						   ATOM_PLL_CNTL_FLAG_FRACTION_DISABLE) ? false : true;
+			dividers->whole_fb_div = le16_to_cpu(args.v3.ulFbDiv.usFbDiv);
+			dividers->frac_fb_div = le16_to_cpu(args.v3.ulFbDiv.usFbDivFrac);
+			dividers->ref_div = args.v3.ucRefDiv;
+			dividers->vco_mode = (args.v3.ucCntlFlag &
+					      ATOM_PLL_CNTL_FLAG_MPLL_VCO_MODE) ? 1 : 0;
+		} else {
+			/* for SI we use ComputeMemoryClockParam for memory plls */
+			if (adev->asic_type >= CHIP_TAHITI)
+				return -EINVAL;
+			args.v5.ulClockParams = cpu_to_le32((clock_type << 24) | clock);
+			if (strobe_mode)
+				args.v5.ucInputFlag = ATOM_PLL_INPUT_FLAG_PLL_STROBE_MODE_EN;
+
+			amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+			dividers->post_div = args.v5.ucPostDiv;
+			dividers->enable_post_div = (args.v5.ucCntlFlag &
+						     ATOM_PLL_CNTL_FLAG_PLL_POST_DIV_EN) ? true : false;
+			dividers->enable_dithen = (args.v5.ucCntlFlag &
+						   ATOM_PLL_CNTL_FLAG_FRACTION_DISABLE) ? false : true;
+			dividers->whole_fb_div = le16_to_cpu(args.v5.ulFbDiv.usFbDiv);
+			dividers->frac_fb_div = le16_to_cpu(args.v5.ulFbDiv.usFbDivFrac);
+			dividers->ref_div = args.v5.ucRefDiv;
+			dividers->vco_mode = (args.v5.ucCntlFlag &
+					      ATOM_PLL_CNTL_FLAG_MPLL_VCO_MODE) ? 1 : 0;
+		}
+		break;
+	case 4:
+		/* fusion */
+		args.v4.ulClock = cpu_to_le32(clock);	/* 10 khz */
+
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		dividers->post_divider = dividers->post_div = args.v4.ucPostDiv;
+		dividers->real_clock = le32_to_cpu(args.v4.ulClock);
+		break;
+	case 6:
+		/* CI */
+		/* COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK, COMPUTE_GPUCLK_INPUT_FLAG_SCLK */
+		args.v6_in.ulClock.ulComputeClockFlag = clock_type;
+		args.v6_in.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
+
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		dividers->whole_fb_div = le16_to_cpu(args.v6_out.ulFbDiv.usFbDiv);
+		dividers->frac_fb_div = le16_to_cpu(args.v6_out.ulFbDiv.usFbDivFrac);
+		dividers->ref_div = args.v6_out.ucPllRefDiv;
+		dividers->post_div = args.v6_out.ucPllPostDiv;
+		dividers->flags = args.v6_out.ucPllCntlFlag;
+		dividers->real_clock = le32_to_cpu(args.v6_out.ulClock.ulClock);
+		dividers->post_divider = args.v6_out.ulClock.ucPostDiv;
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
+int amdgpu_atombios_get_memory_pll_dividers(struct amdgpu_device *adev,
+					    u32 clock,
+					    bool strobe_mode,
+					    struct atom_mpll_param *mpll_param)
+{
+	COMPUTE_MEMORY_CLOCK_PARAM_PARAMETERS_V2_1 args;
+	int index = GetIndexIntoMasterTable(COMMAND, ComputeMemoryClockParam);
+	u8 frev, crev;
+
+	memset(&args, 0, sizeof(args));
+	memset(mpll_param, 0, sizeof(struct atom_mpll_param));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return -EINVAL;
+
+	switch (frev) {
+	case 2:
+		switch (crev) {
+		case 1:
+			/* SI */
+			args.ulClock = cpu_to_le32(clock);	/* 10 khz */
+			args.ucInputFlag = 0;
+			if (strobe_mode)
+				args.ucInputFlag |= MPLL_INPUT_FLAG_STROBE_MODE_EN;
+
+			amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+			mpll_param->clkfrac = le16_to_cpu(args.ulFbDiv.usFbDivFrac);
+			mpll_param->clkf = le16_to_cpu(args.ulFbDiv.usFbDiv);
+			mpll_param->post_div = args.ucPostDiv;
+			mpll_param->dll_speed = args.ucDllSpeed;
+			mpll_param->bwcntl = args.ucBWCntl;
+			mpll_param->vco_mode =
+				(args.ucPllCntlFlag & MPLL_CNTL_FLAG_VCO_MODE_MASK);
+			mpll_param->yclk_sel =
+				(args.ucPllCntlFlag & MPLL_CNTL_FLAG_BYPASS_DQ_PLL) ? 1 : 0;
+			mpll_param->qdr =
+				(args.ucPllCntlFlag & MPLL_CNTL_FLAG_QDR_ENABLE) ? 1 : 0;
+			mpll_param->half_rate =
+				(args.ucPllCntlFlag & MPLL_CNTL_FLAG_AD_HALF_RATE) ? 1 : 0;
+			break;
+		default:
+			return -EINVAL;
+		}
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
+void amdgpu_atombios_set_engine_dram_timings(struct amdgpu_device *adev,
+					     u32 eng_clock, u32 mem_clock)
+{
+	SET_ENGINE_CLOCK_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, DynamicMemorySettings);
+	u32 tmp;
+
+	memset(&args, 0, sizeof(args));
+
+	tmp = eng_clock & SET_CLOCK_FREQ_MASK;
+	tmp |= (COMPUTE_ENGINE_PLL_PARAM << 24);
+
+	args.ulTargetEngineClock = cpu_to_le32(tmp);
+	if (mem_clock)
+		args.sReserved.ulClock = cpu_to_le32(mem_clock & SET_CLOCK_FREQ_MASK);
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_get_default_voltages(struct amdgpu_device *adev,
+					  u16 *vddc, u16 *vddci, u16 *mvdd)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, FirmwareInfo);
+	u8 frev, crev;
+	u16 data_offset;
+	union firmware_info *firmware_info;
+
+	*vddc = 0;
+	*vddci = 0;
+	*mvdd = 0;
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		firmware_info =
+			(union firmware_info *)(mode_info->atom_context->bios +
+						data_offset);
+		*vddc = le16_to_cpu(firmware_info->info_14.usBootUpVDDCVoltage);
+		if ((frev == 2) && (crev >= 2)) {
+			*vddci = le16_to_cpu(firmware_info->info_22.usBootUpVDDCIVoltage);
+			*mvdd = le16_to_cpu(firmware_info->info_22.usBootUpMVDDCVoltage);
+		}
+	}
+}
+
+union set_voltage {
+	struct _SET_VOLTAGE_PS_ALLOCATION alloc;
+	struct _SET_VOLTAGE_PARAMETERS v1;
+	struct _SET_VOLTAGE_PARAMETERS_V2 v2;
+	struct _SET_VOLTAGE_PARAMETERS_V1_3 v3;
+};
+
+int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type,
+			     u16 voltage_id, u16 *voltage)
+{
+	union set_voltage args;
+	int index = GetIndexIntoMasterTable(COMMAND, SetVoltage);
+	u8 frev, crev;
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return -EINVAL;
+
+	switch (crev) {
+	case 1:
+		return -EINVAL;
+	case 2:
+		args.v2.ucVoltageType = SET_VOLTAGE_GET_MAX_VOLTAGE;
+		args.v2.ucVoltageMode = 0;
+		args.v2.usVoltageLevel = 0;
+
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		*voltage = le16_to_cpu(args.v2.usVoltageLevel);
+		break;
+	case 3:
+		args.v3.ucVoltageType = voltage_type;
+		args.v3.ucVoltageMode = ATOM_GET_VOLTAGE_LEVEL;
+		args.v3.usVoltageLevel = cpu_to_le16(voltage_id);
+
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		*voltage = le16_to_cpu(args.v3.usVoltageLevel);
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int amdgpu_atombios_get_leakage_vddc_based_on_leakage_idx(struct amdgpu_device *adev,
+						      u16 *voltage,
+						      u16 leakage_idx)
+{
+	return amdgpu_atombios_get_max_vddc(adev, VOLTAGE_TYPE_VDDC, leakage_idx, voltage);
+}
+
+int amdgpu_atombios_get_leakage_id_from_vbios(struct amdgpu_device *adev,
+					      u16 *leakage_id)
+{
+	union set_voltage args;
+	int index = GetIndexIntoMasterTable(COMMAND, SetVoltage);
+	u8 frev, crev;
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return -EINVAL;
+
+	switch (crev) {
+	case 3:
+	case 4:
+		args.v3.ucVoltageType = 0;
+		args.v3.ucVoltageMode = ATOM_GET_LEAKAGE_ID;
+		args.v3.usVoltageLevel = 0;
+
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		*leakage_id = le16_to_cpu(args.v3.usVoltageLevel);
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int amdgpu_atombios_get_leakage_vddc_based_on_leakage_params(struct amdgpu_device *adev,
+							     u16 *vddc, u16 *vddci,
+							     u16 virtual_voltage_id,
+							     u16 vbios_voltage_id)
+{
+	int index = GetIndexIntoMasterTable(DATA, ASIC_ProfilingInfo);
+	u8 frev, crev;
+	u16 data_offset, size;
+	int i, j;
+	ATOM_ASIC_PROFILING_INFO_V2_1 *profile;
+	u16 *leakage_bin, *vddc_id_buf, *vddc_buf, *vddci_id_buf, *vddci_buf;
+
+	*vddc = 0;
+	*vddci = 0;
+
+	if (!amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+				    &frev, &crev, &data_offset))
+		return -EINVAL;
+
+	profile = (ATOM_ASIC_PROFILING_INFO_V2_1 *)
+		(adev->mode_info.atom_context->bios + data_offset);
+
+	switch (frev) {
+	case 1:
+		return -EINVAL;
+	case 2:
+		switch (crev) {
+		case 1:
+			if (size < sizeof(ATOM_ASIC_PROFILING_INFO_V2_1))
+				return -EINVAL;
+			leakage_bin = (u16 *)
+				(adev->mode_info.atom_context->bios + data_offset +
+				 le16_to_cpu(profile->usLeakageBinArrayOffset));
+			vddc_id_buf = (u16 *)
+				(adev->mode_info.atom_context->bios + data_offset +
+				 le16_to_cpu(profile->usElbVDDC_IdArrayOffset));
+			vddc_buf = (u16 *)
+				(adev->mode_info.atom_context->bios + data_offset +
+				 le16_to_cpu(profile->usElbVDDC_LevelArrayOffset));
+			vddci_id_buf = (u16 *)
+				(adev->mode_info.atom_context->bios + data_offset +
+				 le16_to_cpu(profile->usElbVDDCI_IdArrayOffset));
+			vddci_buf = (u16 *)
+				(adev->mode_info.atom_context->bios + data_offset +
+				 le16_to_cpu(profile->usElbVDDCI_LevelArrayOffset));
+
+			if (profile->ucElbVDDC_Num > 0) {
+				for (i = 0; i < profile->ucElbVDDC_Num; i++) {
+					if (vddc_id_buf[i] == virtual_voltage_id) {
+						for (j = 0; j < profile->ucLeakageBinNum; j++) {
+							if (vbios_voltage_id <= leakage_bin[j]) {
+								*vddc = vddc_buf[j * profile->ucElbVDDC_Num + i];
+								break;
+							}
+						}
+						break;
+					}
+				}
+			}
+			if (profile->ucElbVDDCI_Num > 0) {
+				for (i = 0; i < profile->ucElbVDDCI_Num; i++) {
+					if (vddci_id_buf[i] == virtual_voltage_id) {
+						for (j = 0; j < profile->ucLeakageBinNum; j++) {
+							if (vbios_voltage_id <= leakage_bin[j]) {
+								*vddci = vddci_buf[j * profile->ucElbVDDCI_Num + i];
+								break;
+							}
+						}
+						break;
+					}
+				}
+			}
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+			return -EINVAL;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+union get_voltage_info {
+	struct _GET_VOLTAGE_INFO_INPUT_PARAMETER_V1_2 in;
+	struct _GET_EVV_VOLTAGE_INFO_OUTPUT_PARAMETER_V1_2 evv_out;
+};
+
+int amdgpu_atombios_get_voltage_evv(struct amdgpu_device *adev,
+				    u16 virtual_voltage_id,
+				    u16 *voltage)
+{
+	int index = GetIndexIntoMasterTable(COMMAND, GetVoltageInfo);
+	u32 entry_id;
+	u32 count = adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count;
+	union get_voltage_info args;
+
+	for (entry_id = 0; entry_id < count; entry_id++) {
+		if (adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[entry_id].v ==
+		    virtual_voltage_id)
+			break;
+	}
+
+	if (entry_id >= count)
+		return -EINVAL;
+
+	args.in.ucVoltageType = VOLTAGE_TYPE_VDDC;
+	args.in.ucVoltageMode = ATOM_GET_VOLTAGE_EVV_VOLTAGE;
+	args.in.usVoltageLevel = cpu_to_le16(virtual_voltage_id);
+	args.in.ulSCLKFreq =
+		cpu_to_le32(adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[entry_id].clk);
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+	*voltage = le16_to_cpu(args.evv_out.usVoltageLevel);
+
+	return 0;
+}
+
+union voltage_object_info {
+	struct _ATOM_VOLTAGE_OBJECT_INFO v1;
+	struct _ATOM_VOLTAGE_OBJECT_INFO_V2 v2;
+	struct _ATOM_VOLTAGE_OBJECT_INFO_V3_1 v3;
+};
+
+union voltage_object {
+	struct _ATOM_VOLTAGE_OBJECT v1;
+	struct _ATOM_VOLTAGE_OBJECT_V2 v2;
+	union _ATOM_VOLTAGE_OBJECT_V3 v3;
+};
+
+
+static ATOM_VOLTAGE_OBJECT_V3 *amdgpu_atombios_lookup_voltage_object_v3(ATOM_VOLTAGE_OBJECT_INFO_V3_1 *v3,
+									u8 voltage_type, u8 voltage_mode)
+{
+	u32 size = le16_to_cpu(v3->sHeader.usStructureSize);
+	u32 offset = offsetof(ATOM_VOLTAGE_OBJECT_INFO_V3_1, asVoltageObj[0]);
+	u8 *start = (u8*)v3;
+
+	while (offset < size) {
+		ATOM_VOLTAGE_OBJECT_V3 *vo = (ATOM_VOLTAGE_OBJECT_V3 *)(start + offset);
+		if ((vo->asGpioVoltageObj.sHeader.ucVoltageType == voltage_type) &&
+		    (vo->asGpioVoltageObj.sHeader.ucVoltageMode == voltage_mode))
+			return vo;
+		offset += le16_to_cpu(vo->asGpioVoltageObj.sHeader.usSize);
+	}
+	return NULL;
+}
+
+int amdgpu_atombios_get_svi2_info(struct amdgpu_device *adev,
+			      u8 voltage_type,
+			      u8 *svd_gpio_id, u8 *svc_gpio_id)
+{
+	int index = GetIndexIntoMasterTable(DATA, VoltageObjectInfo);
+	u8 frev, crev;
+	u16 data_offset, size;
+	union voltage_object_info *voltage_info;
+	union voltage_object *voltage_object = NULL;
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		voltage_info = (union voltage_object_info *)
+			(adev->mode_info.atom_context->bios + data_offset);
+
+		switch (frev) {
+		case 3:
+			switch (crev) {
+			case 1:
+				voltage_object = (union voltage_object *)
+					amdgpu_atombios_lookup_voltage_object_v3(&voltage_info->v3,
+								      voltage_type,
+								      VOLTAGE_OBJ_SVID2);
+				if (voltage_object) {
+					*svd_gpio_id = voltage_object->v3.asSVID2Obj.ucSVDGpioId;
+					*svc_gpio_id = voltage_object->v3.asSVID2Obj.ucSVCGpioId;
+				} else {
+					return -EINVAL;
+				}
+				break;
+			default:
+				DRM_ERROR("unknown voltage object table\n");
+				return -EINVAL;
+			}
+			break;
+		default:
+			DRM_ERROR("unknown voltage object table\n");
+			return -EINVAL;
+		}
+
+	}
+	return 0;
+}
+
+bool
+amdgpu_atombios_is_voltage_gpio(struct amdgpu_device *adev,
+				u8 voltage_type, u8 voltage_mode)
+{
+	int index = GetIndexIntoMasterTable(DATA, VoltageObjectInfo);
+	u8 frev, crev;
+	u16 data_offset, size;
+	union voltage_object_info *voltage_info;
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		voltage_info = (union voltage_object_info *)
+			(adev->mode_info.atom_context->bios + data_offset);
+
+		switch (frev) {
+		case 3:
+			switch (crev) {
+			case 1:
+				if (amdgpu_atombios_lookup_voltage_object_v3(&voltage_info->v3,
+								  voltage_type, voltage_mode))
+					return true;
+				break;
+			default:
+				DRM_ERROR("unknown voltage object table\n");
+				return false;
+			}
+			break;
+		default:
+			DRM_ERROR("unknown voltage object table\n");
+			return false;
+		}
+
+	}
+	return false;
+}
+
+int amdgpu_atombios_get_voltage_table(struct amdgpu_device *adev,
+				      u8 voltage_type, u8 voltage_mode,
+				      struct atom_voltage_table *voltage_table)
+{
+	int index = GetIndexIntoMasterTable(DATA, VoltageObjectInfo);
+	u8 frev, crev;
+	u16 data_offset, size;
+	int i;
+	union voltage_object_info *voltage_info;
+	union voltage_object *voltage_object = NULL;
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		voltage_info = (union voltage_object_info *)
+			(adev->mode_info.atom_context->bios + data_offset);
+
+		switch (frev) {
+		case 3:
+			switch (crev) {
+			case 1:
+				voltage_object = (union voltage_object *)
+					amdgpu_atombios_lookup_voltage_object_v3(&voltage_info->v3,
+								      voltage_type, voltage_mode);
+				if (voltage_object) {
+					ATOM_GPIO_VOLTAGE_OBJECT_V3 *gpio =
+						&voltage_object->v3.asGpioVoltageObj;
+					VOLTAGE_LUT_ENTRY_V2 *lut;
+					if (gpio->ucGpioEntryNum > MAX_VOLTAGE_ENTRIES)
+						return -EINVAL;
+					lut = &gpio->asVolGpioLut[0];
+					for (i = 0; i < gpio->ucGpioEntryNum; i++) {
+						voltage_table->entries[i].value =
+							le16_to_cpu(lut->usVoltageValue);
+						voltage_table->entries[i].smio_low =
+							le32_to_cpu(lut->ulVoltageId);
+						lut = (VOLTAGE_LUT_ENTRY_V2 *)
+							((u8 *)lut + sizeof(VOLTAGE_LUT_ENTRY_V2));
+					}
+					voltage_table->mask_low = le32_to_cpu(gpio->ulGpioMaskVal);
+					voltage_table->count = gpio->ucGpioEntryNum;
+					voltage_table->phase_delay = gpio->ucPhaseDelay;
+					return 0;
+				}
+				break;
+			default:
+				DRM_ERROR("unknown voltage object table\n");
+				return -EINVAL;
+			}
+			break;
+		default:
+			DRM_ERROR("unknown voltage object table\n");
+			return -EINVAL;
+		}
+	}
+	return -EINVAL;
+}
+
+union vram_info {
+	struct _ATOM_VRAM_INFO_V3 v1_3;
+	struct _ATOM_VRAM_INFO_V4 v1_4;
+	struct _ATOM_VRAM_INFO_HEADER_V2_1 v2_1;
+};
+
+#define MEM_ID_MASK           0xff000000
+#define MEM_ID_SHIFT          24
+#define CLOCK_RANGE_MASK      0x00ffffff
+#define CLOCK_RANGE_SHIFT     0
+#define LOW_NIBBLE_MASK       0xf
+#define DATA_EQU_PREV         0
+#define DATA_FROM_TABLE       4
+
+int amdgpu_atombios_init_mc_reg_table(struct amdgpu_device *adev,
+				      u8 module_index,
+				      struct atom_mc_reg_table *reg_table)
+{
+	int index = GetIndexIntoMasterTable(DATA, VRAM_Info);
+	u8 frev, crev, num_entries, t_mem_id, num_ranges = 0;
+	u32 i = 0, j;
+	u16 data_offset, size;
+	union vram_info *vram_info;
+
+	memset(reg_table, 0, sizeof(struct atom_mc_reg_table));
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		vram_info = (union vram_info *)
+			(adev->mode_info.atom_context->bios + data_offset);
+		switch (frev) {
+		case 1:
+			DRM_ERROR("old table version %d, %d\n", frev, crev);
+			return -EINVAL;
+		case 2:
+			switch (crev) {
+			case 1:
+				if (module_index < vram_info->v2_1.ucNumOfVRAMModule) {
+					ATOM_INIT_REG_BLOCK *reg_block =
+						(ATOM_INIT_REG_BLOCK *)
+						((u8 *)vram_info + le16_to_cpu(vram_info->v2_1.usMemClkPatchTblOffset));
+					ATOM_MEMORY_SETTING_DATA_BLOCK *reg_data =
+						(ATOM_MEMORY_SETTING_DATA_BLOCK *)
+						((u8 *)reg_block + (2 * sizeof(u16)) +
+						 le16_to_cpu(reg_block->usRegIndexTblSize));
+					ATOM_INIT_REG_INDEX_FORMAT *format = &reg_block->asRegIndexBuf[0];
+					num_entries = (u8)((le16_to_cpu(reg_block->usRegIndexTblSize)) /
+							   sizeof(ATOM_INIT_REG_INDEX_FORMAT)) - 1;
+					if (num_entries > VBIOS_MC_REGISTER_ARRAY_SIZE)
+						return -EINVAL;
+					while (i < num_entries) {
+						if (format->ucPreRegDataLength & ACCESS_PLACEHOLDER)
+							break;
+						reg_table->mc_reg_address[i].s1 =
+							(u16)(le16_to_cpu(format->usRegIndex));
+						reg_table->mc_reg_address[i].pre_reg_data =
+							(u8)(format->ucPreRegDataLength);
+						i++;
+						format = (ATOM_INIT_REG_INDEX_FORMAT *)
+							((u8 *)format + sizeof(ATOM_INIT_REG_INDEX_FORMAT));
+					}
+					reg_table->last = i;
+					while ((le32_to_cpu(*(u32 *)reg_data) != END_OF_REG_DATA_BLOCK) &&
+					       (num_ranges < VBIOS_MAX_AC_TIMING_ENTRIES)) {
+						t_mem_id = (u8)((le32_to_cpu(*(u32 *)reg_data) & MEM_ID_MASK)
+								>> MEM_ID_SHIFT);
+						if (module_index == t_mem_id) {
+							reg_table->mc_reg_table_entry[num_ranges].mclk_max =
+								(u32)((le32_to_cpu(*(u32 *)reg_data) & CLOCK_RANGE_MASK)
+								      >> CLOCK_RANGE_SHIFT);
+							for (i = 0, j = 1; i < reg_table->last; i++) {
+								if ((reg_table->mc_reg_address[i].pre_reg_data & LOW_NIBBLE_MASK) == DATA_FROM_TABLE) {
+									reg_table->mc_reg_table_entry[num_ranges].mc_data[i] =
+										(u32)le32_to_cpu(*((u32 *)reg_data + j));
+									j++;
+								} else if ((reg_table->mc_reg_address[i].pre_reg_data & LOW_NIBBLE_MASK) == DATA_EQU_PREV) {
+									reg_table->mc_reg_table_entry[num_ranges].mc_data[i] =
+										reg_table->mc_reg_table_entry[num_ranges].mc_data[i - 1];
+								}
+							}
+							num_ranges++;
+						}
+						reg_data = (ATOM_MEMORY_SETTING_DATA_BLOCK *)
+							((u8 *)reg_data + le16_to_cpu(reg_block->usRegDataBlkSize));
+					}
+					if (le32_to_cpu(*(u32 *)reg_data) != END_OF_REG_DATA_BLOCK)
+						return -EINVAL;
+					reg_table->num_entries = num_ranges;
+				} else
+					return -EINVAL;
+				break;
+			default:
+				DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+				return -EINVAL;
+			}
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+			return -EINVAL;
+		}
+		return 0;
+	}
+	return -EINVAL;
+}
+
+bool amdgpu_atombios_has_gpu_virtualization_table(struct amdgpu_device *adev)
+{
+	int index = GetIndexIntoMasterTable(DATA, GPUVirtualizationInfo);
+	u8 frev, crev;
+	u16 data_offset, size;
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+					  &frev, &crev, &data_offset))
+		return true;
+
+	return false;
+}
+
+void amdgpu_atombios_scratch_regs_lock(struct amdgpu_device *adev, bool lock)
+{
+	uint32_t bios_6_scratch;
+
+	bios_6_scratch = RREG32(adev->bios_scratch_reg_offset + 6);
+
+	if (lock) {
+		bios_6_scratch |= ATOM_S6_CRITICAL_STATE;
+		bios_6_scratch &= ~ATOM_S6_ACC_MODE;
+	} else {
+		bios_6_scratch &= ~ATOM_S6_CRITICAL_STATE;
+		bios_6_scratch |= ATOM_S6_ACC_MODE;
+	}
+
+	WREG32(adev->bios_scratch_reg_offset + 6, bios_6_scratch);
+}
+
+static void amdgpu_atombios_scratch_regs_init(struct amdgpu_device *adev)
+{
+	uint32_t bios_2_scratch, bios_6_scratch;
+
+	adev->bios_scratch_reg_offset = mmBIOS_SCRATCH_0;
+
+	bios_2_scratch = RREG32(adev->bios_scratch_reg_offset + 2);
+	bios_6_scratch = RREG32(adev->bios_scratch_reg_offset + 6);
+
+	/* let the bios control the backlight */
+	bios_2_scratch &= ~ATOM_S2_VRI_BRIGHT_ENABLE;
+
+	/* tell the bios not to handle mode switching */
+	bios_6_scratch |= ATOM_S6_ACC_BLOCK_DISPLAY_SWITCH;
+
+	/* clear the vbios dpms state */
+	bios_2_scratch &= ~ATOM_S2_DEVICE_DPMS_STATE;
+
+	WREG32(adev->bios_scratch_reg_offset + 2, bios_2_scratch);
+	WREG32(adev->bios_scratch_reg_offset + 6, bios_6_scratch);
+}
+
+void amdgpu_atombios_scratch_regs_engine_hung(struct amdgpu_device *adev,
+					      bool hung)
+{
+	u32 tmp = RREG32(adev->bios_scratch_reg_offset + 3);
+
+	if (hung)
+		tmp |= ATOM_S3_ASIC_GUI_ENGINE_HUNG;
+	else
+		tmp &= ~ATOM_S3_ASIC_GUI_ENGINE_HUNG;
+
+	WREG32(adev->bios_scratch_reg_offset + 3, tmp);
+}
+
+bool amdgpu_atombios_scratch_need_asic_init(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(adev->bios_scratch_reg_offset + 7);
+
+	if (tmp & ATOM_S7_ASIC_INIT_COMPLETE_MASK)
+		return false;
+	else
+		return true;
+}
+
+/* Atom needs data in little endian format so swap as appropriate when copying
+ * data to or from atom. Note that atom operates on dw units.
+ *
+ * Use to_le=true when sending data to atom and provide at least
+ * ALIGN(num_bytes,4) bytes in the dst buffer.
+ *
+ * Use to_le=false when receiving data from atom and provide ALIGN(num_bytes,4)
+ * byes in the src buffer.
+ */
+void amdgpu_atombios_copy_swap(u8 *dst, u8 *src, u8 num_bytes, bool to_le)
+{
+#ifdef __BIG_ENDIAN
+	u32 src_tmp[5], dst_tmp[5];
+	int i;
+	u8 align_num_bytes = roundup2(num_bytes, 4);
+
+	if (to_le) {
+		memcpy(src_tmp, src, num_bytes);
+		for (i = 0; i < align_num_bytes / 4; i++)
+			dst_tmp[i] = cpu_to_le32(src_tmp[i]);
+		memcpy(dst, dst_tmp, align_num_bytes);
+	} else {
+		memcpy(src_tmp, src, align_num_bytes);
+		for (i = 0; i < align_num_bytes / 4; i++)
+			dst_tmp[i] = le32_to_cpu(src_tmp[i]);
+		memcpy(dst, dst_tmp, num_bytes);
+	}
+#else
+	memcpy(dst, src, num_bytes);
+#endif
+}
+
+static int amdgpu_atombios_allocate_fb_scratch(struct amdgpu_device *adev)
+{
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	int index = GetIndexIntoMasterTable(DATA, VRAM_UsageByFirmware);
+	uint16_t data_offset;
+	int usage_bytes = 0;
+	struct _ATOM_VRAM_USAGE_BY_FIRMWARE *firmware_usage;
+	u64 start_addr;
+	u64 size;
+
+	if (amdgpu_atom_parse_data_header(ctx, index, NULL, NULL, NULL, &data_offset)) {
+		firmware_usage = (struct _ATOM_VRAM_USAGE_BY_FIRMWARE *)(ctx->bios + data_offset);
+
+		DRM_DEBUG("atom firmware requested %08x %dkb\n",
+			  le32_to_cpu(firmware_usage->asFirmwareVramReserveInfo[0].ulStartAddrUsedByFirmware),
+			  le16_to_cpu(firmware_usage->asFirmwareVramReserveInfo[0].usFirmwareUseInKb));
+
+		start_addr = firmware_usage->asFirmwareVramReserveInfo[0].ulStartAddrUsedByFirmware;
+		size = firmware_usage->asFirmwareVramReserveInfo[0].usFirmwareUseInKb;
+
+		if ((uint32_t)(start_addr & ATOM_VRAM_OPERATION_FLAGS_MASK) ==
+			(uint32_t)(ATOM_VRAM_BLOCK_SRIOV_MSG_SHARE_RESERVATION <<
+			ATOM_VRAM_OPERATION_FLAGS_SHIFT)) {
+			/* Firmware request VRAM reservation for SR-IOV */
+			adev->fw_vram_usage.start_offset = (start_addr &
+				(~ATOM_VRAM_OPERATION_FLAGS_MASK)) << 10;
+			adev->fw_vram_usage.size = size << 10;
+			/* Use the default scratch size */
+			usage_bytes = 0;
+		} else {
+			usage_bytes = le16_to_cpu(firmware_usage->asFirmwareVramReserveInfo[0].usFirmwareUseInKb) * 1024;
+		}
+	}
+	ctx->scratch_size_bytes = 0;
+	if (usage_bytes == 0)
+		usage_bytes = 20 * 1024;
+	/* allocate some scratch memory */
+	ctx->scratch = kzalloc(usage_bytes, GFP_KERNEL);
+	if (!ctx->scratch)
+		return -ENOMEM;
+	ctx->scratch_size_bytes = usage_bytes;
+	return 0;
+}
+
+/* ATOM accessor methods */
+/*
+ * ATOM is an interpreted byte code stored in tables in the vbios.  The
+ * driver registers callbacks to access registers and the interpreter
+ * in the driver parses the tables and executes then to program specific
+ * actions (set display modes, asic init, etc.).  See amdgpu_atombios.c,
+ * atombios.h, and atom.c
+ */
+
+/**
+ * cail_pll_read - read PLL register
+ *
+ * @info: atom card_info pointer
+ * @reg: PLL register offset
+ *
+ * Provides a PLL register accessor for the atom interpreter (r4xx+).
+ * Returns the value of the PLL register.
+ */
+static uint32_t cail_pll_read(struct card_info *info, uint32_t reg)
+{
+	return 0;
+}
+
+/**
+ * cail_pll_write - write PLL register
+ *
+ * @info: atom card_info pointer
+ * @reg: PLL register offset
+ * @val: value to write to the pll register
+ *
+ * Provides a PLL register accessor for the atom interpreter (r4xx+).
+ */
+static void cail_pll_write(struct card_info *info, uint32_t reg, uint32_t val)
+{
+
+}
+
+/**
+ * cail_mc_read - read MC (Memory Controller) register
+ *
+ * @info: atom card_info pointer
+ * @reg: MC register offset
+ *
+ * Provides an MC register accessor for the atom interpreter (r4xx+).
+ * Returns the value of the MC register.
+ */
+static uint32_t cail_mc_read(struct card_info *info, uint32_t reg)
+{
+	return 0;
+}
+
+/**
+ * cail_mc_write - write MC (Memory Controller) register
+ *
+ * @info: atom card_info pointer
+ * @reg: MC register offset
+ * @val: value to write to the pll register
+ *
+ * Provides a MC register accessor for the atom interpreter (r4xx+).
+ */
+static void cail_mc_write(struct card_info *info, uint32_t reg, uint32_t val)
+{
+
+}
+
+/**
+ * cail_reg_write - write MMIO register
+ *
+ * @info: atom card_info pointer
+ * @reg: MMIO register offset
+ * @val: value to write to the pll register
+ *
+ * Provides a MMIO register accessor for the atom interpreter (r4xx+).
+ */
+static void cail_reg_write(struct card_info *info, uint32_t reg, uint32_t val)
+{
+	struct amdgpu_device *adev = info->dev->dev_private;
+
+	WREG32(reg, val);
+}
+
+/**
+ * cail_reg_read - read MMIO register
+ *
+ * @info: atom card_info pointer
+ * @reg: MMIO register offset
+ *
+ * Provides an MMIO register accessor for the atom interpreter (r4xx+).
+ * Returns the value of the MMIO register.
+ */
+static uint32_t cail_reg_read(struct card_info *info, uint32_t reg)
+{
+	struct amdgpu_device *adev = info->dev->dev_private;
+	uint32_t r;
+
+	r = RREG32(reg);
+	return r;
+}
+
+/**
+ * cail_ioreg_write - write IO register
+ *
+ * @info: atom card_info pointer
+ * @reg: IO register offset
+ * @val: value to write to the pll register
+ *
+ * Provides a IO register accessor for the atom interpreter (r4xx+).
+ */
+static void cail_ioreg_write(struct card_info *info, uint32_t reg, uint32_t val)
+{
+	struct amdgpu_device *adev = info->dev->dev_private;
+
+	WREG32_IO(reg, val);
+}
+
+/**
+ * cail_ioreg_read - read IO register
+ *
+ * @info: atom card_info pointer
+ * @reg: IO register offset
+ *
+ * Provides an IO register accessor for the atom interpreter (r4xx+).
+ * Returns the value of the IO register.
+ */
+static uint32_t cail_ioreg_read(struct card_info *info, uint32_t reg)
+{
+	struct amdgpu_device *adev = info->dev->dev_private;
+	uint32_t r;
+
+	r = RREG32_IO(reg);
+	return r;
+}
+
+static ssize_t amdgpu_atombios_get_vbios_version(struct device *dev,
+						 struct device_attribute *attr,
+						 char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	struct atom_context *ctx = adev->mode_info.atom_context;
+
+	return snprintf(buf, PAGE_SIZE, "%s\n", ctx->vbios_version);
+}
+
+static DEVICE_ATTR(vbios_version, 0444, amdgpu_atombios_get_vbios_version,
+		   NULL);
+
+/**
+ * amdgpu_atombios_fini - free the driver info and callbacks for atombios
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Frees the driver info and register access callbacks for the ATOM
+ * interpreter (r4xx+).
+ * Called at driver shutdown.
+ */
+void amdgpu_atombios_fini(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.atom_context) {
+		kfree(adev->mode_info.atom_context->scratch);
+		kfree(adev->mode_info.atom_context->iio);
+	}
+	kfree(adev->mode_info.atom_context);
+	adev->mode_info.atom_context = NULL;
+	kfree(adev->mode_info.atom_card_info);
+	adev->mode_info.atom_card_info = NULL;
+	device_remove_file(adev->dev, &dev_attr_vbios_version);
+}
+
+/**
+ * amdgpu_atombios_init - init the driver info and callbacks for atombios
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initializes the driver info and register access callbacks for the
+ * ATOM interpreter (r4xx+).
+ * Returns 0 on sucess, -ENOMEM on failure.
+ * Called at driver startup.
+ */
+int amdgpu_atombios_init(struct amdgpu_device *adev)
+{
+	struct card_info *atom_card_info =
+	    kzalloc(sizeof(struct card_info), GFP_KERNEL);
+	int ret;
+
+	if (!atom_card_info)
+		return -ENOMEM;
+
+	adev->mode_info.atom_card_info = atom_card_info;
+	atom_card_info->dev = adev->ddev;
+	atom_card_info->reg_read = cail_reg_read;
+	atom_card_info->reg_write = cail_reg_write;
+	/* needed for iio ops */
+	if (adev->rio_mem_size) {
+		atom_card_info->ioreg_read = cail_ioreg_read;
+		atom_card_info->ioreg_write = cail_ioreg_write;
+	} else {
+		DRM_DEBUG("PCI I/O BAR is not found. Using MMIO to access ATOM BIOS\n");
+		atom_card_info->ioreg_read = cail_reg_read;
+		atom_card_info->ioreg_write = cail_reg_write;
+	}
+	atom_card_info->mc_read = cail_mc_read;
+	atom_card_info->mc_write = cail_mc_write;
+	atom_card_info->pll_read = cail_pll_read;
+	atom_card_info->pll_write = cail_pll_write;
+
+	adev->mode_info.atom_context = amdgpu_atom_parse(atom_card_info, adev->bios);
+	if (!adev->mode_info.atom_context) {
+		amdgpu_atombios_fini(adev);
+		return -ENOMEM;
+	}
+
+	rw_init(&adev->mode_info.atom_context->mutex, "atomcm");
+	if (adev->is_atom_fw) {
+		amdgpu_atomfirmware_scratch_regs_init(adev);
+		amdgpu_atomfirmware_allocate_fb_scratch(adev);
+	} else {
+		amdgpu_atombios_scratch_regs_init(adev);
+		amdgpu_atombios_allocate_fb_scratch(adev);
+	}
+
+	ret = device_create_file(adev->dev, &dev_attr_vbios_version);
+	if (ret) {
+		DRM_ERROR("Failed to create device file for VBIOS version\n");
+		return ret;
+	}
+
+	return 0;
+}
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios.h
new file mode 100644
index 00000000000..fd8f18074f7
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios.h
@@ -0,0 +1,222 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_ATOMBIOS_H__
+#define __AMDGPU_ATOMBIOS_H__
+
+struct atom_clock_dividers {
+	u32 post_div;
+	union {
+		struct {
+#ifdef __BIG_ENDIAN
+			u32 reserved : 6;
+			u32 whole_fb_div : 12;
+			u32 frac_fb_div : 14;
+#else
+			u32 frac_fb_div : 14;
+			u32 whole_fb_div : 12;
+			u32 reserved : 6;
+#endif
+		};
+		u32 fb_div;
+	};
+	u32 ref_div;
+	bool enable_post_div;
+	bool enable_dithen;
+	u32 vco_mode;
+	u32 real_clock;
+	/* added for CI */
+	u32 post_divider;
+	u32 flags;
+};
+
+struct atom_mpll_param {
+	union {
+		struct {
+#ifdef __BIG_ENDIAN
+			u32 reserved : 8;
+			u32 clkfrac : 12;
+			u32 clkf : 12;
+#else
+			u32 clkf : 12;
+			u32 clkfrac : 12;
+			u32 reserved : 8;
+#endif
+		};
+		u32 fb_div;
+	};
+	u32 post_div;
+	u32 bwcntl;
+	u32 dll_speed;
+	u32 vco_mode;
+	u32 yclk_sel;
+	u32 qdr;
+	u32 half_rate;
+};
+
+#define MEM_TYPE_GDDR5  0x50
+#define MEM_TYPE_GDDR4  0x40
+#define MEM_TYPE_GDDR3  0x30
+#define MEM_TYPE_DDR2   0x20
+#define MEM_TYPE_GDDR1  0x10
+#define MEM_TYPE_DDR3   0xb0
+#define MEM_TYPE_MASK   0xf0
+
+struct atom_memory_info {
+	u8 mem_vendor;
+	u8 mem_type;
+};
+
+#define MAX_AC_TIMING_ENTRIES 16
+
+struct atom_memory_clock_range_table
+{
+	u8 num_entries;
+	u8 rsv[3];
+	u32 mclk[MAX_AC_TIMING_ENTRIES];
+};
+
+#define VBIOS_MC_REGISTER_ARRAY_SIZE 32
+#define VBIOS_MAX_AC_TIMING_ENTRIES 20
+
+struct atom_mc_reg_entry {
+	u32 mclk_max;
+	u32 mc_data[VBIOS_MC_REGISTER_ARRAY_SIZE];
+};
+
+struct atom_mc_register_address {
+	u16 s1;
+	u8 pre_reg_data;
+};
+
+struct atom_mc_reg_table {
+	u8 last;
+	u8 num_entries;
+	struct atom_mc_reg_entry mc_reg_table_entry[VBIOS_MAX_AC_TIMING_ENTRIES];
+	struct atom_mc_register_address mc_reg_address[VBIOS_MC_REGISTER_ARRAY_SIZE];
+};
+
+#define MAX_VOLTAGE_ENTRIES 32
+
+struct atom_voltage_table_entry
+{
+	u16 value;
+	u32 smio_low;
+};
+
+struct atom_voltage_table
+{
+	u32 count;
+	u32 mask_low;
+	u32 phase_delay;
+	struct atom_voltage_table_entry entries[MAX_VOLTAGE_ENTRIES];
+};
+
+struct amdgpu_gpio_rec
+amdgpu_atombios_lookup_gpio(struct amdgpu_device *adev,
+			    u8 id);
+
+struct amdgpu_i2c_bus_rec amdgpu_atombios_lookup_i2c_gpio(struct amdgpu_device *adev,
+							  uint8_t id);
+void amdgpu_atombios_i2c_init(struct amdgpu_device *adev);
+
+bool amdgpu_atombios_has_dce_engine_info(struct amdgpu_device *adev);
+
+bool amdgpu_atombios_get_connector_info_from_object_table(struct amdgpu_device *adev);
+
+int amdgpu_atombios_get_clock_info(struct amdgpu_device *adev);
+
+int amdgpu_atombios_get_gfx_info(struct amdgpu_device *adev);
+
+int amdgpu_atombios_get_vram_width(struct amdgpu_device *adev);
+
+bool amdgpu_atombios_get_asic_ss_info(struct amdgpu_device *adev,
+				      struct amdgpu_atom_ss *ss,
+				      int id, u32 clock);
+
+int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
+				       u8 clock_type,
+				       u32 clock,
+				       bool strobe_mode,
+				       struct atom_clock_dividers *dividers);
+
+int amdgpu_atombios_get_memory_pll_dividers(struct amdgpu_device *adev,
+					    u32 clock,
+					    bool strobe_mode,
+					    struct atom_mpll_param *mpll_param);
+
+void amdgpu_atombios_set_engine_dram_timings(struct amdgpu_device *adev,
+					     u32 eng_clock, u32 mem_clock);
+
+int amdgpu_atombios_get_leakage_id_from_vbios(struct amdgpu_device *adev,
+					      u16 *leakage_id);
+
+int amdgpu_atombios_get_leakage_vddc_based_on_leakage_params(struct amdgpu_device *adev,
+							     u16 *vddc, u16 *vddci,
+							     u16 virtual_voltage_id,
+							     u16 vbios_voltage_id);
+
+int amdgpu_atombios_get_voltage_evv(struct amdgpu_device *adev,
+				    u16 virtual_voltage_id,
+				    u16 *voltage);
+
+bool
+amdgpu_atombios_is_voltage_gpio(struct amdgpu_device *adev,
+				u8 voltage_type, u8 voltage_mode);
+
+int amdgpu_atombios_get_voltage_table(struct amdgpu_device *adev,
+				      u8 voltage_type, u8 voltage_mode,
+				      struct atom_voltage_table *voltage_table);
+
+int amdgpu_atombios_init_mc_reg_table(struct amdgpu_device *adev,
+				      u8 module_index,
+				      struct atom_mc_reg_table *reg_table);
+
+bool amdgpu_atombios_has_gpu_virtualization_table(struct amdgpu_device *adev);
+
+void amdgpu_atombios_scratch_regs_lock(struct amdgpu_device *adev, bool lock);
+void amdgpu_atombios_scratch_regs_engine_hung(struct amdgpu_device *adev,
+					      bool hung);
+bool amdgpu_atombios_scratch_need_asic_init(struct amdgpu_device *adev);
+
+void amdgpu_atombios_copy_swap(u8 *dst, u8 *src, u8 num_bytes, bool to_le);
+int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type,
+			     u16 voltage_id, u16 *voltage);
+int amdgpu_atombios_get_leakage_vddc_based_on_leakage_idx(struct amdgpu_device *adev,
+						      u16 *voltage,
+						      u16 leakage_idx);
+void amdgpu_atombios_get_default_voltages(struct amdgpu_device *adev,
+					  u16 *vddc, u16 *vddci, u16 *mvdd);
+int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
+				       u8 clock_type,
+				       u32 clock,
+				       bool strobe_mode,
+				       struct atom_clock_dividers *dividers);
+int amdgpu_atombios_get_svi2_info(struct amdgpu_device *adev,
+			      u8 voltage_type,
+			      u8 *svd_gpio_id, u8 *svc_gpio_id);
+
+void amdgpu_atombios_fini(struct amdgpu_device *adev);
+int amdgpu_atombios_init(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios_crtc.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios_crtc.c
new file mode 100644
index 00000000000..8a0818b23ea
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios_crtc.c
@@ -0,0 +1,886 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <drm/drmP.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/amdgpu_drm.h>
+#include <drm/drm_fixed.h>
+#include "amdgpu.h"
+#include "atom.h"
+#include "atom-bits.h"
+#include "atombios_encoders.h"
+#include "atombios_crtc.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_pll.h"
+#include "amdgpu_connectors.h"
+
+void amdgpu_atombios_crtc_overscan_setup(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	SET_CRTC_OVERSCAN_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_OverScan);
+	int a1, a2;
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucCRTC = amdgpu_crtc->crtc_id;
+
+	switch (amdgpu_crtc->rmx_type) {
+	case RMX_CENTER:
+		args.usOverscanTop = cpu_to_le16((adjusted_mode->crtc_vdisplay - mode->crtc_vdisplay) / 2);
+		args.usOverscanBottom = cpu_to_le16((adjusted_mode->crtc_vdisplay - mode->crtc_vdisplay) / 2);
+		args.usOverscanLeft = cpu_to_le16((adjusted_mode->crtc_hdisplay - mode->crtc_hdisplay) / 2);
+		args.usOverscanRight = cpu_to_le16((adjusted_mode->crtc_hdisplay - mode->crtc_hdisplay) / 2);
+		break;
+	case RMX_ASPECT:
+		a1 = mode->crtc_vdisplay * adjusted_mode->crtc_hdisplay;
+		a2 = adjusted_mode->crtc_vdisplay * mode->crtc_hdisplay;
+
+		if (a1 > a2) {
+			args.usOverscanLeft = cpu_to_le16((adjusted_mode->crtc_hdisplay - (a2 / mode->crtc_vdisplay)) / 2);
+			args.usOverscanRight = cpu_to_le16((adjusted_mode->crtc_hdisplay - (a2 / mode->crtc_vdisplay)) / 2);
+		} else if (a2 > a1) {
+			args.usOverscanTop = cpu_to_le16((adjusted_mode->crtc_vdisplay - (a1 / mode->crtc_hdisplay)) / 2);
+			args.usOverscanBottom = cpu_to_le16((adjusted_mode->crtc_vdisplay - (a1 / mode->crtc_hdisplay)) / 2);
+		}
+		break;
+	case RMX_FULL:
+	default:
+		args.usOverscanRight = cpu_to_le16(amdgpu_crtc->h_border);
+		args.usOverscanLeft = cpu_to_le16(amdgpu_crtc->h_border);
+		args.usOverscanBottom = cpu_to_le16(amdgpu_crtc->v_border);
+		args.usOverscanTop = cpu_to_le16(amdgpu_crtc->v_border);
+		break;
+	}
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_scaler_setup(struct drm_crtc *crtc)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	ENABLE_SCALER_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, EnableScaler);
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucScaler = amdgpu_crtc->crtc_id;
+
+	switch (amdgpu_crtc->rmx_type) {
+	case RMX_FULL:
+		args.ucEnable = ATOM_SCALER_EXPANSION;
+		break;
+	case RMX_CENTER:
+		args.ucEnable = ATOM_SCALER_CENTER;
+		break;
+	case RMX_ASPECT:
+		args.ucEnable = ATOM_SCALER_EXPANSION;
+		break;
+	default:
+		args.ucEnable = ATOM_SCALER_DISABLE;
+		break;
+	}
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_lock(struct drm_crtc *crtc, int lock)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int index =
+	    GetIndexIntoMasterTable(COMMAND, UpdateCRTC_DoubleBufferRegisters);
+	ENABLE_CRTC_PS_ALLOCATION args;
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucCRTC = amdgpu_crtc->crtc_id;
+	args.ucEnable = lock;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_enable(struct drm_crtc *crtc, int state)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int index = GetIndexIntoMasterTable(COMMAND, EnableCRTC);
+	ENABLE_CRTC_PS_ALLOCATION args;
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucCRTC = amdgpu_crtc->crtc_id;
+	args.ucEnable = state;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_blank(struct drm_crtc *crtc, int state)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int index = GetIndexIntoMasterTable(COMMAND, BlankCRTC);
+	BLANK_CRTC_PS_ALLOCATION args;
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucCRTC = amdgpu_crtc->crtc_id;
+	args.ucBlanking = state;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_powergate(struct drm_crtc *crtc, int state)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int index = GetIndexIntoMasterTable(COMMAND, EnableDispPowerGating);
+	ENABLE_DISP_POWER_GATING_PS_ALLOCATION args;
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucDispPipeId = amdgpu_crtc->crtc_id;
+	args.ucEnable = state;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_powergate_init(struct amdgpu_device *adev)
+{
+	int index = GetIndexIntoMasterTable(COMMAND, EnableDispPowerGating);
+	ENABLE_DISP_POWER_GATING_PS_ALLOCATION args;
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucEnable = ATOM_INIT;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_set_dtd_timing(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	SET_CRTC_USING_DTD_TIMING_PARAMETERS args;
+	int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_UsingDTDTiming);
+	u16 misc = 0;
+
+	memset(&args, 0, sizeof(args));
+	args.usH_Size = cpu_to_le16(mode->crtc_hdisplay - (amdgpu_crtc->h_border * 2));
+	args.usH_Blanking_Time =
+		cpu_to_le16(mode->crtc_hblank_end - mode->crtc_hdisplay + (amdgpu_crtc->h_border * 2));
+	args.usV_Size = cpu_to_le16(mode->crtc_vdisplay - (amdgpu_crtc->v_border * 2));
+	args.usV_Blanking_Time =
+		cpu_to_le16(mode->crtc_vblank_end - mode->crtc_vdisplay + (amdgpu_crtc->v_border * 2));
+	args.usH_SyncOffset =
+		cpu_to_le16(mode->crtc_hsync_start - mode->crtc_hdisplay + amdgpu_crtc->h_border);
+	args.usH_SyncWidth =
+		cpu_to_le16(mode->crtc_hsync_end - mode->crtc_hsync_start);
+	args.usV_SyncOffset =
+		cpu_to_le16(mode->crtc_vsync_start - mode->crtc_vdisplay + amdgpu_crtc->v_border);
+	args.usV_SyncWidth =
+		cpu_to_le16(mode->crtc_vsync_end - mode->crtc_vsync_start);
+	args.ucH_Border = amdgpu_crtc->h_border;
+	args.ucV_Border = amdgpu_crtc->v_border;
+
+	if (mode->flags & DRM_MODE_FLAG_NVSYNC)
+		misc |= ATOM_VSYNC_POLARITY;
+	if (mode->flags & DRM_MODE_FLAG_NHSYNC)
+		misc |= ATOM_HSYNC_POLARITY;
+	if (mode->flags & DRM_MODE_FLAG_CSYNC)
+		misc |= ATOM_COMPOSITESYNC;
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		misc |= ATOM_INTERLACE;
+	if (mode->flags & DRM_MODE_FLAG_DBLSCAN)
+		misc |= ATOM_DOUBLE_CLOCK_MODE;
+
+	args.susModeMiscInfo.usAccess = cpu_to_le16(misc);
+	args.ucCRTC = amdgpu_crtc->crtc_id;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+union atom_enable_ss {
+	ENABLE_SPREAD_SPECTRUM_ON_PPLL_PS_ALLOCATION v1;
+	ENABLE_SPREAD_SPECTRUM_ON_PPLL_V2 v2;
+	ENABLE_SPREAD_SPECTRUM_ON_PPLL_V3 v3;
+};
+
+static void amdgpu_atombios_crtc_program_ss(struct amdgpu_device *adev,
+				     int enable,
+				     int pll_id,
+				     int crtc_id,
+				     struct amdgpu_atom_ss *ss)
+{
+	unsigned i;
+	int index = GetIndexIntoMasterTable(COMMAND, EnableSpreadSpectrumOnPPLL);
+	union atom_enable_ss args;
+
+	if (enable) {
+		/* Don't mess with SS if percentage is 0 or external ss.
+		 * SS is already disabled previously, and disabling it
+		 * again can cause display problems if the pll is already
+		 * programmed.
+		 */
+		if (ss->percentage == 0)
+			return;
+		if (ss->type & ATOM_EXTERNAL_SS_MASK)
+			return;
+	} else {
+		for (i = 0; i < adev->mode_info.num_crtc; i++) {
+			if (adev->mode_info.crtcs[i] &&
+			    adev->mode_info.crtcs[i]->enabled &&
+			    i != crtc_id &&
+			    pll_id == adev->mode_info.crtcs[i]->pll_id) {
+				/* one other crtc is using this pll don't turn
+				 * off spread spectrum as it might turn off
+				 * display on active crtc
+				 */
+				return;
+			}
+		}
+	}
+
+	memset(&args, 0, sizeof(args));
+
+	args.v3.usSpreadSpectrumAmountFrac = cpu_to_le16(0);
+	args.v3.ucSpreadSpectrumType = ss->type & ATOM_SS_CENTRE_SPREAD_MODE_MASK;
+	switch (pll_id) {
+	case ATOM_PPLL1:
+		args.v3.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V3_P1PLL;
+		break;
+	case ATOM_PPLL2:
+		args.v3.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V3_P2PLL;
+		break;
+	case ATOM_DCPLL:
+		args.v3.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V3_DCPLL;
+		break;
+	case ATOM_PPLL_INVALID:
+		return;
+	}
+	args.v3.usSpreadSpectrumAmount = cpu_to_le16(ss->amount);
+	args.v3.usSpreadSpectrumStep = cpu_to_le16(ss->step);
+	args.v3.ucEnable = enable;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+union adjust_pixel_clock {
+	ADJUST_DISPLAY_PLL_PS_ALLOCATION v1;
+	ADJUST_DISPLAY_PLL_PS_ALLOCATION_V3 v3;
+};
+
+static u32 amdgpu_atombios_crtc_adjust_pll(struct drm_crtc *crtc,
+				    struct drm_display_mode *mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_encoder *encoder = amdgpu_crtc->encoder;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	u32 adjusted_clock = mode->clock;
+	int encoder_mode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+	u32 dp_clock = mode->clock;
+	u32 clock = mode->clock;
+	int bpc = amdgpu_crtc->bpc;
+	bool is_duallink = amdgpu_dig_monitor_is_duallink(encoder, mode->clock);
+	union adjust_pixel_clock args;
+	u8 frev, crev;
+	int index;
+
+	amdgpu_crtc->pll_flags = AMDGPU_PLL_USE_FRAC_FB_DIV;
+
+	if ((amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT | ATOM_DEVICE_DFP_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) != ENCODER_OBJECT_ID_NONE)) {
+		if (connector) {
+			struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+			struct amdgpu_connector_atom_dig *dig_connector =
+				amdgpu_connector->con_priv;
+
+			dp_clock = dig_connector->dp_clock;
+		}
+	}
+
+	/* use recommended ref_div for ss */
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+		if (amdgpu_crtc->ss_enabled) {
+			if (amdgpu_crtc->ss.refdiv) {
+				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_REF_DIV;
+				amdgpu_crtc->pll_reference_div = amdgpu_crtc->ss.refdiv;
+				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_FRAC_FB_DIV;
+			}
+		}
+	}
+
+	/* DVO wants 2x pixel clock if the DVO chip is in 12 bit mode */
+	if (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1)
+		adjusted_clock = mode->clock * 2;
+	if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+		amdgpu_crtc->pll_flags |= AMDGPU_PLL_PREFER_CLOSEST_LOWER;
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+		amdgpu_crtc->pll_flags |= AMDGPU_PLL_IS_LCD;
+
+
+	/* adjust pll for deep color modes */
+	if (encoder_mode == ATOM_ENCODER_MODE_HDMI) {
+		switch (bpc) {
+		case 8:
+		default:
+			break;
+		case 10:
+			clock = (clock * 5) / 4;
+			break;
+		case 12:
+			clock = (clock * 3) / 2;
+			break;
+		case 16:
+			clock = clock * 2;
+			break;
+		}
+	}
+
+	/* DCE3+ has an AdjustDisplayPll that will adjust the pixel clock
+	 * accordingly based on the encoder/transmitter to work around
+	 * special hw requirements.
+	 */
+	index = GetIndexIntoMasterTable(COMMAND, AdjustDisplayPll);
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev,
+				   &crev))
+		return adjusted_clock;
+
+	memset(&args, 0, sizeof(args));
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 1:
+		case 2:
+			args.v1.usPixelClock = cpu_to_le16(clock / 10);
+			args.v1.ucTransmitterID = amdgpu_encoder->encoder_id;
+			args.v1.ucEncodeMode = encoder_mode;
+			if (amdgpu_crtc->ss_enabled && amdgpu_crtc->ss.percentage)
+				args.v1.ucConfig |=
+					ADJUST_DISPLAY_CONFIG_SS_ENABLE;
+
+			amdgpu_atom_execute_table(adev->mode_info.atom_context,
+					   index, (uint32_t *)&args);
+			adjusted_clock = le16_to_cpu(args.v1.usPixelClock) * 10;
+			break;
+		case 3:
+			args.v3.sInput.usPixelClock = cpu_to_le16(clock / 10);
+			args.v3.sInput.ucTransmitterID = amdgpu_encoder->encoder_id;
+			args.v3.sInput.ucEncodeMode = encoder_mode;
+			args.v3.sInput.ucDispPllConfig = 0;
+			if (amdgpu_crtc->ss_enabled && amdgpu_crtc->ss.percentage)
+				args.v3.sInput.ucDispPllConfig |=
+					DISPPLL_CONFIG_SS_ENABLE;
+			if (ENCODER_MODE_IS_DP(encoder_mode)) {
+				args.v3.sInput.ucDispPllConfig |=
+					DISPPLL_CONFIG_COHERENT_MODE;
+				/* 16200 or 27000 */
+				args.v3.sInput.usPixelClock = cpu_to_le16(dp_clock / 10);
+			} else if (amdgpu_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) {
+				struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+				if (dig->coherent_mode)
+					args.v3.sInput.ucDispPllConfig |=
+						DISPPLL_CONFIG_COHERENT_MODE;
+				if (is_duallink)
+					args.v3.sInput.ucDispPllConfig |=
+						DISPPLL_CONFIG_DUAL_LINK;
+			}
+			if (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) !=
+			    ENCODER_OBJECT_ID_NONE)
+				args.v3.sInput.ucExtTransmitterID =
+					amdgpu_encoder_get_dp_bridge_encoder_id(encoder);
+			else
+				args.v3.sInput.ucExtTransmitterID = 0;
+
+			amdgpu_atom_execute_table(adev->mode_info.atom_context,
+					   index, (uint32_t *)&args);
+			adjusted_clock = le32_to_cpu(args.v3.sOutput.ulDispPllFreq) * 10;
+			if (args.v3.sOutput.ucRefDiv) {
+				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_FRAC_FB_DIV;
+				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_REF_DIV;
+				amdgpu_crtc->pll_reference_div = args.v3.sOutput.ucRefDiv;
+			}
+			if (args.v3.sOutput.ucPostDiv) {
+				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_FRAC_FB_DIV;
+				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_POST_DIV;
+				amdgpu_crtc->pll_post_div = args.v3.sOutput.ucPostDiv;
+			}
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+			return adjusted_clock;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+		return adjusted_clock;
+	}
+
+	return adjusted_clock;
+}
+
+union set_pixel_clock {
+	SET_PIXEL_CLOCK_PS_ALLOCATION base;
+	PIXEL_CLOCK_PARAMETERS v1;
+	PIXEL_CLOCK_PARAMETERS_V2 v2;
+	PIXEL_CLOCK_PARAMETERS_V3 v3;
+	PIXEL_CLOCK_PARAMETERS_V5 v5;
+	PIXEL_CLOCK_PARAMETERS_V6 v6;
+	PIXEL_CLOCK_PARAMETERS_V7 v7;
+};
+
+/* on DCE5, make sure the voltage is high enough to support the
+ * required disp clk.
+ */
+void amdgpu_atombios_crtc_set_disp_eng_pll(struct amdgpu_device *adev,
+					   u32 dispclk)
+{
+	u8 frev, crev;
+	int index;
+	union set_pixel_clock args;
+
+	memset(&args, 0, sizeof(args));
+
+	index = GetIndexIntoMasterTable(COMMAND, SetPixelClock);
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev,
+				   &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 5:
+			/* if the default dcpll clock is specified,
+			 * SetPixelClock provides the dividers
+			 */
+			args.v5.ucCRTC = ATOM_CRTC_INVALID;
+			args.v5.usPixelClock = cpu_to_le16(dispclk);
+			args.v5.ucPpll = ATOM_DCPLL;
+			break;
+		case 6:
+			/* if the default dcpll clock is specified,
+			 * SetPixelClock provides the dividers
+			 */
+			args.v6.ulDispEngClkFreq = cpu_to_le32(dispclk);
+			if (adev->asic_type == CHIP_TAHITI ||
+			    adev->asic_type == CHIP_PITCAIRN ||
+			    adev->asic_type == CHIP_VERDE ||
+			    adev->asic_type == CHIP_OLAND)
+				args.v6.ucPpll = ATOM_PPLL0;
+			else
+				args.v6.ucPpll = ATOM_EXT_PLL1;
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+			return;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+		return;
+	}
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+union set_dce_clock {
+	SET_DCE_CLOCK_PS_ALLOCATION_V1_1 v1_1;
+	SET_DCE_CLOCK_PS_ALLOCATION_V2_1 v2_1;
+};
+
+u32 amdgpu_atombios_crtc_set_dce_clock(struct amdgpu_device *adev,
+				       u32 freq, u8 clk_type, u8 clk_src)
+{
+	u8 frev, crev;
+	int index;
+	union set_dce_clock args;
+	u32 ret_freq = 0;
+
+	memset(&args, 0, sizeof(args));
+
+	index = GetIndexIntoMasterTable(COMMAND, SetDCEClock);
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev,
+				   &crev))
+		return 0;
+
+	switch (frev) {
+	case 2:
+		switch (crev) {
+		case 1:
+			args.v2_1.asParam.ulDCEClkFreq = cpu_to_le32(freq); /* 10kHz units */
+			args.v2_1.asParam.ucDCEClkType = clk_type;
+			args.v2_1.asParam.ucDCEClkSrc = clk_src;
+			amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+			ret_freq = le32_to_cpu(args.v2_1.asParam.ulDCEClkFreq) * 10;
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+			return 0;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+		return 0;
+	}
+
+	return ret_freq;
+}
+
+static bool is_pixel_clock_source_from_pll(u32 encoder_mode, int pll_id)
+{
+	if (ENCODER_MODE_IS_DP(encoder_mode)) {
+		if (pll_id < ATOM_EXT_PLL1)
+			return true;
+		else
+			return false;
+	} else {
+		return true;
+	}
+}
+
+void amdgpu_atombios_crtc_program_pll(struct drm_crtc *crtc,
+				      u32 crtc_id,
+				      int pll_id,
+				      u32 encoder_mode,
+				      u32 encoder_id,
+				      u32 clock,
+				      u32 ref_div,
+				      u32 fb_div,
+				      u32 frac_fb_div,
+				      u32 post_div,
+				      int bpc,
+				      bool ss_enabled,
+				      struct amdgpu_atom_ss *ss)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u8 frev, crev;
+	int index = GetIndexIntoMasterTable(COMMAND, SetPixelClock);
+	union set_pixel_clock args;
+
+	memset(&args, 0, sizeof(args));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev,
+				   &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 1:
+			if (clock == ATOM_DISABLE)
+				return;
+			args.v1.usPixelClock = cpu_to_le16(clock / 10);
+			args.v1.usRefDiv = cpu_to_le16(ref_div);
+			args.v1.usFbDiv = cpu_to_le16(fb_div);
+			args.v1.ucFracFbDiv = frac_fb_div;
+			args.v1.ucPostDiv = post_div;
+			args.v1.ucPpll = pll_id;
+			args.v1.ucCRTC = crtc_id;
+			args.v1.ucRefDivSrc = 1;
+			break;
+		case 2:
+			args.v2.usPixelClock = cpu_to_le16(clock / 10);
+			args.v2.usRefDiv = cpu_to_le16(ref_div);
+			args.v2.usFbDiv = cpu_to_le16(fb_div);
+			args.v2.ucFracFbDiv = frac_fb_div;
+			args.v2.ucPostDiv = post_div;
+			args.v2.ucPpll = pll_id;
+			args.v2.ucCRTC = crtc_id;
+			args.v2.ucRefDivSrc = 1;
+			break;
+		case 3:
+			args.v3.usPixelClock = cpu_to_le16(clock / 10);
+			args.v3.usRefDiv = cpu_to_le16(ref_div);
+			args.v3.usFbDiv = cpu_to_le16(fb_div);
+			args.v3.ucFracFbDiv = frac_fb_div;
+			args.v3.ucPostDiv = post_div;
+			args.v3.ucPpll = pll_id;
+			if (crtc_id == ATOM_CRTC2)
+				args.v3.ucMiscInfo = PIXEL_CLOCK_MISC_CRTC_SEL_CRTC2;
+			else
+				args.v3.ucMiscInfo = PIXEL_CLOCK_MISC_CRTC_SEL_CRTC1;
+			if (ss_enabled && (ss->type & ATOM_EXTERNAL_SS_MASK))
+				args.v3.ucMiscInfo |= PIXEL_CLOCK_MISC_REF_DIV_SRC;
+			args.v3.ucTransmitterId = encoder_id;
+			args.v3.ucEncoderMode = encoder_mode;
+			break;
+		case 5:
+			args.v5.ucCRTC = crtc_id;
+			args.v5.usPixelClock = cpu_to_le16(clock / 10);
+			args.v5.ucRefDiv = ref_div;
+			args.v5.usFbDiv = cpu_to_le16(fb_div);
+			args.v5.ulFbDivDecFrac = cpu_to_le32(frac_fb_div * 100000);
+			args.v5.ucPostDiv = post_div;
+			args.v5.ucMiscInfo = 0; /* HDMI depth, etc. */
+			if ((ss_enabled && (ss->type & ATOM_EXTERNAL_SS_MASK)) &&
+			    (pll_id < ATOM_EXT_PLL1))
+				args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_REF_DIV_SRC;
+			if (encoder_mode == ATOM_ENCODER_MODE_HDMI) {
+				switch (bpc) {
+				case 8:
+				default:
+					args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_HDMI_24BPP;
+					break;
+				case 10:
+					/* yes this is correct, the atom define is wrong */
+					args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_HDMI_32BPP;
+					break;
+				case 12:
+					/* yes this is correct, the atom define is wrong */
+					args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_HDMI_30BPP;
+					break;
+				}
+			}
+			args.v5.ucTransmitterID = encoder_id;
+			args.v5.ucEncoderMode = encoder_mode;
+			args.v5.ucPpll = pll_id;
+			break;
+		case 6:
+			args.v6.ulDispEngClkFreq = cpu_to_le32(crtc_id << 24 | clock / 10);
+			args.v6.ucRefDiv = ref_div;
+			args.v6.usFbDiv = cpu_to_le16(fb_div);
+			args.v6.ulFbDivDecFrac = cpu_to_le32(frac_fb_div * 100000);
+			args.v6.ucPostDiv = post_div;
+			args.v6.ucMiscInfo = 0; /* HDMI depth, etc. */
+			if ((ss_enabled && (ss->type & ATOM_EXTERNAL_SS_MASK)) &&
+			    (pll_id < ATOM_EXT_PLL1) &&
+			    !is_pixel_clock_source_from_pll(encoder_mode, pll_id))
+				args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_REF_DIV_SRC;
+			if (encoder_mode == ATOM_ENCODER_MODE_HDMI) {
+				switch (bpc) {
+				case 8:
+				default:
+					args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_24BPP;
+					break;
+				case 10:
+					args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_30BPP_V6;
+					break;
+				case 12:
+					args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_36BPP_V6;
+					break;
+				case 16:
+					args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_48BPP;
+					break;
+				}
+			}
+			args.v6.ucTransmitterID = encoder_id;
+			args.v6.ucEncoderMode = encoder_mode;
+			args.v6.ucPpll = pll_id;
+			break;
+		case 7:
+			args.v7.ulPixelClock = cpu_to_le32(clock * 10); /* 100 hz units */
+			args.v7.ucMiscInfo = 0;
+			if ((encoder_mode == ATOM_ENCODER_MODE_DVI) &&
+			    (clock > 165000))
+				args.v7.ucMiscInfo |= PIXEL_CLOCK_V7_MISC_DVI_DUALLINK_EN;
+			args.v7.ucCRTC = crtc_id;
+			if (encoder_mode == ATOM_ENCODER_MODE_HDMI) {
+				switch (bpc) {
+				case 8:
+				default:
+					args.v7.ucDeepColorRatio = PIXEL_CLOCK_V7_DEEPCOLOR_RATIO_DIS;
+					break;
+				case 10:
+					args.v7.ucDeepColorRatio = PIXEL_CLOCK_V7_DEEPCOLOR_RATIO_5_4;
+					break;
+				case 12:
+					args.v7.ucDeepColorRatio = PIXEL_CLOCK_V7_DEEPCOLOR_RATIO_3_2;
+					break;
+				case 16:
+					args.v7.ucDeepColorRatio = PIXEL_CLOCK_V7_DEEPCOLOR_RATIO_2_1;
+					break;
+				}
+			}
+			args.v7.ucTransmitterID = encoder_id;
+			args.v7.ucEncoderMode = encoder_mode;
+			args.v7.ucPpll = pll_id;
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+			return;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+		return;
+	}
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+int amdgpu_atombios_crtc_prepare_pll(struct drm_crtc *crtc,
+			      struct drm_display_mode *mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder =
+		to_amdgpu_encoder(amdgpu_crtc->encoder);
+	int encoder_mode = amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder);
+
+	amdgpu_crtc->bpc = 8;
+	amdgpu_crtc->ss_enabled = false;
+
+	if ((amdgpu_encoder->active_device & (ATOM_DEVICE_LCD_SUPPORT | ATOM_DEVICE_DFP_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(amdgpu_crtc->encoder) != ENCODER_OBJECT_ID_NONE)) {
+		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+		struct drm_connector *connector =
+			amdgpu_get_connector_for_encoder(amdgpu_crtc->encoder);
+		struct amdgpu_connector *amdgpu_connector =
+			to_amdgpu_connector(connector);
+		struct amdgpu_connector_atom_dig *dig_connector =
+			amdgpu_connector->con_priv;
+		int dp_clock;
+
+		/* Assign mode clock for hdmi deep color max clock limit check */
+		amdgpu_connector->pixelclock_for_modeset = mode->clock;
+		amdgpu_crtc->bpc = amdgpu_connector_get_monitor_bpc(connector);
+
+		switch (encoder_mode) {
+		case ATOM_ENCODER_MODE_DP_MST:
+		case ATOM_ENCODER_MODE_DP:
+			/* DP/eDP */
+			dp_clock = dig_connector->dp_clock / 10;
+			amdgpu_crtc->ss_enabled =
+				amdgpu_atombios_get_asic_ss_info(adev, &amdgpu_crtc->ss,
+								 ASIC_INTERNAL_SS_ON_DP,
+								 dp_clock);
+			break;
+		case ATOM_ENCODER_MODE_LVDS:
+			amdgpu_crtc->ss_enabled =
+				amdgpu_atombios_get_asic_ss_info(adev,
+								 &amdgpu_crtc->ss,
+								 dig->lcd_ss_id,
+								 mode->clock / 10);
+			break;
+		case ATOM_ENCODER_MODE_DVI:
+			amdgpu_crtc->ss_enabled =
+				amdgpu_atombios_get_asic_ss_info(adev,
+								 &amdgpu_crtc->ss,
+								 ASIC_INTERNAL_SS_ON_TMDS,
+								 mode->clock / 10);
+			break;
+		case ATOM_ENCODER_MODE_HDMI:
+			amdgpu_crtc->ss_enabled =
+				amdgpu_atombios_get_asic_ss_info(adev,
+								 &amdgpu_crtc->ss,
+								 ASIC_INTERNAL_SS_ON_HDMI,
+								 mode->clock / 10);
+			break;
+		default:
+			break;
+		}
+	}
+
+	/* adjust pixel clock as needed */
+	amdgpu_crtc->adjusted_clock = amdgpu_atombios_crtc_adjust_pll(crtc, mode);
+
+	return 0;
+}
+
+void amdgpu_atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder =
+		to_amdgpu_encoder(amdgpu_crtc->encoder);
+	u32 pll_clock = mode->clock;
+	u32 clock = mode->clock;
+	u32 ref_div = 0, fb_div = 0, frac_fb_div = 0, post_div = 0;
+	struct amdgpu_pll *pll;
+	int encoder_mode = amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder);
+
+	/* pass the actual clock to amdgpu_atombios_crtc_program_pll for HDMI */
+	if ((encoder_mode == ATOM_ENCODER_MODE_HDMI) &&
+	    (amdgpu_crtc->bpc > 8))
+		clock = amdgpu_crtc->adjusted_clock;
+
+	switch (amdgpu_crtc->pll_id) {
+	case ATOM_PPLL1:
+		pll = &adev->clock.ppll[0];
+		break;
+	case ATOM_PPLL2:
+		pll = &adev->clock.ppll[1];
+		break;
+	case ATOM_PPLL0:
+	case ATOM_PPLL_INVALID:
+	default:
+		pll = &adev->clock.ppll[2];
+		break;
+	}
+
+	/* update pll params */
+	pll->flags = amdgpu_crtc->pll_flags;
+	pll->reference_div = amdgpu_crtc->pll_reference_div;
+	pll->post_div = amdgpu_crtc->pll_post_div;
+
+	amdgpu_pll_compute(pll, amdgpu_crtc->adjusted_clock, &pll_clock,
+			    &fb_div, &frac_fb_div, &ref_div, &post_div);
+
+	amdgpu_atombios_crtc_program_ss(adev, ATOM_DISABLE, amdgpu_crtc->pll_id,
+				 amdgpu_crtc->crtc_id, &amdgpu_crtc->ss);
+
+	amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
+				  encoder_mode, amdgpu_encoder->encoder_id, clock,
+				  ref_div, fb_div, frac_fb_div, post_div,
+				  amdgpu_crtc->bpc, amdgpu_crtc->ss_enabled, &amdgpu_crtc->ss);
+
+	if (amdgpu_crtc->ss_enabled) {
+		/* calculate ss amount and step size */
+		u32 step_size;
+		u32 amount = (((fb_div * 10) + frac_fb_div) *
+			      (u32)amdgpu_crtc->ss.percentage) /
+			(100 * (u32)amdgpu_crtc->ss.percentage_divider);
+		amdgpu_crtc->ss.amount = (amount / 10) & ATOM_PPLL_SS_AMOUNT_V2_FBDIV_MASK;
+		amdgpu_crtc->ss.amount |= ((amount - (amount / 10)) << ATOM_PPLL_SS_AMOUNT_V2_NFRAC_SHIFT) &
+			ATOM_PPLL_SS_AMOUNT_V2_NFRAC_MASK;
+		if (amdgpu_crtc->ss.type & ATOM_PPLL_SS_TYPE_V2_CENTRE_SPREAD)
+			step_size = (4 * amount * ref_div * ((u32)amdgpu_crtc->ss.rate * 2048)) /
+				(125 * 25 * pll->reference_freq / 100);
+		else
+			step_size = (2 * amount * ref_div * ((u32)amdgpu_crtc->ss.rate * 2048)) /
+				(125 * 25 * pll->reference_freq / 100);
+		amdgpu_crtc->ss.step = step_size;
+
+		amdgpu_atombios_crtc_program_ss(adev, ATOM_ENABLE, amdgpu_crtc->pll_id,
+					 amdgpu_crtc->crtc_id, &amdgpu_crtc->ss);
+	}
+}
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios_dp.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios_dp.c
new file mode 100644
index 00000000000..f81068ba4cc
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios_dp.c
@@ -0,0 +1,762 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+#include "atom.h"
+#include "atom-bits.h"
+#include "atombios_encoders.h"
+#include "atombios_dp.h"
+#include "amdgpu_connectors.h"
+#include "amdgpu_atombios.h"
+#include <drm/drm_dp_helper.h>
+
+/* move these to drm_dp_helper.c/h */
+#define DP_LINK_CONFIGURATION_SIZE 9
+#define DP_DPCD_SIZE DP_RECEIVER_CAP_SIZE
+
+static char *voltage_names[] = {
+	"0.4V", "0.6V", "0.8V", "1.2V"
+};
+static char *pre_emph_names[] = {
+	"0dB", "3.5dB", "6dB", "9.5dB"
+};
+
+/***** amdgpu AUX functions *****/
+
+union aux_channel_transaction {
+	PROCESS_AUX_CHANNEL_TRANSACTION_PS_ALLOCATION v1;
+	PROCESS_AUX_CHANNEL_TRANSACTION_PARAMETERS_V2 v2;
+};
+
+static int amdgpu_atombios_dp_process_aux_ch(struct amdgpu_i2c_chan *chan,
+				      u8 *send, int send_bytes,
+				      u8 *recv, int recv_size,
+				      u8 delay, u8 *ack)
+{
+	struct drm_device *dev = chan->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	union aux_channel_transaction args;
+	int index = GetIndexIntoMasterTable(COMMAND, ProcessAuxChannelTransaction);
+	unsigned char *base;
+	int recv_bytes;
+	int r = 0;
+
+	memset(&args, 0, sizeof(args));
+
+	mutex_lock(&chan->mutex);
+
+	base = (unsigned char *)(adev->mode_info.atom_context->scratch + 1);
+
+	amdgpu_atombios_copy_swap(base, send, send_bytes, true);
+
+	args.v2.lpAuxRequest = cpu_to_le16((u16)(0 + 4));
+	args.v2.lpDataOut = cpu_to_le16((u16)(16 + 4));
+	args.v2.ucDataOutLen = 0;
+	args.v2.ucChannelID = chan->rec.i2c_id;
+	args.v2.ucDelay = delay / 10;
+	args.v2.ucHPD_ID = chan->rec.hpd;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+	*ack = args.v2.ucReplyStatus;
+
+	/* timeout */
+	if (args.v2.ucReplyStatus == 1) {
+		r = -ETIMEDOUT;
+		goto done;
+	}
+
+	/* flags not zero */
+	if (args.v2.ucReplyStatus == 2) {
+		DRM_DEBUG_KMS("dp_aux_ch flags not zero\n");
+		r = -EIO;
+		goto done;
+	}
+
+	/* error */
+	if (args.v2.ucReplyStatus == 3) {
+		DRM_DEBUG_KMS("dp_aux_ch error\n");
+		r = -EIO;
+		goto done;
+	}
+
+	recv_bytes = args.v1.ucDataOutLen;
+	if (recv_bytes > recv_size)
+		recv_bytes = recv_size;
+
+	if (recv && recv_size)
+		amdgpu_atombios_copy_swap(recv, base + 16, recv_bytes, false);
+
+	r = recv_bytes;
+done:
+	mutex_unlock(&chan->mutex);
+
+	return r;
+}
+
+#define BARE_ADDRESS_SIZE 3
+#define HEADER_SIZE (BARE_ADDRESS_SIZE + 1)
+
+static ssize_t
+amdgpu_atombios_dp_aux_transfer(struct drm_dp_aux *aux, struct drm_dp_aux_msg *msg)
+{
+	struct amdgpu_i2c_chan *chan =
+		container_of(aux, struct amdgpu_i2c_chan, aux);
+	int ret;
+	u8 tx_buf[20];
+	size_t tx_size;
+	u8 ack, delay = 0;
+
+	if (WARN_ON(msg->size > 16))
+		return -E2BIG;
+
+	tx_buf[0] = msg->address & 0xff;
+	tx_buf[1] = msg->address >> 8;
+	tx_buf[2] = (msg->request << 4) |
+		((msg->address >> 16) & 0xf);
+	tx_buf[3] = msg->size ? (msg->size - 1) : 0;
+
+	switch (msg->request & ~DP_AUX_I2C_MOT) {
+	case DP_AUX_NATIVE_WRITE:
+	case DP_AUX_I2C_WRITE:
+		/* tx_size needs to be 4 even for bare address packets since the atom
+		 * table needs the info in tx_buf[3].
+		 */
+		tx_size = HEADER_SIZE + msg->size;
+		if (msg->size == 0)
+			tx_buf[3] |= BARE_ADDRESS_SIZE << 4;
+		else
+			tx_buf[3] |= tx_size << 4;
+		memcpy(tx_buf + HEADER_SIZE, msg->buffer, msg->size);
+		ret = amdgpu_atombios_dp_process_aux_ch(chan,
+						 tx_buf, tx_size, NULL, 0, delay, &ack);
+		if (ret >= 0)
+			/* Return payload size. */
+			ret = msg->size;
+		break;
+	case DP_AUX_NATIVE_READ:
+	case DP_AUX_I2C_READ:
+		/* tx_size needs to be 4 even for bare address packets since the atom
+		 * table needs the info in tx_buf[3].
+		 */
+		tx_size = HEADER_SIZE;
+		if (msg->size == 0)
+			tx_buf[3] |= BARE_ADDRESS_SIZE << 4;
+		else
+			tx_buf[3] |= tx_size << 4;
+		ret = amdgpu_atombios_dp_process_aux_ch(chan,
+						 tx_buf, tx_size, msg->buffer, msg->size, delay, &ack);
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
+
+	if (ret >= 0)
+		msg->reply = ack >> 4;
+
+	return ret;
+}
+
+void amdgpu_atombios_dp_aux_init(struct amdgpu_connector *amdgpu_connector)
+{
+	int ret;
+
+	amdgpu_connector->ddc_bus->rec.hpd = amdgpu_connector->hpd.hpd;
+	amdgpu_connector->ddc_bus->aux.dev = amdgpu_connector->base.kdev;
+	amdgpu_connector->ddc_bus->aux.transfer = amdgpu_atombios_dp_aux_transfer;
+	ret = drm_dp_aux_register(&amdgpu_connector->ddc_bus->aux);
+	if (!ret)
+		amdgpu_connector->ddc_bus->has_aux = true;
+
+	WARN(ret, "drm_dp_aux_register_i2c_bus() failed with error %d\n", ret);
+}
+
+/***** general DP utility functions *****/
+
+#define DP_VOLTAGE_MAX         DP_TRAIN_VOLTAGE_SWING_LEVEL_3
+#define DP_PRE_EMPHASIS_MAX    DP_TRAIN_PRE_EMPH_LEVEL_3
+
+static void amdgpu_atombios_dp_get_adjust_train(const u8 link_status[DP_LINK_STATUS_SIZE],
+						int lane_count,
+						u8 train_set[4])
+{
+	u8 v = 0;
+	u8 p = 0;
+	int lane;
+
+	for (lane = 0; lane < lane_count; lane++) {
+		u8 this_v = drm_dp_get_adjust_request_voltage(link_status, lane);
+		u8 this_p = drm_dp_get_adjust_request_pre_emphasis(link_status, lane);
+
+		DRM_DEBUG_KMS("requested signal parameters: lane %d voltage %s pre_emph %s\n",
+			  lane,
+			  voltage_names[this_v >> DP_TRAIN_VOLTAGE_SWING_SHIFT],
+			  pre_emph_names[this_p >> DP_TRAIN_PRE_EMPHASIS_SHIFT]);
+
+		if (this_v > v)
+			v = this_v;
+		if (this_p > p)
+			p = this_p;
+	}
+
+	if (v >= DP_VOLTAGE_MAX)
+		v |= DP_TRAIN_MAX_SWING_REACHED;
+
+	if (p >= DP_PRE_EMPHASIS_MAX)
+		p |= DP_TRAIN_MAX_PRE_EMPHASIS_REACHED;
+
+	DRM_DEBUG_KMS("using signal parameters: voltage %s pre_emph %s\n",
+		  voltage_names[(v & DP_TRAIN_VOLTAGE_SWING_MASK) >> DP_TRAIN_VOLTAGE_SWING_SHIFT],
+		  pre_emph_names[(p & DP_TRAIN_PRE_EMPHASIS_MASK) >> DP_TRAIN_PRE_EMPHASIS_SHIFT]);
+
+	for (lane = 0; lane < 4; lane++)
+		train_set[lane] = v | p;
+}
+
+/* convert bits per color to bits per pixel */
+/* get bpc from the EDID */
+static unsigned amdgpu_atombios_dp_convert_bpc_to_bpp(int bpc)
+{
+	if (bpc == 0)
+		return 24;
+	else
+		return bpc * 3;
+}
+
+/***** amdgpu specific DP functions *****/
+
+static int amdgpu_atombios_dp_get_dp_link_config(struct drm_connector *connector,
+						 const u8 dpcd[DP_DPCD_SIZE],
+						 unsigned pix_clock,
+						 unsigned *dp_lanes, unsigned *dp_rate)
+{
+	unsigned bpp =
+		amdgpu_atombios_dp_convert_bpc_to_bpp(amdgpu_connector_get_monitor_bpc(connector));
+	static const unsigned link_rates[3] = { 162000, 270000, 540000 };
+	unsigned max_link_rate = drm_dp_max_link_rate(dpcd);
+	unsigned max_lane_num = drm_dp_max_lane_count(dpcd);
+	unsigned lane_num, i, max_pix_clock;
+
+	if (amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector) ==
+	    ENCODER_OBJECT_ID_NUTMEG) {
+		for (lane_num = 1; lane_num <= max_lane_num; lane_num <<= 1) {
+			max_pix_clock = (lane_num * 270000 * 8) / bpp;
+			if (max_pix_clock >= pix_clock) {
+				*dp_lanes = lane_num;
+				*dp_rate = 270000;
+				return 0;
+			}
+		}
+	} else {
+		for (i = 0; i < ARRAY_SIZE(link_rates) && link_rates[i] <= max_link_rate; i++) {
+			for (lane_num = 1; lane_num <= max_lane_num; lane_num <<= 1) {
+				max_pix_clock = (lane_num * link_rates[i] * 8) / bpp;
+				if (max_pix_clock >= pix_clock) {
+					*dp_lanes = lane_num;
+					*dp_rate = link_rates[i];
+					return 0;
+				}
+			}
+		}
+	}
+
+	return -EINVAL;
+}
+
+static u8 amdgpu_atombios_dp_encoder_service(struct amdgpu_device *adev,
+				      int action, int dp_clock,
+				      u8 ucconfig, u8 lane_num)
+{
+	DP_ENCODER_SERVICE_PARAMETERS args;
+	int index = GetIndexIntoMasterTable(COMMAND, DPEncoderService);
+
+	memset(&args, 0, sizeof(args));
+	args.ucLinkClock = dp_clock / 10;
+	args.ucConfig = ucconfig;
+	args.ucAction = action;
+	args.ucLaneNum = lane_num;
+	args.ucStatus = 0;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	return args.ucStatus;
+}
+
+u8 amdgpu_atombios_dp_get_sinktype(struct amdgpu_connector *amdgpu_connector)
+{
+	struct drm_device *dev = amdgpu_connector->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	return amdgpu_atombios_dp_encoder_service(adev, ATOM_DP_ACTION_GET_SINK_TYPE, 0,
+					   amdgpu_connector->ddc_bus->rec.i2c_id, 0);
+}
+
+static void amdgpu_atombios_dp_probe_oui(struct amdgpu_connector *amdgpu_connector)
+{
+	struct amdgpu_connector_atom_dig *dig_connector = amdgpu_connector->con_priv;
+	u8 buf[3];
+
+	if (!(dig_connector->dpcd[DP_DOWN_STREAM_PORT_COUNT] & DP_OUI_SUPPORT))
+		return;
+
+	if (drm_dp_dpcd_read(&amdgpu_connector->ddc_bus->aux, DP_SINK_OUI, buf, 3) == 3)
+		DRM_DEBUG_KMS("Sink OUI: %02hx%02hx%02hx\n",
+			      buf[0], buf[1], buf[2]);
+
+	if (drm_dp_dpcd_read(&amdgpu_connector->ddc_bus->aux, DP_BRANCH_OUI, buf, 3) == 3)
+		DRM_DEBUG_KMS("Branch OUI: %02hx%02hx%02hx\n",
+			      buf[0], buf[1], buf[2]);
+}
+
+int amdgpu_atombios_dp_get_dpcd(struct amdgpu_connector *amdgpu_connector)
+{
+	struct amdgpu_connector_atom_dig *dig_connector = amdgpu_connector->con_priv;
+	u8 msg[DP_DPCD_SIZE];
+	int ret;
+
+	ret = drm_dp_dpcd_read(&amdgpu_connector->ddc_bus->aux, DP_DPCD_REV,
+			       msg, DP_DPCD_SIZE);
+	if (ret == DP_DPCD_SIZE) {
+		memcpy(dig_connector->dpcd, msg, DP_DPCD_SIZE);
+
+		DRM_DEBUG_KMS("DPCD: %*ph\n", (int)sizeof(dig_connector->dpcd),
+			      dig_connector->dpcd);
+
+		amdgpu_atombios_dp_probe_oui(amdgpu_connector);
+
+		return 0;
+	}
+
+	dig_connector->dpcd[0] = 0;
+	return -EINVAL;
+}
+
+int amdgpu_atombios_dp_get_panel_mode(struct drm_encoder *encoder,
+			       struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *dig_connector;
+	int panel_mode = DP_PANEL_MODE_EXTERNAL_DP_MODE;
+	u16 dp_bridge = amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector);
+	u8 tmp;
+
+	if (!amdgpu_connector->con_priv)
+		return panel_mode;
+
+	dig_connector = amdgpu_connector->con_priv;
+
+	if (dp_bridge != ENCODER_OBJECT_ID_NONE) {
+		/* DP bridge chips */
+		if (drm_dp_dpcd_readb(&amdgpu_connector->ddc_bus->aux,
+				      DP_EDP_CONFIGURATION_CAP, &tmp) == 1) {
+			if (tmp & 1)
+				panel_mode = DP_PANEL_MODE_INTERNAL_DP2_MODE;
+			else if ((dp_bridge == ENCODER_OBJECT_ID_NUTMEG) ||
+				 (dp_bridge == ENCODER_OBJECT_ID_TRAVIS))
+				panel_mode = DP_PANEL_MODE_INTERNAL_DP1_MODE;
+			else
+				panel_mode = DP_PANEL_MODE_EXTERNAL_DP_MODE;
+		}
+	} else if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) {
+		/* eDP */
+		if (drm_dp_dpcd_readb(&amdgpu_connector->ddc_bus->aux,
+				      DP_EDP_CONFIGURATION_CAP, &tmp) == 1) {
+			if (tmp & 1)
+				panel_mode = DP_PANEL_MODE_INTERNAL_DP2_MODE;
+		}
+	}
+
+	return panel_mode;
+}
+
+void amdgpu_atombios_dp_set_link_config(struct drm_connector *connector,
+				 const struct drm_display_mode *mode)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *dig_connector;
+	int ret;
+
+	if (!amdgpu_connector->con_priv)
+		return;
+	dig_connector = amdgpu_connector->con_priv;
+
+	if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) ||
+	    (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) {
+		ret = amdgpu_atombios_dp_get_dp_link_config(connector, dig_connector->dpcd,
+							    mode->clock,
+							    &dig_connector->dp_lane_count,
+							    &dig_connector->dp_clock);
+		if (ret) {
+			dig_connector->dp_clock = 0;
+			dig_connector->dp_lane_count = 0;
+		}
+	}
+}
+
+int amdgpu_atombios_dp_mode_valid_helper(struct drm_connector *connector,
+				  struct drm_display_mode *mode)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *dig_connector;
+	unsigned dp_lanes, dp_clock;
+	int ret;
+
+	if (!amdgpu_connector->con_priv)
+		return MODE_CLOCK_HIGH;
+	dig_connector = amdgpu_connector->con_priv;
+
+	ret = amdgpu_atombios_dp_get_dp_link_config(connector, dig_connector->dpcd,
+						    mode->clock, &dp_lanes, &dp_clock);
+	if (ret)
+		return MODE_CLOCK_HIGH;
+
+	if ((dp_clock == 540000) &&
+	    (!amdgpu_connector_is_dp12_capable(connector)))
+		return MODE_CLOCK_HIGH;
+
+	return MODE_OK;
+}
+
+bool amdgpu_atombios_dp_needs_link_train(struct amdgpu_connector *amdgpu_connector)
+{
+	u8 link_status[DP_LINK_STATUS_SIZE];
+	struct amdgpu_connector_atom_dig *dig = amdgpu_connector->con_priv;
+
+	if (drm_dp_dpcd_read_link_status(&amdgpu_connector->ddc_bus->aux, link_status)
+	    <= 0)
+		return false;
+	if (drm_dp_channel_eq_ok(link_status, dig->dp_lane_count))
+		return false;
+	return true;
+}
+
+void amdgpu_atombios_dp_set_rx_power_state(struct drm_connector *connector,
+				    u8 power_state)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *dig_connector;
+
+	if (!amdgpu_connector->con_priv)
+		return;
+
+	dig_connector = amdgpu_connector->con_priv;
+
+	/* power up/down the sink */
+	if (dig_connector->dpcd[0] >= 0x11) {
+		drm_dp_dpcd_writeb(&amdgpu_connector->ddc_bus->aux,
+				   DP_SET_POWER, power_state);
+		usleep_range(1000, 2000);
+	}
+}
+
+struct amdgpu_atombios_dp_link_train_info {
+	struct amdgpu_device *adev;
+	struct drm_encoder *encoder;
+	struct drm_connector *connector;
+	int dp_clock;
+	int dp_lane_count;
+	bool tp3_supported;
+	u8 dpcd[DP_RECEIVER_CAP_SIZE];
+	u8 train_set[4];
+	u8 link_status[DP_LINK_STATUS_SIZE];
+	u8 tries;
+	struct drm_dp_aux *aux;
+};
+
+static void
+amdgpu_atombios_dp_update_vs_emph(struct amdgpu_atombios_dp_link_train_info *dp_info)
+{
+	/* set the initial vs/emph on the source */
+	amdgpu_atombios_encoder_setup_dig_transmitter(dp_info->encoder,
+					       ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH,
+					       0, dp_info->train_set[0]); /* sets all lanes at once */
+
+	/* set the vs/emph on the sink */
+	drm_dp_dpcd_write(dp_info->aux, DP_TRAINING_LANE0_SET,
+			  dp_info->train_set, dp_info->dp_lane_count);
+}
+
+static void
+amdgpu_atombios_dp_set_tp(struct amdgpu_atombios_dp_link_train_info *dp_info, int tp)
+{
+	int rtp = 0;
+
+	/* set training pattern on the source */
+	switch (tp) {
+	case DP_TRAINING_PATTERN_1:
+		rtp = ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN1;
+		break;
+	case DP_TRAINING_PATTERN_2:
+		rtp = ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN2;
+		break;
+	case DP_TRAINING_PATTERN_3:
+		rtp = ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN3;
+			break;
+	}
+	amdgpu_atombios_encoder_setup_dig_encoder(dp_info->encoder, rtp, 0);
+
+	/* enable training pattern on the sink */
+	drm_dp_dpcd_writeb(dp_info->aux, DP_TRAINING_PATTERN_SET, tp);
+}
+
+static int
+amdgpu_atombios_dp_link_train_init(struct amdgpu_atombios_dp_link_train_info *dp_info)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(dp_info->encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u8 tmp;
+
+	/* power up the sink */
+	amdgpu_atombios_dp_set_rx_power_state(dp_info->connector, DP_SET_POWER_D0);
+
+	/* possibly enable downspread on the sink */
+	if (dp_info->dpcd[3] & 0x1)
+		drm_dp_dpcd_writeb(dp_info->aux,
+				   DP_DOWNSPREAD_CTRL, DP_SPREAD_AMP_0_5);
+	else
+		drm_dp_dpcd_writeb(dp_info->aux,
+				   DP_DOWNSPREAD_CTRL, 0);
+
+	if (dig->panel_mode == DP_PANEL_MODE_INTERNAL_DP2_MODE)
+		drm_dp_dpcd_writeb(dp_info->aux, DP_EDP_CONFIGURATION_SET, 1);
+
+	/* set the lane count on the sink */
+	tmp = dp_info->dp_lane_count;
+	if (drm_dp_enhanced_frame_cap(dp_info->dpcd))
+		tmp |= DP_LANE_COUNT_ENHANCED_FRAME_EN;
+	drm_dp_dpcd_writeb(dp_info->aux, DP_LANE_COUNT_SET, tmp);
+
+	/* set the link rate on the sink */
+	tmp = drm_dp_link_rate_to_bw_code(dp_info->dp_clock);
+	drm_dp_dpcd_writeb(dp_info->aux, DP_LINK_BW_SET, tmp);
+
+	/* start training on the source */
+	amdgpu_atombios_encoder_setup_dig_encoder(dp_info->encoder,
+					   ATOM_ENCODER_CMD_DP_LINK_TRAINING_START, 0);
+
+	/* disable the training pattern on the sink */
+	drm_dp_dpcd_writeb(dp_info->aux,
+			   DP_TRAINING_PATTERN_SET,
+			   DP_TRAINING_PATTERN_DISABLE);
+
+	return 0;
+}
+
+static int
+amdgpu_atombios_dp_link_train_finish(struct amdgpu_atombios_dp_link_train_info *dp_info)
+{
+	udelay(400);
+
+	/* disable the training pattern on the sink */
+	drm_dp_dpcd_writeb(dp_info->aux,
+			   DP_TRAINING_PATTERN_SET,
+			   DP_TRAINING_PATTERN_DISABLE);
+
+	/* disable the training pattern on the source */
+	amdgpu_atombios_encoder_setup_dig_encoder(dp_info->encoder,
+					   ATOM_ENCODER_CMD_DP_LINK_TRAINING_COMPLETE, 0);
+
+	return 0;
+}
+
+static int
+amdgpu_atombios_dp_link_train_cr(struct amdgpu_atombios_dp_link_train_info *dp_info)
+{
+	bool clock_recovery;
+	u8 voltage;
+	int i;
+
+	amdgpu_atombios_dp_set_tp(dp_info, DP_TRAINING_PATTERN_1);
+	memset(dp_info->train_set, 0, 4);
+	amdgpu_atombios_dp_update_vs_emph(dp_info);
+
+	udelay(400);
+
+	/* clock recovery loop */
+	clock_recovery = false;
+	dp_info->tries = 0;
+	voltage = 0xff;
+	while (1) {
+		drm_dp_link_train_clock_recovery_delay(dp_info->dpcd);
+
+		if (drm_dp_dpcd_read_link_status(dp_info->aux,
+						 dp_info->link_status) <= 0) {
+			DRM_ERROR("displayport link status failed\n");
+			break;
+		}
+
+		if (drm_dp_clock_recovery_ok(dp_info->link_status, dp_info->dp_lane_count)) {
+			clock_recovery = true;
+			break;
+		}
+
+		for (i = 0; i < dp_info->dp_lane_count; i++) {
+			if ((dp_info->train_set[i] & DP_TRAIN_MAX_SWING_REACHED) == 0)
+				break;
+		}
+		if (i == dp_info->dp_lane_count) {
+			DRM_ERROR("clock recovery reached max voltage\n");
+			break;
+		}
+
+		if ((dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK) == voltage) {
+			++dp_info->tries;
+			if (dp_info->tries == 5) {
+				DRM_ERROR("clock recovery tried 5 times\n");
+				break;
+			}
+		} else
+			dp_info->tries = 0;
+
+		voltage = dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK;
+
+		/* Compute new train_set as requested by sink */
+		amdgpu_atombios_dp_get_adjust_train(dp_info->link_status, dp_info->dp_lane_count,
+					     dp_info->train_set);
+
+		amdgpu_atombios_dp_update_vs_emph(dp_info);
+	}
+	if (!clock_recovery) {
+		DRM_ERROR("clock recovery failed\n");
+		return -1;
+	} else {
+		DRM_DEBUG_KMS("clock recovery at voltage %d pre-emphasis %d\n",
+			  dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK,
+			  (dp_info->train_set[0] & DP_TRAIN_PRE_EMPHASIS_MASK) >>
+			  DP_TRAIN_PRE_EMPHASIS_SHIFT);
+		return 0;
+	}
+}
+
+static int
+amdgpu_atombios_dp_link_train_ce(struct amdgpu_atombios_dp_link_train_info *dp_info)
+{
+	bool channel_eq;
+
+	if (dp_info->tp3_supported)
+		amdgpu_atombios_dp_set_tp(dp_info, DP_TRAINING_PATTERN_3);
+	else
+		amdgpu_atombios_dp_set_tp(dp_info, DP_TRAINING_PATTERN_2);
+
+	/* channel equalization loop */
+	dp_info->tries = 0;
+	channel_eq = false;
+	while (1) {
+		drm_dp_link_train_channel_eq_delay(dp_info->dpcd);
+
+		if (drm_dp_dpcd_read_link_status(dp_info->aux,
+						 dp_info->link_status) <= 0) {
+			DRM_ERROR("displayport link status failed\n");
+			break;
+		}
+
+		if (drm_dp_channel_eq_ok(dp_info->link_status, dp_info->dp_lane_count)) {
+			channel_eq = true;
+			break;
+		}
+
+		/* Try 5 times */
+		if (dp_info->tries > 5) {
+			DRM_ERROR("channel eq failed: 5 tries\n");
+			break;
+		}
+
+		/* Compute new train_set as requested by sink */
+		amdgpu_atombios_dp_get_adjust_train(dp_info->link_status, dp_info->dp_lane_count,
+					     dp_info->train_set);
+
+		amdgpu_atombios_dp_update_vs_emph(dp_info);
+		dp_info->tries++;
+	}
+
+	if (!channel_eq) {
+		DRM_ERROR("channel eq failed\n");
+		return -1;
+	} else {
+		DRM_DEBUG_KMS("channel eq at voltage %d pre-emphasis %d\n",
+			  dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK,
+			  (dp_info->train_set[0] & DP_TRAIN_PRE_EMPHASIS_MASK)
+			  >> DP_TRAIN_PRE_EMPHASIS_SHIFT);
+		return 0;
+	}
+}
+
+void amdgpu_atombios_dp_link_train(struct drm_encoder *encoder,
+			    struct drm_connector *connector)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig;
+	struct amdgpu_connector *amdgpu_connector;
+	struct amdgpu_connector_atom_dig *dig_connector;
+	struct amdgpu_atombios_dp_link_train_info dp_info;
+	u8 tmp;
+
+	if (!amdgpu_encoder->enc_priv)
+		return;
+	dig = amdgpu_encoder->enc_priv;
+
+	amdgpu_connector = to_amdgpu_connector(connector);
+	if (!amdgpu_connector->con_priv)
+		return;
+	dig_connector = amdgpu_connector->con_priv;
+
+	if ((dig_connector->dp_sink_type != CONNECTOR_OBJECT_ID_DISPLAYPORT) &&
+	    (dig_connector->dp_sink_type != CONNECTOR_OBJECT_ID_eDP))
+		return;
+
+	if (drm_dp_dpcd_readb(&amdgpu_connector->ddc_bus->aux, DP_MAX_LANE_COUNT, &tmp)
+	    == 1) {
+		if (tmp & DP_TPS3_SUPPORTED)
+			dp_info.tp3_supported = true;
+		else
+			dp_info.tp3_supported = false;
+	} else {
+		dp_info.tp3_supported = false;
+	}
+
+	memcpy(dp_info.dpcd, dig_connector->dpcd, DP_RECEIVER_CAP_SIZE);
+	dp_info.adev = adev;
+	dp_info.encoder = encoder;
+	dp_info.connector = connector;
+	dp_info.dp_lane_count = dig_connector->dp_lane_count;
+	dp_info.dp_clock = dig_connector->dp_clock;
+	dp_info.aux = &amdgpu_connector->ddc_bus->aux;
+
+	if (amdgpu_atombios_dp_link_train_init(&dp_info))
+		goto done;
+	if (amdgpu_atombios_dp_link_train_cr(&dp_info))
+		goto done;
+	if (amdgpu_atombios_dp_link_train_ce(&dp_info))
+		goto done;
+done:
+	if (amdgpu_atombios_dp_link_train_finish(&dp_info))
+		return;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios_encoders.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios_encoders.c
new file mode 100644
index 00000000000..d43b9faaef6
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios_encoders.c
@@ -0,0 +1,2160 @@
+/*
+ * Copyright 2007-11 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <drm/drmP.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_connectors.h"
+#include "atom.h"
+#include "atombios_encoders.h"
+#include "atombios_dp.h"
+#include <linux/backlight.h>
+#include "bif/bif_4_1_d.h"
+
+u8
+amdgpu_atombios_encoder_get_backlight_level_from_reg(struct amdgpu_device *adev)
+{
+	u8 backlight_level;
+	u32 bios_2_scratch;
+
+	bios_2_scratch = RREG32(mmBIOS_SCRATCH_2);
+
+	backlight_level = ((bios_2_scratch & ATOM_S2_CURRENT_BL_LEVEL_MASK) >>
+			   ATOM_S2_CURRENT_BL_LEVEL_SHIFT);
+
+	return backlight_level;
+}
+
+void
+amdgpu_atombios_encoder_set_backlight_level_to_reg(struct amdgpu_device *adev,
+					    u8 backlight_level)
+{
+	u32 bios_2_scratch;
+
+	bios_2_scratch = RREG32(mmBIOS_SCRATCH_2);
+
+	bios_2_scratch &= ~ATOM_S2_CURRENT_BL_LEVEL_MASK;
+	bios_2_scratch |= ((backlight_level << ATOM_S2_CURRENT_BL_LEVEL_SHIFT) &
+			   ATOM_S2_CURRENT_BL_LEVEL_MASK);
+
+	WREG32(mmBIOS_SCRATCH_2, bios_2_scratch);
+}
+
+u8
+amdgpu_atombios_encoder_get_backlight_level(struct amdgpu_encoder *amdgpu_encoder)
+{
+	struct drm_device *dev = amdgpu_encoder->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU))
+		return 0;
+
+	return amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+}
+
+void
+amdgpu_atombios_encoder_set_backlight_level(struct amdgpu_encoder *amdgpu_encoder,
+				     u8 level)
+{
+	struct drm_encoder *encoder = &amdgpu_encoder->base;
+	struct drm_device *dev = amdgpu_encoder->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder_atom_dig *dig;
+
+	if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU))
+		return;
+
+	if ((amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) &&
+	    amdgpu_encoder->enc_priv) {
+		dig = amdgpu_encoder->enc_priv;
+		dig->backlight_level = level;
+		amdgpu_atombios_encoder_set_backlight_level_to_reg(adev, dig->backlight_level);
+
+		switch (amdgpu_encoder->encoder_id) {
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+			if (dig->backlight_level == 0)
+				amdgpu_atombios_encoder_setup_dig_transmitter(encoder,
+								       ATOM_TRANSMITTER_ACTION_LCD_BLOFF, 0, 0);
+			else {
+				amdgpu_atombios_encoder_setup_dig_transmitter(encoder,
+								       ATOM_TRANSMITTER_ACTION_BL_BRIGHTNESS_CONTROL, 0, 0);
+				amdgpu_atombios_encoder_setup_dig_transmitter(encoder,
+								       ATOM_TRANSMITTER_ACTION_LCD_BLON, 0, 0);
+			}
+			break;
+		default:
+			break;
+		}
+	}
+}
+
+#if defined(CONFIG_BACKLIGHT_CLASS_DEVICE) || defined(CONFIG_BACKLIGHT_CLASS_DEVICE_MODULE)
+
+static u8 amdgpu_atombios_encoder_backlight_level(struct backlight_device *bd)
+{
+	u8 level;
+
+	/* Convert brightness to hardware level */
+	if (bd->props.brightness < 0)
+		level = 0;
+	else if (bd->props.brightness > AMDGPU_MAX_BL_LEVEL)
+		level = AMDGPU_MAX_BL_LEVEL;
+	else
+		level = bd->props.brightness;
+
+	return level;
+}
+
+static int amdgpu_atombios_encoder_update_backlight_status(struct backlight_device *bd)
+{
+	struct amdgpu_backlight_privdata *pdata = bl_get_data(bd);
+	struct amdgpu_encoder *amdgpu_encoder = pdata->encoder;
+
+	amdgpu_atombios_encoder_set_backlight_level(amdgpu_encoder,
+					     amdgpu_atombios_encoder_backlight_level(bd));
+
+	return 0;
+}
+
+static int
+amdgpu_atombios_encoder_get_backlight_brightness(struct backlight_device *bd)
+{
+	struct amdgpu_backlight_privdata *pdata = bl_get_data(bd);
+	struct amdgpu_encoder *amdgpu_encoder = pdata->encoder;
+	struct drm_device *dev = amdgpu_encoder->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	return amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+}
+
+static const struct backlight_ops amdgpu_atombios_encoder_backlight_ops = {
+	.get_brightness = amdgpu_atombios_encoder_get_backlight_brightness,
+	.update_status	= amdgpu_atombios_encoder_update_backlight_status,
+};
+
+void amdgpu_atombios_encoder_init_backlight(struct amdgpu_encoder *amdgpu_encoder,
+				     struct drm_connector *drm_connector)
+{
+	struct drm_device *dev = amdgpu_encoder->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct backlight_device *bd;
+	struct backlight_properties props;
+	struct amdgpu_backlight_privdata *pdata;
+	struct amdgpu_encoder_atom_dig *dig;
+	u8 backlight_level;
+	char bl_name[16];
+
+	/* Mac laptops with multiple GPUs use the gmux driver for backlight
+	 * so don't register a backlight device
+	 */
+	if ((adev->pdev->subsystem_vendor == PCI_VENDOR_ID_APPLE) &&
+	    (adev->pdev->device == 0x6741))
+		return;
+
+	if (!amdgpu_encoder->enc_priv)
+		return;
+
+	if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU))
+		return;
+
+	pdata = kmalloc(sizeof(struct amdgpu_backlight_privdata), GFP_KERNEL);
+	if (!pdata) {
+		DRM_ERROR("Memory allocation failed\n");
+		goto error;
+	}
+
+	memset(&props, 0, sizeof(props));
+	props.max_brightness = AMDGPU_MAX_BL_LEVEL;
+	props.type = BACKLIGHT_RAW;
+#ifdef notyet
+	snprintf(bl_name, sizeof(bl_name),
+		 "amdgpu_bl%d", dev->primary->index);
+#else
+	snprintf(bl_name, sizeof(bl_name),
+		 "amdgpu_bl%d", 0);
+#endif
+	bd = backlight_device_register(bl_name, drm_connector->kdev,
+				       pdata, &amdgpu_atombios_encoder_backlight_ops, &props);
+	if (IS_ERR(bd)) {
+		DRM_ERROR("Backlight registration failed\n");
+		goto error;
+	}
+
+	pdata->encoder = amdgpu_encoder;
+
+	backlight_level = amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+
+	dig = amdgpu_encoder->enc_priv;
+	dig->bl_dev = bd;
+
+	bd->props.brightness = amdgpu_atombios_encoder_get_backlight_brightness(bd);
+	bd->props.power = FB_BLANK_UNBLANK;
+	backlight_update_status(bd);
+
+	DRM_INFO("amdgpu atom DIG backlight initialized\n");
+
+	return;
+
+error:
+	kfree(pdata);
+	return;
+}
+
+void
+amdgpu_atombios_encoder_fini_backlight(struct amdgpu_encoder *amdgpu_encoder)
+{
+	struct drm_device *dev = amdgpu_encoder->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct backlight_device *bd = NULL;
+	struct amdgpu_encoder_atom_dig *dig;
+
+	if (!amdgpu_encoder->enc_priv)
+		return;
+
+	if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU))
+		return;
+
+	dig = amdgpu_encoder->enc_priv;
+	bd = dig->bl_dev;
+	dig->bl_dev = NULL;
+
+	if (bd) {
+		struct amdgpu_legacy_backlight_privdata *pdata;
+
+		pdata = bl_get_data(bd);
+		backlight_device_unregister(bd);
+		kfree(pdata);
+
+		DRM_INFO("amdgpu atom LVDS backlight unloaded\n");
+	}
+}
+
+#else /* !CONFIG_BACKLIGHT_CLASS_DEVICE */
+
+void amdgpu_atombios_encoder_init_backlight(struct amdgpu_encoder *encoder)
+{
+}
+
+void amdgpu_atombios_encoder_fini_backlight(struct amdgpu_encoder *encoder)
+{
+}
+
+#endif
+
+bool amdgpu_atombios_encoder_is_digital(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		return true;
+	default:
+		return false;
+	}
+}
+
+bool amdgpu_atombios_encoder_mode_fixup(struct drm_encoder *encoder,
+				 const struct drm_display_mode *mode,
+				 struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	/* set the active encoder to connector routing */
+	amdgpu_encoder_set_active_device(encoder);
+	drm_mode_set_crtcinfo(adjusted_mode, 0);
+
+	/* hw bug */
+	if ((mode->flags & DRM_MODE_FLAG_INTERLACE)
+	    && (mode->crtc_vsync_start < (mode->crtc_vdisplay + 2)))
+		adjusted_mode->crtc_vsync_start = adjusted_mode->crtc_vdisplay + 2;
+
+	/* vertical FP must be at least 1 */
+	if (mode->crtc_vsync_start == mode->crtc_vdisplay)
+		adjusted_mode->crtc_vsync_start++;
+
+	/* get the native mode for scaling */
+	if (amdgpu_encoder->active_device & (ATOM_DEVICE_LCD_SUPPORT))
+		amdgpu_panel_mode_fixup(encoder, adjusted_mode);
+	else if (amdgpu_encoder->rmx_type != RMX_OFF)
+		amdgpu_panel_mode_fixup(encoder, adjusted_mode);
+
+	if ((amdgpu_encoder->active_device & (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) != ENCODER_OBJECT_ID_NONE)) {
+		struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+		amdgpu_atombios_dp_set_link_config(connector, adjusted_mode);
+	}
+
+	return true;
+}
+
+static void
+amdgpu_atombios_encoder_setup_dac(struct drm_encoder *encoder, int action)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	DAC_ENCODER_CONTROL_PS_ALLOCATION args;
+	int index = 0;
+
+	memset(&args, 0, sizeof(args));
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_DAC1:
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+		index = GetIndexIntoMasterTable(COMMAND, DAC1EncoderControl);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_DAC2:
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+		index = GetIndexIntoMasterTable(COMMAND, DAC2EncoderControl);
+		break;
+	}
+
+	args.ucAction = action;
+	args.ucDacStandard = ATOM_DAC1_PS2;
+	args.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+}
+
+static u8 amdgpu_atombios_encoder_get_bpc(struct drm_encoder *encoder)
+{
+	int bpc = 8;
+
+	if (encoder->crtc) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+		bpc = amdgpu_crtc->bpc;
+	}
+
+	switch (bpc) {
+	case 0:
+		return PANEL_BPC_UNDEFINE;
+	case 6:
+		return PANEL_6BIT_PER_COLOR;
+	case 8:
+	default:
+		return PANEL_8BIT_PER_COLOR;
+	case 10:
+		return PANEL_10BIT_PER_COLOR;
+	case 12:
+		return PANEL_12BIT_PER_COLOR;
+	case 16:
+		return PANEL_16BIT_PER_COLOR;
+	}
+}
+
+union dvo_encoder_control {
+	ENABLE_EXTERNAL_TMDS_ENCODER_PS_ALLOCATION ext_tmds;
+	DVO_ENCODER_CONTROL_PS_ALLOCATION dvo;
+	DVO_ENCODER_CONTROL_PS_ALLOCATION_V3 dvo_v3;
+	DVO_ENCODER_CONTROL_PS_ALLOCATION_V1_4 dvo_v4;
+};
+
+static void
+amdgpu_atombios_encoder_setup_dvo(struct drm_encoder *encoder, int action)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	union dvo_encoder_control args;
+	int index = GetIndexIntoMasterTable(COMMAND, DVOEncoderControl);
+	uint8_t frev, crev;
+
+	memset(&args, 0, sizeof(args));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 1:
+			/* R4xx, R5xx */
+			args.ext_tmds.sXTmdsEncoder.ucEnable = action;
+
+			if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.ext_tmds.sXTmdsEncoder.ucMisc |= PANEL_ENCODER_MISC_DUAL;
+
+			args.ext_tmds.sXTmdsEncoder.ucMisc |= ATOM_PANEL_MISC_888RGB;
+			break;
+		case 2:
+			/* RS600/690/740 */
+			args.dvo.sDVOEncoder.ucAction = action;
+			args.dvo.sDVOEncoder.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			/* DFP1, CRT1, TV1 depending on the type of port */
+			args.dvo.sDVOEncoder.ucDeviceType = ATOM_DEVICE_DFP1_INDEX;
+
+			if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.dvo.sDVOEncoder.usDevAttr.sDigAttrib.ucAttribute |= PANEL_ENCODER_MISC_DUAL;
+			break;
+		case 3:
+			/* R6xx */
+			args.dvo_v3.ucAction = action;
+			args.dvo_v3.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			args.dvo_v3.ucDVOConfig = 0; /* XXX */
+			break;
+		case 4:
+			/* DCE8 */
+			args.dvo_v4.ucAction = action;
+			args.dvo_v4.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			args.dvo_v4.ucDVOConfig = 0; /* XXX */
+			args.dvo_v4.ucBitPerColor = amdgpu_atombios_encoder_get_bpc(encoder);
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+			break;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		break;
+	}
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+	struct amdgpu_connector_atom_dig *dig_connector;
+
+	/* dp bridges are always DP */
+	if (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) != ENCODER_OBJECT_ID_NONE)
+		return ATOM_ENCODER_MODE_DP;
+
+	/* DVO is always DVO */
+	if ((amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_DVO1) ||
+	    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1))
+		return ATOM_ENCODER_MODE_DVO;
+
+	connector = amdgpu_get_connector_for_encoder(encoder);
+	/* if we don't have an active device yet, just use one of
+	 * the connectors tied to the encoder.
+	 */
+	if (!connector)
+		connector = amdgpu_get_connector_for_encoder_init(encoder);
+	amdgpu_connector = to_amdgpu_connector(connector);
+
+	switch (connector->connector_type) {
+	case DRM_MODE_CONNECTOR_DVII:
+	case DRM_MODE_CONNECTOR_HDMIB: /* HDMI-B is basically DL-DVI; analog works fine */
+		if (amdgpu_audio != 0) {
+			if (amdgpu_connector->use_digital &&
+			    (amdgpu_connector->audio == AMDGPU_AUDIO_ENABLE))
+				return ATOM_ENCODER_MODE_HDMI;
+			else if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector)) &&
+				 (amdgpu_connector->audio == AMDGPU_AUDIO_AUTO))
+				return ATOM_ENCODER_MODE_HDMI;
+			else if (amdgpu_connector->use_digital)
+				return ATOM_ENCODER_MODE_DVI;
+			else
+				return ATOM_ENCODER_MODE_CRT;
+		} else if (amdgpu_connector->use_digital) {
+			return ATOM_ENCODER_MODE_DVI;
+		} else {
+			return ATOM_ENCODER_MODE_CRT;
+		}
+		break;
+	case DRM_MODE_CONNECTOR_DVID:
+	case DRM_MODE_CONNECTOR_HDMIA:
+	default:
+		if (amdgpu_audio != 0) {
+			if (amdgpu_connector->audio == AMDGPU_AUDIO_ENABLE)
+				return ATOM_ENCODER_MODE_HDMI;
+			else if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector)) &&
+				 (amdgpu_connector->audio == AMDGPU_AUDIO_AUTO))
+				return ATOM_ENCODER_MODE_HDMI;
+			else
+				return ATOM_ENCODER_MODE_DVI;
+		} else {
+			return ATOM_ENCODER_MODE_DVI;
+		}
+		break;
+	case DRM_MODE_CONNECTOR_LVDS:
+		return ATOM_ENCODER_MODE_LVDS;
+		break;
+	case DRM_MODE_CONNECTOR_DisplayPort:
+		dig_connector = amdgpu_connector->con_priv;
+		if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) ||
+		    (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) {
+			return ATOM_ENCODER_MODE_DP;
+		} else if (amdgpu_audio != 0) {
+			if (amdgpu_connector->audio == AMDGPU_AUDIO_ENABLE)
+				return ATOM_ENCODER_MODE_HDMI;
+			else if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector)) &&
+				 (amdgpu_connector->audio == AMDGPU_AUDIO_AUTO))
+				return ATOM_ENCODER_MODE_HDMI;
+			else
+				return ATOM_ENCODER_MODE_DVI;
+		} else {
+			return ATOM_ENCODER_MODE_DVI;
+		}
+		break;
+	case DRM_MODE_CONNECTOR_eDP:
+		return ATOM_ENCODER_MODE_DP;
+	case DRM_MODE_CONNECTOR_DVIA:
+	case DRM_MODE_CONNECTOR_VGA:
+		return ATOM_ENCODER_MODE_CRT;
+		break;
+	case DRM_MODE_CONNECTOR_Composite:
+	case DRM_MODE_CONNECTOR_SVIDEO:
+	case DRM_MODE_CONNECTOR_9PinDIN:
+		/* fix me */
+		return ATOM_ENCODER_MODE_TV;
+		/*return ATOM_ENCODER_MODE_CV;*/
+		break;
+	}
+}
+
+/*
+ * DIG Encoder/Transmitter Setup
+ *
+ * DCE 6.0
+ * - 3 DIG transmitter blocks UNIPHY0/1/2 (links A and B).
+ * Supports up to 6 digital outputs
+ * - 6 DIG encoder blocks.
+ * - DIG to PHY mapping is hardcoded
+ * DIG1 drives UNIPHY0 link A, A+B
+ * DIG2 drives UNIPHY0 link B
+ * DIG3 drives UNIPHY1 link A, A+B
+ * DIG4 drives UNIPHY1 link B
+ * DIG5 drives UNIPHY2 link A, A+B
+ * DIG6 drives UNIPHY2 link B
+ *
+ * Routing
+ * crtc -> dig encoder -> UNIPHY/LVTMA (1 or 2 links)
+ * Examples:
+ * crtc0 -> dig2 -> LVTMA   links A+B -> TMDS/HDMI
+ * crtc1 -> dig1 -> UNIPHY0 link  B   -> DP
+ * crtc0 -> dig1 -> UNIPHY2 link  A   -> LVDS
+ * crtc1 -> dig2 -> UNIPHY1 link  B+A -> TMDS/HDMI
+ */
+
+union dig_encoder_control {
+	DIG_ENCODER_CONTROL_PS_ALLOCATION v1;
+	DIG_ENCODER_CONTROL_PARAMETERS_V2 v2;
+	DIG_ENCODER_CONTROL_PARAMETERS_V3 v3;
+	DIG_ENCODER_CONTROL_PARAMETERS_V4 v4;
+	DIG_ENCODER_CONTROL_PARAMETERS_V5 v5;
+};
+
+void
+amdgpu_atombios_encoder_setup_dig_encoder(struct drm_encoder *encoder,
+				   int action, int panel_mode)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	union dig_encoder_control args;
+	int index = GetIndexIntoMasterTable(COMMAND, DIGxEncoderControl);
+	uint8_t frev, crev;
+	int dp_clock = 0;
+	int dp_lane_count = 0;
+	int hpd_id = AMDGPU_HPD_NONE;
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		struct amdgpu_connector_atom_dig *dig_connector =
+			amdgpu_connector->con_priv;
+
+		dp_clock = dig_connector->dp_clock;
+		dp_lane_count = dig_connector->dp_lane_count;
+		hpd_id = amdgpu_connector->hpd.hpd;
+	}
+
+	/* no dig encoder assigned */
+	if (dig->dig_encoder == -1)
+		return;
+
+	memset(&args, 0, sizeof(args));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 1:
+			args.v1.ucAction = action;
+			args.v1.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			if (action == ATOM_ENCODER_CMD_SETUP_PANEL_MODE)
+				args.v3.ucPanelMode = panel_mode;
+			else
+				args.v1.ucEncoderMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (ENCODER_MODE_IS_DP(args.v1.ucEncoderMode))
+				args.v1.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v1.ucLaneNum = 8;
+			else
+				args.v1.ucLaneNum = 4;
+
+			if (ENCODER_MODE_IS_DP(args.v1.ucEncoderMode) && (dp_clock == 270000))
+				args.v1.ucConfig |= ATOM_ENCODER_CONFIG_DPLINKRATE_2_70GHZ;
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+				args.v1.ucConfig = ATOM_ENCODER_CONFIG_V2_TRANSMITTER1;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+				args.v1.ucConfig = ATOM_ENCODER_CONFIG_V2_TRANSMITTER2;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+				args.v1.ucConfig = ATOM_ENCODER_CONFIG_V2_TRANSMITTER3;
+				break;
+			}
+			if (dig->linkb)
+				args.v1.ucConfig |= ATOM_ENCODER_CONFIG_LINKB;
+			else
+				args.v1.ucConfig |= ATOM_ENCODER_CONFIG_LINKA;
+			break;
+		case 2:
+		case 3:
+			args.v3.ucAction = action;
+			args.v3.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			if (action == ATOM_ENCODER_CMD_SETUP_PANEL_MODE)
+				args.v3.ucPanelMode = panel_mode;
+			else
+				args.v3.ucEncoderMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (ENCODER_MODE_IS_DP(args.v3.ucEncoderMode))
+				args.v3.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v3.ucLaneNum = 8;
+			else
+				args.v3.ucLaneNum = 4;
+
+			if (ENCODER_MODE_IS_DP(args.v3.ucEncoderMode) && (dp_clock == 270000))
+				args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V3_DPLINKRATE_2_70GHZ;
+			args.v3.acConfig.ucDigSel = dig->dig_encoder;
+			args.v3.ucBitPerColor = amdgpu_atombios_encoder_get_bpc(encoder);
+			break;
+		case 4:
+			args.v4.ucAction = action;
+			args.v4.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			if (action == ATOM_ENCODER_CMD_SETUP_PANEL_MODE)
+				args.v4.ucPanelMode = panel_mode;
+			else
+				args.v4.ucEncoderMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (ENCODER_MODE_IS_DP(args.v4.ucEncoderMode))
+				args.v4.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v4.ucLaneNum = 8;
+			else
+				args.v4.ucLaneNum = 4;
+
+			if (ENCODER_MODE_IS_DP(args.v4.ucEncoderMode)) {
+				if (dp_clock == 540000)
+					args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V4_DPLINKRATE_5_40GHZ;
+				else if (dp_clock == 324000)
+					args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V4_DPLINKRATE_3_24GHZ;
+				else if (dp_clock == 270000)
+					args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V4_DPLINKRATE_2_70GHZ;
+				else
+					args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V4_DPLINKRATE_1_62GHZ;
+			}
+			args.v4.acConfig.ucDigSel = dig->dig_encoder;
+			args.v4.ucBitPerColor = amdgpu_atombios_encoder_get_bpc(encoder);
+			if (hpd_id == AMDGPU_HPD_NONE)
+				args.v4.ucHPD_ID = 0;
+			else
+				args.v4.ucHPD_ID = hpd_id + 1;
+			break;
+		case 5:
+			switch (action) {
+			case ATOM_ENCODER_CMD_SETUP_PANEL_MODE:
+				args.v5.asDPPanelModeParam.ucAction = action;
+				args.v5.asDPPanelModeParam.ucPanelMode = panel_mode;
+				args.v5.asDPPanelModeParam.ucDigId = dig->dig_encoder;
+				break;
+			case ATOM_ENCODER_CMD_STREAM_SETUP:
+				args.v5.asStreamParam.ucAction = action;
+				args.v5.asStreamParam.ucDigId = dig->dig_encoder;
+				args.v5.asStreamParam.ucDigMode =
+					amdgpu_atombios_encoder_get_encoder_mode(encoder);
+				if (ENCODER_MODE_IS_DP(args.v5.asStreamParam.ucDigMode))
+					args.v5.asStreamParam.ucLaneNum = dp_lane_count;
+				else if (amdgpu_dig_monitor_is_duallink(encoder,
+									amdgpu_encoder->pixel_clock))
+					args.v5.asStreamParam.ucLaneNum = 8;
+				else
+					args.v5.asStreamParam.ucLaneNum = 4;
+				args.v5.asStreamParam.ulPixelClock =
+					cpu_to_le32(amdgpu_encoder->pixel_clock / 10);
+				args.v5.asStreamParam.ucBitPerColor =
+					amdgpu_atombios_encoder_get_bpc(encoder);
+				args.v5.asStreamParam.ucLinkRateIn270Mhz = dp_clock / 27000;
+				break;
+			case ATOM_ENCODER_CMD_DP_LINK_TRAINING_START:
+			case ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN1:
+			case ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN2:
+			case ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN3:
+			case ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN4:
+			case ATOM_ENCODER_CMD_DP_LINK_TRAINING_COMPLETE:
+			case ATOM_ENCODER_CMD_DP_VIDEO_OFF:
+			case ATOM_ENCODER_CMD_DP_VIDEO_ON:
+				args.v5.asCmdParam.ucAction = action;
+				args.v5.asCmdParam.ucDigId = dig->dig_encoder;
+				break;
+			default:
+				DRM_ERROR("Unsupported action 0x%x\n", action);
+				break;
+			}
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+			break;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		break;
+	}
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+}
+
+union dig_transmitter_control {
+	DIG_TRANSMITTER_CONTROL_PS_ALLOCATION v1;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V2 v2;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V3 v3;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V4 v4;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V1_5 v5;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V1_6 v6;
+};
+
+void
+amdgpu_atombios_encoder_setup_dig_transmitter(struct drm_encoder *encoder, int action,
+					      uint8_t lane_num, uint8_t lane_set)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	union dig_transmitter_control args;
+	int index = 0;
+	uint8_t frev, crev;
+	bool is_dp = false;
+	int pll_id = 0;
+	int dp_clock = 0;
+	int dp_lane_count = 0;
+	int connector_object_id = 0;
+	int igp_lane_info = 0;
+	int dig_encoder = dig->dig_encoder;
+	int hpd_id = AMDGPU_HPD_NONE;
+
+	if (action == ATOM_TRANSMITTER_ACTION_INIT) {
+		connector = amdgpu_get_connector_for_encoder_init(encoder);
+		/* just needed to avoid bailing in the encoder check.  the encoder
+		 * isn't used for init
+		 */
+		dig_encoder = 0;
+	} else
+		connector = amdgpu_get_connector_for_encoder(encoder);
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		struct amdgpu_connector_atom_dig *dig_connector =
+			amdgpu_connector->con_priv;
+
+		hpd_id = amdgpu_connector->hpd.hpd;
+		dp_clock = dig_connector->dp_clock;
+		dp_lane_count = dig_connector->dp_lane_count;
+		connector_object_id =
+			(amdgpu_connector->connector_object_id & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+	}
+
+	if (encoder->crtc) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+		pll_id = amdgpu_crtc->pll_id;
+	}
+
+	/* no dig encoder assigned */
+	if (dig_encoder == -1)
+		return;
+
+	if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(encoder)))
+		is_dp = true;
+
+	memset(&args, 0, sizeof(args));
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+		index = GetIndexIntoMasterTable(COMMAND, DVOOutputControl);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		index = GetIndexIntoMasterTable(COMMAND, UNIPHYTransmitterControl);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+		index = GetIndexIntoMasterTable(COMMAND, LVTMATransmitterControl);
+		break;
+	}
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 1:
+			args.v1.ucAction = action;
+			if (action == ATOM_TRANSMITTER_ACTION_INIT) {
+				args.v1.usInitInfo = cpu_to_le16(connector_object_id);
+			} else if (action == ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH) {
+				args.v1.asMode.ucLaneSel = lane_num;
+				args.v1.asMode.ucLaneSet = lane_set;
+			} else {
+				if (is_dp)
+					args.v1.usPixelClock = cpu_to_le16(dp_clock / 10);
+				else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v1.usPixelClock = cpu_to_le16((amdgpu_encoder->pixel_clock / 2) / 10);
+				else
+					args.v1.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			}
+
+			args.v1.ucConfig = ATOM_TRANSMITTER_CONFIG_CLKSRC_PPLL;
+
+			if (dig_encoder)
+				args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG2_ENCODER;
+			else
+				args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG1_ENCODER;
+
+			if ((adev->flags & AMD_IS_APU) &&
+			    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_UNIPHY)) {
+				if (is_dp ||
+				    !amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock)) {
+					if (igp_lane_info & 0x1)
+						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_0_3;
+					else if (igp_lane_info & 0x2)
+						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_4_7;
+					else if (igp_lane_info & 0x4)
+						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_8_11;
+					else if (igp_lane_info & 0x8)
+						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_12_15;
+				} else {
+					if (igp_lane_info & 0x3)
+						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_0_7;
+					else if (igp_lane_info & 0xc)
+						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_8_15;
+				}
+			}
+
+			if (dig->linkb)
+				args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LINKB;
+			else
+				args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LINKA;
+
+			if (is_dp)
+				args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_COHERENT;
+			else if (amdgpu_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) {
+				if (dig->coherent_mode)
+					args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_COHERENT;
+				if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_8LANE_LINK;
+			}
+			break;
+		case 2:
+			args.v2.ucAction = action;
+			if (action == ATOM_TRANSMITTER_ACTION_INIT) {
+				args.v2.usInitInfo = cpu_to_le16(connector_object_id);
+			} else if (action == ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH) {
+				args.v2.asMode.ucLaneSel = lane_num;
+				args.v2.asMode.ucLaneSet = lane_set;
+			} else {
+				if (is_dp)
+					args.v2.usPixelClock = cpu_to_le16(dp_clock / 10);
+				else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v2.usPixelClock = cpu_to_le16((amdgpu_encoder->pixel_clock / 2) / 10);
+				else
+					args.v2.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			}
+
+			args.v2.acConfig.ucEncoderSel = dig_encoder;
+			if (dig->linkb)
+				args.v2.acConfig.ucLinkSel = 1;
+
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+				args.v2.acConfig.ucTransmitterSel = 0;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+				args.v2.acConfig.ucTransmitterSel = 1;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+				args.v2.acConfig.ucTransmitterSel = 2;
+				break;
+			}
+
+			if (is_dp) {
+				args.v2.acConfig.fCoherentMode = 1;
+				args.v2.acConfig.fDPConnector = 1;
+			} else if (amdgpu_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) {
+				if (dig->coherent_mode)
+					args.v2.acConfig.fCoherentMode = 1;
+				if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v2.acConfig.fDualLinkConnector = 1;
+			}
+			break;
+		case 3:
+			args.v3.ucAction = action;
+			if (action == ATOM_TRANSMITTER_ACTION_INIT) {
+				args.v3.usInitInfo = cpu_to_le16(connector_object_id);
+			} else if (action == ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH) {
+				args.v3.asMode.ucLaneSel = lane_num;
+				args.v3.asMode.ucLaneSet = lane_set;
+			} else {
+				if (is_dp)
+					args.v3.usPixelClock = cpu_to_le16(dp_clock / 10);
+				else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v3.usPixelClock = cpu_to_le16((amdgpu_encoder->pixel_clock / 2) / 10);
+				else
+					args.v3.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			}
+
+			if (is_dp)
+				args.v3.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v3.ucLaneNum = 8;
+			else
+				args.v3.ucLaneNum = 4;
+
+			if (dig->linkb)
+				args.v3.acConfig.ucLinkSel = 1;
+			if (dig_encoder & 1)
+				args.v3.acConfig.ucEncoderSel = 1;
+
+			/* Select the PLL for the PHY
+			 * DP PHY should be clocked from external src if there is
+			 * one.
+			 */
+			/* On DCE4, if there is an external clock, it generates the DP ref clock */
+			if (is_dp && adev->clock.dp_extclk)
+				args.v3.acConfig.ucRefClkSource = 2; /* external src */
+			else
+				args.v3.acConfig.ucRefClkSource = pll_id;
+
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+				args.v3.acConfig.ucTransmitterSel = 0;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+				args.v3.acConfig.ucTransmitterSel = 1;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+				args.v3.acConfig.ucTransmitterSel = 2;
+				break;
+			}
+
+			if (is_dp)
+				args.v3.acConfig.fCoherentMode = 1; /* DP requires coherent */
+			else if (amdgpu_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) {
+				if (dig->coherent_mode)
+					args.v3.acConfig.fCoherentMode = 1;
+				if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v3.acConfig.fDualLinkConnector = 1;
+			}
+			break;
+		case 4:
+			args.v4.ucAction = action;
+			if (action == ATOM_TRANSMITTER_ACTION_INIT) {
+				args.v4.usInitInfo = cpu_to_le16(connector_object_id);
+			} else if (action == ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH) {
+				args.v4.asMode.ucLaneSel = lane_num;
+				args.v4.asMode.ucLaneSet = lane_set;
+			} else {
+				if (is_dp)
+					args.v4.usPixelClock = cpu_to_le16(dp_clock / 10);
+				else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v4.usPixelClock = cpu_to_le16((amdgpu_encoder->pixel_clock / 2) / 10);
+				else
+					args.v4.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			}
+
+			if (is_dp)
+				args.v4.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v4.ucLaneNum = 8;
+			else
+				args.v4.ucLaneNum = 4;
+
+			if (dig->linkb)
+				args.v4.acConfig.ucLinkSel = 1;
+			if (dig_encoder & 1)
+				args.v4.acConfig.ucEncoderSel = 1;
+
+			/* Select the PLL for the PHY
+			 * DP PHY should be clocked from external src if there is
+			 * one.
+			 */
+			/* On DCE5 DCPLL usually generates the DP ref clock */
+			if (is_dp) {
+				if (adev->clock.dp_extclk)
+					args.v4.acConfig.ucRefClkSource = ENCODER_REFCLK_SRC_EXTCLK;
+				else
+					args.v4.acConfig.ucRefClkSource = ENCODER_REFCLK_SRC_DCPLL;
+			} else
+				args.v4.acConfig.ucRefClkSource = pll_id;
+
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+				args.v4.acConfig.ucTransmitterSel = 0;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+				args.v4.acConfig.ucTransmitterSel = 1;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+				args.v4.acConfig.ucTransmitterSel = 2;
+				break;
+			}
+
+			if (is_dp)
+				args.v4.acConfig.fCoherentMode = 1; /* DP requires coherent */
+			else if (amdgpu_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) {
+				if (dig->coherent_mode)
+					args.v4.acConfig.fCoherentMode = 1;
+				if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v4.acConfig.fDualLinkConnector = 1;
+			}
+			break;
+		case 5:
+			args.v5.ucAction = action;
+			if (is_dp)
+				args.v5.usSymClock = cpu_to_le16(dp_clock / 10);
+			else
+				args.v5.usSymClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+				if (dig->linkb)
+					args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYB;
+				else
+					args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYA;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+				if (dig->linkb)
+					args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYD;
+				else
+					args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYC;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+				if (dig->linkb)
+					args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYF;
+				else
+					args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYE;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+				args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYG;
+				break;
+			}
+			if (is_dp)
+				args.v5.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v5.ucLaneNum = 8;
+			else
+				args.v5.ucLaneNum = 4;
+			args.v5.ucConnObjId = connector_object_id;
+			args.v5.ucDigMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (is_dp && adev->clock.dp_extclk)
+				args.v5.asConfig.ucPhyClkSrcId = ENCODER_REFCLK_SRC_EXTCLK;
+			else
+				args.v5.asConfig.ucPhyClkSrcId = pll_id;
+
+			if (is_dp)
+				args.v5.asConfig.ucCoherentMode = 1; /* DP requires coherent */
+			else if (amdgpu_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) {
+				if (dig->coherent_mode)
+					args.v5.asConfig.ucCoherentMode = 1;
+			}
+			if (hpd_id == AMDGPU_HPD_NONE)
+				args.v5.asConfig.ucHPDSel = 0;
+			else
+				args.v5.asConfig.ucHPDSel = hpd_id + 1;
+			args.v5.ucDigEncoderSel = 1 << dig_encoder;
+			args.v5.ucDPLaneSet = lane_set;
+			break;
+		case 6:
+			args.v6.ucAction = action;
+			if (is_dp)
+				args.v6.ulSymClock = cpu_to_le32(dp_clock / 10);
+			else
+				args.v6.ulSymClock = cpu_to_le32(amdgpu_encoder->pixel_clock / 10);
+
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+				if (dig->linkb)
+					args.v6.ucPhyId = ATOM_PHY_ID_UNIPHYB;
+				else
+					args.v6.ucPhyId = ATOM_PHY_ID_UNIPHYA;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+				if (dig->linkb)
+					args.v6.ucPhyId = ATOM_PHY_ID_UNIPHYD;
+				else
+					args.v6.ucPhyId = ATOM_PHY_ID_UNIPHYC;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+				if (dig->linkb)
+					args.v6.ucPhyId = ATOM_PHY_ID_UNIPHYF;
+				else
+					args.v6.ucPhyId = ATOM_PHY_ID_UNIPHYE;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+				args.v6.ucPhyId = ATOM_PHY_ID_UNIPHYG;
+				break;
+			}
+			if (is_dp)
+				args.v6.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v6.ucLaneNum = 8;
+			else
+				args.v6.ucLaneNum = 4;
+			args.v6.ucConnObjId = connector_object_id;
+			if (action == ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH)
+				args.v6.ucDPLaneSet = lane_set;
+			else
+				args.v6.ucDigMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (hpd_id == AMDGPU_HPD_NONE)
+				args.v6.ucHPDSel = 0;
+			else
+				args.v6.ucHPDSel = hpd_id + 1;
+			args.v6.ucDigEncoderSel = 1 << dig_encoder;
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+			break;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		break;
+	}
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+bool
+amdgpu_atombios_encoder_set_edp_panel_power(struct drm_connector *connector,
+				     int action)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_device *dev = amdgpu_connector->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	union dig_transmitter_control args;
+	int index = GetIndexIntoMasterTable(COMMAND, UNIPHYTransmitterControl);
+	uint8_t frev, crev;
+
+	if (connector->connector_type != DRM_MODE_CONNECTOR_eDP)
+		goto done;
+
+	if ((action != ATOM_TRANSMITTER_ACTION_POWER_ON) &&
+	    (action != ATOM_TRANSMITTER_ACTION_POWER_OFF))
+		goto done;
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		goto done;
+
+	memset(&args, 0, sizeof(args));
+
+	args.v1.ucAction = action;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+	/* wait for the panel to power up */
+	if (action == ATOM_TRANSMITTER_ACTION_POWER_ON) {
+		int i;
+
+		for (i = 0; i < 300; i++) {
+			if (amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd))
+				return true;
+			mdelay(1);
+		}
+		return false;
+	}
+done:
+	return true;
+}
+
+union external_encoder_control {
+	EXTERNAL_ENCODER_CONTROL_PS_ALLOCATION v1;
+	EXTERNAL_ENCODER_CONTROL_PS_ALLOCATION_V3 v3;
+};
+
+static void
+amdgpu_atombios_encoder_setup_external_encoder(struct drm_encoder *encoder,
+					struct drm_encoder *ext_encoder,
+					int action)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder *ext_amdgpu_encoder = to_amdgpu_encoder(ext_encoder);
+	union external_encoder_control args;
+	struct drm_connector *connector;
+	int index = GetIndexIntoMasterTable(COMMAND, ExternalEncoderControl);
+	u8 frev, crev;
+	int dp_clock = 0;
+	int dp_lane_count = 0;
+	int connector_object_id = 0;
+	u32 ext_enum = (ext_amdgpu_encoder->encoder_enum & ENUM_ID_MASK) >> ENUM_ID_SHIFT;
+
+	if (action == EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT)
+		connector = amdgpu_get_connector_for_encoder_init(encoder);
+	else
+		connector = amdgpu_get_connector_for_encoder(encoder);
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		struct amdgpu_connector_atom_dig *dig_connector =
+			amdgpu_connector->con_priv;
+
+		dp_clock = dig_connector->dp_clock;
+		dp_lane_count = dig_connector->dp_lane_count;
+		connector_object_id =
+			(amdgpu_connector->connector_object_id & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+	}
+
+	memset(&args, 0, sizeof(args));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		/* no params on frev 1 */
+		break;
+	case 2:
+		switch (crev) {
+		case 1:
+		case 2:
+			args.v1.sDigEncoder.ucAction = action;
+			args.v1.sDigEncoder.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			args.v1.sDigEncoder.ucEncoderMode =
+				amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (ENCODER_MODE_IS_DP(args.v1.sDigEncoder.ucEncoderMode)) {
+				if (dp_clock == 270000)
+					args.v1.sDigEncoder.ucConfig |= ATOM_ENCODER_CONFIG_DPLINKRATE_2_70GHZ;
+				args.v1.sDigEncoder.ucLaneNum = dp_lane_count;
+			} else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v1.sDigEncoder.ucLaneNum = 8;
+			else
+				args.v1.sDigEncoder.ucLaneNum = 4;
+			break;
+		case 3:
+			args.v3.sExtEncoder.ucAction = action;
+			if (action == EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT)
+				args.v3.sExtEncoder.usConnectorId = cpu_to_le16(connector_object_id);
+			else
+				args.v3.sExtEncoder.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			args.v3.sExtEncoder.ucEncoderMode =
+				amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (ENCODER_MODE_IS_DP(args.v3.sExtEncoder.ucEncoderMode)) {
+				if (dp_clock == 270000)
+					args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_DPLINKRATE_2_70GHZ;
+				else if (dp_clock == 540000)
+					args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_DPLINKRATE_5_40GHZ;
+				args.v3.sExtEncoder.ucLaneNum = dp_lane_count;
+			} else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v3.sExtEncoder.ucLaneNum = 8;
+			else
+				args.v3.sExtEncoder.ucLaneNum = 4;
+			switch (ext_enum) {
+			case GRAPH_OBJECT_ENUM_ID1:
+				args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_ENCODER1;
+				break;
+			case GRAPH_OBJECT_ENUM_ID2:
+				args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_ENCODER2;
+				break;
+			case GRAPH_OBJECT_ENUM_ID3:
+				args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_ENCODER3;
+				break;
+			}
+			args.v3.sExtEncoder.ucBitPerColor = amdgpu_atombios_encoder_get_bpc(encoder);
+			break;
+		default:
+			DRM_ERROR("Unknown table version: %d, %d\n", frev, crev);
+			return;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version: %d, %d\n", frev, crev);
+		return;
+	}
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+static void
+amdgpu_atombios_encoder_setup_dig(struct drm_encoder *encoder, int action)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_encoder *ext_encoder = amdgpu_get_external_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	struct amdgpu_connector_atom_dig *amdgpu_dig_connector = NULL;
+
+	if (connector) {
+		amdgpu_connector = to_amdgpu_connector(connector);
+		amdgpu_dig_connector = amdgpu_connector->con_priv;
+	}
+
+	if (action == ATOM_ENABLE) {
+		if (!connector)
+			dig->panel_mode = DP_PANEL_MODE_EXTERNAL_DP_MODE;
+		else
+			dig->panel_mode = amdgpu_atombios_dp_get_panel_mode(encoder, connector);
+
+		/* setup and enable the encoder */
+		amdgpu_atombios_encoder_setup_dig_encoder(encoder, ATOM_ENCODER_CMD_SETUP, 0);
+		amdgpu_atombios_encoder_setup_dig_encoder(encoder,
+						   ATOM_ENCODER_CMD_SETUP_PANEL_MODE,
+						   dig->panel_mode);
+		if (ext_encoder)
+			amdgpu_atombios_encoder_setup_external_encoder(encoder, ext_encoder,
+								EXTERNAL_ENCODER_ACTION_V3_ENCODER_SETUP);
+		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(encoder)) &&
+		    connector) {
+			if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) {
+				amdgpu_atombios_encoder_set_edp_panel_power(connector,
+								     ATOM_TRANSMITTER_ACTION_POWER_ON);
+				amdgpu_dig_connector->edp_on = true;
+			}
+		}
+		/* enable the transmitter */
+		amdgpu_atombios_encoder_setup_dig_transmitter(encoder,
+						       ATOM_TRANSMITTER_ACTION_ENABLE,
+						       0, 0);
+		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(encoder)) &&
+		    connector) {
+			/* DP_SET_POWER_D0 is set in amdgpu_atombios_dp_link_train */
+			amdgpu_atombios_dp_link_train(encoder, connector);
+			amdgpu_atombios_encoder_setup_dig_encoder(encoder, ATOM_ENCODER_CMD_DP_VIDEO_ON, 0);
+		}
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+			amdgpu_atombios_encoder_set_backlight_level(amdgpu_encoder, dig->backlight_level);
+		if (ext_encoder)
+			amdgpu_atombios_encoder_setup_external_encoder(encoder, ext_encoder, ATOM_ENABLE);
+	} else {
+		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(encoder)) &&
+		    connector)
+			amdgpu_atombios_encoder_setup_dig_encoder(encoder,
+							   ATOM_ENCODER_CMD_DP_VIDEO_OFF, 0);
+		if (ext_encoder)
+			amdgpu_atombios_encoder_setup_external_encoder(encoder, ext_encoder, ATOM_DISABLE);
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+			amdgpu_atombios_encoder_setup_dig_transmitter(encoder,
+							       ATOM_TRANSMITTER_ACTION_LCD_BLOFF, 0, 0);
+
+		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(encoder)) &&
+		    connector)
+			amdgpu_atombios_dp_set_rx_power_state(connector, DP_SET_POWER_D3);
+		/* disable the transmitter */
+		amdgpu_atombios_encoder_setup_dig_transmitter(encoder,
+						       ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0);
+		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(encoder)) &&
+		    connector) {
+			if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) {
+				amdgpu_atombios_encoder_set_edp_panel_power(connector,
+								     ATOM_TRANSMITTER_ACTION_POWER_OFF);
+				amdgpu_dig_connector->edp_on = false;
+			}
+		}
+	}
+}
+
+void
+amdgpu_atombios_encoder_dpms(struct drm_encoder *encoder, int mode)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	DRM_DEBUG_KMS("encoder dpms %d to mode %d, devices %08x, active_devices %08x\n",
+		  amdgpu_encoder->encoder_id, mode, amdgpu_encoder->devices,
+		  amdgpu_encoder->active_device);
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		switch (mode) {
+		case DRM_MODE_DPMS_ON:
+			amdgpu_atombios_encoder_setup_dig(encoder, ATOM_ENABLE);
+			break;
+		case DRM_MODE_DPMS_STANDBY:
+		case DRM_MODE_DPMS_SUSPEND:
+		case DRM_MODE_DPMS_OFF:
+			amdgpu_atombios_encoder_setup_dig(encoder, ATOM_DISABLE);
+			break;
+		}
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+		switch (mode) {
+		case DRM_MODE_DPMS_ON:
+			amdgpu_atombios_encoder_setup_dvo(encoder, ATOM_ENABLE);
+			break;
+		case DRM_MODE_DPMS_STANDBY:
+		case DRM_MODE_DPMS_SUSPEND:
+		case DRM_MODE_DPMS_OFF:
+			amdgpu_atombios_encoder_setup_dvo(encoder, ATOM_DISABLE);
+			break;
+		}
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+		switch (mode) {
+		case DRM_MODE_DPMS_ON:
+			amdgpu_atombios_encoder_setup_dac(encoder, ATOM_ENABLE);
+			break;
+		case DRM_MODE_DPMS_STANDBY:
+		case DRM_MODE_DPMS_SUSPEND:
+		case DRM_MODE_DPMS_OFF:
+			amdgpu_atombios_encoder_setup_dac(encoder, ATOM_DISABLE);
+			break;
+		}
+		break;
+	default:
+		return;
+	}
+}
+
+union crtc_source_param {
+	SELECT_CRTC_SOURCE_PS_ALLOCATION v1;
+	SELECT_CRTC_SOURCE_PARAMETERS_V2 v2;
+	SELECT_CRTC_SOURCE_PARAMETERS_V3 v3;
+};
+
+void
+amdgpu_atombios_encoder_set_crtc_source(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	union crtc_source_param args;
+	int index = GetIndexIntoMasterTable(COMMAND, SelectCRTC_Source);
+	uint8_t frev, crev;
+	struct amdgpu_encoder_atom_dig *dig;
+
+	memset(&args, 0, sizeof(args));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 1:
+		default:
+			args.v1.ucCRTC = amdgpu_crtc->crtc_id;
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_TMDS1:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1:
+				args.v1.ucDevice = ATOM_DEVICE_DFP1_INDEX;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_LVDS:
+			case ENCODER_OBJECT_ID_INTERNAL_LVTM1:
+				if (amdgpu_encoder->devices & ATOM_DEVICE_LCD1_SUPPORT)
+					args.v1.ucDevice = ATOM_DEVICE_LCD1_INDEX;
+				else
+					args.v1.ucDevice = ATOM_DEVICE_DFP3_INDEX;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_DVO1:
+			case ENCODER_OBJECT_ID_INTERNAL_DDI:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+				args.v1.ucDevice = ATOM_DEVICE_DFP2_INDEX;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_DAC1:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+				if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+					args.v1.ucDevice = ATOM_DEVICE_TV1_INDEX;
+				else if (amdgpu_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT))
+					args.v1.ucDevice = ATOM_DEVICE_CV_INDEX;
+				else
+					args.v1.ucDevice = ATOM_DEVICE_CRT1_INDEX;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_DAC2:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+				if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+					args.v1.ucDevice = ATOM_DEVICE_TV1_INDEX;
+				else if (amdgpu_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT))
+					args.v1.ucDevice = ATOM_DEVICE_CV_INDEX;
+				else
+					args.v1.ucDevice = ATOM_DEVICE_CRT2_INDEX;
+				break;
+			}
+			break;
+		case 2:
+			args.v2.ucCRTC = amdgpu_crtc->crtc_id;
+			if (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) != ENCODER_OBJECT_ID_NONE) {
+				struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+
+				if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)
+					args.v2.ucEncodeMode = ATOM_ENCODER_MODE_LVDS;
+				else if (connector->connector_type == DRM_MODE_CONNECTOR_VGA)
+					args.v2.ucEncodeMode = ATOM_ENCODER_MODE_CRT;
+				else
+					args.v2.ucEncodeMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+			} else if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+				args.v2.ucEncodeMode = ATOM_ENCODER_MODE_LVDS;
+			} else {
+				args.v2.ucEncodeMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+			}
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+				dig = amdgpu_encoder->enc_priv;
+				switch (dig->dig_encoder) {
+				case 0:
+					args.v2.ucEncoderID = ASIC_INT_DIG1_ENCODER_ID;
+					break;
+				case 1:
+					args.v2.ucEncoderID = ASIC_INT_DIG2_ENCODER_ID;
+					break;
+				case 2:
+					args.v2.ucEncoderID = ASIC_INT_DIG3_ENCODER_ID;
+					break;
+				case 3:
+					args.v2.ucEncoderID = ASIC_INT_DIG4_ENCODER_ID;
+					break;
+				case 4:
+					args.v2.ucEncoderID = ASIC_INT_DIG5_ENCODER_ID;
+					break;
+				case 5:
+					args.v2.ucEncoderID = ASIC_INT_DIG6_ENCODER_ID;
+					break;
+				case 6:
+					args.v2.ucEncoderID = ASIC_INT_DIG7_ENCODER_ID;
+					break;
+				}
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+				args.v2.ucEncoderID = ASIC_INT_DVO_ENCODER_ID;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+				if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+					args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else if (amdgpu_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT))
+					args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else
+					args.v2.ucEncoderID = ASIC_INT_DAC1_ENCODER_ID;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+				if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+					args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else if (amdgpu_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT))
+					args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else
+					args.v2.ucEncoderID = ASIC_INT_DAC2_ENCODER_ID;
+				break;
+			}
+			break;
+		case 3:
+			args.v3.ucCRTC = amdgpu_crtc->crtc_id;
+			if (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) != ENCODER_OBJECT_ID_NONE) {
+				struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+
+				if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)
+					args.v2.ucEncodeMode = ATOM_ENCODER_MODE_LVDS;
+				else if (connector->connector_type == DRM_MODE_CONNECTOR_VGA)
+					args.v2.ucEncodeMode = ATOM_ENCODER_MODE_CRT;
+				else
+					args.v2.ucEncodeMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+			} else if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+				args.v2.ucEncodeMode = ATOM_ENCODER_MODE_LVDS;
+			} else {
+				args.v2.ucEncodeMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+			}
+			args.v3.ucDstBpc = amdgpu_atombios_encoder_get_bpc(encoder);
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+				dig = amdgpu_encoder->enc_priv;
+				switch (dig->dig_encoder) {
+				case 0:
+					args.v3.ucEncoderID = ASIC_INT_DIG1_ENCODER_ID;
+					break;
+				case 1:
+					args.v3.ucEncoderID = ASIC_INT_DIG2_ENCODER_ID;
+					break;
+				case 2:
+					args.v3.ucEncoderID = ASIC_INT_DIG3_ENCODER_ID;
+					break;
+				case 3:
+					args.v3.ucEncoderID = ASIC_INT_DIG4_ENCODER_ID;
+					break;
+				case 4:
+					args.v3.ucEncoderID = ASIC_INT_DIG5_ENCODER_ID;
+					break;
+				case 5:
+					args.v3.ucEncoderID = ASIC_INT_DIG6_ENCODER_ID;
+					break;
+				case 6:
+					args.v3.ucEncoderID = ASIC_INT_DIG7_ENCODER_ID;
+					break;
+				}
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+				args.v3.ucEncoderID = ASIC_INT_DVO_ENCODER_ID;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+				if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+					args.v3.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else if (amdgpu_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT))
+					args.v3.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else
+					args.v3.ucEncoderID = ASIC_INT_DAC1_ENCODER_ID;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+				if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+					args.v3.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else if (amdgpu_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT))
+					args.v3.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else
+					args.v3.ucEncoderID = ASIC_INT_DAC2_ENCODER_ID;
+				break;
+			}
+			break;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version: %d, %d\n", frev, crev);
+		return;
+	}
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+/* This only needs to be called once at startup */
+void
+amdgpu_atombios_encoder_init_dig(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_encoder *encoder;
+
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+		struct drm_encoder *ext_encoder = amdgpu_get_external_encoder(encoder);
+
+		switch (amdgpu_encoder->encoder_id) {
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+			amdgpu_atombios_encoder_setup_dig_transmitter(encoder, ATOM_TRANSMITTER_ACTION_INIT,
+							       0, 0);
+			break;
+		}
+
+		if (ext_encoder)
+			amdgpu_atombios_encoder_setup_external_encoder(encoder, ext_encoder,
+								EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT);
+	}
+}
+
+static bool
+amdgpu_atombios_encoder_dac_load_detect(struct drm_encoder *encoder,
+				 struct drm_connector *connector)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_TV_SUPPORT |
+				       ATOM_DEVICE_CV_SUPPORT |
+				       ATOM_DEVICE_CRT_SUPPORT)) {
+		DAC_LOAD_DETECTION_PS_ALLOCATION args;
+		int index = GetIndexIntoMasterTable(COMMAND, DAC_LoadDetection);
+		uint8_t frev, crev;
+
+		memset(&args, 0, sizeof(args));
+
+		if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+			return false;
+
+		args.sDacload.ucMisc = 0;
+
+		if ((amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_DAC1) ||
+		    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1))
+			args.sDacload.ucDacType = ATOM_DAC_A;
+		else
+			args.sDacload.ucDacType = ATOM_DAC_B;
+
+		if (amdgpu_connector->devices & ATOM_DEVICE_CRT1_SUPPORT)
+			args.sDacload.usDeviceID = cpu_to_le16(ATOM_DEVICE_CRT1_SUPPORT);
+		else if (amdgpu_connector->devices & ATOM_DEVICE_CRT2_SUPPORT)
+			args.sDacload.usDeviceID = cpu_to_le16(ATOM_DEVICE_CRT2_SUPPORT);
+		else if (amdgpu_connector->devices & ATOM_DEVICE_CV_SUPPORT) {
+			args.sDacload.usDeviceID = cpu_to_le16(ATOM_DEVICE_CV_SUPPORT);
+			if (crev >= 3)
+				args.sDacload.ucMisc = DAC_LOAD_MISC_YPrPb;
+		} else if (amdgpu_connector->devices & ATOM_DEVICE_TV1_SUPPORT) {
+			args.sDacload.usDeviceID = cpu_to_le16(ATOM_DEVICE_TV1_SUPPORT);
+			if (crev >= 3)
+				args.sDacload.ucMisc = DAC_LOAD_MISC_YPrPb;
+		}
+
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		return true;
+	} else
+		return false;
+}
+
+enum drm_connector_status
+amdgpu_atombios_encoder_dac_detect(struct drm_encoder *encoder,
+			    struct drm_connector *connector)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	uint32_t bios_0_scratch;
+
+	if (!amdgpu_atombios_encoder_dac_load_detect(encoder, connector)) {
+		DRM_DEBUG_KMS("detect returned false \n");
+		return connector_status_unknown;
+	}
+
+	bios_0_scratch = RREG32(mmBIOS_SCRATCH_0);
+
+	DRM_DEBUG_KMS("Bios 0 scratch %x %08x\n", bios_0_scratch, amdgpu_encoder->devices);
+	if (amdgpu_connector->devices & ATOM_DEVICE_CRT1_SUPPORT) {
+		if (bios_0_scratch & ATOM_S0_CRT1_MASK)
+			return connector_status_connected;
+	}
+	if (amdgpu_connector->devices & ATOM_DEVICE_CRT2_SUPPORT) {
+		if (bios_0_scratch & ATOM_S0_CRT2_MASK)
+			return connector_status_connected;
+	}
+	if (amdgpu_connector->devices & ATOM_DEVICE_CV_SUPPORT) {
+		if (bios_0_scratch & (ATOM_S0_CV_MASK|ATOM_S0_CV_MASK_A))
+			return connector_status_connected;
+	}
+	if (amdgpu_connector->devices & ATOM_DEVICE_TV1_SUPPORT) {
+		if (bios_0_scratch & (ATOM_S0_TV1_COMPOSITE | ATOM_S0_TV1_COMPOSITE_A))
+			return connector_status_connected; /* CTV */
+		else if (bios_0_scratch & (ATOM_S0_TV1_SVIDEO | ATOM_S0_TV1_SVIDEO_A))
+			return connector_status_connected; /* STV */
+	}
+	return connector_status_disconnected;
+}
+
+enum drm_connector_status
+amdgpu_atombios_encoder_dig_detect(struct drm_encoder *encoder,
+			    struct drm_connector *connector)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_encoder *ext_encoder = amdgpu_get_external_encoder(encoder);
+	u32 bios_0_scratch;
+
+	if (!ext_encoder)
+		return connector_status_unknown;
+
+	if ((amdgpu_connector->devices & ATOM_DEVICE_CRT_SUPPORT) == 0)
+		return connector_status_unknown;
+
+	/* load detect on the dp bridge */
+	amdgpu_atombios_encoder_setup_external_encoder(encoder, ext_encoder,
+						EXTERNAL_ENCODER_ACTION_V3_DACLOAD_DETECTION);
+
+	bios_0_scratch = RREG32(mmBIOS_SCRATCH_0);
+
+	DRM_DEBUG_KMS("Bios 0 scratch %x %08x\n", bios_0_scratch, amdgpu_encoder->devices);
+	if (amdgpu_connector->devices & ATOM_DEVICE_CRT1_SUPPORT) {
+		if (bios_0_scratch & ATOM_S0_CRT1_MASK)
+			return connector_status_connected;
+	}
+	if (amdgpu_connector->devices & ATOM_DEVICE_CRT2_SUPPORT) {
+		if (bios_0_scratch & ATOM_S0_CRT2_MASK)
+			return connector_status_connected;
+	}
+	if (amdgpu_connector->devices & ATOM_DEVICE_CV_SUPPORT) {
+		if (bios_0_scratch & (ATOM_S0_CV_MASK|ATOM_S0_CV_MASK_A))
+			return connector_status_connected;
+	}
+	if (amdgpu_connector->devices & ATOM_DEVICE_TV1_SUPPORT) {
+		if (bios_0_scratch & (ATOM_S0_TV1_COMPOSITE | ATOM_S0_TV1_COMPOSITE_A))
+			return connector_status_connected; /* CTV */
+		else if (bios_0_scratch & (ATOM_S0_TV1_SVIDEO | ATOM_S0_TV1_SVIDEO_A))
+			return connector_status_connected; /* STV */
+	}
+	return connector_status_disconnected;
+}
+
+void
+amdgpu_atombios_encoder_setup_ext_encoder_ddc(struct drm_encoder *encoder)
+{
+	struct drm_encoder *ext_encoder = amdgpu_get_external_encoder(encoder);
+
+	if (ext_encoder)
+		/* ddc_setup on the dp bridge */
+		amdgpu_atombios_encoder_setup_external_encoder(encoder, ext_encoder,
+							EXTERNAL_ENCODER_ACTION_V3_DDC_SETUP);
+
+}
+
+void
+amdgpu_atombios_encoder_set_bios_scratch_regs(struct drm_connector *connector,
+				       struct drm_encoder *encoder,
+				       bool connected)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector =
+	    to_amdgpu_connector(connector);
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	uint32_t bios_0_scratch, bios_3_scratch, bios_6_scratch;
+
+	bios_0_scratch = RREG32(mmBIOS_SCRATCH_0);
+	bios_3_scratch = RREG32(mmBIOS_SCRATCH_3);
+	bios_6_scratch = RREG32(mmBIOS_SCRATCH_6);
+
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_LCD1_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_LCD1_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("LCD1 connected\n");
+			bios_0_scratch |= ATOM_S0_LCD1;
+			bios_3_scratch |= ATOM_S3_LCD1_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_LCD1;
+		} else {
+			DRM_DEBUG_KMS("LCD1 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_LCD1;
+			bios_3_scratch &= ~ATOM_S3_LCD1_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_LCD1;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_CRT1_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_CRT1_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("CRT1 connected\n");
+			bios_0_scratch |= ATOM_S0_CRT1_COLOR;
+			bios_3_scratch |= ATOM_S3_CRT1_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_CRT1;
+		} else {
+			DRM_DEBUG_KMS("CRT1 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_CRT1_MASK;
+			bios_3_scratch &= ~ATOM_S3_CRT1_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_CRT1;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_CRT2_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_CRT2_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("CRT2 connected\n");
+			bios_0_scratch |= ATOM_S0_CRT2_COLOR;
+			bios_3_scratch |= ATOM_S3_CRT2_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_CRT2;
+		} else {
+			DRM_DEBUG_KMS("CRT2 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_CRT2_MASK;
+			bios_3_scratch &= ~ATOM_S3_CRT2_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_CRT2;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_DFP1_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_DFP1_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("DFP1 connected\n");
+			bios_0_scratch |= ATOM_S0_DFP1;
+			bios_3_scratch |= ATOM_S3_DFP1_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_DFP1;
+		} else {
+			DRM_DEBUG_KMS("DFP1 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_DFP1;
+			bios_3_scratch &= ~ATOM_S3_DFP1_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP1;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_DFP2_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_DFP2_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("DFP2 connected\n");
+			bios_0_scratch |= ATOM_S0_DFP2;
+			bios_3_scratch |= ATOM_S3_DFP2_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_DFP2;
+		} else {
+			DRM_DEBUG_KMS("DFP2 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_DFP2;
+			bios_3_scratch &= ~ATOM_S3_DFP2_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP2;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_DFP3_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_DFP3_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("DFP3 connected\n");
+			bios_0_scratch |= ATOM_S0_DFP3;
+			bios_3_scratch |= ATOM_S3_DFP3_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_DFP3;
+		} else {
+			DRM_DEBUG_KMS("DFP3 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_DFP3;
+			bios_3_scratch &= ~ATOM_S3_DFP3_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP3;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_DFP4_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_DFP4_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("DFP4 connected\n");
+			bios_0_scratch |= ATOM_S0_DFP4;
+			bios_3_scratch |= ATOM_S3_DFP4_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_DFP4;
+		} else {
+			DRM_DEBUG_KMS("DFP4 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_DFP4;
+			bios_3_scratch &= ~ATOM_S3_DFP4_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP4;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_DFP5_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_DFP5_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("DFP5 connected\n");
+			bios_0_scratch |= ATOM_S0_DFP5;
+			bios_3_scratch |= ATOM_S3_DFP5_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_DFP5;
+		} else {
+			DRM_DEBUG_KMS("DFP5 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_DFP5;
+			bios_3_scratch &= ~ATOM_S3_DFP5_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP5;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_DFP6_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_DFP6_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("DFP6 connected\n");
+			bios_0_scratch |= ATOM_S0_DFP6;
+			bios_3_scratch |= ATOM_S3_DFP6_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_DFP6;
+		} else {
+			DRM_DEBUG_KMS("DFP6 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_DFP6;
+			bios_3_scratch &= ~ATOM_S3_DFP6_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP6;
+		}
+	}
+
+	WREG32(mmBIOS_SCRATCH_0, bios_0_scratch);
+	WREG32(mmBIOS_SCRATCH_3, bios_3_scratch);
+	WREG32(mmBIOS_SCRATCH_6, bios_6_scratch);
+}
+
+union lvds_info {
+	struct _ATOM_LVDS_INFO info;
+	struct _ATOM_LVDS_INFO_V12 info_12;
+};
+
+struct amdgpu_encoder_atom_dig *
+amdgpu_atombios_encoder_get_lcd_info(struct amdgpu_encoder *encoder)
+{
+	struct drm_device *dev = encoder->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, LVDS_Info);
+	uint16_t data_offset, misc;
+	union lvds_info *lvds_info;
+	uint8_t frev, crev;
+	struct amdgpu_encoder_atom_dig *lvds = NULL;
+	int encoder_enum = (encoder->encoder_enum & ENUM_ID_MASK) >> ENUM_ID_SHIFT;
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		lvds_info =
+			(union lvds_info *)(mode_info->atom_context->bios + data_offset);
+		lvds =
+		    kzalloc(sizeof(struct amdgpu_encoder_atom_dig), GFP_KERNEL);
+
+		if (!lvds)
+			return NULL;
+
+		lvds->native_mode.clock =
+		    le16_to_cpu(lvds_info->info.sLCDTiming.usPixClk) * 10;
+		lvds->native_mode.hdisplay =
+		    le16_to_cpu(lvds_info->info.sLCDTiming.usHActive);
+		lvds->native_mode.vdisplay =
+		    le16_to_cpu(lvds_info->info.sLCDTiming.usVActive);
+		lvds->native_mode.htotal = lvds->native_mode.hdisplay +
+			le16_to_cpu(lvds_info->info.sLCDTiming.usHBlanking_Time);
+		lvds->native_mode.hsync_start = lvds->native_mode.hdisplay +
+			le16_to_cpu(lvds_info->info.sLCDTiming.usHSyncOffset);
+		lvds->native_mode.hsync_end = lvds->native_mode.hsync_start +
+			le16_to_cpu(lvds_info->info.sLCDTiming.usHSyncWidth);
+		lvds->native_mode.vtotal = lvds->native_mode.vdisplay +
+			le16_to_cpu(lvds_info->info.sLCDTiming.usVBlanking_Time);
+		lvds->native_mode.vsync_start = lvds->native_mode.vdisplay +
+			le16_to_cpu(lvds_info->info.sLCDTiming.usVSyncOffset);
+		lvds->native_mode.vsync_end = lvds->native_mode.vsync_start +
+			le16_to_cpu(lvds_info->info.sLCDTiming.usVSyncWidth);
+		lvds->panel_pwr_delay =
+		    le16_to_cpu(lvds_info->info.usOffDelayInMs);
+		lvds->lcd_misc = lvds_info->info.ucLVDS_Misc;
+
+		misc = le16_to_cpu(lvds_info->info.sLCDTiming.susModeMiscInfo.usAccess);
+		if (misc & ATOM_VSYNC_POLARITY)
+			lvds->native_mode.flags |= DRM_MODE_FLAG_NVSYNC;
+		if (misc & ATOM_HSYNC_POLARITY)
+			lvds->native_mode.flags |= DRM_MODE_FLAG_NHSYNC;
+		if (misc & ATOM_COMPOSITESYNC)
+			lvds->native_mode.flags |= DRM_MODE_FLAG_CSYNC;
+		if (misc & ATOM_INTERLACE)
+			lvds->native_mode.flags |= DRM_MODE_FLAG_INTERLACE;
+		if (misc & ATOM_DOUBLE_CLOCK_MODE)
+			lvds->native_mode.flags |= DRM_MODE_FLAG_DBLSCAN;
+
+		lvds->native_mode.width_mm = le16_to_cpu(lvds_info->info.sLCDTiming.usImageHSize);
+		lvds->native_mode.height_mm = le16_to_cpu(lvds_info->info.sLCDTiming.usImageVSize);
+
+		/* set crtc values */
+		drm_mode_set_crtcinfo(&lvds->native_mode, CRTC_INTERLACE_HALVE_V);
+
+		lvds->lcd_ss_id = lvds_info->info.ucSS_Id;
+
+		encoder->native_mode = lvds->native_mode;
+
+		if (encoder_enum == 2)
+			lvds->linkb = true;
+		else
+			lvds->linkb = false;
+
+		/* parse the lcd record table */
+		if (le16_to_cpu(lvds_info->info.usModePatchTableOffset)) {
+			ATOM_FAKE_EDID_PATCH_RECORD *fake_edid_record;
+			ATOM_PANEL_RESOLUTION_PATCH_RECORD *panel_res_record;
+			bool bad_record = false;
+			u8 *record;
+
+			if ((frev == 1) && (crev < 2))
+				/* absolute */
+				record = (u8 *)(mode_info->atom_context->bios +
+						le16_to_cpu(lvds_info->info.usModePatchTableOffset));
+			else
+				/* relative */
+				record = (u8 *)(mode_info->atom_context->bios +
+						data_offset +
+						le16_to_cpu(lvds_info->info.usModePatchTableOffset));
+			while (*record != ATOM_RECORD_END_TYPE) {
+				switch (*record) {
+				case LCD_MODE_PATCH_RECORD_MODE_TYPE:
+					record += sizeof(ATOM_PATCH_RECORD_MODE);
+					break;
+				case LCD_RTS_RECORD_TYPE:
+					record += sizeof(ATOM_LCD_RTS_RECORD);
+					break;
+				case LCD_CAP_RECORD_TYPE:
+					record += sizeof(ATOM_LCD_MODE_CONTROL_CAP);
+					break;
+				case LCD_FAKE_EDID_PATCH_RECORD_TYPE:
+					fake_edid_record = (ATOM_FAKE_EDID_PATCH_RECORD *)record;
+					if (fake_edid_record->ucFakeEDIDLength) {
+						struct edid *edid;
+						int edid_size =
+							max((int)EDID_LENGTH, (int)fake_edid_record->ucFakeEDIDLength);
+						edid = kmalloc(edid_size, GFP_KERNEL);
+						if (edid) {
+							memcpy((u8 *)edid, (u8 *)&fake_edid_record->ucFakeEDIDString[0],
+							       fake_edid_record->ucFakeEDIDLength);
+
+							if (drm_edid_is_valid(edid)) {
+								adev->mode_info.bios_hardcoded_edid = edid;
+								adev->mode_info.bios_hardcoded_edid_size = edid_size;
+							} else
+								kfree(edid);
+						}
+					}
+					record += fake_edid_record->ucFakeEDIDLength ?
+						fake_edid_record->ucFakeEDIDLength + 2 :
+						sizeof(ATOM_FAKE_EDID_PATCH_RECORD);
+					break;
+				case LCD_PANEL_RESOLUTION_RECORD_TYPE:
+					panel_res_record = (ATOM_PANEL_RESOLUTION_PATCH_RECORD *)record;
+					lvds->native_mode.width_mm = panel_res_record->usHSize;
+					lvds->native_mode.height_mm = panel_res_record->usVSize;
+					record += sizeof(ATOM_PANEL_RESOLUTION_PATCH_RECORD);
+					break;
+				default:
+					DRM_ERROR("Bad LCD record %d\n", *record);
+					bad_record = true;
+					break;
+				}
+				if (bad_record)
+					break;
+			}
+		}
+	}
+	return lvds;
+}
+
+struct amdgpu_encoder_atom_dig *
+amdgpu_atombios_encoder_get_dig_info(struct amdgpu_encoder *amdgpu_encoder)
+{
+	int encoder_enum = (amdgpu_encoder->encoder_enum & ENUM_ID_MASK) >> ENUM_ID_SHIFT;
+	struct amdgpu_encoder_atom_dig *dig = kzalloc(sizeof(struct amdgpu_encoder_atom_dig), GFP_KERNEL);
+
+	if (!dig)
+		return NULL;
+
+	/* coherent mode by default */
+	dig->coherent_mode = true;
+	dig->dig_encoder = -1;
+
+	if (encoder_enum == 2)
+		dig->linkb = true;
+	else
+		dig->linkb = false;
+
+	return dig;
+}
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios_i2c.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios_i2c.c
new file mode 100644
index 00000000000..f9b2ce9a98f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atombios_i2c.c
@@ -0,0 +1,181 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ *
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+#include "atombios_i2c.h"
+
+#define TARGET_HW_I2C_CLOCK 50
+
+/* these are a limitation of ProcessI2cChannelTransaction not the hw */
+#define ATOM_MAX_HW_I2C_WRITE 3
+#define ATOM_MAX_HW_I2C_READ  255
+
+static int amdgpu_atombios_i2c_process_i2c_ch(struct amdgpu_i2c_chan *chan,
+				       u8 slave_addr, u8 flags,
+				       u8 *buf, u8 num)
+{
+	struct drm_device *dev = chan->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	PROCESS_I2C_CHANNEL_TRANSACTION_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, ProcessI2cChannelTransaction);
+	unsigned char *base;
+	u16 out = cpu_to_le16(0);
+	int r = 0;
+
+	memset(&args, 0, sizeof(args));
+
+	mutex_lock(&chan->mutex);
+
+	base = (unsigned char *)adev->mode_info.atom_context->scratch;
+
+	if (flags & HW_I2C_WRITE) {
+		if (num > ATOM_MAX_HW_I2C_WRITE) {
+			DRM_ERROR("hw i2c: tried to write too many bytes (%d vs 3)\n", num);
+			r = -EINVAL;
+			goto done;
+		}
+		if (buf == NULL)
+			args.ucRegIndex = 0;
+		else
+			args.ucRegIndex = buf[0];
+		if (num)
+			num--;
+		if (num) {
+			if (buf) {
+				memcpy(&out, &buf[1], num);
+			} else {
+				DRM_ERROR("hw i2c: missing buf with num > 1\n");
+				r = -EINVAL;
+				goto done;
+			}
+		}
+		args.lpI2CDataOut = cpu_to_le16(out);
+	} else {
+		if (num > ATOM_MAX_HW_I2C_READ) {
+			DRM_ERROR("hw i2c: tried to read too many bytes (%d vs 255)\n", num);
+			r = -EINVAL;
+			goto done;
+		}
+		args.ucRegIndex = 0;
+		args.lpI2CDataOut = 0;
+	}
+
+	args.ucFlag = flags;
+	args.ucI2CSpeed = TARGET_HW_I2C_CLOCK;
+	args.ucTransBytes = num;
+	args.ucSlaveAddr = slave_addr << 1;
+	args.ucLineNumber = chan->rec.i2c_id;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+	/* error */
+	if (args.ucStatus != HW_ASSISTED_I2C_STATUS_SUCCESS) {
+		DRM_DEBUG_KMS("hw_i2c error\n");
+		r = -EIO;
+		goto done;
+	}
+
+	if (!(flags & HW_I2C_WRITE))
+		amdgpu_atombios_copy_swap(buf, base, num, false);
+
+done:
+	mutex_unlock(&chan->mutex);
+
+	return r;
+}
+
+int amdgpu_atombios_i2c_xfer(struct i2c_adapter *i2c_adap,
+		      struct i2c_msg *msgs, int num)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_get_adapdata(i2c_adap);
+	struct i2c_msg *p;
+	int i, remaining, current_count, buffer_offset, max_bytes, ret;
+	u8 flags;
+
+	/* check for bus probe */
+	p = &msgs[0];
+	if ((num == 1) && (p->len == 0)) {
+		ret = amdgpu_atombios_i2c_process_i2c_ch(i2c,
+						  p->addr, HW_I2C_WRITE,
+						  NULL, 0);
+		if (ret)
+			return ret;
+		else
+			return num;
+	}
+
+	for (i = 0; i < num; i++) {
+		p = &msgs[i];
+		remaining = p->len;
+		buffer_offset = 0;
+		/* max_bytes are a limitation of ProcessI2cChannelTransaction not the hw */
+		if (p->flags & I2C_M_RD) {
+			max_bytes = ATOM_MAX_HW_I2C_READ;
+			flags = HW_I2C_READ;
+		} else {
+			max_bytes = ATOM_MAX_HW_I2C_WRITE;
+			flags = HW_I2C_WRITE;
+		}
+		while (remaining) {
+			if (remaining > max_bytes)
+				current_count = max_bytes;
+			else
+				current_count = remaining;
+			ret = amdgpu_atombios_i2c_process_i2c_ch(i2c,
+							  p->addr, flags,
+							  &p->buf[buffer_offset], current_count);
+			if (ret)
+				return ret;
+			remaining -= current_count;
+			buffer_offset += current_count;
+		}
+	}
+
+	return num;
+}
+
+u32 amdgpu_atombios_i2c_func(struct i2c_adapter *adap)
+{
+	return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL;
+}
+
+void amdgpu_atombios_i2c_channel_trans(struct amdgpu_device* adev, u8 slave_addr, u8 line_number, u8 offset, u8 data)
+{
+	PROCESS_I2C_CHANNEL_TRANSACTION_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, ProcessI2cChannelTransaction);
+
+	args.ucRegIndex = offset;
+	args.lpI2CDataOut = data;
+	args.ucFlag = 1;
+	args.ucI2CSpeed = TARGET_HW_I2C_CLOCK;
+	args.ucTransBytes = 1;
+	args.ucSlaveAddr = slave_addr;
+	args.ucLineNumber = line_number;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_atomfirmware.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atomfirmware.c
new file mode 100644
index 00000000000..236915849cf
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atomfirmware.c
@@ -0,0 +1,368 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "atomfirmware.h"
+#include "amdgpu_atomfirmware.h"
+#include "atom.h"
+#include "atombios.h"
+
+#define get_index_into_master_table(master_table, table_name) (offsetof(struct master_table, table_name) / sizeof(uint16_t))
+
+bool amdgpu_atomfirmware_gpu_supports_virtualization(struct amdgpu_device *adev)
+{
+	int index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+						firmwareinfo);
+	uint16_t data_offset;
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, NULL,
+					  NULL, NULL, &data_offset)) {
+		struct atom_firmware_info_v3_1 *firmware_info =
+			(struct atom_firmware_info_v3_1 *)(adev->mode_info.atom_context->bios +
+							   data_offset);
+
+		if (le32_to_cpu(firmware_info->firmware_capability) &
+		    ATOM_FIRMWARE_CAP_GPU_VIRTUALIZATION)
+			return true;
+	}
+	return false;
+}
+
+void amdgpu_atomfirmware_scratch_regs_init(struct amdgpu_device *adev)
+{
+	int index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+						firmwareinfo);
+	uint16_t data_offset;
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, NULL,
+					  NULL, NULL, &data_offset)) {
+		struct atom_firmware_info_v3_1 *firmware_info =
+			(struct atom_firmware_info_v3_1 *)(adev->mode_info.atom_context->bios +
+							   data_offset);
+
+		adev->bios_scratch_reg_offset =
+			le32_to_cpu(firmware_info->bios_scratch_reg_startaddr);
+	}
+}
+
+int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev)
+{
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	int index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+						vram_usagebyfirmware);
+	struct vram_usagebyfirmware_v2_1 *	firmware_usage;
+	uint32_t start_addr, size;
+	uint16_t data_offset;
+	int usage_bytes = 0;
+
+	if (amdgpu_atom_parse_data_header(ctx, index, NULL, NULL, NULL, &data_offset)) {
+		firmware_usage = (struct vram_usagebyfirmware_v2_1 *)(ctx->bios + data_offset);
+		DRM_DEBUG("atom firmware requested %08x %dkb fw %dkb drv\n",
+			  le32_to_cpu(firmware_usage->start_address_in_kb),
+			  le16_to_cpu(firmware_usage->used_by_firmware_in_kb),
+			  le16_to_cpu(firmware_usage->used_by_driver_in_kb));
+
+		start_addr = le32_to_cpu(firmware_usage->start_address_in_kb);
+		size = le16_to_cpu(firmware_usage->used_by_firmware_in_kb);
+
+		if ((uint32_t)(start_addr & ATOM_VRAM_OPERATION_FLAGS_MASK) ==
+			(uint32_t)(ATOM_VRAM_BLOCK_SRIOV_MSG_SHARE_RESERVATION <<
+			ATOM_VRAM_OPERATION_FLAGS_SHIFT)) {
+			/* Firmware request VRAM reservation for SR-IOV */
+			adev->fw_vram_usage.start_offset = (start_addr &
+				(~ATOM_VRAM_OPERATION_FLAGS_MASK)) << 10;
+			adev->fw_vram_usage.size = size << 10;
+			/* Use the default scratch size */
+			usage_bytes = 0;
+		} else {
+			usage_bytes = le16_to_cpu(firmware_usage->used_by_driver_in_kb) << 10;
+		}
+	}
+	ctx->scratch_size_bytes = 0;
+	if (usage_bytes == 0)
+		usage_bytes = 20 * 1024;
+	/* allocate some scratch memory */
+	ctx->scratch = kzalloc(usage_bytes, GFP_KERNEL);
+	if (!ctx->scratch)
+		return -ENOMEM;
+	ctx->scratch_size_bytes = usage_bytes;
+	return 0;
+}
+
+union igp_info {
+	struct atom_integrated_system_info_v1_11 v11;
+};
+
+union umc_info {
+	struct atom_umc_info_v3_1 v31;
+};
+/*
+ * Return vram width from integrated system info table, if available,
+ * or 0 if not.
+ */
+int amdgpu_atomfirmware_get_vram_width(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+						integratedsysteminfo);
+	u16 data_offset, size;
+	union igp_info *igp_info;
+	u8 frev, crev;
+
+	/* get any igp specific overrides */
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		igp_info = (union igp_info *)
+			(mode_info->atom_context->bios + data_offset);
+		switch (crev) {
+		case 11:
+			return igp_info->v11.umachannelnumber * 64;
+		default:
+			return 0;
+		}
+	}
+
+	return 0;
+}
+
+static int convert_atom_mem_type_to_vram_type (struct amdgpu_device *adev,
+					       int atom_mem_type)
+{
+	int vram_type;
+
+	if (adev->flags & AMD_IS_APU) {
+		switch (atom_mem_type) {
+		case Ddr2MemType:
+		case LpDdr2MemType:
+			vram_type = AMDGPU_VRAM_TYPE_DDR2;
+			break;
+		case Ddr3MemType:
+		case LpDdr3MemType:
+			vram_type = AMDGPU_VRAM_TYPE_DDR3;
+			break;
+		case Ddr4MemType:
+		case LpDdr4MemType:
+			vram_type = AMDGPU_VRAM_TYPE_DDR4;
+			break;
+		default:
+			vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+			break;
+		}
+	} else {
+		switch (atom_mem_type) {
+		case ATOM_DGPU_VRAM_TYPE_GDDR5:
+			vram_type = AMDGPU_VRAM_TYPE_GDDR5;
+			break;
+		case ATOM_DGPU_VRAM_TYPE_HBM:
+			vram_type = AMDGPU_VRAM_TYPE_HBM;
+			break;
+		default:
+			vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+			break;
+		}
+	}
+
+	return vram_type;
+}
+/*
+ * Return vram type from either integrated system info table
+ * or umc info table, if available, or 0 (TYPE_UNKNOWN) if not
+ */
+int amdgpu_atomfirmware_get_vram_type(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index;
+	u16 data_offset, size;
+	union igp_info *igp_info;
+	union umc_info *umc_info;
+	u8 frev, crev;
+	u8 mem_type;
+
+	if (adev->flags & AMD_IS_APU)
+		index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+						    integratedsysteminfo);
+	else
+		index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+						    umc_info);
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context,
+					  index, &size,
+					  &frev, &crev, &data_offset)) {
+		if (adev->flags & AMD_IS_APU) {
+			igp_info = (union igp_info *)
+				(mode_info->atom_context->bios + data_offset);
+			switch (crev) {
+			case 11:
+				mem_type = igp_info->v11.memorytype;
+				return convert_atom_mem_type_to_vram_type(adev, mem_type);
+			default:
+				return 0;
+			}
+		} else {
+			umc_info = (union umc_info *)
+				(mode_info->atom_context->bios + data_offset);
+			switch (crev) {
+			case 1:
+				mem_type = umc_info->v31.vram_type;
+				return convert_atom_mem_type_to_vram_type(adev, mem_type);
+			default:
+				return 0;
+			}
+		}
+	}
+
+	return 0;
+}
+
+union firmware_info {
+	struct atom_firmware_info_v3_1 v31;
+};
+
+union smu_info {
+	struct atom_smu_info_v3_1 v31;
+};
+
+int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	struct amdgpu_pll *spll = &adev->clock.spll;
+	struct amdgpu_pll *mpll = &adev->clock.mpll;
+	uint8_t frev, crev;
+	uint16_t data_offset;
+	int ret = -EINVAL, index;
+
+	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+					    firmwareinfo);
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		union firmware_info *firmware_info =
+			(union firmware_info *)(mode_info->atom_context->bios +
+						data_offset);
+
+		adev->clock.default_sclk =
+			le32_to_cpu(firmware_info->v31.bootup_sclk_in10khz);
+		adev->clock.default_mclk =
+			le32_to_cpu(firmware_info->v31.bootup_mclk_in10khz);
+
+		adev->pm.current_sclk = adev->clock.default_sclk;
+		adev->pm.current_mclk = adev->clock.default_mclk;
+
+		/* not technically a clock, but... */
+		adev->mode_info.firmware_flags =
+			le32_to_cpu(firmware_info->v31.firmware_capability);
+
+		ret = 0;
+	}
+
+	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+					    smu_info);
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		union smu_info *smu_info =
+			(union smu_info *)(mode_info->atom_context->bios +
+					   data_offset);
+
+		/* system clock */
+		spll->reference_freq = le32_to_cpu(smu_info->v31.core_refclk_10khz);
+
+		spll->reference_div = 0;
+		spll->min_post_div = 1;
+		spll->max_post_div = 1;
+		spll->min_ref_div = 2;
+		spll->max_ref_div = 0xff;
+		spll->min_feedback_div = 4;
+		spll->max_feedback_div = 0xff;
+		spll->best_vco = 0;
+
+		ret = 0;
+	}
+
+	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+					    umc_info);
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		union umc_info *umc_info =
+			(union umc_info *)(mode_info->atom_context->bios +
+					   data_offset);
+
+		/* memory clock */
+		mpll->reference_freq = le32_to_cpu(umc_info->v31.mem_refclk_10khz);
+
+		mpll->reference_div = 0;
+		mpll->min_post_div = 1;
+		mpll->max_post_div = 1;
+		mpll->min_ref_div = 2;
+		mpll->max_ref_div = 0xff;
+		mpll->min_feedback_div = 4;
+		mpll->max_feedback_div = 0xff;
+		mpll->best_vco = 0;
+
+		ret = 0;
+	}
+
+	return ret;
+}
+
+union gfx_info {
+	struct  atom_gfx_info_v2_4 v24;
+};
+
+int amdgpu_atomfirmware_get_gfx_info(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index;
+	uint8_t frev, crev;
+	uint16_t data_offset;
+
+	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+					    gfx_info);
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		union gfx_info *gfx_info = (union gfx_info *)
+			(mode_info->atom_context->bios + data_offset);
+		switch (crev) {
+		case 4:
+			adev->gfx.config.max_shader_engines = gfx_info->v24.gc_num_se;
+			adev->gfx.config.max_cu_per_sh = gfx_info->v24.gc_num_cu_per_sh;
+			adev->gfx.config.max_sh_per_se = gfx_info->v24.gc_num_sh_per_se;
+			adev->gfx.config.max_backends_per_se = gfx_info->v24.gc_num_rb_per_se;
+			adev->gfx.config.max_texture_channel_caches = gfx_info->v24.gc_num_tccs;
+			adev->gfx.config.max_gprs = le16_to_cpu(gfx_info->v24.gc_num_gprs);
+			adev->gfx.config.max_gs_threads = gfx_info->v24.gc_num_max_gs_thds;
+			adev->gfx.config.gs_vgt_table_depth = gfx_info->v24.gc_gs_table_depth;
+			adev->gfx.config.gs_prim_buffer_depth =
+				le16_to_cpu(gfx_info->v24.gc_gsprim_buff_depth);
+			adev->gfx.config.double_offchip_lds_buf =
+				gfx_info->v24.gc_double_offchip_lds_buffer;
+			adev->gfx.cu_info.wave_front_size = le16_to_cpu(gfx_info->v24.gc_wave_size);
+			adev->gfx.cu_info.max_waves_per_simd = le16_to_cpu(gfx_info->v24.gc_max_waves_per_simd);
+			adev->gfx.cu_info.max_scratch_slots_per_cu = gfx_info->v24.gc_max_scratch_slots_per_cu;
+			adev->gfx.cu_info.lds_size = le16_to_cpu(gfx_info->v24.gc_lds_size);
+			return 0;
+		default:
+			return -EINVAL;
+		}
+
+	}
+	return -EINVAL;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_atomfirmware.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atomfirmware.h
new file mode 100644
index 00000000000..20f158fd3b7
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_atomfirmware.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_ATOMFIRMWARE_H__
+#define __AMDGPU_ATOMFIRMWARE_H__
+
+bool amdgpu_atomfirmware_gpu_supports_virtualization(struct amdgpu_device *adev);
+void amdgpu_atomfirmware_scratch_regs_init(struct amdgpu_device *adev);
+int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev);
+int amdgpu_atomfirmware_get_vram_width(struct amdgpu_device *adev);
+int amdgpu_atomfirmware_get_vram_type(struct amdgpu_device *adev);
+int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev);
+int amdgpu_atomfirmware_get_gfx_info(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_benchmark.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_benchmark.c
new file mode 100644
index 00000000000..3079ea8523c
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_benchmark.c
@@ -0,0 +1,244 @@
+/*
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Jerome Glisse
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+#define AMDGPU_BENCHMARK_ITERATIONS 1024
+#define AMDGPU_BENCHMARK_COMMON_MODES_N 17
+
+static int amdgpu_benchmark_do_move(struct amdgpu_device *adev, unsigned size,
+				    uint64_t saddr, uint64_t daddr, int n)
+{
+	unsigned long start_jiffies;
+	unsigned long end_jiffies;
+	struct dma_fence *fence = NULL;
+	int i, r;
+
+	start_jiffies = jiffies;
+	for (i = 0; i < n; i++) {
+		struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+		r = amdgpu_copy_buffer(ring, saddr, daddr, size, NULL, &fence,
+				       false, false);
+		if (r)
+			goto exit_do_move;
+		r = dma_fence_wait(fence, false);
+		if (r)
+			goto exit_do_move;
+		dma_fence_put(fence);
+	}
+	end_jiffies = jiffies;
+	r = jiffies_to_msecs(end_jiffies - start_jiffies);
+
+exit_do_move:
+	if (fence)
+		dma_fence_put(fence);
+	return r;
+}
+
+
+static void amdgpu_benchmark_log_results(int n, unsigned size,
+					 unsigned int time,
+					 unsigned sdomain, unsigned ddomain,
+					 char *kind)
+{
+	unsigned int throughput = (n * (size >> 10)) / time;
+	DRM_INFO("amdgpu: %s %u bo moves of %u kB from"
+		 " %d to %d in %u ms, throughput: %u Mb/s or %u MB/s\n",
+		 kind, n, size >> 10, sdomain, ddomain, time,
+		 throughput * 8, throughput);
+}
+
+static void amdgpu_benchmark_move(struct amdgpu_device *adev, unsigned size,
+				  unsigned sdomain, unsigned ddomain)
+{
+	struct amdgpu_bo *dobj = NULL;
+	struct amdgpu_bo *sobj = NULL;
+	struct amdgpu_bo_param bp;
+	uint64_t saddr, daddr;
+	int r, n;
+	int time;
+
+	memset(&bp, 0, sizeof(bp));
+	bp.size = size;
+	bp.byte_align = PAGE_SIZE;
+	bp.domain = sdomain;
+	bp.flags = 0;
+	bp.type = ttm_bo_type_kernel;
+	bp.resv = NULL;
+	n = AMDGPU_BENCHMARK_ITERATIONS;
+	r = amdgpu_bo_create(adev, &bp, &sobj);
+	if (r) {
+		goto out_cleanup;
+	}
+	r = amdgpu_bo_reserve(sobj, false);
+	if (unlikely(r != 0))
+		goto out_cleanup;
+	r = amdgpu_bo_pin(sobj, sdomain);
+	if (r) {
+		amdgpu_bo_unreserve(sobj);
+		goto out_cleanup;
+	}
+	r = amdgpu_ttm_alloc_gart(&sobj->tbo);
+	amdgpu_bo_unreserve(sobj);
+	if (r) {
+		goto out_cleanup;
+	}
+	saddr = amdgpu_bo_gpu_offset(sobj);
+	bp.domain = ddomain;
+	r = amdgpu_bo_create(adev, &bp, &dobj);
+	if (r) {
+		goto out_cleanup;
+	}
+	r = amdgpu_bo_reserve(dobj, false);
+	if (unlikely(r != 0))
+		goto out_cleanup;
+	r = amdgpu_bo_pin(dobj, ddomain);
+	if (r) {
+		amdgpu_bo_unreserve(sobj);
+		goto out_cleanup;
+	}
+	r = amdgpu_ttm_alloc_gart(&dobj->tbo);
+	amdgpu_bo_unreserve(dobj);
+	if (r) {
+		goto out_cleanup;
+	}
+	daddr = amdgpu_bo_gpu_offset(dobj);
+
+	if (adev->mman.buffer_funcs) {
+		time = amdgpu_benchmark_do_move(adev, size, saddr, daddr, n);
+		if (time < 0)
+			goto out_cleanup;
+		if (time > 0)
+			amdgpu_benchmark_log_results(n, size, time,
+						     sdomain, ddomain, "dma");
+	}
+
+out_cleanup:
+	/* Check error value now. The value can be overwritten when clean up.*/
+	if (r) {
+		DRM_ERROR("Error while benchmarking BO move.\n");
+	}
+
+	if (sobj) {
+		r = amdgpu_bo_reserve(sobj, true);
+		if (likely(r == 0)) {
+			amdgpu_bo_unpin(sobj);
+			amdgpu_bo_unreserve(sobj);
+		}
+		amdgpu_bo_unref(&sobj);
+	}
+	if (dobj) {
+		r = amdgpu_bo_reserve(dobj, true);
+		if (likely(r == 0)) {
+			amdgpu_bo_unpin(dobj);
+			amdgpu_bo_unreserve(dobj);
+		}
+		amdgpu_bo_unref(&dobj);
+	}
+}
+
+void amdgpu_benchmark(struct amdgpu_device *adev, int test_number)
+{
+	int i;
+	static const int common_modes[AMDGPU_BENCHMARK_COMMON_MODES_N] = {
+		640 * 480 * 4,
+		720 * 480 * 4,
+		800 * 600 * 4,
+		848 * 480 * 4,
+		1024 * 768 * 4,
+		1152 * 768 * 4,
+		1280 * 720 * 4,
+		1280 * 800 * 4,
+		1280 * 854 * 4,
+		1280 * 960 * 4,
+		1280 * 1024 * 4,
+		1440 * 900 * 4,
+		1400 * 1050 * 4,
+		1680 * 1050 * 4,
+		1600 * 1200 * 4,
+		1920 * 1080 * 4,
+		1920 * 1200 * 4
+	};
+
+	switch (test_number) {
+	case 1:
+		/* simple test, VRAM to GTT and GTT to VRAM */
+		amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_GTT,
+				      AMDGPU_GEM_DOMAIN_VRAM);
+		amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_VRAM,
+				      AMDGPU_GEM_DOMAIN_GTT);
+		break;
+	case 2:
+		/* simple test, VRAM to VRAM */
+		amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_VRAM,
+				      AMDGPU_GEM_DOMAIN_VRAM);
+		break;
+	case 3:
+		/* GTT to VRAM, buffer size sweep, powers of 2 */
+		for (i = 1; i <= 16384; i <<= 1)
+			amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_GTT,
+					      AMDGPU_GEM_DOMAIN_VRAM);
+		break;
+	case 4:
+		/* VRAM to GTT, buffer size sweep, powers of 2 */
+		for (i = 1; i <= 16384; i <<= 1)
+			amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      AMDGPU_GEM_DOMAIN_GTT);
+		break;
+	case 5:
+		/* VRAM to VRAM, buffer size sweep, powers of 2 */
+		for (i = 1; i <= 16384; i <<= 1)
+			amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      AMDGPU_GEM_DOMAIN_VRAM);
+		break;
+	case 6:
+		/* GTT to VRAM, buffer size sweep, common modes */
+		for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++)
+			amdgpu_benchmark_move(adev, common_modes[i],
+					      AMDGPU_GEM_DOMAIN_GTT,
+					      AMDGPU_GEM_DOMAIN_VRAM);
+		break;
+	case 7:
+		/* VRAM to GTT, buffer size sweep, common modes */
+		for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++)
+			amdgpu_benchmark_move(adev, common_modes[i],
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      AMDGPU_GEM_DOMAIN_GTT);
+		break;
+	case 8:
+		/* VRAM to VRAM, buffer size sweep, common modes */
+		for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++)
+			amdgpu_benchmark_move(adev, common_modes[i],
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      AMDGPU_GEM_DOMAIN_VRAM);
+		break;
+
+	default:
+		DRM_ERROR("Unknown benchmark\n");
+	}
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_bios.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_bios.c
new file mode 100644
index 00000000000..0dfe4041015
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_bios.c
@@ -0,0 +1,577 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "atom.h"
+
+#include <linux/slab.h>
+#include <linux/acpi.h>
+
+#if defined(__amd64__) || defined(__i386__)
+#include <dev/isa/isareg.h>
+#include <dev/isa/isavar.h>
+#endif
+
+#ifdef __HAVE_ACPI
+#include "acpi.h"
+#endif
+
+/*
+ * BIOS.
+ */
+
+#define AMD_VBIOS_SIGNATURE " 761295520"
+#define AMD_VBIOS_SIGNATURE_OFFSET 0x30
+#define AMD_VBIOS_SIGNATURE_SIZE sizeof(AMD_VBIOS_SIGNATURE)
+#define AMD_VBIOS_SIGNATURE_END (AMD_VBIOS_SIGNATURE_OFFSET + AMD_VBIOS_SIGNATURE_SIZE)
+#define AMD_IS_VALID_VBIOS(p) ((p)[0] == 0x55 && (p)[1] == 0xAA)
+#define AMD_VBIOS_LENGTH(p) ((p)[2] << 9)
+
+/* Check if current bios is an ATOM BIOS.
+ * Return true if it is ATOM BIOS. Otherwise, return false.
+ */
+static bool check_atom_bios(uint8_t *bios, size_t size)
+{
+	uint16_t tmp, bios_header_start;
+
+	if (!bios || size < 0x49) {
+		DRM_INFO("vbios mem is null or mem size is wrong\n");
+		return false;
+	}
+
+	if (!AMD_IS_VALID_VBIOS(bios)) {
+		DRM_INFO("BIOS signature incorrect %x %x\n", bios[0], bios[1]);
+		return false;
+	}
+
+	bios_header_start = bios[0x48] | (bios[0x49] << 8);
+	if (!bios_header_start) {
+		DRM_INFO("Can't locate bios header\n");
+		return false;
+	}
+
+	tmp = bios_header_start + 4;
+	if (size < tmp) {
+		DRM_INFO("BIOS header is broken\n");
+		return false;
+	}
+
+	if (!memcmp(bios + tmp, "ATOM", 4) ||
+	    !memcmp(bios + tmp, "MOTA", 4)) {
+		DRM_DEBUG("ATOMBIOS detected\n");
+		return true;
+	}
+
+	return false;
+}
+
+/* If you boot an IGP board with a discrete card as the primary,
+ * the IGP rom is not accessible via the rom bar as the IGP rom is
+ * part of the system bios.  On boot, the system bios puts a
+ * copy of the igp rom at the start of vram if a discrete card is
+ * present.
+ */
+#ifdef __linux__
+static bool igp_read_bios_from_vram(struct amdgpu_device *adev)
+{
+	uint8_t __iomem *bios;
+	resource_size_t vram_base;
+	resource_size_t size = 256 * 1024; /* ??? */
+
+	if (!(adev->flags & AMD_IS_APU))
+		if (amdgpu_device_need_post(adev))
+			return false;
+
+	adev->bios = NULL;
+	vram_base = pci_resource_start(adev->pdev, 0);
+	bios = ioremap_wc(vram_base, size);
+	if (!bios) {
+		return false;
+	}
+
+	adev->bios = kmalloc(size, GFP_KERNEL);
+	if (!adev->bios) {
+		iounmap(bios);
+		return false;
+	}
+	adev->bios_size = size;
+	memcpy_fromio(adev->bios, bios, size);
+	iounmap(bios);
+
+	if (!check_atom_bios(adev->bios, size)) {
+		kfree(adev->bios);
+		return false;
+	}
+
+	return true;
+}
+#else
+static bool igp_read_bios_from_vram(struct amdgpu_device *adev)
+{
+	uint8_t __iomem *bios;
+	resource_size_t size = 256 * 1024; /* ??? */
+	bus_space_handle_t bsh;
+	bus_space_tag_t bst = adev->memt;
+
+	if (!(adev->flags & AMD_IS_APU))
+		if (amdgpu_device_need_post(adev))
+			return false;
+
+	adev->bios = NULL;
+
+	if (bus_space_map(bst, adev->fb_aper_offset, size, BUS_SPACE_MAP_LINEAR, &bsh) != 0)
+		return false;
+
+	bios = bus_space_vaddr(adev->memt, bsh);
+	if (bios == NULL) {
+		bus_space_unmap(bst, bsh, size);
+		return false;
+	}
+
+	adev->bios = kmalloc(size, GFP_KERNEL);
+	if (!adev->bios) {
+		bus_space_unmap(bst, bsh, size);
+		return false;
+	}
+	adev->bios_size = size;
+	memcpy_fromio(adev->bios, bios, size);
+	bus_space_unmap(bst, bsh, size);
+
+	if (!check_atom_bios(adev->bios, size)) {
+		kfree(adev->bios);
+		return false;
+	}
+
+	return true;
+}
+#endif
+
+#ifdef __linux__
+bool amdgpu_read_bios(struct amdgpu_device *adev)
+{
+	uint8_t __iomem *bios;
+	size_t size;
+
+	adev->bios = NULL;
+	/* XXX: some cards may return 0 for rom size? ddx has a workaround */
+	bios = pci_map_rom(adev->pdev, &size);
+	if (!bios) {
+		return false;
+	}
+
+	adev->bios = kzalloc(size, GFP_KERNEL);
+	if (adev->bios == NULL) {
+		pci_unmap_rom(adev->pdev, bios);
+		return false;
+	}
+	adev->bios_size = size;
+	memcpy_fromio(adev->bios, bios, size);
+	pci_unmap_rom(adev->pdev, bios);
+
+	if (!check_atom_bios(adev->bios, size)) {
+		kfree(adev->bios);
+		return false;
+	}
+
+	return true;
+}
+#else
+bool amdgpu_read_bios(struct amdgpu_device *adev)
+{
+	uint8_t __iomem *bios;
+	size_t size;
+	pcireg_t address, mask;
+	bus_space_handle_t romh;
+	int rc;
+
+	adev->bios = NULL;
+	/* XXX: some cards may return 0 for rom size? ddx has a workaround */
+
+	address = pci_conf_read(adev->pc, adev->pa_tag, PCI_ROM_REG);
+	pci_conf_write(adev->pc, adev->pa_tag, PCI_ROM_REG, ~PCI_ROM_ENABLE);
+	mask = pci_conf_read(adev->pc, adev->pa_tag, PCI_ROM_REG);
+	address |= PCI_ROM_ENABLE;
+	pci_conf_write(adev->pc, adev->pa_tag, PCI_ROM_REG, address);
+
+	size = PCI_ROM_SIZE(mask);
+	if (size == 0)
+		return false;
+	rc = bus_space_map(adev->memt, PCI_ROM_ADDR(address), size,
+	    BUS_SPACE_MAP_LINEAR, &romh);
+	if (rc != 0) {
+		printf(": can't map PCI ROM (%d)\n", rc);
+		return false;
+	}
+	bios = (uint8_t *)bus_space_vaddr(adev->memt, romh);
+	if (!bios) {
+		printf(": bus_space_vaddr failed\n");
+		return false;
+	}
+
+	adev->bios = kzalloc(size, GFP_KERNEL);
+	if (adev->bios == NULL) {
+		bus_space_unmap(adev->memt, romh, size);
+		return false;
+	}
+	adev->bios_size = size;
+	memcpy_fromio(adev->bios, bios, size);
+	bus_space_unmap(adev->memt, romh, size);
+
+	if (!check_atom_bios(adev->bios, size)) {
+		kfree(adev->bios);
+		return false;
+	}
+
+	return true;
+}
+#endif
+
+static bool amdgpu_read_bios_from_rom(struct amdgpu_device *adev)
+{
+	u8 header[AMD_VBIOS_SIGNATURE_END+1] = {0};
+	int len;
+
+	if (!adev->asic_funcs->read_bios_from_rom)
+		return false;
+
+	/* validate VBIOS signature */
+	if (amdgpu_asic_read_bios_from_rom(adev, &header[0], sizeof(header)) == false)
+		return false;
+	header[AMD_VBIOS_SIGNATURE_END] = 0;
+
+	if ((!AMD_IS_VALID_VBIOS(header)) ||
+	    0 != memcmp((char *)&header[AMD_VBIOS_SIGNATURE_OFFSET],
+			AMD_VBIOS_SIGNATURE,
+			strlen(AMD_VBIOS_SIGNATURE)))
+		return false;
+
+	/* valid vbios, go on */
+	len = AMD_VBIOS_LENGTH(header);
+	len = roundup2(len, 4);
+	adev->bios = kmalloc(len, GFP_KERNEL);
+	if (!adev->bios) {
+		DRM_ERROR("no memory to allocate for BIOS\n");
+		return false;
+	}
+	adev->bios_size = len;
+
+	/* read complete BIOS */
+	amdgpu_asic_read_bios_from_rom(adev, adev->bios, len);
+
+	if (!check_atom_bios(adev->bios, len)) {
+		kfree(adev->bios);
+		return false;
+	}
+
+	return true;
+}
+
+#ifdef __linux__
+static bool amdgpu_read_platform_bios(struct amdgpu_device *adev)
+{
+	uint8_t __iomem *bios;
+	size_t size;
+
+	adev->bios = NULL;
+
+	bios = pci_platform_rom(adev->pdev, &size);
+	if (!bios) {
+		return false;
+	}
+
+	adev->bios = kzalloc(size, GFP_KERNEL);
+	if (adev->bios == NULL)
+		return false;
+
+	memcpy_fromio(adev->bios, bios, size);
+
+	if (!check_atom_bios(adev->bios, size)) {
+		kfree(adev->bios);
+		return false;
+	}
+
+	adev->bios_size = size;
+
+	return true;
+}
+#else
+static bool amdgpu_read_platform_bios(struct amdgpu_device *adev)
+{
+#if defined(__amd64__) || defined(__i386__)
+	uint8_t __iomem *bios;
+	bus_size_t size = 256 * 1024; /* ??? */
+
+	adev->bios = NULL;
+
+	bios = (u8 *)ISA_HOLE_VADDR(0xc0000);
+
+	adev->bios = kzalloc(size, GFP_KERNEL);
+	if (adev->bios == NULL)
+		return false;
+
+	memcpy_fromio(adev->bios, bios, size);
+
+	if (!check_atom_bios(adev->bios, size)) {
+		kfree(adev->bios);
+		return false;
+	}
+
+	adev->bios_size = size;
+
+	return true;
+#endif
+	return false;
+}
+#endif
+
+#ifdef CONFIG_ACPI
+/* ATRM is used to get the BIOS on the discrete cards in
+ * dual-gpu systems.
+ */
+/* retrieve the ROM in 4k blocks */
+#define ATRM_BIOS_PAGE 4096
+/**
+ * amdgpu_atrm_call - fetch a chunk of the vbios
+ *
+ * @atrm_handle: acpi ATRM handle
+ * @bios: vbios image pointer
+ * @offset: offset of vbios image data to fetch
+ * @len: length of vbios image data to fetch
+ *
+ * Executes ATRM to fetch a chunk of the discrete
+ * vbios image on PX systems (all asics).
+ * Returns the length of the buffer fetched.
+ */
+static int amdgpu_atrm_call(acpi_handle atrm_handle, uint8_t *bios,
+			    int offset, int len)
+{
+	acpi_status status;
+	union acpi_object atrm_arg_elements[2], *obj;
+	struct acpi_object_list atrm_arg;
+	struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL};
+
+	atrm_arg.count = 2;
+	atrm_arg.pointer = &atrm_arg_elements[0];
+
+	atrm_arg_elements[0].type = ACPI_TYPE_INTEGER;
+	atrm_arg_elements[0].integer.value = offset;
+
+	atrm_arg_elements[1].type = ACPI_TYPE_INTEGER;
+	atrm_arg_elements[1].integer.value = len;
+
+	status = acpi_evaluate_object(atrm_handle, NULL, &atrm_arg, &buffer);
+	if (ACPI_FAILURE(status)) {
+		printk("failed to evaluate ATRM got %s\n", acpi_format_exception(status));
+		return -ENODEV;
+	}
+
+	obj = (union acpi_object *)buffer.pointer;
+	memcpy(bios+offset, obj->buffer.pointer, obj->buffer.length);
+	len = obj->buffer.length;
+	kfree(buffer.pointer);
+	return len;
+}
+
+static bool amdgpu_atrm_get_bios(struct amdgpu_device *adev)
+{
+	int ret;
+	int size = 256 * 1024;
+	int i;
+	struct pci_dev *pdev = NULL;
+	acpi_handle dhandle, atrm_handle;
+	acpi_status status;
+	bool found = false;
+
+	/* ATRM is for the discrete card only */
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, pdev)) != NULL) {
+		dhandle = ACPI_HANDLE(&pdev->dev);
+		if (!dhandle)
+			continue;
+
+		status = acpi_get_handle(dhandle, "ATRM", &atrm_handle);
+		if (!ACPI_FAILURE(status)) {
+			found = true;
+			break;
+		}
+	}
+
+	if (!found) {
+		while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_OTHER << 8, pdev)) != NULL) {
+			dhandle = ACPI_HANDLE(&pdev->dev);
+			if (!dhandle)
+				continue;
+
+			status = acpi_get_handle(dhandle, "ATRM", &atrm_handle);
+			if (!ACPI_FAILURE(status)) {
+				found = true;
+				break;
+			}
+		}
+	}
+
+	if (!found)
+		return false;
+
+	adev->bios = kmalloc(size, GFP_KERNEL);
+	if (!adev->bios) {
+		DRM_ERROR("Unable to allocate bios\n");
+		return false;
+	}
+
+	for (i = 0; i < size / ATRM_BIOS_PAGE; i++) {
+		ret = amdgpu_atrm_call(atrm_handle,
+				       adev->bios,
+				       (i * ATRM_BIOS_PAGE),
+				       ATRM_BIOS_PAGE);
+		if (ret < ATRM_BIOS_PAGE)
+			break;
+	}
+
+	if (!check_atom_bios(adev->bios, size)) {
+		kfree(adev->bios);
+		return false;
+	}
+	adev->bios_size = size;
+	return true;
+}
+#else
+static inline bool amdgpu_atrm_get_bios(struct amdgpu_device *adev)
+{
+	return false;
+}
+#endif
+
+static bool amdgpu_read_disabled_bios(struct amdgpu_device *adev)
+{
+	if (adev->flags & AMD_IS_APU)
+		return igp_read_bios_from_vram(adev);
+	else
+		return amdgpu_asic_read_disabled_bios(adev);
+}
+
+#if NACPI > 0
+#define CONFIG_ACPI
+#endif
+
+#ifdef CONFIG_ACPI
+static bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev)
+{
+	struct acpi_table_header *hdr;
+	acpi_size tbl_size;
+	UEFI_ACPI_VFCT *vfct;
+	unsigned offset;
+
+	if (!ACPI_SUCCESS(acpi_get_table("VFCT", 1, &hdr)))
+		return false;
+	tbl_size = hdr->length;
+	if (tbl_size < sizeof(UEFI_ACPI_VFCT)) {
+		DRM_ERROR("ACPI VFCT table present but broken (too short #1)\n");
+		return false;
+	}
+
+	vfct = (UEFI_ACPI_VFCT *)hdr;
+	offset = vfct->VBIOSImageOffset;
+
+	while (offset < tbl_size) {
+		GOP_VBIOS_CONTENT *vbios = (GOP_VBIOS_CONTENT *)((char *)hdr + offset);
+		VFCT_IMAGE_HEADER *vhdr = &vbios->VbiosHeader;
+
+		offset += sizeof(VFCT_IMAGE_HEADER);
+		if (offset > tbl_size) {
+			DRM_ERROR("ACPI VFCT image header truncated\n");
+			return false;
+		}
+
+		offset += vhdr->ImageLength;
+		if (offset > tbl_size) {
+			DRM_ERROR("ACPI VFCT image truncated\n");
+			return false;
+		}
+
+		if (vhdr->ImageLength &&
+		    vhdr->PCIBus == adev->pdev->bus->number &&
+		    vhdr->PCIDevice == PCI_SLOT(adev->pdev->devfn) &&
+		    vhdr->PCIFunction == PCI_FUNC(adev->pdev->devfn) &&
+		    vhdr->VendorID == adev->pdev->vendor &&
+		    vhdr->DeviceID == adev->pdev->device) {
+			adev->bios = kmemdup(&vbios->VbiosContent,
+					     vhdr->ImageLength,
+					     GFP_KERNEL);
+
+			if (!check_atom_bios(adev->bios, vhdr->ImageLength)) {
+				kfree(adev->bios);
+				return false;
+			}
+			adev->bios_size = vhdr->ImageLength;
+			return true;
+		}
+	}
+
+	DRM_ERROR("ACPI VFCT table present but broken (too short #2)\n");
+	return false;
+}
+#else
+static inline bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev)
+{
+	return false;
+}
+#endif
+
+bool amdgpu_get_bios(struct amdgpu_device *adev)
+{
+	if (amdgpu_atrm_get_bios(adev))
+		goto success;
+
+	if (amdgpu_acpi_vfct_bios(adev))
+		goto success;
+
+	if (igp_read_bios_from_vram(adev))
+		goto success;
+
+	if (amdgpu_read_bios(adev))
+		goto success;
+
+	if (amdgpu_read_bios_from_rom(adev))
+		goto success;
+
+	if (amdgpu_read_disabled_bios(adev))
+		goto success;
+
+	if (amdgpu_read_platform_bios(adev))
+		goto success;
+
+	DRM_ERROR("Unable to locate a BIOS ROM\n");
+	return false;
+
+success:
+	adev->is_atom_fw = (adev->asic_type >= CHIP_VEGA10) ? true : false;
+	return true;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_bo_list.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_bo_list.c
new file mode 100644
index 00000000000..ed081d608d2
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_bo_list.c
@@ -0,0 +1,330 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Christian König <deathsimple@vodafone.de>
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+#define AMDGPU_BO_LIST_MAX_PRIORITY	32u
+#define AMDGPU_BO_LIST_NUM_BUCKETS	(AMDGPU_BO_LIST_MAX_PRIORITY + 1)
+
+static void amdgpu_bo_list_free_rcu(struct rcu_head *rcu)
+{
+	struct amdgpu_bo_list *list = container_of(rcu, struct amdgpu_bo_list,
+						   rhead);
+
+	kvfree(list);
+}
+
+static void amdgpu_bo_list_free(struct kref *ref)
+{
+	struct amdgpu_bo_list *list = container_of(ref, struct amdgpu_bo_list,
+						   refcount);
+	struct amdgpu_bo_list_entry *e;
+
+	amdgpu_bo_list_for_each_entry(e, list)
+		amdgpu_bo_unref(&e->robj);
+
+	call_rcu(&list->rhead, amdgpu_bo_list_free_rcu);
+}
+
+int amdgpu_bo_list_create(struct amdgpu_device *adev, struct drm_file *filp,
+			  struct drm_amdgpu_bo_list_entry *info,
+			  unsigned num_entries, struct amdgpu_bo_list **result)
+{
+	STUB();
+	unsigned last_entry = 0, first_userptr = num_entries;
+	struct amdgpu_bo_list_entry *array;
+	struct amdgpu_bo_list *list;
+	uint64_t total_size = 0;
+	size_t size;
+	unsigned i;
+	int r;
+
+	if (num_entries > (SIZE_MAX - sizeof(struct amdgpu_bo_list))
+				/ sizeof(struct amdgpu_bo_list_entry))
+		return -EINVAL;
+
+	size = sizeof(struct amdgpu_bo_list);
+	size += num_entries * sizeof(struct amdgpu_bo_list_entry);
+	list = kvmalloc(size, GFP_KERNEL);
+	if (!list)
+		return -ENOMEM;
+
+	kref_init(&list->refcount);
+	list->gds_obj = adev->gds.gds_gfx_bo;
+	list->gws_obj = adev->gds.gws_gfx_bo;
+	list->oa_obj = adev->gds.oa_gfx_bo;
+
+	array = amdgpu_bo_list_array_entry(list, 0);
+	memset(array, 0, num_entries * sizeof(struct amdgpu_bo_list_entry));
+
+	for (i = 0; i < num_entries; ++i) {
+		struct amdgpu_bo_list_entry *entry;
+		struct drm_gem_object *gobj;
+		struct amdgpu_bo *bo;
+		struct mm_struct *usermm;
+
+		gobj = drm_gem_object_lookup(filp, info[i].bo_handle);
+		if (!gobj) {
+			r = -ENOENT;
+			goto error_free;
+		}
+
+		bo = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
+		drm_gem_object_put_unlocked(gobj);
+
+#ifdef notyet
+		usermm = amdgpu_ttm_tt_get_usermm(bo->tbo.ttm);
+		if (usermm) {
+			if (usermm != current->mm) {
+				amdgpu_bo_unref(&bo);
+				r = -EPERM;
+				goto error_free;
+			}
+			entry = &array[--first_userptr];
+		} else {
+			entry = &array[last_entry++];
+		}
+#else
+			entry = &array[last_entry++];
+#endif
+
+		entry->robj = bo;
+		entry->priority = min(info[i].bo_priority,
+				      AMDGPU_BO_LIST_MAX_PRIORITY);
+		entry->tv.bo = &entry->robj->tbo;
+		entry->tv.shared = !entry->robj->prime_shared_count;
+
+		if (entry->robj->preferred_domains == AMDGPU_GEM_DOMAIN_GDS)
+			list->gds_obj = entry->robj;
+		if (entry->robj->preferred_domains == AMDGPU_GEM_DOMAIN_GWS)
+			list->gws_obj = entry->robj;
+		if (entry->robj->preferred_domains == AMDGPU_GEM_DOMAIN_OA)
+			list->oa_obj = entry->robj;
+
+		total_size += amdgpu_bo_size(entry->robj);
+		trace_amdgpu_bo_list_set(list, entry->robj);
+	}
+
+	list->first_userptr = first_userptr;
+	list->num_entries = num_entries;
+
+	trace_amdgpu_cs_bo_status(list->num_entries, total_size);
+
+	*result = list;
+	return 0;
+
+error_free:
+	while (i--)
+		amdgpu_bo_unref(&array[i].robj);
+	kvfree(list);
+	return r;
+}
+
+static void amdgpu_bo_list_destroy(struct amdgpu_fpriv *fpriv, int id)
+{
+	struct amdgpu_bo_list *list;
+
+	mutex_lock(&fpriv->bo_list_lock);
+	list = idr_remove(&fpriv->bo_list_handles, id);
+	mutex_unlock(&fpriv->bo_list_lock);
+	if (list)
+		kref_put(&list->refcount, amdgpu_bo_list_free);
+}
+
+int amdgpu_bo_list_get(struct amdgpu_fpriv *fpriv, int id,
+		       struct amdgpu_bo_list **result)
+{
+	rcu_read_lock();
+	*result = idr_find(&fpriv->bo_list_handles, id);
+
+	if (*result && kref_get_unless_zero(&(*result)->refcount)) {
+		rcu_read_unlock();
+		return 0;
+	}
+
+	rcu_read_unlock();
+	return -ENOENT;
+}
+
+void amdgpu_bo_list_get_list(struct amdgpu_bo_list *list,
+			     struct list_head *validated)
+{
+	/* This is based on the bucket sort with O(n) time complexity.
+	 * An item with priority "i" is added to bucket[i]. The lists are then
+	 * concatenated in descending order.
+	 */
+	struct list_head bucket[AMDGPU_BO_LIST_NUM_BUCKETS];
+	struct amdgpu_bo_list_entry *e;
+	unsigned i;
+
+	for (i = 0; i < AMDGPU_BO_LIST_NUM_BUCKETS; i++)
+		INIT_LIST_HEAD(&bucket[i]);
+
+	/* Since buffers which appear sooner in the relocation list are
+	 * likely to be used more often than buffers which appear later
+	 * in the list, the sort mustn't change the ordering of buffers
+	 * with the same priority, i.e. it must be stable.
+	 */
+	amdgpu_bo_list_for_each_entry(e, list) {
+		unsigned priority = e->priority;
+
+		if (!e->robj->parent)
+			list_add_tail(&e->tv.head, &bucket[priority]);
+
+		e->user_pages = NULL;
+	}
+
+	/* Connect the sorted buckets in the output list. */
+	for (i = 0; i < AMDGPU_BO_LIST_NUM_BUCKETS; i++)
+		list_splice(&bucket[i], validated);
+}
+
+void amdgpu_bo_list_put(struct amdgpu_bo_list *list)
+{
+	kref_put(&list->refcount, amdgpu_bo_list_free);
+}
+
+int amdgpu_bo_create_list_entry_array(struct drm_amdgpu_bo_list_in *in,
+				      struct drm_amdgpu_bo_list_entry **info_param)
+{
+	const void __user *uptr = u64_to_user_ptr(in->bo_info_ptr);
+	const uint32_t info_size = sizeof(struct drm_amdgpu_bo_list_entry);
+	struct drm_amdgpu_bo_list_entry *info;
+	int r;
+
+	info = kvmalloc_array(in->bo_number, info_size, GFP_KERNEL);
+	if (!info)
+		return -ENOMEM;
+
+	/* copy the handle array from userspace to a kernel buffer */
+	r = -EFAULT;
+	if (likely(info_size == in->bo_info_size)) {
+		unsigned long bytes = in->bo_number *
+			in->bo_info_size;
+
+		if (copy_from_user(info, uptr, bytes))
+			goto error_free;
+
+	} else {
+		unsigned long bytes = min(in->bo_info_size, info_size);
+		unsigned i;
+
+		memset(info, 0, in->bo_number * info_size);
+		for (i = 0; i < in->bo_number; ++i) {
+			if (copy_from_user(&info[i], uptr, bytes))
+				goto error_free;
+
+			uptr += in->bo_info_size;
+		}
+	}
+
+	*info_param = info;
+	return 0;
+
+error_free:
+	kvfree(info);
+	return r;
+}
+
+int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	union drm_amdgpu_bo_list *args = data;
+	uint32_t handle = args->in.list_handle;
+	struct drm_amdgpu_bo_list_entry *info = NULL;
+	struct amdgpu_bo_list *list, *old;
+	int r;
+
+	r = amdgpu_bo_create_list_entry_array(&args->in, &info);
+	if (r)
+		goto error_free;
+
+	switch (args->in.operation) {
+	case AMDGPU_BO_LIST_OP_CREATE:
+		r = amdgpu_bo_list_create(adev, filp, info, args->in.bo_number,
+					  &list);
+		if (r)
+			goto error_free;
+
+		mutex_lock(&fpriv->bo_list_lock);
+		r = idr_alloc(&fpriv->bo_list_handles, list, 1, 0, GFP_KERNEL);
+		mutex_unlock(&fpriv->bo_list_lock);
+		if (r < 0) {
+			amdgpu_bo_list_put(list);
+			return r;
+		}
+
+		handle = r;
+		break;
+
+	case AMDGPU_BO_LIST_OP_DESTROY:
+		amdgpu_bo_list_destroy(fpriv, handle);
+		handle = 0;
+		break;
+
+	case AMDGPU_BO_LIST_OP_UPDATE:
+		r = amdgpu_bo_list_create(adev, filp, info, args->in.bo_number,
+					  &list);
+		if (r)
+			goto error_free;
+
+		mutex_lock(&fpriv->bo_list_lock);
+		old = idr_replace(&fpriv->bo_list_handles, list, handle);
+		mutex_unlock(&fpriv->bo_list_lock);
+
+		if (IS_ERR(old)) {
+			amdgpu_bo_list_put(list);
+			r = PTR_ERR(old);
+			goto error_free;
+		}
+
+		amdgpu_bo_list_put(old);
+		break;
+
+	default:
+		r = -EINVAL;
+		goto error_free;
+	}
+
+	memset(args, 0, sizeof(*args));
+	args->out.list_handle = handle;
+	kvfree(info);
+
+	return 0;
+
+error_free:
+	if (info)
+		kvfree(info);
+	return r;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_bo_list.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_bo_list.h
new file mode 100644
index 00000000000..0c126c3822b
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_bo_list.h
@@ -0,0 +1,85 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_BO_LIST_H__
+#define __AMDGPU_BO_LIST_H__
+
+#include <drm/ttm/ttm_execbuf_util.h>
+#include <drm/amdgpu_drm.h>
+
+struct amdgpu_device;
+struct amdgpu_bo;
+struct amdgpu_bo_va;
+struct amdgpu_fpriv;
+
+struct amdgpu_bo_list_entry {
+	struct amdgpu_bo		*robj;
+	struct ttm_validate_buffer	tv;
+	struct amdgpu_bo_va		*bo_va;
+	uint32_t			priority;
+	struct vm_page			**user_pages;
+	int				user_invalidated;
+};
+
+struct amdgpu_bo_list {
+	struct rcu_head rhead;
+	struct kref refcount;
+	struct amdgpu_bo *gds_obj;
+	struct amdgpu_bo *gws_obj;
+	struct amdgpu_bo *oa_obj;
+	unsigned first_userptr;
+	unsigned num_entries;
+};
+
+int amdgpu_bo_list_get(struct amdgpu_fpriv *fpriv, int id,
+		       struct amdgpu_bo_list **result);
+void amdgpu_bo_list_get_list(struct amdgpu_bo_list *list,
+			     struct list_head *validated);
+void amdgpu_bo_list_put(struct amdgpu_bo_list *list);
+int amdgpu_bo_create_list_entry_array(struct drm_amdgpu_bo_list_in *in,
+				      struct drm_amdgpu_bo_list_entry **info_param);
+
+int amdgpu_bo_list_create(struct amdgpu_device *adev,
+				 struct drm_file *filp,
+				 struct drm_amdgpu_bo_list_entry *info,
+				 unsigned num_entries,
+				 struct amdgpu_bo_list **list);
+
+static inline struct amdgpu_bo_list_entry *
+amdgpu_bo_list_array_entry(struct amdgpu_bo_list *list, unsigned index)
+{
+	struct amdgpu_bo_list_entry *array = (void *)&list[1];
+
+	return &array[index];
+}
+
+#define amdgpu_bo_list_for_each_entry(e, list) \
+	for (e = amdgpu_bo_list_array_entry(list, 0); \
+	     e != amdgpu_bo_list_array_entry(list, (list)->num_entries); \
+	     ++e)
+
+#define amdgpu_bo_list_for_each_userptr_entry(e, list) \
+	for (e = amdgpu_bo_list_array_entry(list, (list)->first_userptr); \
+	     e != amdgpu_bo_list_array_entry(list, (list)->num_entries); \
+	     ++e)
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_cgs.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_cgs.c
new file mode 100644
index 00000000000..aa0bddc495e
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_cgs.c
@@ -0,0 +1,504 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ */
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <drm/drmP.h>
+#include <linux/firmware.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "atom.h"
+#include "amdgpu_ucode.h"
+
+struct amdgpu_cgs_device {
+	struct cgs_device base;
+	struct amdgpu_device *adev;
+};
+
+#define CGS_FUNC_ADEV							\
+	struct amdgpu_device *adev =					\
+		((struct amdgpu_cgs_device *)cgs_device)->adev
+
+
+static uint32_t amdgpu_cgs_read_register(struct cgs_device *cgs_device, unsigned offset)
+{
+	CGS_FUNC_ADEV;
+	return RREG32(offset);
+}
+
+static void amdgpu_cgs_write_register(struct cgs_device *cgs_device, unsigned offset,
+				      uint32_t value)
+{
+	CGS_FUNC_ADEV;
+	WREG32(offset, value);
+}
+
+static uint32_t amdgpu_cgs_read_ind_register(struct cgs_device *cgs_device,
+					     enum cgs_ind_reg space,
+					     unsigned index)
+{
+	CGS_FUNC_ADEV;
+	switch (space) {
+	case CGS_IND_REG__MMIO:
+		return RREG32_IDX(index);
+	case CGS_IND_REG__PCIE:
+		return RREG32_PCIE(index);
+	case CGS_IND_REG__SMC:
+		return RREG32_SMC(index);
+	case CGS_IND_REG__UVD_CTX:
+		return RREG32_UVD_CTX(index);
+	case CGS_IND_REG__DIDT:
+		return RREG32_DIDT(index);
+	case CGS_IND_REG_GC_CAC:
+		return RREG32_GC_CAC(index);
+	case CGS_IND_REG_SE_CAC:
+		return RREG32_SE_CAC(index);
+	case CGS_IND_REG__AUDIO_ENDPT:
+		DRM_ERROR("audio endpt register access not implemented.\n");
+		return 0;
+	}
+	WARN(1, "Invalid indirect register space");
+	return 0;
+}
+
+static void amdgpu_cgs_write_ind_register(struct cgs_device *cgs_device,
+					  enum cgs_ind_reg space,
+					  unsigned index, uint32_t value)
+{
+	CGS_FUNC_ADEV;
+	switch (space) {
+	case CGS_IND_REG__MMIO:
+		return WREG32_IDX(index, value);
+	case CGS_IND_REG__PCIE:
+		return WREG32_PCIE(index, value);
+	case CGS_IND_REG__SMC:
+		return WREG32_SMC(index, value);
+	case CGS_IND_REG__UVD_CTX:
+		return WREG32_UVD_CTX(index, value);
+	case CGS_IND_REG__DIDT:
+		return WREG32_DIDT(index, value);
+	case CGS_IND_REG_GC_CAC:
+		return WREG32_GC_CAC(index, value);
+	case CGS_IND_REG_SE_CAC:
+		return WREG32_SE_CAC(index, value);
+	case CGS_IND_REG__AUDIO_ENDPT:
+		DRM_ERROR("audio endpt register access not implemented.\n");
+		return;
+	}
+	WARN(1, "Invalid indirect register space");
+}
+
+static uint32_t fw_type_convert(struct cgs_device *cgs_device, uint32_t fw_type)
+{
+	CGS_FUNC_ADEV;
+	enum AMDGPU_UCODE_ID result = AMDGPU_UCODE_ID_MAXIMUM;
+
+	switch (fw_type) {
+	case CGS_UCODE_ID_SDMA0:
+		result = AMDGPU_UCODE_ID_SDMA0;
+		break;
+	case CGS_UCODE_ID_SDMA1:
+		result = AMDGPU_UCODE_ID_SDMA1;
+		break;
+	case CGS_UCODE_ID_CP_CE:
+		result = AMDGPU_UCODE_ID_CP_CE;
+		break;
+	case CGS_UCODE_ID_CP_PFP:
+		result = AMDGPU_UCODE_ID_CP_PFP;
+		break;
+	case CGS_UCODE_ID_CP_ME:
+		result = AMDGPU_UCODE_ID_CP_ME;
+		break;
+	case CGS_UCODE_ID_CP_MEC:
+	case CGS_UCODE_ID_CP_MEC_JT1:
+		result = AMDGPU_UCODE_ID_CP_MEC1;
+		break;
+	case CGS_UCODE_ID_CP_MEC_JT2:
+		/* for VI. JT2 should be the same as JT1, because:
+			1, MEC2 and MEC1 use exactly same FW.
+			2, JT2 is not pached but JT1 is.
+		*/
+		if (adev->asic_type >= CHIP_TOPAZ)
+			result = AMDGPU_UCODE_ID_CP_MEC1;
+		else
+			result = AMDGPU_UCODE_ID_CP_MEC2;
+		break;
+	case CGS_UCODE_ID_RLC_G:
+		result = AMDGPU_UCODE_ID_RLC_G;
+		break;
+	case CGS_UCODE_ID_STORAGE:
+		result = AMDGPU_UCODE_ID_STORAGE;
+		break;
+	default:
+		DRM_ERROR("Firmware type not supported\n");
+	}
+	return result;
+}
+
+static uint16_t amdgpu_get_firmware_version(struct cgs_device *cgs_device,
+					enum cgs_ucode_id type)
+{
+	CGS_FUNC_ADEV;
+	uint16_t fw_version = 0;
+
+	switch (type) {
+		case CGS_UCODE_ID_SDMA0:
+			fw_version = adev->sdma.instance[0].fw_version;
+			break;
+		case CGS_UCODE_ID_SDMA1:
+			fw_version = adev->sdma.instance[1].fw_version;
+			break;
+		case CGS_UCODE_ID_CP_CE:
+			fw_version = adev->gfx.ce_fw_version;
+			break;
+		case CGS_UCODE_ID_CP_PFP:
+			fw_version = adev->gfx.pfp_fw_version;
+			break;
+		case CGS_UCODE_ID_CP_ME:
+			fw_version = adev->gfx.me_fw_version;
+			break;
+		case CGS_UCODE_ID_CP_MEC:
+			fw_version = adev->gfx.mec_fw_version;
+			break;
+		case CGS_UCODE_ID_CP_MEC_JT1:
+			fw_version = adev->gfx.mec_fw_version;
+			break;
+		case CGS_UCODE_ID_CP_MEC_JT2:
+			fw_version = adev->gfx.mec_fw_version;
+			break;
+		case CGS_UCODE_ID_RLC_G:
+			fw_version = adev->gfx.rlc_fw_version;
+			break;
+		case CGS_UCODE_ID_STORAGE:
+			break;
+		default:
+			DRM_ERROR("firmware type %d do not have version\n", type);
+			break;
+	}
+	return fw_version;
+}
+
+static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
+					enum cgs_ucode_id type,
+					struct cgs_firmware_info *info)
+{
+	CGS_FUNC_ADEV;
+
+	if ((CGS_UCODE_ID_SMU != type) && (CGS_UCODE_ID_SMU_SK != type)) {
+		uint64_t gpu_addr;
+		uint32_t data_size;
+		const struct gfx_firmware_header_v1_0 *header;
+		enum AMDGPU_UCODE_ID id;
+		struct amdgpu_firmware_info *ucode;
+
+		id = fw_type_convert(cgs_device, type);
+		ucode = &adev->firmware.ucode[id];
+		if (ucode->fw == NULL)
+			return -EINVAL;
+
+		gpu_addr  = ucode->mc_addr;
+		header = (const struct gfx_firmware_header_v1_0 *)ucode->fw->data;
+		data_size = le32_to_cpu(header->header.ucode_size_bytes);
+
+		if ((type == CGS_UCODE_ID_CP_MEC_JT1) ||
+		    (type == CGS_UCODE_ID_CP_MEC_JT2)) {
+			gpu_addr += roundup2(le32_to_cpu(header->header.ucode_size_bytes), PAGE_SIZE);
+			data_size = le32_to_cpu(header->jt_size) << 2;
+		}
+
+		info->kptr = ucode->kaddr;
+		info->image_size = data_size;
+		info->mc_addr = gpu_addr;
+		info->version = (uint16_t)le32_to_cpu(header->header.ucode_version);
+
+		if (CGS_UCODE_ID_CP_MEC == type)
+			info->image_size = le32_to_cpu(header->jt_offset) << 2;
+
+		info->fw_version = amdgpu_get_firmware_version(cgs_device, type);
+		info->feature_version = (uint16_t)le32_to_cpu(header->ucode_feature_version);
+	} else {
+		char fw_name[30] = {0};
+		int err = 0;
+		uint32_t ucode_size;
+		uint32_t ucode_start_address;
+		const uint8_t *src;
+		const struct smc_firmware_header_v1_0 *hdr;
+		const struct common_firmware_header *header;
+		struct amdgpu_firmware_info *ucode = NULL;
+
+		if (!adev->pm.fw) {
+			switch (adev->asic_type) {
+			case CHIP_TAHITI:
+				strlcpy(fw_name, "radeon/tahiti_smc.bin", sizeof(fw_name));
+				break;
+			case CHIP_PITCAIRN:
+				if ((adev->pdev->revision == 0x81) &&
+				    ((adev->pdev->device == 0x6810) ||
+				    (adev->pdev->device == 0x6811))) {
+					info->is_kicker = true;
+					strlcpy(fw_name, "radeon/pitcairn_k_smc.bin", sizeof(fw_name));
+				} else {
+					strlcpy(fw_name, "radeon/pitcairn_smc.bin", sizeof(fw_name));
+				}
+				break;
+			case CHIP_VERDE:
+				if (((adev->pdev->device == 0x6820) &&
+					((adev->pdev->revision == 0x81) ||
+					(adev->pdev->revision == 0x83))) ||
+				    ((adev->pdev->device == 0x6821) &&
+					((adev->pdev->revision == 0x83) ||
+					(adev->pdev->revision == 0x87))) ||
+				    ((adev->pdev->revision == 0x87) &&
+					((adev->pdev->device == 0x6823) ||
+					(adev->pdev->device == 0x682b)))) {
+					info->is_kicker = true;
+					strlcpy(fw_name, "radeon/verde_k_smc.bin", sizeof(fw_name));
+				} else {
+					strlcpy(fw_name, "radeon/verde_smc.bin", sizeof(fw_name));
+				}
+				break;
+			case CHIP_OLAND:
+				if (((adev->pdev->revision == 0x81) &&
+					((adev->pdev->device == 0x6600) ||
+					(adev->pdev->device == 0x6604) ||
+					(adev->pdev->device == 0x6605) ||
+					(adev->pdev->device == 0x6610))) ||
+				    ((adev->pdev->revision == 0x83) &&
+					(adev->pdev->device == 0x6610))) {
+					info->is_kicker = true;
+					strlcpy(fw_name, "radeon/oland_k_smc.bin", sizeof(fw_name));
+				} else {
+					strlcpy(fw_name, "radeon/oland_smc.bin", sizeof(fw_name));
+				}
+				break;
+			case CHIP_HAINAN:
+				if (((adev->pdev->revision == 0x81) &&
+					(adev->pdev->device == 0x6660)) ||
+				    ((adev->pdev->revision == 0x83) &&
+					((adev->pdev->device == 0x6660) ||
+					(adev->pdev->device == 0x6663) ||
+					(adev->pdev->device == 0x6665) ||
+					 (adev->pdev->device == 0x6667)))) {
+					info->is_kicker = true;
+					strlcpy(fw_name, "radeon/hainan_k_smc.bin", sizeof(fw_name));
+				} else if ((adev->pdev->revision == 0xc3) &&
+					 (adev->pdev->device == 0x6665)) {
+					info->is_kicker = true;
+					strlcpy(fw_name, "radeon/banks_k_2_smc.bin", sizeof(fw_name));
+				} else {
+					strlcpy(fw_name, "radeon/hainan_smc.bin", sizeof(fw_name));
+				}
+				break;
+			case CHIP_BONAIRE:
+				if ((adev->pdev->revision == 0x80) ||
+					(adev->pdev->revision == 0x81) ||
+					(adev->pdev->device == 0x665f)) {
+					info->is_kicker = true;
+					strlcpy(fw_name, "amdgpu/bonaire_k_smc.bin", sizeof(fw_name));
+				} else {
+					strlcpy(fw_name, "amdgpu/bonaire_smc.bin", sizeof(fw_name));
+				}
+				break;
+			case CHIP_HAWAII:
+				if (adev->pdev->revision == 0x80) {
+					info->is_kicker = true;
+					strlcpy(fw_name, "amdgpu/hawaii_k_smc.bin", sizeof(fw_name));
+				} else {
+					strlcpy(fw_name, "amdgpu/hawaii_smc.bin", sizeof(fw_name));
+				}
+				break;
+			case CHIP_TOPAZ:
+				if (((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0x81)) ||
+				    ((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0x83)) ||
+				    ((adev->pdev->device == 0x6907) && (adev->pdev->revision == 0x87)) ||
+				    ((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0xD1)) ||
+				    ((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0xD3))) {
+					info->is_kicker = true;
+					strlcpy(fw_name, "amdgpu/topaz_k_smc.bin", sizeof(fw_name));
+				} else
+					strlcpy(fw_name, "amdgpu/topaz_smc.bin", sizeof(fw_name));
+				break;
+			case CHIP_TONGA:
+				if (((adev->pdev->device == 0x6939) && (adev->pdev->revision == 0xf1)) ||
+				    ((adev->pdev->device == 0x6938) && (adev->pdev->revision == 0xf1))) {
+					info->is_kicker = true;
+					strlcpy(fw_name, "amdgpu/tonga_k_smc.bin", sizeof(fw_name));
+				} else
+					strlcpy(fw_name, "amdgpu/tonga_smc.bin", sizeof(fw_name));
+				break;
+			case CHIP_FIJI:
+				strlcpy(fw_name, "amdgpu/fiji_smc.bin", sizeof(fw_name));
+				break;
+			case CHIP_POLARIS11:
+				if (type == CGS_UCODE_ID_SMU) {
+					if (((adev->pdev->device == 0x67ef) &&
+					     ((adev->pdev->revision == 0xe0) ||
+					      (adev->pdev->revision == 0xe5))) ||
+					    ((adev->pdev->device == 0x67ff) &&
+					     ((adev->pdev->revision == 0xcf) ||
+					      (adev->pdev->revision == 0xef) ||
+					      (adev->pdev->revision == 0xff)))) {
+						info->is_kicker = true;
+						strlcpy(fw_name, "amdgpu/polaris11_k_smc.bin", sizeof(fw_name));
+					} else if ((adev->pdev->device == 0x67ef) &&
+						   (adev->pdev->revision == 0xe2)) {
+						info->is_kicker = true;
+						strlcpy(fw_name, "amdgpu/polaris11_k2_smc.bin", sizeof(fw_name));
+					} else {
+						strlcpy(fw_name, "amdgpu/polaris11_smc.bin", sizeof(fw_name));
+					}
+				} else if (type == CGS_UCODE_ID_SMU_SK) {
+					strlcpy(fw_name, "amdgpu/polaris11_smc_sk.bin", sizeof(fw_name));
+				}
+				break;
+			case CHIP_POLARIS10:
+				if (type == CGS_UCODE_ID_SMU) {
+					if (((adev->pdev->device == 0x67df) &&
+					     ((adev->pdev->revision == 0xe0) ||
+					      (adev->pdev->revision == 0xe3) ||
+					      (adev->pdev->revision == 0xe4) ||
+					      (adev->pdev->revision == 0xe5) ||
+					      (adev->pdev->revision == 0xe7) ||
+					      (adev->pdev->revision == 0xef))) ||
+					    ((adev->pdev->device == 0x6fdf) &&
+					     ((adev->pdev->revision == 0xef) ||
+					      (adev->pdev->revision == 0xff)))) {
+						info->is_kicker = true;
+						strlcpy(fw_name, "amdgpu/polaris10_k_smc.bin", sizeof(fw_name));
+					} else if ((adev->pdev->device == 0x67df) &&
+						   ((adev->pdev->revision == 0xe1) ||
+						    (adev->pdev->revision == 0xf7))) {
+						info->is_kicker = true;
+						strlcpy(fw_name, "amdgpu/polaris10_k2_smc.bin", sizeof(fw_name));
+					} else {
+						strlcpy(fw_name, "amdgpu/polaris10_smc.bin", sizeof(fw_name));
+					}
+				} else if (type == CGS_UCODE_ID_SMU_SK) {
+					strlcpy(fw_name, "amdgpu/polaris10_smc_sk.bin", sizeof(fw_name));
+				}
+				break;
+			case CHIP_POLARIS12:
+				if (((adev->pdev->device == 0x6987) &&
+				     ((adev->pdev->revision == 0xc0) ||
+				      (adev->pdev->revision == 0xc3))) ||
+				    ((adev->pdev->device == 0x6981) &&
+				     ((adev->pdev->revision == 0x00) ||
+				      (adev->pdev->revision == 0x01) ||
+				      (adev->pdev->revision == 0x10)))) {
+					info->is_kicker = true;
+					strlcpy(fw_name, "amdgpu/polaris12_k_smc.bin", sizeof(fw_name));
+				} else {
+					strlcpy(fw_name, "amdgpu/polaris12_smc.bin", sizeof(fw_name));
+				}
+				break;
+			case CHIP_VEGAM:
+				strlcpy(fw_name, "amdgpu/vegam_smc.bin", sizeof(fw_name));
+				break;
+			case CHIP_VEGA10:
+				if ((adev->pdev->device == 0x687f) &&
+					((adev->pdev->revision == 0xc0) ||
+					(adev->pdev->revision == 0xc1) ||
+					(adev->pdev->revision == 0xc3)))
+					strlcpy(fw_name, "amdgpu/vega10_acg_smc.bin", sizeof(fw_name));
+				else
+					strlcpy(fw_name, "amdgpu/vega10_smc.bin", sizeof(fw_name));
+				break;
+			case CHIP_VEGA12:
+				strlcpy(fw_name, "amdgpu/vega12_smc.bin", sizeof(fw_name));
+				break;
+			case CHIP_VEGA20:
+				strlcpy(fw_name, "amdgpu/vega20_smc.bin", sizeof(fw_name));
+				break;
+			default:
+				DRM_ERROR("SMC firmware not supported\n");
+				return -EINVAL;
+			}
+
+			err = request_firmware(&adev->pm.fw, fw_name, adev->dev);
+			if (err) {
+				DRM_ERROR("Failed to request firmware\n");
+				return err;
+			}
+
+			err = amdgpu_ucode_validate(adev->pm.fw);
+			if (err) {
+				DRM_ERROR("Failed to load firmware \"%s\"", fw_name);
+				release_firmware(adev->pm.fw);
+				adev->pm.fw = NULL;
+				return err;
+			}
+
+			if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+				ucode = &adev->firmware.ucode[AMDGPU_UCODE_ID_SMC];
+				ucode->ucode_id = AMDGPU_UCODE_ID_SMC;
+				ucode->fw = adev->pm.fw;
+				header = (const struct common_firmware_header *)ucode->fw->data;
+				adev->firmware.fw_size +=
+					roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+			}
+		}
+
+		hdr = (const struct smc_firmware_header_v1_0 *)	adev->pm.fw->data;
+		amdgpu_ucode_print_smc_hdr(&hdr->header);
+		adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version);
+		ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
+		ucode_start_address = le32_to_cpu(hdr->ucode_start_addr);
+		src = (const uint8_t *)(adev->pm.fw->data +
+		       le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+		info->version = adev->pm.fw_version;
+		info->image_size = ucode_size;
+		info->ucode_start_address = ucode_start_address;
+		info->kptr = (void *)src;
+	}
+	return 0;
+}
+
+static const struct cgs_ops amdgpu_cgs_ops = {
+	.read_register = amdgpu_cgs_read_register,
+	.write_register = amdgpu_cgs_write_register,
+	.read_ind_register = amdgpu_cgs_read_ind_register,
+	.write_ind_register = amdgpu_cgs_write_ind_register,
+	.get_firmware_info = amdgpu_cgs_get_firmware_info,
+};
+
+struct cgs_device *amdgpu_cgs_create_device(struct amdgpu_device *adev)
+{
+	struct amdgpu_cgs_device *cgs_device =
+		kmalloc(sizeof(*cgs_device), GFP_KERNEL);
+
+	if (!cgs_device) {
+		DRM_ERROR("Couldn't allocate CGS device structure\n");
+		return NULL;
+	}
+
+	cgs_device->base.ops = &amdgpu_cgs_ops;
+	cgs_device->adev = adev;
+
+	return (struct cgs_device *)cgs_device;
+}
+
+void amdgpu_cgs_destroy_device(struct cgs_device *cgs_device)
+{
+	kfree(cgs_device);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ci_dpm.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ci_dpm.c
new file mode 100644
index 00000000000..46326cf1e9a
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ci_dpm.c
@@ -0,0 +1,6844 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_ucode.h"
+#include "cikd.h"
+#include "amdgpu_dpm.h"
+#include "ci_dpm.h"
+#include "gfx_v7_0.h"
+#include "atom.h"
+#include "amd_pcie.h"
+#include <linux/seq_file.h>
+
+#include "smu/smu_7_0_1_d.h"
+#include "smu/smu_7_0_1_sh_mask.h"
+
+#include "dce/dce_8_0_d.h"
+#include "dce/dce_8_0_sh_mask.h"
+
+#include "bif/bif_4_1_d.h"
+#include "bif/bif_4_1_sh_mask.h"
+
+#include "gca/gfx_7_2_d.h"
+#include "gca/gfx_7_2_sh_mask.h"
+
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+
+MODULE_FIRMWARE("amdgpu/bonaire_smc.bin");
+MODULE_FIRMWARE("amdgpu/bonaire_k_smc.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_smc.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_k_smc.bin");
+
+#define MC_CG_ARB_FREQ_F0           0x0a
+#define MC_CG_ARB_FREQ_F1           0x0b
+#define MC_CG_ARB_FREQ_F2           0x0c
+#define MC_CG_ARB_FREQ_F3           0x0d
+
+#define SMC_RAM_END 0x40000
+
+#define VOLTAGE_SCALE               4
+#define VOLTAGE_VID_OFFSET_SCALE1    625
+#define VOLTAGE_VID_OFFSET_SCALE2    100
+
+static const struct amd_pm_funcs ci_dpm_funcs;
+
+static const struct ci_pt_defaults defaults_hawaii_xt =
+{
+	1, 0xF, 0xFD, 0x19, 5, 0x14, 0, 0xB0000,
+	{ 0x2E,  0x00,  0x00,  0x88,  0x00,  0x00,  0x72,  0x60,  0x51,  0xA7,  0x79,  0x6B,  0x90,  0xBD,  0x79  },
+	{ 0x217, 0x217, 0x217, 0x242, 0x242, 0x242, 0x269, 0x269, 0x269, 0x2A1, 0x2A1, 0x2A1, 0x2C9, 0x2C9, 0x2C9 }
+};
+
+static const struct ci_pt_defaults defaults_hawaii_pro =
+{
+	1, 0xF, 0xFD, 0x19, 5, 0x14, 0, 0x65062,
+	{ 0x2E,  0x00,  0x00,  0x88,  0x00,  0x00,  0x72,  0x60,  0x51,  0xA7,  0x79,  0x6B,  0x90,  0xBD,  0x79  },
+	{ 0x217, 0x217, 0x217, 0x242, 0x242, 0x242, 0x269, 0x269, 0x269, 0x2A1, 0x2A1, 0x2A1, 0x2C9, 0x2C9, 0x2C9 }
+};
+
+static const struct ci_pt_defaults defaults_bonaire_xt =
+{
+	1, 0xF, 0xFD, 0x19, 5, 45, 0, 0xB0000,
+	{ 0x79,  0x253, 0x25D, 0xAE,  0x72,  0x80,  0x83,  0x86,  0x6F,  0xC8,  0xC9,  0xC9,  0x2F,  0x4D,  0x61  },
+	{ 0x17C, 0x172, 0x180, 0x1BC, 0x1B3, 0x1BD, 0x206, 0x200, 0x203, 0x25D, 0x25A, 0x255, 0x2C3, 0x2C5, 0x2B4 }
+};
+
+#if 0
+static const struct ci_pt_defaults defaults_bonaire_pro =
+{
+	1, 0xF, 0xFD, 0x19, 5, 45, 0, 0x65062,
+	{ 0x8C,  0x23F, 0x244, 0xA6,  0x83,  0x85,  0x86,  0x86,  0x83,  0xDB,  0xDB,  0xDA,  0x67,  0x60,  0x5F  },
+	{ 0x187, 0x193, 0x193, 0x1C7, 0x1D1, 0x1D1, 0x210, 0x219, 0x219, 0x266, 0x26C, 0x26C, 0x2C9, 0x2CB, 0x2CB }
+};
+#endif
+
+static const struct ci_pt_defaults defaults_saturn_xt =
+{
+	1, 0xF, 0xFD, 0x19, 5, 55, 0, 0x70000,
+	{ 0x8C,  0x247, 0x249, 0xA6,  0x80,  0x81,  0x8B,  0x89,  0x86,  0xC9,  0xCA,  0xC9,  0x4D,  0x4D,  0x4D  },
+	{ 0x187, 0x187, 0x187, 0x1C7, 0x1C7, 0x1C7, 0x210, 0x210, 0x210, 0x266, 0x266, 0x266, 0x2C9, 0x2C9, 0x2C9 }
+};
+
+#if 0
+static const struct ci_pt_defaults defaults_saturn_pro =
+{
+	1, 0xF, 0xFD, 0x19, 5, 55, 0, 0x30000,
+	{ 0x96,  0x21D, 0x23B, 0xA1,  0x85,  0x87,  0x83,  0x84,  0x81,  0xE6,  0xE6,  0xE6,  0x71,  0x6A,  0x6A  },
+	{ 0x193, 0x19E, 0x19E, 0x1D2, 0x1DC, 0x1DC, 0x21A, 0x223, 0x223, 0x26E, 0x27E, 0x274, 0x2CF, 0x2D2, 0x2D2 }
+};
+#endif
+
+static const struct ci_pt_config_reg didt_config_ci[] =
+{
+	{ 0x10, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x10, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x10, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x10, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x11, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x11, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x11, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x11, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x12, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x12, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x12, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x12, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x2, 0x00003fff, 0, 0x4, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x2, 0x03ff0000, 16, 0x80, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x2, 0x78000000, 27, 0x3, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x1, 0x0000ffff, 0, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x1, 0xffff0000, 16, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x0, 0x00000001, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x30, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x30, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x30, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x30, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x31, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x31, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x31, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x31, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x32, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x32, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x32, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x32, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x22, 0x00003fff, 0, 0x4, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x22, 0x03ff0000, 16, 0x80, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x22, 0x78000000, 27, 0x3, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x21, 0x0000ffff, 0, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x21, 0xffff0000, 16, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x20, 0x00000001, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x50, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x50, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x50, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x50, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x51, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x51, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x51, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x51, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x52, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x52, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x52, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x52, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x42, 0x00003fff, 0, 0x4, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x42, 0x03ff0000, 16, 0x80, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x42, 0x78000000, 27, 0x3, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x41, 0x0000ffff, 0, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x41, 0xffff0000, 16, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x40, 0x00000001, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x70, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x70, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x70, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x70, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x71, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x71, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x71, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x71, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x72, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x72, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x72, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x72, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x62, 0x00003fff, 0, 0x4, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x62, 0x03ff0000, 16, 0x80, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x62, 0x78000000, 27, 0x3, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x61, 0x0000ffff, 0, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x61, 0xffff0000, 16, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x60, 0x00000001, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0xFFFFFFFF }
+};
+
+static u8 ci_get_memory_module_index(struct amdgpu_device *adev)
+{
+	return (u8) ((RREG32(mmBIOS_SCRATCH_4) >> 16) & 0xff);
+}
+
+#define MC_CG_ARB_FREQ_F0           0x0a
+#define MC_CG_ARB_FREQ_F1           0x0b
+#define MC_CG_ARB_FREQ_F2           0x0c
+#define MC_CG_ARB_FREQ_F3           0x0d
+
+static int ci_copy_and_switch_arb_sets(struct amdgpu_device *adev,
+				       u32 arb_freq_src, u32 arb_freq_dest)
+{
+	u32 mc_arb_dram_timing;
+	u32 mc_arb_dram_timing2;
+	u32 burst_time;
+	u32 mc_cg_config;
+
+	switch (arb_freq_src) {
+	case MC_CG_ARB_FREQ_F0:
+		mc_arb_dram_timing  = RREG32(mmMC_ARB_DRAM_TIMING);
+		mc_arb_dram_timing2 = RREG32(mmMC_ARB_DRAM_TIMING2);
+		burst_time = (RREG32(mmMC_ARB_BURST_TIME) & MC_ARB_BURST_TIME__STATE0_MASK) >>
+			 MC_ARB_BURST_TIME__STATE0__SHIFT;
+		break;
+	case MC_CG_ARB_FREQ_F1:
+		mc_arb_dram_timing  = RREG32(mmMC_ARB_DRAM_TIMING_1);
+		mc_arb_dram_timing2 = RREG32(mmMC_ARB_DRAM_TIMING2_1);
+		burst_time = (RREG32(mmMC_ARB_BURST_TIME) & MC_ARB_BURST_TIME__STATE1_MASK) >>
+			 MC_ARB_BURST_TIME__STATE1__SHIFT;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	switch (arb_freq_dest) {
+	case MC_CG_ARB_FREQ_F0:
+		WREG32(mmMC_ARB_DRAM_TIMING, mc_arb_dram_timing);
+		WREG32(mmMC_ARB_DRAM_TIMING2, mc_arb_dram_timing2);
+		WREG32_P(mmMC_ARB_BURST_TIME, (burst_time << MC_ARB_BURST_TIME__STATE0__SHIFT),
+			~MC_ARB_BURST_TIME__STATE0_MASK);
+		break;
+	case MC_CG_ARB_FREQ_F1:
+		WREG32(mmMC_ARB_DRAM_TIMING_1, mc_arb_dram_timing);
+		WREG32(mmMC_ARB_DRAM_TIMING2_1, mc_arb_dram_timing2);
+		WREG32_P(mmMC_ARB_BURST_TIME, (burst_time << MC_ARB_BURST_TIME__STATE1__SHIFT),
+			~MC_ARB_BURST_TIME__STATE1_MASK);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	mc_cg_config = RREG32(mmMC_CG_CONFIG) | 0x0000000F;
+	WREG32(mmMC_CG_CONFIG, mc_cg_config);
+	WREG32_P(mmMC_ARB_CG, (arb_freq_dest) << MC_ARB_CG__CG_ARB_REQ__SHIFT,
+		~MC_ARB_CG__CG_ARB_REQ_MASK);
+
+	return 0;
+}
+
+static u8 ci_get_ddr3_mclk_frequency_ratio(u32 memory_clock)
+{
+	u8 mc_para_index;
+
+	if (memory_clock < 10000)
+		mc_para_index = 0;
+	else if (memory_clock >= 80000)
+		mc_para_index = 0x0f;
+	else
+		mc_para_index = (u8)((memory_clock - 10000) / 5000 + 1);
+	return mc_para_index;
+}
+
+static u8 ci_get_mclk_frequency_ratio(u32 memory_clock, bool strobe_mode)
+{
+	u8 mc_para_index;
+
+	if (strobe_mode) {
+		if (memory_clock < 12500)
+			mc_para_index = 0x00;
+		else if (memory_clock > 47500)
+			mc_para_index = 0x0f;
+		else
+			mc_para_index = (u8)((memory_clock - 10000) / 2500);
+	} else {
+		if (memory_clock < 65000)
+			mc_para_index = 0x00;
+		else if (memory_clock > 135000)
+			mc_para_index = 0x0f;
+		else
+			mc_para_index = (u8)((memory_clock - 60000) / 5000);
+	}
+	return mc_para_index;
+}
+
+static void ci_trim_voltage_table_to_fit_state_table(struct amdgpu_device *adev,
+						     u32 max_voltage_steps,
+						     struct atom_voltage_table *voltage_table)
+{
+	unsigned int i, diff;
+
+	if (voltage_table->count <= max_voltage_steps)
+		return;
+
+	diff = voltage_table->count - max_voltage_steps;
+
+	for (i = 0; i < max_voltage_steps; i++)
+		voltage_table->entries[i] = voltage_table->entries[i + diff];
+
+	voltage_table->count = max_voltage_steps;
+}
+
+static int ci_get_std_voltage_value_sidd(struct amdgpu_device *adev,
+					 struct atom_voltage_table_entry *voltage_table,
+					 u16 *std_voltage_hi_sidd, u16 *std_voltage_lo_sidd);
+static int ci_set_power_limit(struct amdgpu_device *adev, u32 n);
+static int ci_set_overdrive_target_tdp(struct amdgpu_device *adev,
+				       u32 target_tdp);
+static int ci_update_uvd_dpm(struct amdgpu_device *adev, bool gate);
+static void ci_dpm_set_irq_funcs(struct amdgpu_device *adev);
+
+static PPSMC_Result amdgpu_ci_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+							     PPSMC_Msg msg, u32 parameter);
+static void ci_thermal_start_smc_fan_control(struct amdgpu_device *adev);
+static void ci_fan_ctrl_set_default_mode(struct amdgpu_device *adev);
+
+static struct ci_power_info *ci_get_pi(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = adev->pm.dpm.priv;
+
+	return pi;
+}
+
+static struct ci_ps *ci_get_ps(struct amdgpu_ps *rps)
+{
+	struct ci_ps *ps = rps->ps_priv;
+
+	return ps;
+}
+
+static void ci_initialize_powertune_defaults(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	switch (adev->pdev->device) {
+	case 0x6649:
+	case 0x6650:
+	case 0x6651:
+	case 0x6658:
+	case 0x665C:
+	case 0x665D:
+	default:
+		pi->powertune_defaults = &defaults_bonaire_xt;
+		break;
+	case 0x6640:
+	case 0x6641:
+	case 0x6646:
+	case 0x6647:
+		pi->powertune_defaults = &defaults_saturn_xt;
+		break;
+	case 0x67B8:
+	case 0x67B0:
+		pi->powertune_defaults = &defaults_hawaii_xt;
+		break;
+	case 0x67BA:
+	case 0x67B1:
+		pi->powertune_defaults = &defaults_hawaii_pro;
+		break;
+	case 0x67A0:
+	case 0x67A1:
+	case 0x67A2:
+	case 0x67A8:
+	case 0x67A9:
+	case 0x67AA:
+	case 0x67B9:
+	case 0x67BE:
+		pi->powertune_defaults = &defaults_bonaire_xt;
+		break;
+	}
+
+	pi->dte_tj_offset = 0;
+
+	pi->caps_power_containment = true;
+	pi->caps_cac = false;
+	pi->caps_sq_ramping = false;
+	pi->caps_db_ramping = false;
+	pi->caps_td_ramping = false;
+	pi->caps_tcp_ramping = false;
+
+	if (pi->caps_power_containment) {
+		pi->caps_cac = true;
+		if (adev->asic_type == CHIP_HAWAII)
+			pi->enable_bapm_feature = false;
+		else
+			pi->enable_bapm_feature = true;
+		pi->enable_tdc_limit_feature = true;
+		pi->enable_pkg_pwr_tracking_feature = true;
+	}
+}
+
+static u8 ci_convert_to_vid(u16 vddc)
+{
+	return (6200 - (vddc * VOLTAGE_SCALE)) / 25;
+}
+
+static int ci_populate_bapm_vddc_vid_sidd(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u8 *hi_vid = pi->smc_powertune_table.BapmVddCVidHiSidd;
+	u8 *lo_vid = pi->smc_powertune_table.BapmVddCVidLoSidd;
+	u8 *hi2_vid = pi->smc_powertune_table.BapmVddCVidHiSidd2;
+	u32 i;
+
+	if (adev->pm.dpm.dyn_state.cac_leakage_table.entries == NULL)
+		return -EINVAL;
+	if (adev->pm.dpm.dyn_state.cac_leakage_table.count > 8)
+		return -EINVAL;
+	if (adev->pm.dpm.dyn_state.cac_leakage_table.count !=
+	    adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count)
+		return -EINVAL;
+
+	for (i = 0; i < adev->pm.dpm.dyn_state.cac_leakage_table.count; i++) {
+		if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_EVV) {
+			lo_vid[i] = ci_convert_to_vid(adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc1);
+			hi_vid[i] = ci_convert_to_vid(adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc2);
+			hi2_vid[i] = ci_convert_to_vid(adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc3);
+		} else {
+			lo_vid[i] = ci_convert_to_vid(adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc);
+			hi_vid[i] = ci_convert_to_vid((u16)adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].leakage);
+		}
+	}
+	return 0;
+}
+
+static int ci_populate_vddc_vid(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u8 *vid = pi->smc_powertune_table.VddCVid;
+	u32 i;
+
+	if (pi->vddc_voltage_table.count > 8)
+		return -EINVAL;
+
+	for (i = 0; i < pi->vddc_voltage_table.count; i++)
+		vid[i] = ci_convert_to_vid(pi->vddc_voltage_table.entries[i].value);
+
+	return 0;
+}
+
+static int ci_populate_svi_load_line(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct ci_pt_defaults *pt_defaults = pi->powertune_defaults;
+
+	pi->smc_powertune_table.SviLoadLineEn = pt_defaults->svi_load_line_en;
+	pi->smc_powertune_table.SviLoadLineVddC = pt_defaults->svi_load_line_vddc;
+	pi->smc_powertune_table.SviLoadLineTrimVddC = 3;
+	pi->smc_powertune_table.SviLoadLineOffsetVddC = 0;
+
+	return 0;
+}
+
+static int ci_populate_tdc_limit(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct ci_pt_defaults *pt_defaults = pi->powertune_defaults;
+	u16 tdc_limit;
+
+	tdc_limit = adev->pm.dpm.dyn_state.cac_tdp_table->tdc * 256;
+	pi->smc_powertune_table.TDC_VDDC_PkgLimit = cpu_to_be16(tdc_limit);
+	pi->smc_powertune_table.TDC_VDDC_ThrottleReleaseLimitPerc =
+		pt_defaults->tdc_vddc_throttle_release_limit_perc;
+	pi->smc_powertune_table.TDC_MAWt = pt_defaults->tdc_mawt;
+
+	return 0;
+}
+
+static int ci_populate_dw8(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct ci_pt_defaults *pt_defaults = pi->powertune_defaults;
+	int ret;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev,
+				     SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, PmFuseTable) +
+				     offsetof(SMU7_Discrete_PmFuses, TdcWaterfallCtl),
+				     (u32 *)&pi->smc_powertune_table.TdcWaterfallCtl,
+				     pi->sram_end);
+	if (ret)
+		return -EINVAL;
+	else
+		pi->smc_powertune_table.TdcWaterfallCtl = pt_defaults->tdc_waterfall_ctl;
+
+	return 0;
+}
+
+static int ci_populate_fuzzy_fan(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if ((adev->pm.dpm.fan.fan_output_sensitivity & (1 << 15)) ||
+	    (adev->pm.dpm.fan.fan_output_sensitivity == 0))
+		adev->pm.dpm.fan.fan_output_sensitivity =
+			adev->pm.dpm.fan.default_fan_output_sensitivity;
+
+	pi->smc_powertune_table.FuzzyFan_PwmSetDelta =
+		cpu_to_be16(adev->pm.dpm.fan.fan_output_sensitivity);
+
+	return 0;
+}
+
+static int ci_min_max_v_gnbl_pm_lid_from_bapm_vddc(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u8 *hi_vid = pi->smc_powertune_table.BapmVddCVidHiSidd;
+	u8 *lo_vid = pi->smc_powertune_table.BapmVddCVidLoSidd;
+	int i, min, max;
+
+	min = max = hi_vid[0];
+	for (i = 0; i < 8; i++) {
+		if (0 != hi_vid[i]) {
+			if (min > hi_vid[i])
+				min = hi_vid[i];
+			if (max < hi_vid[i])
+				max = hi_vid[i];
+		}
+
+		if (0 != lo_vid[i]) {
+			if (min > lo_vid[i])
+				min = lo_vid[i];
+			if (max < lo_vid[i])
+				max = lo_vid[i];
+		}
+	}
+
+	if ((min == 0) || (max == 0))
+		return -EINVAL;
+	pi->smc_powertune_table.GnbLPMLMaxVid = (u8)max;
+	pi->smc_powertune_table.GnbLPMLMinVid = (u8)min;
+
+	return 0;
+}
+
+static int ci_populate_bapm_vddc_base_leakage_sidd(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u16 hi_sidd = pi->smc_powertune_table.BapmVddCBaseLeakageHiSidd;
+	u16 lo_sidd = pi->smc_powertune_table.BapmVddCBaseLeakageLoSidd;
+	struct amdgpu_cac_tdp_table *cac_tdp_table =
+		adev->pm.dpm.dyn_state.cac_tdp_table;
+
+	hi_sidd = cac_tdp_table->high_cac_leakage / 100 * 256;
+	lo_sidd = cac_tdp_table->low_cac_leakage / 100 * 256;
+
+	pi->smc_powertune_table.BapmVddCBaseLeakageHiSidd = cpu_to_be16(hi_sidd);
+	pi->smc_powertune_table.BapmVddCBaseLeakageLoSidd = cpu_to_be16(lo_sidd);
+
+	return 0;
+}
+
+static int ci_populate_bapm_parameters_in_dpm_table(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct ci_pt_defaults *pt_defaults = pi->powertune_defaults;
+	SMU7_Discrete_DpmTable  *dpm_table = &pi->smc_state_table;
+	struct amdgpu_cac_tdp_table *cac_tdp_table =
+		adev->pm.dpm.dyn_state.cac_tdp_table;
+	struct amdgpu_ppm_table *ppm = adev->pm.dpm.dyn_state.ppm_table;
+	int i, j, k;
+	const u16 *def1;
+	const u16 *def2;
+
+	dpm_table->DefaultTdp = cac_tdp_table->tdp * 256;
+	dpm_table->TargetTdp = cac_tdp_table->configurable_tdp * 256;
+
+	dpm_table->DTETjOffset = (u8)pi->dte_tj_offset;
+	dpm_table->GpuTjMax =
+		(u8)(pi->thermal_temp_setting.temperature_high / 1000);
+	dpm_table->GpuTjHyst = 8;
+
+	dpm_table->DTEAmbientTempBase = pt_defaults->dte_ambient_temp_base;
+
+	if (ppm) {
+		dpm_table->PPM_PkgPwrLimit = cpu_to_be16((u16)ppm->dgpu_tdp * 256 / 1000);
+		dpm_table->PPM_TemperatureLimit = cpu_to_be16((u16)ppm->tj_max * 256);
+	} else {
+		dpm_table->PPM_PkgPwrLimit = cpu_to_be16(0);
+		dpm_table->PPM_TemperatureLimit = cpu_to_be16(0);
+	}
+
+	dpm_table->BAPM_TEMP_GRADIENT = cpu_to_be32(pt_defaults->bapm_temp_gradient);
+	def1 = pt_defaults->bapmti_r;
+	def2 = pt_defaults->bapmti_rc;
+
+	for (i = 0; i < SMU7_DTE_ITERATIONS; i++) {
+		for (j = 0; j < SMU7_DTE_SOURCES; j++) {
+			for (k = 0; k < SMU7_DTE_SINKS; k++) {
+				dpm_table->BAPMTI_R[i][j][k] = cpu_to_be16(*def1);
+				dpm_table->BAPMTI_RC[i][j][k] = cpu_to_be16(*def2);
+				def1++;
+				def2++;
+			}
+		}
+	}
+
+	return 0;
+}
+
+static int ci_populate_pm_base(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 pm_fuse_table_offset;
+	int ret;
+
+	if (pi->caps_power_containment) {
+		ret = amdgpu_ci_read_smc_sram_dword(adev,
+					     SMU7_FIRMWARE_HEADER_LOCATION +
+					     offsetof(SMU7_Firmware_Header, PmFuseTable),
+					     &pm_fuse_table_offset, pi->sram_end);
+		if (ret)
+			return ret;
+		ret = ci_populate_bapm_vddc_vid_sidd(adev);
+		if (ret)
+			return ret;
+		ret = ci_populate_vddc_vid(adev);
+		if (ret)
+			return ret;
+		ret = ci_populate_svi_load_line(adev);
+		if (ret)
+			return ret;
+		ret = ci_populate_tdc_limit(adev);
+		if (ret)
+			return ret;
+		ret = ci_populate_dw8(adev);
+		if (ret)
+			return ret;
+		ret = ci_populate_fuzzy_fan(adev);
+		if (ret)
+			return ret;
+		ret = ci_min_max_v_gnbl_pm_lid_from_bapm_vddc(adev);
+		if (ret)
+			return ret;
+		ret = ci_populate_bapm_vddc_base_leakage_sidd(adev);
+		if (ret)
+			return ret;
+		ret = amdgpu_ci_copy_bytes_to_smc(adev, pm_fuse_table_offset,
+					   (u8 *)&pi->smc_powertune_table,
+					   sizeof(SMU7_Discrete_PmFuses), pi->sram_end);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static void ci_do_enable_didt(struct amdgpu_device *adev, const bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 data;
+
+	if (pi->caps_sq_ramping) {
+		data = RREG32_DIDT(ixDIDT_SQ_CTRL0);
+		if (enable)
+			data |= DIDT_SQ_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_SQ_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_SQ_CTRL0, data);
+	}
+
+	if (pi->caps_db_ramping) {
+		data = RREG32_DIDT(ixDIDT_DB_CTRL0);
+		if (enable)
+			data |= DIDT_DB_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_DB_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_DB_CTRL0, data);
+	}
+
+	if (pi->caps_td_ramping) {
+		data = RREG32_DIDT(ixDIDT_TD_CTRL0);
+		if (enable)
+			data |= DIDT_TD_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_TD_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_TD_CTRL0, data);
+	}
+
+	if (pi->caps_tcp_ramping) {
+		data = RREG32_DIDT(ixDIDT_TCP_CTRL0);
+		if (enable)
+			data |= DIDT_TCP_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_TCP_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_TCP_CTRL0, data);
+	}
+}
+
+static int ci_program_pt_config_registers(struct amdgpu_device *adev,
+					  const struct ci_pt_config_reg *cac_config_regs)
+{
+	const struct ci_pt_config_reg *config_regs = cac_config_regs;
+	u32 data;
+	u32 cache = 0;
+
+	if (config_regs == NULL)
+		return -EINVAL;
+
+	while (config_regs->offset != 0xFFFFFFFF) {
+		if (config_regs->type == CISLANDS_CONFIGREG_CACHE) {
+			cache |= ((config_regs->value << config_regs->shift) & config_regs->mask);
+		} else {
+			switch (config_regs->type) {
+			case CISLANDS_CONFIGREG_SMC_IND:
+				data = RREG32_SMC(config_regs->offset);
+				break;
+			case CISLANDS_CONFIGREG_DIDT_IND:
+				data = RREG32_DIDT(config_regs->offset);
+				break;
+			default:
+				data = RREG32(config_regs->offset);
+				break;
+			}
+
+			data &= ~config_regs->mask;
+			data |= ((config_regs->value << config_regs->shift) & config_regs->mask);
+			data |= cache;
+
+			switch (config_regs->type) {
+			case CISLANDS_CONFIGREG_SMC_IND:
+				WREG32_SMC(config_regs->offset, data);
+				break;
+			case CISLANDS_CONFIGREG_DIDT_IND:
+				WREG32_DIDT(config_regs->offset, data);
+				break;
+			default:
+				WREG32(config_regs->offset, data);
+				break;
+			}
+			cache = 0;
+		}
+		config_regs++;
+	}
+	return 0;
+}
+
+static int ci_enable_didt(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret;
+
+	if (pi->caps_sq_ramping || pi->caps_db_ramping ||
+	    pi->caps_td_ramping || pi->caps_tcp_ramping) {
+		adev->gfx.rlc.funcs->enter_safe_mode(adev);
+
+		if (enable) {
+			ret = ci_program_pt_config_registers(adev, didt_config_ci);
+			if (ret) {
+				adev->gfx.rlc.funcs->exit_safe_mode(adev);
+				return ret;
+			}
+		}
+
+		ci_do_enable_didt(adev, enable);
+
+		adev->gfx.rlc.funcs->exit_safe_mode(adev);
+	}
+
+	return 0;
+}
+
+static int ci_enable_power_containment(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+	int ret = 0;
+
+	if (enable) {
+		pi->power_containment_features = 0;
+		if (pi->caps_power_containment) {
+			if (pi->enable_bapm_feature) {
+				smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableDTE);
+				if (smc_result != PPSMC_Result_OK)
+					ret = -EINVAL;
+				else
+					pi->power_containment_features |= POWERCONTAINMENT_FEATURE_BAPM;
+			}
+
+			if (pi->enable_tdc_limit_feature) {
+				smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_TDCLimitEnable);
+				if (smc_result != PPSMC_Result_OK)
+					ret = -EINVAL;
+				else
+					pi->power_containment_features |= POWERCONTAINMENT_FEATURE_TDCLimit;
+			}
+
+			if (pi->enable_pkg_pwr_tracking_feature) {
+				smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PkgPwrLimitEnable);
+				if (smc_result != PPSMC_Result_OK) {
+					ret = -EINVAL;
+				} else {
+					struct amdgpu_cac_tdp_table *cac_tdp_table =
+						adev->pm.dpm.dyn_state.cac_tdp_table;
+					u32 default_pwr_limit =
+						(u32)(cac_tdp_table->maximum_power_delivery_limit * 256);
+
+					pi->power_containment_features |= POWERCONTAINMENT_FEATURE_PkgPwrLimit;
+
+					ci_set_power_limit(adev, default_pwr_limit);
+				}
+			}
+		}
+	} else {
+		if (pi->caps_power_containment && pi->power_containment_features) {
+			if (pi->power_containment_features & POWERCONTAINMENT_FEATURE_TDCLimit)
+				amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_TDCLimitDisable);
+
+			if (pi->power_containment_features & POWERCONTAINMENT_FEATURE_BAPM)
+				amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DisableDTE);
+
+			if (pi->power_containment_features & POWERCONTAINMENT_FEATURE_PkgPwrLimit)
+				amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PkgPwrLimitDisable);
+			pi->power_containment_features = 0;
+		}
+	}
+
+	return ret;
+}
+
+static int ci_enable_smc_cac(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+	int ret = 0;
+
+	if (pi->caps_cac) {
+		if (enable) {
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableCac);
+			if (smc_result != PPSMC_Result_OK) {
+				ret = -EINVAL;
+				pi->cac_enabled = false;
+			} else {
+				pi->cac_enabled = true;
+			}
+		} else if (pi->cac_enabled) {
+			amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DisableCac);
+			pi->cac_enabled = false;
+		}
+	}
+
+	return ret;
+}
+
+static int ci_enable_thermal_based_sclk_dpm(struct amdgpu_device *adev,
+					    bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result = PPSMC_Result_OK;
+
+	if (pi->thermal_sclk_dpm_enabled) {
+		if (enable)
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_ENABLE_THERMAL_DPM);
+		else
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DISABLE_THERMAL_DPM);
+	}
+
+	if (smc_result == PPSMC_Result_OK)
+		return 0;
+	else
+		return -EINVAL;
+}
+
+static int ci_power_control_set_level(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_cac_tdp_table *cac_tdp_table =
+		adev->pm.dpm.dyn_state.cac_tdp_table;
+	s32 adjust_percent;
+	s32 target_tdp;
+	int ret = 0;
+	bool adjust_polarity = false; /* ??? */
+
+	if (pi->caps_power_containment) {
+		adjust_percent = adjust_polarity ?
+			adev->pm.dpm.tdp_adjustment : (-1 * adev->pm.dpm.tdp_adjustment);
+		target_tdp = ((100 + adjust_percent) *
+			      (s32)cac_tdp_table->configurable_tdp) / 100;
+
+		ret = ci_set_overdrive_target_tdp(adev, (u32)target_tdp);
+	}
+
+	return ret;
+}
+
+static void ci_dpm_powergate_uvd(void *handle, bool gate)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	pi->uvd_power_gated = gate;
+
+	if (gate) {
+		/* stop the UVD block */
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+						       AMD_PG_STATE_GATE);
+		ci_update_uvd_dpm(adev, gate);
+	} else {
+		amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+						       AMD_PG_STATE_UNGATE);
+		ci_update_uvd_dpm(adev, gate);
+	}
+}
+
+static bool ci_dpm_vblank_too_short(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 vblank_time = amdgpu_dpm_get_vblank_time(adev);
+	u32 switch_limit = adev->gmc.vram_type == AMDGPU_VRAM_TYPE_GDDR5 ? 450 : 300;
+
+	/* disable mclk switching if the refresh is >120Hz, even if the
+	 * blanking period would allow it
+	 */
+	if (amdgpu_dpm_get_vrefresh(adev) > 120)
+		return true;
+
+	if (vblank_time < switch_limit)
+		return true;
+	else
+		return false;
+
+}
+
+static void ci_apply_state_adjust_rules(struct amdgpu_device *adev,
+					struct amdgpu_ps *rps)
+{
+	struct ci_ps *ps = ci_get_ps(rps);
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_clock_and_voltage_limits *max_limits;
+	bool disable_mclk_switching;
+	u32 sclk, mclk;
+	int i;
+
+	if (rps->vce_active) {
+		rps->evclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].evclk;
+		rps->ecclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].ecclk;
+	} else {
+		rps->evclk = 0;
+		rps->ecclk = 0;
+	}
+
+	if ((adev->pm.dpm.new_active_crtc_count > 1) ||
+	    ci_dpm_vblank_too_short(adev))
+		disable_mclk_switching = true;
+	else
+		disable_mclk_switching = false;
+
+	if ((rps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) == ATOM_PPLIB_CLASSIFICATION_UI_BATTERY)
+		pi->battery_state = true;
+	else
+		pi->battery_state = false;
+
+	if (adev->pm.ac_power)
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	else
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
+
+	if (adev->pm.ac_power == false) {
+		for (i = 0; i < ps->performance_level_count; i++) {
+			if (ps->performance_levels[i].mclk > max_limits->mclk)
+				ps->performance_levels[i].mclk = max_limits->mclk;
+			if (ps->performance_levels[i].sclk > max_limits->sclk)
+				ps->performance_levels[i].sclk = max_limits->sclk;
+		}
+	}
+
+	/* XXX validate the min clocks required for display */
+
+	if (disable_mclk_switching) {
+		mclk  = ps->performance_levels[ps->performance_level_count - 1].mclk;
+		sclk = ps->performance_levels[0].sclk;
+	} else {
+		mclk = ps->performance_levels[0].mclk;
+		sclk = ps->performance_levels[0].sclk;
+	}
+
+	if (adev->pm.pm_display_cfg.min_core_set_clock > sclk)
+		sclk = adev->pm.pm_display_cfg.min_core_set_clock;
+
+	if (adev->pm.pm_display_cfg.min_mem_set_clock > mclk)
+		mclk = adev->pm.pm_display_cfg.min_mem_set_clock;
+
+	if (rps->vce_active) {
+		if (sclk < adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].sclk)
+			sclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].sclk;
+		if (mclk < adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].mclk)
+			mclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].mclk;
+	}
+
+	ps->performance_levels[0].sclk = sclk;
+	ps->performance_levels[0].mclk = mclk;
+
+	if (ps->performance_levels[1].sclk < ps->performance_levels[0].sclk)
+		ps->performance_levels[1].sclk = ps->performance_levels[0].sclk;
+
+	if (disable_mclk_switching) {
+		if (ps->performance_levels[0].mclk < ps->performance_levels[1].mclk)
+			ps->performance_levels[0].mclk = ps->performance_levels[1].mclk;
+	} else {
+		if (ps->performance_levels[1].mclk < ps->performance_levels[0].mclk)
+			ps->performance_levels[1].mclk = ps->performance_levels[0].mclk;
+	}
+}
+
+static int ci_thermal_set_temperature_range(struct amdgpu_device *adev,
+					    int min_temp, int max_temp)
+{
+	int low_temp = 0 * 1000;
+	int high_temp = 255 * 1000;
+	u32 tmp;
+
+	if (low_temp < min_temp)
+		low_temp = min_temp;
+	if (high_temp > max_temp)
+		high_temp = max_temp;
+	if (high_temp < low_temp) {
+		DRM_ERROR("invalid thermal range: %d - %d\n", low_temp, high_temp);
+		return -EINVAL;
+	}
+
+	tmp = RREG32_SMC(ixCG_THERMAL_INT);
+	tmp &= ~(CG_THERMAL_INT__DIG_THERM_INTH_MASK | CG_THERMAL_INT__DIG_THERM_INTL_MASK);
+	tmp |= ((high_temp / 1000) << CG_THERMAL_INT__DIG_THERM_INTH__SHIFT) |
+		((low_temp / 1000)) << CG_THERMAL_INT__DIG_THERM_INTL__SHIFT;
+	WREG32_SMC(ixCG_THERMAL_INT, tmp);
+
+#if 0
+	/* XXX: need to figure out how to handle this properly */
+	tmp = RREG32_SMC(ixCG_THERMAL_CTRL);
+	tmp &= DIG_THERM_DPM_MASK;
+	tmp |= DIG_THERM_DPM(high_temp / 1000);
+	WREG32_SMC(ixCG_THERMAL_CTRL, tmp);
+#endif
+
+	adev->pm.dpm.thermal.min_temp = low_temp;
+	adev->pm.dpm.thermal.max_temp = high_temp;
+	return 0;
+}
+
+static int ci_thermal_enable_alert(struct amdgpu_device *adev,
+				   bool enable)
+{
+	u32 thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
+	PPSMC_Result result;
+
+	if (enable) {
+		thermal_int &= ~(CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK |
+				 CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK);
+		WREG32_SMC(ixCG_THERMAL_INT, thermal_int);
+		result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Thermal_Cntl_Enable);
+		if (result != PPSMC_Result_OK) {
+			DRM_DEBUG_KMS("Could not enable thermal interrupts.\n");
+			return -EINVAL;
+		}
+	} else {
+		thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK |
+			CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
+		WREG32_SMC(ixCG_THERMAL_INT, thermal_int);
+		result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Thermal_Cntl_Disable);
+		if (result != PPSMC_Result_OK) {
+			DRM_DEBUG_KMS("Could not disable thermal interrupts.\n");
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static void ci_fan_ctrl_set_static_mode(struct amdgpu_device *adev, u32 mode)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+
+	if (pi->fan_ctrl_is_in_default_mode) {
+		tmp = (RREG32_SMC(ixCG_FDO_CTRL2) & CG_FDO_CTRL2__FDO_PWM_MODE_MASK)
+			>> CG_FDO_CTRL2__FDO_PWM_MODE__SHIFT;
+		pi->fan_ctrl_default_mode = tmp;
+		tmp = (RREG32_SMC(ixCG_FDO_CTRL2) & CG_FDO_CTRL2__TMIN_MASK)
+			>> CG_FDO_CTRL2__TMIN__SHIFT;
+		pi->t_min = tmp;
+		pi->fan_ctrl_is_in_default_mode = false;
+	}
+
+	tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__TMIN_MASK;
+	tmp |= 0 << CG_FDO_CTRL2__TMIN__SHIFT;
+	WREG32_SMC(ixCG_FDO_CTRL2, tmp);
+
+	tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__FDO_PWM_MODE_MASK;
+	tmp |= mode << CG_FDO_CTRL2__FDO_PWM_MODE__SHIFT;
+	WREG32_SMC(ixCG_FDO_CTRL2, tmp);
+}
+
+static int ci_thermal_setup_fan_table(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	SMU7_Discrete_FanTable fan_table = { FDO_MODE_HARDWARE };
+	u32 duty100;
+	u32 t_diff1, t_diff2, pwm_diff1, pwm_diff2;
+	u16 fdo_min, slope1, slope2;
+	u32 reference_clock, tmp;
+	int ret;
+	u64 tmp64;
+
+	if (!pi->fan_table_start) {
+		adev->pm.dpm.fan.ucode_fan_control = false;
+		return 0;
+	}
+
+	duty100 = (RREG32_SMC(ixCG_FDO_CTRL1) & CG_FDO_CTRL1__FMAX_DUTY100_MASK)
+		>> CG_FDO_CTRL1__FMAX_DUTY100__SHIFT;
+
+	if (duty100 == 0) {
+		adev->pm.dpm.fan.ucode_fan_control = false;
+		return 0;
+	}
+
+	tmp64 = (u64)adev->pm.dpm.fan.pwm_min * duty100;
+	do_div(tmp64, 10000);
+	fdo_min = (u16)tmp64;
+
+	t_diff1 = adev->pm.dpm.fan.t_med - adev->pm.dpm.fan.t_min;
+	t_diff2 = adev->pm.dpm.fan.t_high - adev->pm.dpm.fan.t_med;
+
+	pwm_diff1 = adev->pm.dpm.fan.pwm_med - adev->pm.dpm.fan.pwm_min;
+	pwm_diff2 = adev->pm.dpm.fan.pwm_high - adev->pm.dpm.fan.pwm_med;
+
+	slope1 = (u16)((50 + ((16 * duty100 * pwm_diff1) / t_diff1)) / 100);
+	slope2 = (u16)((50 + ((16 * duty100 * pwm_diff2) / t_diff2)) / 100);
+
+	fan_table.TempMin = cpu_to_be16((50 + adev->pm.dpm.fan.t_min) / 100);
+	fan_table.TempMed = cpu_to_be16((50 + adev->pm.dpm.fan.t_med) / 100);
+	fan_table.TempMax = cpu_to_be16((50 + adev->pm.dpm.fan.t_max) / 100);
+
+	fan_table.Slope1 = cpu_to_be16(slope1);
+	fan_table.Slope2 = cpu_to_be16(slope2);
+
+	fan_table.FdoMin = cpu_to_be16(fdo_min);
+
+	fan_table.HystDown = cpu_to_be16(adev->pm.dpm.fan.t_hyst);
+
+	fan_table.HystUp = cpu_to_be16(1);
+
+	fan_table.HystSlope = cpu_to_be16(1);
+
+	fan_table.TempRespLim = cpu_to_be16(5);
+
+	reference_clock = amdgpu_asic_get_xclk(adev);
+
+	fan_table.RefreshPeriod = cpu_to_be32((adev->pm.dpm.fan.cycle_delay *
+					       reference_clock) / 1600);
+
+	fan_table.FdoMax = cpu_to_be16((u16)duty100);
+
+	tmp = (RREG32_SMC(ixCG_MULT_THERMAL_CTRL) & CG_MULT_THERMAL_CTRL__TEMP_SEL_MASK)
+		>> CG_MULT_THERMAL_CTRL__TEMP_SEL__SHIFT;
+	fan_table.TempSrc = (uint8_t)tmp;
+
+	ret = amdgpu_ci_copy_bytes_to_smc(adev,
+					  pi->fan_table_start,
+					  (u8 *)(&fan_table),
+					  sizeof(fan_table),
+					  pi->sram_end);
+
+	if (ret) {
+		DRM_ERROR("Failed to load fan table to the SMC.");
+		adev->pm.dpm.fan.ucode_fan_control = false;
+	}
+
+	return 0;
+}
+
+static int ci_fan_ctrl_start_smc_fan_control(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result ret;
+
+	if (pi->caps_od_fuzzy_fan_control_support) {
+		ret = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+							       PPSMC_StartFanControl,
+							       FAN_CONTROL_FUZZY);
+		if (ret != PPSMC_Result_OK)
+			return -EINVAL;
+		ret = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+							       PPSMC_MSG_SetFanPwmMax,
+							       adev->pm.dpm.fan.default_max_fan_pwm);
+		if (ret != PPSMC_Result_OK)
+			return -EINVAL;
+	} else {
+		ret = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+							       PPSMC_StartFanControl,
+							       FAN_CONTROL_TABLE);
+		if (ret != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	pi->fan_is_controlled_by_smc = true;
+	return 0;
+}
+
+
+static int ci_fan_ctrl_stop_smc_fan_control(struct amdgpu_device *adev)
+{
+	PPSMC_Result ret;
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	ret = amdgpu_ci_send_msg_to_smc(adev, PPSMC_StopFanControl);
+	if (ret == PPSMC_Result_OK) {
+		pi->fan_is_controlled_by_smc = false;
+		return 0;
+	} else {
+		return -EINVAL;
+	}
+}
+
+static int ci_dpm_get_fan_speed_percent(void *handle,
+					u32 *speed)
+{
+	u32 duty, duty100;
+	u64 tmp64;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.no_fan)
+		return -ENOENT;
+
+	duty100 = (RREG32_SMC(ixCG_FDO_CTRL1) & CG_FDO_CTRL1__FMAX_DUTY100_MASK)
+		>> CG_FDO_CTRL1__FMAX_DUTY100__SHIFT;
+	duty = (RREG32_SMC(ixCG_THERMAL_STATUS) & CG_THERMAL_STATUS__FDO_PWM_DUTY_MASK)
+		>> CG_THERMAL_STATUS__FDO_PWM_DUTY__SHIFT;
+
+	if (duty100 == 0)
+		return -EINVAL;
+
+	tmp64 = (u64)duty * 100;
+	do_div(tmp64, duty100);
+	*speed = (u32)tmp64;
+
+	if (*speed > 100)
+		*speed = 100;
+
+	return 0;
+}
+
+static int ci_dpm_set_fan_speed_percent(void *handle,
+					u32 speed)
+{
+	u32 tmp;
+	u32 duty, duty100;
+	u64 tmp64;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (adev->pm.no_fan)
+		return -ENOENT;
+
+	if (pi->fan_is_controlled_by_smc)
+		return -EINVAL;
+
+	if (speed > 100)
+		return -EINVAL;
+
+	duty100 = (RREG32_SMC(ixCG_FDO_CTRL1) & CG_FDO_CTRL1__FMAX_DUTY100_MASK)
+		>> CG_FDO_CTRL1__FMAX_DUTY100__SHIFT;
+
+	if (duty100 == 0)
+		return -EINVAL;
+
+	tmp64 = (u64)speed * duty100;
+	do_div(tmp64, 100);
+	duty = (u32)tmp64;
+
+	tmp = RREG32_SMC(ixCG_FDO_CTRL0) & ~CG_FDO_CTRL0__FDO_STATIC_DUTY_MASK;
+	tmp |= duty << CG_FDO_CTRL0__FDO_STATIC_DUTY__SHIFT;
+	WREG32_SMC(ixCG_FDO_CTRL0, tmp);
+
+	return 0;
+}
+
+static void ci_dpm_set_fan_control_mode(void *handle, u32 mode)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	switch (mode) {
+	case AMD_FAN_CTRL_NONE:
+		if (adev->pm.dpm.fan.ucode_fan_control)
+			ci_fan_ctrl_stop_smc_fan_control(adev);
+		ci_dpm_set_fan_speed_percent(adev, 100);
+		break;
+	case AMD_FAN_CTRL_MANUAL:
+		if (adev->pm.dpm.fan.ucode_fan_control)
+			ci_fan_ctrl_stop_smc_fan_control(adev);
+		break;
+	case AMD_FAN_CTRL_AUTO:
+		if (adev->pm.dpm.fan.ucode_fan_control)
+			ci_thermal_start_smc_fan_control(adev);
+		break;
+	default:
+		break;
+	}
+}
+
+static u32 ci_dpm_get_fan_control_mode(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (pi->fan_is_controlled_by_smc)
+		return AMD_FAN_CTRL_AUTO;
+	else
+		return AMD_FAN_CTRL_MANUAL;
+}
+
+#if 0
+static int ci_fan_ctrl_get_fan_speed_rpm(struct amdgpu_device *adev,
+					 u32 *speed)
+{
+	u32 tach_period;
+	u32 xclk = amdgpu_asic_get_xclk(adev);
+
+	if (adev->pm.no_fan)
+		return -ENOENT;
+
+	if (adev->pm.fan_pulses_per_revolution == 0)
+		return -ENOENT;
+
+	tach_period = (RREG32_SMC(ixCG_TACH_STATUS) & CG_TACH_STATUS__TACH_PERIOD_MASK)
+		>> CG_TACH_STATUS__TACH_PERIOD__SHIFT;
+	if (tach_period == 0)
+		return -ENOENT;
+
+	*speed = 60 * xclk * 10000 / tach_period;
+
+	return 0;
+}
+
+static int ci_fan_ctrl_set_fan_speed_rpm(struct amdgpu_device *adev,
+					 u32 speed)
+{
+	u32 tach_period, tmp;
+	u32 xclk = amdgpu_asic_get_xclk(adev);
+
+	if (adev->pm.no_fan)
+		return -ENOENT;
+
+	if (adev->pm.fan_pulses_per_revolution == 0)
+		return -ENOENT;
+
+	if ((speed < adev->pm.fan_min_rpm) ||
+	    (speed > adev->pm.fan_max_rpm))
+		return -EINVAL;
+
+	if (adev->pm.dpm.fan.ucode_fan_control)
+		ci_fan_ctrl_stop_smc_fan_control(adev);
+
+	tach_period = 60 * xclk * 10000 / (8 * speed);
+	tmp = RREG32_SMC(ixCG_TACH_CTRL) & ~CG_TACH_CTRL__TARGET_PERIOD_MASK;
+	tmp |= tach_period << CG_TACH_CTRL__TARGET_PERIOD__SHIFT;
+	WREG32_SMC(CG_TACH_CTRL, tmp);
+
+	ci_fan_ctrl_set_static_mode(adev, FDO_PWM_MODE_STATIC_RPM);
+
+	return 0;
+}
+#endif
+
+static void ci_fan_ctrl_set_default_mode(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+
+	if (!pi->fan_ctrl_is_in_default_mode) {
+		tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__FDO_PWM_MODE_MASK;
+		tmp |= pi->fan_ctrl_default_mode << CG_FDO_CTRL2__FDO_PWM_MODE__SHIFT;
+		WREG32_SMC(ixCG_FDO_CTRL2, tmp);
+
+		tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__TMIN_MASK;
+		tmp |= pi->t_min << CG_FDO_CTRL2__TMIN__SHIFT;
+		WREG32_SMC(ixCG_FDO_CTRL2, tmp);
+		pi->fan_ctrl_is_in_default_mode = true;
+	}
+}
+
+static void ci_thermal_start_smc_fan_control(struct amdgpu_device *adev)
+{
+	if (adev->pm.dpm.fan.ucode_fan_control) {
+		ci_fan_ctrl_start_smc_fan_control(adev);
+		ci_fan_ctrl_set_static_mode(adev, FDO_PWM_MODE_STATIC);
+	}
+}
+
+static void ci_thermal_initialize(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	if (adev->pm.fan_pulses_per_revolution) {
+		tmp = RREG32_SMC(ixCG_TACH_CTRL) & ~CG_TACH_CTRL__EDGE_PER_REV_MASK;
+		tmp |= (adev->pm.fan_pulses_per_revolution - 1)
+			<< CG_TACH_CTRL__EDGE_PER_REV__SHIFT;
+		WREG32_SMC(ixCG_TACH_CTRL, tmp);
+	}
+
+	tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__TACH_PWM_RESP_RATE_MASK;
+	tmp |= 0x28 << CG_FDO_CTRL2__TACH_PWM_RESP_RATE__SHIFT;
+	WREG32_SMC(ixCG_FDO_CTRL2, tmp);
+}
+
+static int ci_thermal_start_thermal_controller(struct amdgpu_device *adev)
+{
+	int ret;
+
+	ci_thermal_initialize(adev);
+	ret = ci_thermal_set_temperature_range(adev, CISLANDS_TEMP_RANGE_MIN, CISLANDS_TEMP_RANGE_MAX);
+	if (ret)
+		return ret;
+	ret = ci_thermal_enable_alert(adev, true);
+	if (ret)
+		return ret;
+	if (adev->pm.dpm.fan.ucode_fan_control) {
+		ret = ci_thermal_setup_fan_table(adev);
+		if (ret)
+			return ret;
+		ci_thermal_start_smc_fan_control(adev);
+	}
+
+	return 0;
+}
+
+static void ci_thermal_stop_thermal_controller(struct amdgpu_device *adev)
+{
+	if (!adev->pm.no_fan)
+		ci_fan_ctrl_set_default_mode(adev);
+}
+
+static int ci_read_smc_soft_register(struct amdgpu_device *adev,
+				     u16 reg_offset, u32 *value)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	return amdgpu_ci_read_smc_sram_dword(adev,
+				      pi->soft_regs_start + reg_offset,
+				      value, pi->sram_end);
+}
+
+static int ci_write_smc_soft_register(struct amdgpu_device *adev,
+				      u16 reg_offset, u32 value)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	return amdgpu_ci_write_smc_sram_dword(adev,
+				       pi->soft_regs_start + reg_offset,
+				       value, pi->sram_end);
+}
+
+static void ci_init_fps_limits(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	SMU7_Discrete_DpmTable *table = &pi->smc_state_table;
+
+	if (pi->caps_fps) {
+		u16 tmp;
+
+		tmp = 45;
+		table->FpsHighT = cpu_to_be16(tmp);
+
+		tmp = 30;
+		table->FpsLowT = cpu_to_be16(tmp);
+	}
+}
+
+static int ci_update_sclk_t(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret = 0;
+	u32 low_sclk_interrupt_t = 0;
+
+	if (pi->caps_sclk_throttle_low_notification) {
+		low_sclk_interrupt_t = cpu_to_be32(pi->low_sclk_interrupt_t);
+
+		ret = amdgpu_ci_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Discrete_DpmTable, LowSclkInterruptT),
+					   (u8 *)&low_sclk_interrupt_t,
+					   sizeof(u32), pi->sram_end);
+
+	}
+
+	return ret;
+}
+
+static void ci_get_leakage_voltages(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u16 leakage_id, virtual_voltage_id;
+	u16 vddc, vddci;
+	int i;
+
+	pi->vddc_leakage.count = 0;
+	pi->vddci_leakage.count = 0;
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_EVV) {
+		for (i = 0; i < CISLANDS_MAX_LEAKAGE_COUNT; i++) {
+			virtual_voltage_id = ATOM_VIRTUAL_VOLTAGE_ID0 + i;
+			if (amdgpu_atombios_get_voltage_evv(adev, virtual_voltage_id, &vddc) != 0)
+				continue;
+			if (vddc != 0 && vddc != virtual_voltage_id) {
+				pi->vddc_leakage.actual_voltage[pi->vddc_leakage.count] = vddc;
+				pi->vddc_leakage.leakage_id[pi->vddc_leakage.count] = virtual_voltage_id;
+				pi->vddc_leakage.count++;
+			}
+		}
+	} else if (amdgpu_atombios_get_leakage_id_from_vbios(adev, &leakage_id) == 0) {
+		for (i = 0; i < CISLANDS_MAX_LEAKAGE_COUNT; i++) {
+			virtual_voltage_id = ATOM_VIRTUAL_VOLTAGE_ID0 + i;
+			if (amdgpu_atombios_get_leakage_vddc_based_on_leakage_params(adev, &vddc, &vddci,
+										     virtual_voltage_id,
+										     leakage_id) == 0) {
+				if (vddc != 0 && vddc != virtual_voltage_id) {
+					pi->vddc_leakage.actual_voltage[pi->vddc_leakage.count] = vddc;
+					pi->vddc_leakage.leakage_id[pi->vddc_leakage.count] = virtual_voltage_id;
+					pi->vddc_leakage.count++;
+				}
+				if (vddci != 0 && vddci != virtual_voltage_id) {
+					pi->vddci_leakage.actual_voltage[pi->vddci_leakage.count] = vddci;
+					pi->vddci_leakage.leakage_id[pi->vddci_leakage.count] = virtual_voltage_id;
+					pi->vddci_leakage.count++;
+				}
+			}
+		}
+	}
+}
+
+static void ci_set_dpm_event_sources(struct amdgpu_device *adev, u32 sources)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	bool want_thermal_protection;
+	enum amdgpu_dpm_event_src dpm_event_src;
+	u32 tmp;
+
+	switch (sources) {
+	case 0:
+	default:
+		want_thermal_protection = false;
+		break;
+	case (1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL):
+		want_thermal_protection = true;
+		dpm_event_src = AMDGPU_DPM_EVENT_SRC_DIGITAL;
+		break;
+	case (1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_EXTERNAL):
+		want_thermal_protection = true;
+		dpm_event_src = AMDGPU_DPM_EVENT_SRC_EXTERNAL;
+		break;
+	case ((1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_EXTERNAL) |
+	      (1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL)):
+		want_thermal_protection = true;
+		dpm_event_src = AMDGPU_DPM_EVENT_SRC_DIGIAL_OR_EXTERNAL;
+		break;
+	}
+
+	if (want_thermal_protection) {
+#if 0
+		/* XXX: need to figure out how to handle this properly */
+		tmp = RREG32_SMC(ixCG_THERMAL_CTRL);
+		tmp &= DPM_EVENT_SRC_MASK;
+		tmp |= DPM_EVENT_SRC(dpm_event_src);
+		WREG32_SMC(ixCG_THERMAL_CTRL, tmp);
+#endif
+
+		tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+		if (pi->thermal_protection)
+			tmp &= ~GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
+		else
+			tmp |= GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
+		WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+	} else {
+		tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+		tmp |= GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
+		WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+	}
+}
+
+static void ci_enable_auto_throttle_source(struct amdgpu_device *adev,
+					   enum amdgpu_dpm_auto_throttle_src source,
+					   bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (enable) {
+		if (!(pi->active_auto_throttle_sources & (1 << source))) {
+			pi->active_auto_throttle_sources |= 1 << source;
+			ci_set_dpm_event_sources(adev, pi->active_auto_throttle_sources);
+		}
+	} else {
+		if (pi->active_auto_throttle_sources & (1 << source)) {
+			pi->active_auto_throttle_sources &= ~(1 << source);
+			ci_set_dpm_event_sources(adev, pi->active_auto_throttle_sources);
+		}
+	}
+}
+
+static void ci_enable_vr_hot_gpio_interrupt(struct amdgpu_device *adev)
+{
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_REGULATOR_HOT)
+		amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableVRHotGPIOInterrupt);
+}
+
+static int ci_unfreeze_sclk_mclk_dpm(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+
+	if (!pi->need_update_smu7_dpm_table)
+		return 0;
+
+	if ((!pi->sclk_dpm_key_disabled) &&
+	    (pi->need_update_smu7_dpm_table & (DPMTABLE_OD_UPDATE_SCLK | DPMTABLE_UPDATE_SCLK))) {
+		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_SCLKDPM_UnfreezeLevel);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	if ((!pi->mclk_dpm_key_disabled) &&
+	    (pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_MCLK)) {
+		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MCLKDPM_UnfreezeLevel);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	pi->need_update_smu7_dpm_table = 0;
+	return 0;
+}
+
+static int ci_enable_sclk_mclk_dpm(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+
+	if (enable) {
+		if (!pi->sclk_dpm_key_disabled) {
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DPM_Enable);
+			if (smc_result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+
+		if (!pi->mclk_dpm_key_disabled) {
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MCLKDPM_Enable);
+			if (smc_result != PPSMC_Result_OK)
+				return -EINVAL;
+
+			WREG32_P(mmMC_SEQ_CNTL_3, MC_SEQ_CNTL_3__CAC_EN_MASK,
+					~MC_SEQ_CNTL_3__CAC_EN_MASK);
+
+			WREG32_SMC(ixLCAC_MC0_CNTL, 0x05);
+			WREG32_SMC(ixLCAC_MC1_CNTL, 0x05);
+			WREG32_SMC(ixLCAC_CPL_CNTL, 0x100005);
+
+			udelay(10);
+
+			WREG32_SMC(ixLCAC_MC0_CNTL, 0x400005);
+			WREG32_SMC(ixLCAC_MC1_CNTL, 0x400005);
+			WREG32_SMC(ixLCAC_CPL_CNTL, 0x500005);
+		}
+	} else {
+		if (!pi->sclk_dpm_key_disabled) {
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DPM_Disable);
+			if (smc_result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+
+		if (!pi->mclk_dpm_key_disabled) {
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MCLKDPM_Disable);
+			if (smc_result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static int ci_start_dpm(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+	int ret;
+	u32 tmp;
+
+	tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+	tmp |= GENERAL_PWRMGT__GLOBAL_PWRMGT_EN_MASK;
+	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+
+	tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+	tmp |= SCLK_PWRMGT_CNTL__DYNAMIC_PM_EN_MASK;
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
+
+	ci_write_smc_soft_register(adev, offsetof(SMU7_SoftRegisters, VoltageChangeTimeout), 0x1000);
+
+	WREG32_P(mmBIF_LNCNT_RESET, 0, ~BIF_LNCNT_RESET__RESET_LNCNT_EN_MASK);
+
+	smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Voltage_Cntl_Enable);
+	if (smc_result != PPSMC_Result_OK)
+		return -EINVAL;
+
+	ret = ci_enable_sclk_mclk_dpm(adev, true);
+	if (ret)
+		return ret;
+
+	if (!pi->pcie_dpm_key_disabled) {
+		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PCIeDPM_Enable);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_freeze_sclk_mclk_dpm(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+
+	if (!pi->need_update_smu7_dpm_table)
+		return 0;
+
+	if ((!pi->sclk_dpm_key_disabled) &&
+	    (pi->need_update_smu7_dpm_table & (DPMTABLE_OD_UPDATE_SCLK | DPMTABLE_UPDATE_SCLK))) {
+		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_SCLKDPM_FreezeLevel);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	if ((!pi->mclk_dpm_key_disabled) &&
+	    (pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_MCLK)) {
+		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MCLKDPM_FreezeLevel);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_stop_dpm(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+	int ret;
+	u32 tmp;
+
+	tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+	tmp &= ~GENERAL_PWRMGT__GLOBAL_PWRMGT_EN_MASK;
+	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+
+	tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+	tmp &= ~SCLK_PWRMGT_CNTL__DYNAMIC_PM_EN_MASK;
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
+
+	if (!pi->pcie_dpm_key_disabled) {
+		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PCIeDPM_Disable);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	ret = ci_enable_sclk_mclk_dpm(adev, false);
+	if (ret)
+		return ret;
+
+	smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Voltage_Cntl_Disable);
+	if (smc_result != PPSMC_Result_OK)
+		return -EINVAL;
+
+	return 0;
+}
+
+static void ci_enable_sclk_control(struct amdgpu_device *adev, bool enable)
+{
+	u32 tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+
+	if (enable)
+		tmp &= ~SCLK_PWRMGT_CNTL__SCLK_PWRMGT_OFF_MASK;
+	else
+		tmp |= SCLK_PWRMGT_CNTL__SCLK_PWRMGT_OFF_MASK;
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
+}
+
+#if 0
+static int ci_notify_hw_of_power_source(struct amdgpu_device *adev,
+					bool ac_power)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_cac_tdp_table *cac_tdp_table =
+		adev->pm.dpm.dyn_state.cac_tdp_table;
+	u32 power_limit;
+
+	if (ac_power)
+		power_limit = (u32)(cac_tdp_table->maximum_power_delivery_limit * 256);
+	else
+		power_limit = (u32)(cac_tdp_table->battery_power_limit * 256);
+
+	ci_set_power_limit(adev, power_limit);
+
+	if (pi->caps_automatic_dc_transition) {
+		if (ac_power)
+			amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_RunningOnAC);
+		else
+			amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Remove_DC_Clamp);
+	}
+
+	return 0;
+}
+#endif
+
+static PPSMC_Result amdgpu_ci_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+						      PPSMC_Msg msg, u32 parameter)
+{
+	WREG32(mmSMC_MSG_ARG_0, parameter);
+	return amdgpu_ci_send_msg_to_smc(adev, msg);
+}
+
+static PPSMC_Result amdgpu_ci_send_msg_to_smc_return_parameter(struct amdgpu_device *adev,
+							PPSMC_Msg msg, u32 *parameter)
+{
+	PPSMC_Result smc_result;
+
+	smc_result = amdgpu_ci_send_msg_to_smc(adev, msg);
+
+	if ((smc_result == PPSMC_Result_OK) && parameter)
+		*parameter = RREG32(mmSMC_MSG_ARG_0);
+
+	return smc_result;
+}
+
+static int ci_dpm_force_state_sclk(struct amdgpu_device *adev, u32 n)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (!pi->sclk_dpm_key_disabled) {
+		PPSMC_Result smc_result =
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SCLKDPM_SetEnabledMask, 1 << n);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_dpm_force_state_mclk(struct amdgpu_device *adev, u32 n)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (!pi->mclk_dpm_key_disabled) {
+		PPSMC_Result smc_result =
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_MCLKDPM_SetEnabledMask, 1 << n);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_dpm_force_state_pcie(struct amdgpu_device *adev, u32 n)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (!pi->pcie_dpm_key_disabled) {
+		PPSMC_Result smc_result =
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_PCIeDPM_ForceLevel, n);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_set_power_limit(struct amdgpu_device *adev, u32 n)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (pi->power_containment_features & POWERCONTAINMENT_FEATURE_PkgPwrLimit) {
+		PPSMC_Result smc_result =
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_PkgPwrSetLimit, n);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_set_overdrive_target_tdp(struct amdgpu_device *adev,
+				       u32 target_tdp)
+{
+	PPSMC_Result smc_result =
+		amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_OverDriveSetTargetTdp, target_tdp);
+	if (smc_result != PPSMC_Result_OK)
+		return -EINVAL;
+	return 0;
+}
+
+#if 0
+static int ci_set_boot_state(struct amdgpu_device *adev)
+{
+	return ci_enable_sclk_mclk_dpm(adev, false);
+}
+#endif
+
+static u32 ci_get_average_sclk_freq(struct amdgpu_device *adev)
+{
+	u32 sclk_freq;
+	PPSMC_Result smc_result =
+		amdgpu_ci_send_msg_to_smc_return_parameter(adev,
+						    PPSMC_MSG_API_GetSclkFrequency,
+						    &sclk_freq);
+	if (smc_result != PPSMC_Result_OK)
+		sclk_freq = 0;
+
+	return sclk_freq;
+}
+
+static u32 ci_get_average_mclk_freq(struct amdgpu_device *adev)
+{
+	u32 mclk_freq;
+	PPSMC_Result smc_result =
+		amdgpu_ci_send_msg_to_smc_return_parameter(adev,
+						    PPSMC_MSG_API_GetMclkFrequency,
+						    &mclk_freq);
+	if (smc_result != PPSMC_Result_OK)
+		mclk_freq = 0;
+
+	return mclk_freq;
+}
+
+static void ci_dpm_start_smc(struct amdgpu_device *adev)
+{
+	int i;
+
+	amdgpu_ci_program_jump_on_start(adev);
+	amdgpu_ci_start_smc_clock(adev);
+	amdgpu_ci_start_smc(adev);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32_SMC(ixFIRMWARE_FLAGS) & FIRMWARE_FLAGS__INTERRUPTS_ENABLED_MASK)
+			break;
+	}
+}
+
+static void ci_dpm_stop_smc(struct amdgpu_device *adev)
+{
+	amdgpu_ci_reset_smc(adev);
+	amdgpu_ci_stop_smc_clock(adev);
+}
+
+static int ci_process_firmware_header(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+	int ret;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev,
+				     SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, DpmTable),
+				     &tmp, pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->dpm_table_start = tmp;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev,
+				     SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, SoftRegisters),
+				     &tmp, pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->soft_regs_start = tmp;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev,
+				     SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, mcRegisterTable),
+				     &tmp, pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->mc_reg_table_start = tmp;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev,
+				     SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, FanTable),
+				     &tmp, pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->fan_table_start = tmp;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev,
+				     SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, mcArbDramTimingTable),
+				     &tmp, pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->arb_table_start = tmp;
+
+	return 0;
+}
+
+static void ci_read_clock_registers(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	pi->clock_registers.cg_spll_func_cntl =
+		RREG32_SMC(ixCG_SPLL_FUNC_CNTL);
+	pi->clock_registers.cg_spll_func_cntl_2 =
+		RREG32_SMC(ixCG_SPLL_FUNC_CNTL_2);
+	pi->clock_registers.cg_spll_func_cntl_3 =
+		RREG32_SMC(ixCG_SPLL_FUNC_CNTL_3);
+	pi->clock_registers.cg_spll_func_cntl_4 =
+		RREG32_SMC(ixCG_SPLL_FUNC_CNTL_4);
+	pi->clock_registers.cg_spll_spread_spectrum =
+		RREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM);
+	pi->clock_registers.cg_spll_spread_spectrum_2 =
+		RREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM_2);
+	pi->clock_registers.dll_cntl = RREG32(mmDLL_CNTL);
+	pi->clock_registers.mclk_pwrmgt_cntl = RREG32(mmMCLK_PWRMGT_CNTL);
+	pi->clock_registers.mpll_ad_func_cntl = RREG32(mmMPLL_AD_FUNC_CNTL);
+	pi->clock_registers.mpll_dq_func_cntl = RREG32(mmMPLL_DQ_FUNC_CNTL);
+	pi->clock_registers.mpll_func_cntl = RREG32(mmMPLL_FUNC_CNTL);
+	pi->clock_registers.mpll_func_cntl_1 = RREG32(mmMPLL_FUNC_CNTL_1);
+	pi->clock_registers.mpll_func_cntl_2 = RREG32(mmMPLL_FUNC_CNTL_2);
+	pi->clock_registers.mpll_ss1 = RREG32(mmMPLL_SS1);
+	pi->clock_registers.mpll_ss2 = RREG32(mmMPLL_SS2);
+}
+
+static void ci_init_sclk_t(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	pi->low_sclk_interrupt_t = 0;
+}
+
+static void ci_enable_thermal_protection(struct amdgpu_device *adev,
+					 bool enable)
+{
+	u32 tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+
+	if (enable)
+		tmp &= ~GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
+	else
+		tmp |= GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
+	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+}
+
+static void ci_enable_acpi_power_management(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+
+	tmp |= GENERAL_PWRMGT__STATIC_PM_EN_MASK;
+
+	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+}
+
+#if 0
+static int ci_enter_ulp_state(struct amdgpu_device *adev)
+{
+
+	WREG32(mmSMC_MESSAGE_0, PPSMC_MSG_SwitchToMinimumPower);
+
+	udelay(25000);
+
+	return 0;
+}
+
+static int ci_exit_ulp_state(struct amdgpu_device *adev)
+{
+	int i;
+
+	WREG32(mmSMC_MESSAGE_0, PPSMC_MSG_ResumeFromMinimumPower);
+
+	udelay(7000);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmSMC_RESP_0) == 1)
+			break;
+		udelay(1000);
+	}
+
+	return 0;
+}
+#endif
+
+static int ci_notify_smc_display_change(struct amdgpu_device *adev,
+					bool has_display)
+{
+	PPSMC_Msg msg = has_display ? PPSMC_MSG_HasDisplay : PPSMC_MSG_NoDisplay;
+
+	return (amdgpu_ci_send_msg_to_smc(adev, msg) == PPSMC_Result_OK) ?  0 : -EINVAL;
+}
+
+static int ci_enable_ds_master_switch(struct amdgpu_device *adev,
+				      bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (enable) {
+		if (pi->caps_sclk_ds) {
+			if (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MASTER_DeepSleep_ON) != PPSMC_Result_OK)
+				return -EINVAL;
+		} else {
+			if (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MASTER_DeepSleep_OFF) != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+	} else {
+		if (pi->caps_sclk_ds) {
+			if (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MASTER_DeepSleep_OFF) != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static void ci_program_display_gap(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixCG_DISPLAY_GAP_CNTL);
+	u32 pre_vbi_time_in_us;
+	u32 frame_time_in_us;
+	u32 ref_clock = adev->clock.spll.reference_freq;
+	u32 refresh_rate = amdgpu_dpm_get_vrefresh(adev);
+	u32 vblank_time = amdgpu_dpm_get_vblank_time(adev);
+
+	tmp &= ~CG_DISPLAY_GAP_CNTL__DISP_GAP_MASK;
+	if (adev->pm.dpm.new_active_crtc_count > 0)
+		tmp |= (AMDGPU_PM_DISPLAY_GAP_VBLANK_OR_WM << CG_DISPLAY_GAP_CNTL__DISP_GAP__SHIFT);
+	else
+		tmp |= (AMDGPU_PM_DISPLAY_GAP_IGNORE << CG_DISPLAY_GAP_CNTL__DISP_GAP__SHIFT);
+	WREG32_SMC(ixCG_DISPLAY_GAP_CNTL, tmp);
+
+	if (refresh_rate == 0)
+		refresh_rate = 60;
+	if (vblank_time == 0xffffffff)
+		vblank_time = 500;
+	frame_time_in_us = 1000000 / refresh_rate;
+	pre_vbi_time_in_us =
+		frame_time_in_us - 200 - vblank_time;
+	tmp = pre_vbi_time_in_us * (ref_clock / 100);
+
+	WREG32_SMC(ixCG_DISPLAY_GAP_CNTL2, tmp);
+	ci_write_smc_soft_register(adev, offsetof(SMU7_SoftRegisters, PreVBlankGap), 0x64);
+	ci_write_smc_soft_register(adev, offsetof(SMU7_SoftRegisters, VBlankTimeout), (frame_time_in_us - pre_vbi_time_in_us));
+
+
+	ci_notify_smc_display_change(adev, (adev->pm.dpm.new_active_crtc_count == 1));
+
+}
+
+static void ci_enable_spread_spectrum(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+
+	if (enable) {
+		if (pi->caps_sclk_ss_support) {
+			tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+			tmp |= GENERAL_PWRMGT__DYN_SPREAD_SPECTRUM_EN_MASK;
+			WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+		}
+	} else {
+		tmp = RREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM);
+		tmp &= ~CG_SPLL_SPREAD_SPECTRUM__SSEN_MASK;
+		WREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM, tmp);
+
+		tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+		tmp &= ~GENERAL_PWRMGT__DYN_SPREAD_SPECTRUM_EN_MASK;
+		WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+	}
+}
+
+static void ci_program_sstp(struct amdgpu_device *adev)
+{
+	WREG32_SMC(ixCG_STATIC_SCREEN_PARAMETER,
+	((CISLANDS_SSTU_DFLT << CG_STATIC_SCREEN_PARAMETER__STATIC_SCREEN_THRESHOLD_UNIT__SHIFT) |
+	 (CISLANDS_SST_DFLT << CG_STATIC_SCREEN_PARAMETER__STATIC_SCREEN_THRESHOLD__SHIFT)));
+}
+
+static void ci_enable_display_gap(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixCG_DISPLAY_GAP_CNTL);
+
+	tmp &= ~(CG_DISPLAY_GAP_CNTL__DISP_GAP_MASK |
+			CG_DISPLAY_GAP_CNTL__DISP_GAP_MCHG_MASK);
+	tmp |= ((AMDGPU_PM_DISPLAY_GAP_IGNORE << CG_DISPLAY_GAP_CNTL__DISP_GAP__SHIFT) |
+		(AMDGPU_PM_DISPLAY_GAP_VBLANK << CG_DISPLAY_GAP_CNTL__DISP_GAP_MCHG__SHIFT));
+
+	WREG32_SMC(ixCG_DISPLAY_GAP_CNTL, tmp);
+}
+
+static void ci_program_vc(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+	tmp &= ~(SCLK_PWRMGT_CNTL__RESET_SCLK_CNT_MASK | SCLK_PWRMGT_CNTL__RESET_BUSY_CNT_MASK);
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
+
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_0, CISLANDS_VRC_DFLT0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_1, CISLANDS_VRC_DFLT1);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_2, CISLANDS_VRC_DFLT2);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_3, CISLANDS_VRC_DFLT3);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_4, CISLANDS_VRC_DFLT4);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_5, CISLANDS_VRC_DFLT5);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_6, CISLANDS_VRC_DFLT6);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_7, CISLANDS_VRC_DFLT7);
+}
+
+static void ci_clear_vc(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+	tmp |= (SCLK_PWRMGT_CNTL__RESET_SCLK_CNT_MASK | SCLK_PWRMGT_CNTL__RESET_BUSY_CNT_MASK);
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
+
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_0, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_1, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_2, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_3, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_4, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_5, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_6, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_7, 0);
+}
+
+static int ci_upload_firmware(struct amdgpu_device *adev)
+{
+	int i, ret;
+
+	if (amdgpu_ci_is_smc_running(adev)) {
+		DRM_INFO("smc is running, no need to load smc firmware\n");
+		return 0;
+	}
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32_SMC(ixRCU_UC_EVENTS) & RCU_UC_EVENTS__boot_seq_done_MASK)
+			break;
+	}
+	WREG32_SMC(ixSMC_SYSCON_MISC_CNTL, 1);
+
+	amdgpu_ci_stop_smc_clock(adev);
+	amdgpu_ci_reset_smc(adev);
+
+	ret = amdgpu_ci_load_smc_ucode(adev, SMC_RAM_END);
+
+	return ret;
+
+}
+
+static int ci_get_svi2_voltage_table(struct amdgpu_device *adev,
+				     struct amdgpu_clock_voltage_dependency_table *voltage_dependency_table,
+				     struct atom_voltage_table *voltage_table)
+{
+	u32 i;
+
+	if (voltage_dependency_table == NULL)
+		return -EINVAL;
+
+	voltage_table->mask_low = 0;
+	voltage_table->phase_delay = 0;
+
+	voltage_table->count = voltage_dependency_table->count;
+	for (i = 0; i < voltage_table->count; i++) {
+		voltage_table->entries[i].value = voltage_dependency_table->entries[i].v;
+		voltage_table->entries[i].smio_low = 0;
+	}
+
+	return 0;
+}
+
+static int ci_construct_voltage_tables(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret;
+
+	if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO) {
+		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_VDDC,
+							VOLTAGE_OBJ_GPIO_LUT,
+							&pi->vddc_voltage_table);
+		if (ret)
+			return ret;
+	} else if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2) {
+		ret = ci_get_svi2_voltage_table(adev,
+						&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk,
+						&pi->vddc_voltage_table);
+		if (ret)
+			return ret;
+	}
+
+	if (pi->vddc_voltage_table.count > SMU7_MAX_LEVELS_VDDC)
+		ci_trim_voltage_table_to_fit_state_table(adev, SMU7_MAX_LEVELS_VDDC,
+							 &pi->vddc_voltage_table);
+
+	if (pi->vddci_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO) {
+		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_VDDCI,
+							VOLTAGE_OBJ_GPIO_LUT,
+							&pi->vddci_voltage_table);
+		if (ret)
+			return ret;
+	} else if (pi->vddci_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2) {
+		ret = ci_get_svi2_voltage_table(adev,
+						&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk,
+						&pi->vddci_voltage_table);
+		if (ret)
+			return ret;
+	}
+
+	if (pi->vddci_voltage_table.count > SMU7_MAX_LEVELS_VDDCI)
+		ci_trim_voltage_table_to_fit_state_table(adev, SMU7_MAX_LEVELS_VDDCI,
+							 &pi->vddci_voltage_table);
+
+	if (pi->mvdd_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO) {
+		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_MVDDC,
+							VOLTAGE_OBJ_GPIO_LUT,
+							&pi->mvdd_voltage_table);
+		if (ret)
+			return ret;
+	} else if (pi->mvdd_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2) {
+		ret = ci_get_svi2_voltage_table(adev,
+						&adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk,
+						&pi->mvdd_voltage_table);
+		if (ret)
+			return ret;
+	}
+
+	if (pi->mvdd_voltage_table.count > SMU7_MAX_LEVELS_MVDD)
+		ci_trim_voltage_table_to_fit_state_table(adev, SMU7_MAX_LEVELS_MVDD,
+							 &pi->mvdd_voltage_table);
+
+	return 0;
+}
+
+static void ci_populate_smc_voltage_table(struct amdgpu_device *adev,
+					  struct atom_voltage_table_entry *voltage_table,
+					  SMU7_Discrete_VoltageLevel *smc_voltage_table)
+{
+	int ret;
+
+	ret = ci_get_std_voltage_value_sidd(adev, voltage_table,
+					    &smc_voltage_table->StdVoltageHiSidd,
+					    &smc_voltage_table->StdVoltageLoSidd);
+
+	if (ret) {
+		smc_voltage_table->StdVoltageHiSidd = voltage_table->value * VOLTAGE_SCALE;
+		smc_voltage_table->StdVoltageLoSidd = voltage_table->value * VOLTAGE_SCALE;
+	}
+
+	smc_voltage_table->Voltage = cpu_to_be16(voltage_table->value * VOLTAGE_SCALE);
+	smc_voltage_table->StdVoltageHiSidd =
+		cpu_to_be16(smc_voltage_table->StdVoltageHiSidd);
+	smc_voltage_table->StdVoltageLoSidd =
+		cpu_to_be16(smc_voltage_table->StdVoltageLoSidd);
+}
+
+static int ci_populate_smc_vddc_table(struct amdgpu_device *adev,
+				      SMU7_Discrete_DpmTable *table)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	unsigned int count;
+
+	table->VddcLevelCount = pi->vddc_voltage_table.count;
+	for (count = 0; count < table->VddcLevelCount; count++) {
+		ci_populate_smc_voltage_table(adev,
+					      &pi->vddc_voltage_table.entries[count],
+					      &table->VddcLevel[count]);
+
+		if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO)
+			table->VddcLevel[count].Smio |=
+				pi->vddc_voltage_table.entries[count].smio_low;
+		else
+			table->VddcLevel[count].Smio = 0;
+	}
+	table->VddcLevelCount = cpu_to_be32(table->VddcLevelCount);
+
+	return 0;
+}
+
+static int ci_populate_smc_vddci_table(struct amdgpu_device *adev,
+				       SMU7_Discrete_DpmTable *table)
+{
+	unsigned int count;
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	table->VddciLevelCount = pi->vddci_voltage_table.count;
+	for (count = 0; count < table->VddciLevelCount; count++) {
+		ci_populate_smc_voltage_table(adev,
+					      &pi->vddci_voltage_table.entries[count],
+					      &table->VddciLevel[count]);
+
+		if (pi->vddci_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO)
+			table->VddciLevel[count].Smio |=
+				pi->vddci_voltage_table.entries[count].smio_low;
+		else
+			table->VddciLevel[count].Smio = 0;
+	}
+	table->VddciLevelCount = cpu_to_be32(table->VddciLevelCount);
+
+	return 0;
+}
+
+static int ci_populate_smc_mvdd_table(struct amdgpu_device *adev,
+				      SMU7_Discrete_DpmTable *table)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	unsigned int count;
+
+	table->MvddLevelCount = pi->mvdd_voltage_table.count;
+	for (count = 0; count < table->MvddLevelCount; count++) {
+		ci_populate_smc_voltage_table(adev,
+					      &pi->mvdd_voltage_table.entries[count],
+					      &table->MvddLevel[count]);
+
+		if (pi->mvdd_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO)
+			table->MvddLevel[count].Smio |=
+				pi->mvdd_voltage_table.entries[count].smio_low;
+		else
+			table->MvddLevel[count].Smio = 0;
+	}
+	table->MvddLevelCount = cpu_to_be32(table->MvddLevelCount);
+
+	return 0;
+}
+
+static int ci_populate_smc_voltage_tables(struct amdgpu_device *adev,
+					  SMU7_Discrete_DpmTable *table)
+{
+	int ret;
+
+	ret = ci_populate_smc_vddc_table(adev, table);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_smc_vddci_table(adev, table);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_smc_mvdd_table(adev, table);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int ci_populate_mvdd_value(struct amdgpu_device *adev, u32 mclk,
+				  SMU7_Discrete_VoltageLevel *voltage)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 i = 0;
+
+	if (pi->mvdd_control != CISLANDS_VOLTAGE_CONTROL_NONE) {
+		for (i = 0; i < adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk.count; i++) {
+			if (mclk <= adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk.entries[i].clk) {
+				voltage->Voltage = pi->mvdd_voltage_table.entries[i].value;
+				break;
+			}
+		}
+
+		if (i >= adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk.count)
+			return -EINVAL;
+	}
+
+	return -EINVAL;
+}
+
+static int ci_get_std_voltage_value_sidd(struct amdgpu_device *adev,
+					 struct atom_voltage_table_entry *voltage_table,
+					 u16 *std_voltage_hi_sidd, u16 *std_voltage_lo_sidd)
+{
+	u16 v_index, idx;
+	bool voltage_found = false;
+	*std_voltage_hi_sidd = voltage_table->value * VOLTAGE_SCALE;
+	*std_voltage_lo_sidd = voltage_table->value * VOLTAGE_SCALE;
+
+	if (adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries == NULL)
+		return -EINVAL;
+
+	if (adev->pm.dpm.dyn_state.cac_leakage_table.entries) {
+		for (v_index = 0; (u32)v_index < adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count; v_index++) {
+			if (voltage_table->value ==
+			    adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[v_index].v) {
+				voltage_found = true;
+				if ((u32)v_index < adev->pm.dpm.dyn_state.cac_leakage_table.count)
+					idx = v_index;
+				else
+					idx = adev->pm.dpm.dyn_state.cac_leakage_table.count - 1;
+				*std_voltage_lo_sidd =
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[idx].vddc * VOLTAGE_SCALE;
+				*std_voltage_hi_sidd =
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[idx].leakage * VOLTAGE_SCALE;
+				break;
+			}
+		}
+
+		if (!voltage_found) {
+			for (v_index = 0; (u32)v_index < adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count; v_index++) {
+				if (voltage_table->value <=
+				    adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[v_index].v) {
+					voltage_found = true;
+					if ((u32)v_index < adev->pm.dpm.dyn_state.cac_leakage_table.count)
+						idx = v_index;
+					else
+						idx = adev->pm.dpm.dyn_state.cac_leakage_table.count - 1;
+					*std_voltage_lo_sidd =
+						adev->pm.dpm.dyn_state.cac_leakage_table.entries[idx].vddc * VOLTAGE_SCALE;
+					*std_voltage_hi_sidd =
+						adev->pm.dpm.dyn_state.cac_leakage_table.entries[idx].leakage * VOLTAGE_SCALE;
+					break;
+				}
+			}
+		}
+	}
+
+	return 0;
+}
+
+static void ci_populate_phase_value_based_on_sclk(struct amdgpu_device *adev,
+						  const struct amdgpu_phase_shedding_limits_table *limits,
+						  u32 sclk,
+						  u32 *phase_shedding)
+{
+	unsigned int i;
+
+	*phase_shedding = 1;
+
+	for (i = 0; i < limits->count; i++) {
+		if (sclk < limits->entries[i].sclk) {
+			*phase_shedding = i;
+			break;
+		}
+	}
+}
+
+static void ci_populate_phase_value_based_on_mclk(struct amdgpu_device *adev,
+						  const struct amdgpu_phase_shedding_limits_table *limits,
+						  u32 mclk,
+						  u32 *phase_shedding)
+{
+	unsigned int i;
+
+	*phase_shedding = 1;
+
+	for (i = 0; i < limits->count; i++) {
+		if (mclk < limits->entries[i].mclk) {
+			*phase_shedding = i;
+			break;
+		}
+	}
+}
+
+static int ci_init_arb_table_index(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+	int ret;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev, pi->arb_table_start,
+				     &tmp, pi->sram_end);
+	if (ret)
+		return ret;
+
+	tmp &= 0x00FFFFFF;
+	tmp |= MC_CG_ARB_FREQ_F1 << 24;
+
+	return amdgpu_ci_write_smc_sram_dword(adev, pi->arb_table_start,
+				       tmp, pi->sram_end);
+}
+
+static int ci_get_dependency_volt_by_clk(struct amdgpu_device *adev,
+					 struct amdgpu_clock_voltage_dependency_table *allowed_clock_voltage_table,
+					 u32 clock, u32 *voltage)
+{
+	u32 i = 0;
+
+	if (allowed_clock_voltage_table->count == 0)
+		return -EINVAL;
+
+	for (i = 0; i < allowed_clock_voltage_table->count; i++) {
+		if (allowed_clock_voltage_table->entries[i].clk >= clock) {
+			*voltage = allowed_clock_voltage_table->entries[i].v;
+			return 0;
+		}
+	}
+
+	*voltage = allowed_clock_voltage_table->entries[i-1].v;
+
+	return 0;
+}
+
+static u8 ci_get_sleep_divider_id_from_clock(u32 sclk, u32 min_sclk_in_sr)
+{
+	u32 i;
+	u32 tmp;
+	u32 min = max(min_sclk_in_sr, (u32)CISLAND_MINIMUM_ENGINE_CLOCK);
+
+	if (sclk < min)
+		return 0;
+
+	for (i = CISLAND_MAX_DEEPSLEEP_DIVIDER_ID;  ; i--) {
+		tmp = sclk >> i;
+		if (tmp >= min || i == 0)
+			break;
+	}
+
+	return (u8)i;
+}
+
+static int ci_initial_switch_from_arb_f0_to_f1(struct amdgpu_device *adev)
+{
+	return ci_copy_and_switch_arb_sets(adev, MC_CG_ARB_FREQ_F0, MC_CG_ARB_FREQ_F1);
+}
+
+static int ci_reset_to_default(struct amdgpu_device *adev)
+{
+	return (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_ResetToDefaults) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+static int ci_force_switch_to_arb_f0(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = (RREG32_SMC(ixSMC_SCRATCH9) & 0x0000ff00) >> 8;
+
+	if (tmp == MC_CG_ARB_FREQ_F0)
+		return 0;
+
+	return ci_copy_and_switch_arb_sets(adev, tmp, MC_CG_ARB_FREQ_F0);
+}
+
+static void ci_register_patching_mc_arb(struct amdgpu_device *adev,
+					const u32 engine_clock,
+					const u32 memory_clock,
+					u32 *dram_timimg2)
+{
+	bool patch;
+	u32 tmp, tmp2;
+
+	tmp = RREG32(mmMC_SEQ_MISC0);
+	patch = ((tmp & 0x0000f00) == 0x300) ? true : false;
+
+	if (patch &&
+	    ((adev->pdev->device == 0x67B0) ||
+	     (adev->pdev->device == 0x67B1))) {
+		if ((memory_clock > 100000) && (memory_clock <= 125000)) {
+			tmp2 = (((0x31 * engine_clock) / 125000) - 1) & 0xff;
+			*dram_timimg2 &= ~0x00ff0000;
+			*dram_timimg2 |= tmp2 << 16;
+		} else if ((memory_clock > 125000) && (memory_clock <= 137500)) {
+			tmp2 = (((0x36 * engine_clock) / 137500) - 1) & 0xff;
+			*dram_timimg2 &= ~0x00ff0000;
+			*dram_timimg2 |= tmp2 << 16;
+		}
+	}
+}
+
+static int ci_populate_memory_timing_parameters(struct amdgpu_device *adev,
+						u32 sclk,
+						u32 mclk,
+						SMU7_Discrete_MCArbDramTimingTableEntry *arb_regs)
+{
+	u32 dram_timing;
+	u32 dram_timing2;
+	u32 burst_time;
+
+	amdgpu_atombios_set_engine_dram_timings(adev, sclk, mclk);
+
+	dram_timing  = RREG32(mmMC_ARB_DRAM_TIMING);
+	dram_timing2 = RREG32(mmMC_ARB_DRAM_TIMING2);
+	burst_time = RREG32(mmMC_ARB_BURST_TIME) & MC_ARB_BURST_TIME__STATE0_MASK;
+
+	ci_register_patching_mc_arb(adev, sclk, mclk, &dram_timing2);
+
+	arb_regs->McArbDramTiming  = cpu_to_be32(dram_timing);
+	arb_regs->McArbDramTiming2 = cpu_to_be32(dram_timing2);
+	arb_regs->McArbBurstTime = (u8)burst_time;
+
+	return 0;
+}
+
+static int ci_do_program_memory_timing_parameters(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	SMU7_Discrete_MCArbDramTimingTable arb_regs;
+	u32 i, j;
+	int ret =  0;
+
+	memset(&arb_regs, 0, sizeof(SMU7_Discrete_MCArbDramTimingTable));
+
+	for (i = 0; i < pi->dpm_table.sclk_table.count; i++) {
+		for (j = 0; j < pi->dpm_table.mclk_table.count; j++) {
+			ret = ci_populate_memory_timing_parameters(adev,
+								   pi->dpm_table.sclk_table.dpm_levels[i].value,
+								   pi->dpm_table.mclk_table.dpm_levels[j].value,
+								   &arb_regs.entries[i][j]);
+			if (ret)
+				break;
+		}
+	}
+
+	if (ret == 0)
+		ret = amdgpu_ci_copy_bytes_to_smc(adev,
+					   pi->arb_table_start,
+					   (u8 *)&arb_regs,
+					   sizeof(SMU7_Discrete_MCArbDramTimingTable),
+					   pi->sram_end);
+
+	return ret;
+}
+
+static int ci_program_memory_timing_parameters(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (pi->need_update_smu7_dpm_table == 0)
+		return 0;
+
+	return ci_do_program_memory_timing_parameters(adev);
+}
+
+static void ci_populate_smc_initial_state(struct amdgpu_device *adev,
+					  struct amdgpu_ps *amdgpu_boot_state)
+{
+	struct ci_ps *boot_state = ci_get_ps(amdgpu_boot_state);
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 level = 0;
+
+	for (level = 0; level < adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count; level++) {
+		if (adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[level].clk >=
+		    boot_state->performance_levels[0].sclk) {
+			pi->smc_state_table.GraphicsBootLevel = level;
+			break;
+		}
+	}
+
+	for (level = 0; level < adev->pm.dpm.dyn_state.vddc_dependency_on_mclk.count; level++) {
+		if (adev->pm.dpm.dyn_state.vddc_dependency_on_mclk.entries[level].clk >=
+		    boot_state->performance_levels[0].mclk) {
+			pi->smc_state_table.MemoryBootLevel = level;
+			break;
+		}
+	}
+}
+
+static u32 ci_get_dpm_level_enable_mask_value(struct ci_single_dpm_table *dpm_table)
+{
+	u32 i;
+	u32 mask_value = 0;
+
+	for (i = dpm_table->count; i > 0; i--) {
+		mask_value = mask_value << 1;
+		if (dpm_table->dpm_levels[i-1].enabled)
+			mask_value |= 0x1;
+		else
+			mask_value &= 0xFFFFFFFE;
+	}
+
+	return mask_value;
+}
+
+static void ci_populate_smc_link_level(struct amdgpu_device *adev,
+				       SMU7_Discrete_DpmTable *table)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_dpm_table *dpm_table = &pi->dpm_table;
+	u32 i;
+
+	for (i = 0; i < dpm_table->pcie_speed_table.count; i++) {
+		table->LinkLevel[i].PcieGenSpeed =
+			(u8)dpm_table->pcie_speed_table.dpm_levels[i].value;
+		table->LinkLevel[i].PcieLaneCount =
+			amdgpu_encode_pci_lane_width(dpm_table->pcie_speed_table.dpm_levels[i].param1);
+		table->LinkLevel[i].EnabledForActivity = 1;
+		table->LinkLevel[i].DownT = cpu_to_be32(5);
+		table->LinkLevel[i].UpT = cpu_to_be32(30);
+	}
+
+	pi->smc_state_table.LinkLevelCount = (u8)dpm_table->pcie_speed_table.count;
+	pi->dpm_level_enable_mask.pcie_dpm_enable_mask =
+		ci_get_dpm_level_enable_mask_value(&dpm_table->pcie_speed_table);
+}
+
+static int ci_populate_smc_uvd_level(struct amdgpu_device *adev,
+				     SMU7_Discrete_DpmTable *table)
+{
+	u32 count;
+	struct atom_clock_dividers dividers;
+	int ret = -EINVAL;
+
+	table->UvdLevelCount =
+		adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count;
+
+	for (count = 0; count < table->UvdLevelCount; count++) {
+		table->UvdLevel[count].VclkFrequency =
+			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[count].vclk;
+		table->UvdLevel[count].DclkFrequency =
+			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[count].dclk;
+		table->UvdLevel[count].MinVddc =
+			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[count].v * VOLTAGE_SCALE;
+		table->UvdLevel[count].MinVddcPhases = 1;
+
+		ret = amdgpu_atombios_get_clock_dividers(adev,
+							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+							 table->UvdLevel[count].VclkFrequency, false, &dividers);
+		if (ret)
+			return ret;
+
+		table->UvdLevel[count].VclkDivider = (u8)dividers.post_divider;
+
+		ret = amdgpu_atombios_get_clock_dividers(adev,
+							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+							 table->UvdLevel[count].DclkFrequency, false, &dividers);
+		if (ret)
+			return ret;
+
+		table->UvdLevel[count].DclkDivider = (u8)dividers.post_divider;
+
+		table->UvdLevel[count].VclkFrequency = cpu_to_be32(table->UvdLevel[count].VclkFrequency);
+		table->UvdLevel[count].DclkFrequency = cpu_to_be32(table->UvdLevel[count].DclkFrequency);
+		table->UvdLevel[count].MinVddc = cpu_to_be16(table->UvdLevel[count].MinVddc);
+	}
+
+	return ret;
+}
+
+static int ci_populate_smc_vce_level(struct amdgpu_device *adev,
+				     SMU7_Discrete_DpmTable *table)
+{
+	u32 count;
+	struct atom_clock_dividers dividers;
+	int ret = -EINVAL;
+
+	table->VceLevelCount =
+		adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.count;
+
+	for (count = 0; count < table->VceLevelCount; count++) {
+		table->VceLevel[count].Frequency =
+			adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[count].evclk;
+		table->VceLevel[count].MinVoltage =
+			(u16)adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[count].v * VOLTAGE_SCALE;
+		table->VceLevel[count].MinPhases = 1;
+
+		ret = amdgpu_atombios_get_clock_dividers(adev,
+							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+							 table->VceLevel[count].Frequency, false, &dividers);
+		if (ret)
+			return ret;
+
+		table->VceLevel[count].Divider = (u8)dividers.post_divider;
+
+		table->VceLevel[count].Frequency = cpu_to_be32(table->VceLevel[count].Frequency);
+		table->VceLevel[count].MinVoltage = cpu_to_be16(table->VceLevel[count].MinVoltage);
+	}
+
+	return ret;
+
+}
+
+static int ci_populate_smc_acp_level(struct amdgpu_device *adev,
+				     SMU7_Discrete_DpmTable *table)
+{
+	u32 count;
+	struct atom_clock_dividers dividers;
+	int ret = -EINVAL;
+
+	table->AcpLevelCount = (u8)
+		(adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.count);
+
+	for (count = 0; count < table->AcpLevelCount; count++) {
+		table->AcpLevel[count].Frequency =
+			adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[count].clk;
+		table->AcpLevel[count].MinVoltage =
+			adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[count].v;
+		table->AcpLevel[count].MinPhases = 1;
+
+		ret = amdgpu_atombios_get_clock_dividers(adev,
+							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+							 table->AcpLevel[count].Frequency, false, &dividers);
+		if (ret)
+			return ret;
+
+		table->AcpLevel[count].Divider = (u8)dividers.post_divider;
+
+		table->AcpLevel[count].Frequency = cpu_to_be32(table->AcpLevel[count].Frequency);
+		table->AcpLevel[count].MinVoltage = cpu_to_be16(table->AcpLevel[count].MinVoltage);
+	}
+
+	return ret;
+}
+
+static int ci_populate_smc_samu_level(struct amdgpu_device *adev,
+				      SMU7_Discrete_DpmTable *table)
+{
+	u32 count;
+	struct atom_clock_dividers dividers;
+	int ret = -EINVAL;
+
+	table->SamuLevelCount =
+		adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.count;
+
+	for (count = 0; count < table->SamuLevelCount; count++) {
+		table->SamuLevel[count].Frequency =
+			adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[count].clk;
+		table->SamuLevel[count].MinVoltage =
+			adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[count].v * VOLTAGE_SCALE;
+		table->SamuLevel[count].MinPhases = 1;
+
+		ret = amdgpu_atombios_get_clock_dividers(adev,
+							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+							 table->SamuLevel[count].Frequency, false, &dividers);
+		if (ret)
+			return ret;
+
+		table->SamuLevel[count].Divider = (u8)dividers.post_divider;
+
+		table->SamuLevel[count].Frequency = cpu_to_be32(table->SamuLevel[count].Frequency);
+		table->SamuLevel[count].MinVoltage = cpu_to_be16(table->SamuLevel[count].MinVoltage);
+	}
+
+	return ret;
+}
+
+static int ci_calculate_mclk_params(struct amdgpu_device *adev,
+				    u32 memory_clock,
+				    SMU7_Discrete_MemoryLevel *mclk,
+				    bool strobe_mode,
+				    bool dll_state_on)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32  dll_cntl = pi->clock_registers.dll_cntl;
+	u32  mclk_pwrmgt_cntl = pi->clock_registers.mclk_pwrmgt_cntl;
+	u32  mpll_ad_func_cntl = pi->clock_registers.mpll_ad_func_cntl;
+	u32  mpll_dq_func_cntl = pi->clock_registers.mpll_dq_func_cntl;
+	u32  mpll_func_cntl = pi->clock_registers.mpll_func_cntl;
+	u32  mpll_func_cntl_1 = pi->clock_registers.mpll_func_cntl_1;
+	u32  mpll_func_cntl_2 = pi->clock_registers.mpll_func_cntl_2;
+	u32  mpll_ss1 = pi->clock_registers.mpll_ss1;
+	u32  mpll_ss2 = pi->clock_registers.mpll_ss2;
+	struct atom_mpll_param mpll_param;
+	int ret;
+
+	ret = amdgpu_atombios_get_memory_pll_dividers(adev, memory_clock, strobe_mode, &mpll_param);
+	if (ret)
+		return ret;
+
+	mpll_func_cntl &= ~MPLL_FUNC_CNTL__BWCTRL_MASK;
+	mpll_func_cntl |= (mpll_param.bwcntl << MPLL_FUNC_CNTL__BWCTRL__SHIFT);
+
+	mpll_func_cntl_1 &= ~(MPLL_FUNC_CNTL_1__CLKF_MASK | MPLL_FUNC_CNTL_1__CLKFRAC_MASK |
+			MPLL_FUNC_CNTL_1__VCO_MODE_MASK);
+	mpll_func_cntl_1 |= (mpll_param.clkf) << MPLL_FUNC_CNTL_1__CLKF__SHIFT |
+		(mpll_param.clkfrac << MPLL_FUNC_CNTL_1__CLKFRAC__SHIFT) |
+		(mpll_param.vco_mode << MPLL_FUNC_CNTL_1__VCO_MODE__SHIFT);
+
+	mpll_ad_func_cntl &= ~MPLL_AD_FUNC_CNTL__YCLK_POST_DIV_MASK;
+	mpll_ad_func_cntl |= (mpll_param.post_div << MPLL_AD_FUNC_CNTL__YCLK_POST_DIV__SHIFT);
+
+	if (adev->gmc.vram_type == AMDGPU_VRAM_TYPE_GDDR5) {
+		mpll_dq_func_cntl &= ~(MPLL_DQ_FUNC_CNTL__YCLK_SEL_MASK |
+				MPLL_AD_FUNC_CNTL__YCLK_POST_DIV_MASK);
+		mpll_dq_func_cntl |= (mpll_param.yclk_sel << MPLL_DQ_FUNC_CNTL__YCLK_SEL__SHIFT) |
+				(mpll_param.post_div << MPLL_AD_FUNC_CNTL__YCLK_POST_DIV__SHIFT);
+	}
+
+	if (pi->caps_mclk_ss_support) {
+		struct amdgpu_atom_ss ss;
+		u32 freq_nom;
+		u32 tmp;
+		u32 reference_clock = adev->clock.mpll.reference_freq;
+
+		if (mpll_param.qdr == 1)
+			freq_nom = memory_clock * 4 * (1 << mpll_param.post_div);
+		else
+			freq_nom = memory_clock * 2 * (1 << mpll_param.post_div);
+
+		tmp = (freq_nom / reference_clock);
+		tmp = tmp * tmp;
+		if (amdgpu_atombios_get_asic_ss_info(adev, &ss,
+						     ASIC_INTERNAL_MEMORY_SS, freq_nom)) {
+			u32 clks = reference_clock * 5 / ss.rate;
+			u32 clkv = (u32)((((131 * ss.percentage * ss.rate) / 100) * tmp) / freq_nom);
+
+			mpll_ss1 &= ~MPLL_SS1__CLKV_MASK;
+			mpll_ss1 |= (clkv << MPLL_SS1__CLKV__SHIFT);
+
+			mpll_ss2 &= ~MPLL_SS2__CLKS_MASK;
+			mpll_ss2 |= (clks << MPLL_SS2__CLKS__SHIFT);
+		}
+	}
+
+	mclk_pwrmgt_cntl &= ~MCLK_PWRMGT_CNTL__DLL_SPEED_MASK;
+	mclk_pwrmgt_cntl |= (mpll_param.dll_speed << MCLK_PWRMGT_CNTL__DLL_SPEED__SHIFT);
+
+	if (dll_state_on)
+		mclk_pwrmgt_cntl |= MCLK_PWRMGT_CNTL__MRDCK0_PDNB_MASK |
+			MCLK_PWRMGT_CNTL__MRDCK1_PDNB_MASK;
+	else
+		mclk_pwrmgt_cntl &= ~(MCLK_PWRMGT_CNTL__MRDCK0_PDNB_MASK |
+			MCLK_PWRMGT_CNTL__MRDCK1_PDNB_MASK);
+
+	mclk->MclkFrequency = memory_clock;
+	mclk->MpllFuncCntl = mpll_func_cntl;
+	mclk->MpllFuncCntl_1 = mpll_func_cntl_1;
+	mclk->MpllFuncCntl_2 = mpll_func_cntl_2;
+	mclk->MpllAdFuncCntl = mpll_ad_func_cntl;
+	mclk->MpllDqFuncCntl = mpll_dq_func_cntl;
+	mclk->MclkPwrmgtCntl = mclk_pwrmgt_cntl;
+	mclk->DllCntl = dll_cntl;
+	mclk->MpllSs1 = mpll_ss1;
+	mclk->MpllSs2 = mpll_ss2;
+
+	return 0;
+}
+
+static int ci_populate_single_memory_level(struct amdgpu_device *adev,
+					   u32 memory_clock,
+					   SMU7_Discrete_MemoryLevel *memory_level)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret;
+	bool dll_state_on;
+
+	if (adev->pm.dpm.dyn_state.vddc_dependency_on_mclk.entries) {
+		ret = ci_get_dependency_volt_by_clk(adev,
+						    &adev->pm.dpm.dyn_state.vddc_dependency_on_mclk,
+						    memory_clock, &memory_level->MinVddc);
+		if (ret)
+			return ret;
+	}
+
+	if (adev->pm.dpm.dyn_state.vddci_dependency_on_mclk.entries) {
+		ret = ci_get_dependency_volt_by_clk(adev,
+						    &adev->pm.dpm.dyn_state.vddci_dependency_on_mclk,
+						    memory_clock, &memory_level->MinVddci);
+		if (ret)
+			return ret;
+	}
+
+	if (adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk.entries) {
+		ret = ci_get_dependency_volt_by_clk(adev,
+						    &adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk,
+						    memory_clock, &memory_level->MinMvdd);
+		if (ret)
+			return ret;
+	}
+
+	memory_level->MinVddcPhases = 1;
+
+	if (pi->vddc_phase_shed_control)
+		ci_populate_phase_value_based_on_mclk(adev,
+						      &adev->pm.dpm.dyn_state.phase_shedding_limits_table,
+						      memory_clock,
+						      &memory_level->MinVddcPhases);
+
+	memory_level->EnabledForActivity = 1;
+	memory_level->EnabledForThrottle = 1;
+	memory_level->UpH = 0;
+	memory_level->DownH = 100;
+	memory_level->VoltageDownH = 0;
+	memory_level->ActivityLevel = (u16)pi->mclk_activity_target;
+
+	memory_level->StutterEnable = false;
+	memory_level->StrobeEnable = false;
+	memory_level->EdcReadEnable = false;
+	memory_level->EdcWriteEnable = false;
+	memory_level->RttEnable = false;
+
+	memory_level->DisplayWatermark = PPSMC_DISPLAY_WATERMARK_LOW;
+
+	if (pi->mclk_stutter_mode_threshold &&
+	    (memory_clock <= pi->mclk_stutter_mode_threshold) &&
+	    (!pi->uvd_enabled) &&
+	    (RREG32(mmDPG_PIPE_STUTTER_CONTROL) & DPG_PIPE_STUTTER_CONTROL__STUTTER_ENABLE_MASK) &&
+	    (adev->pm.dpm.new_active_crtc_count <= 2))
+		memory_level->StutterEnable = true;
+
+	if (pi->mclk_strobe_mode_threshold &&
+	    (memory_clock <= pi->mclk_strobe_mode_threshold))
+		memory_level->StrobeEnable = 1;
+
+	if (adev->gmc.vram_type == AMDGPU_VRAM_TYPE_GDDR5) {
+		memory_level->StrobeRatio =
+			ci_get_mclk_frequency_ratio(memory_clock, memory_level->StrobeEnable);
+		if (pi->mclk_edc_enable_threshold &&
+		    (memory_clock > pi->mclk_edc_enable_threshold))
+			memory_level->EdcReadEnable = true;
+
+		if (pi->mclk_edc_wr_enable_threshold &&
+		    (memory_clock > pi->mclk_edc_wr_enable_threshold))
+			memory_level->EdcWriteEnable = true;
+
+		if (memory_level->StrobeEnable) {
+			if (ci_get_mclk_frequency_ratio(memory_clock, true) >=
+			    ((RREG32(mmMC_SEQ_MISC7) >> 16) & 0xf))
+				dll_state_on = ((RREG32(mmMC_SEQ_MISC5) >> 1) & 0x1) ? true : false;
+			else
+				dll_state_on = ((RREG32(mmMC_SEQ_MISC6) >> 1) & 0x1) ? true : false;
+		} else {
+			dll_state_on = pi->dll_default_on;
+		}
+	} else {
+		memory_level->StrobeRatio = ci_get_ddr3_mclk_frequency_ratio(memory_clock);
+		dll_state_on = ((RREG32(mmMC_SEQ_MISC5) >> 1) & 0x1) ? true : false;
+	}
+
+	ret = ci_calculate_mclk_params(adev, memory_clock, memory_level, memory_level->StrobeEnable, dll_state_on);
+	if (ret)
+		return ret;
+
+	memory_level->MinVddc = cpu_to_be32(memory_level->MinVddc * VOLTAGE_SCALE);
+	memory_level->MinVddcPhases = cpu_to_be32(memory_level->MinVddcPhases);
+	memory_level->MinVddci = cpu_to_be32(memory_level->MinVddci * VOLTAGE_SCALE);
+	memory_level->MinMvdd = cpu_to_be32(memory_level->MinMvdd * VOLTAGE_SCALE);
+
+	memory_level->MclkFrequency = cpu_to_be32(memory_level->MclkFrequency);
+	memory_level->ActivityLevel = cpu_to_be16(memory_level->ActivityLevel);
+	memory_level->MpllFuncCntl = cpu_to_be32(memory_level->MpllFuncCntl);
+	memory_level->MpllFuncCntl_1 = cpu_to_be32(memory_level->MpllFuncCntl_1);
+	memory_level->MpllFuncCntl_2 = cpu_to_be32(memory_level->MpllFuncCntl_2);
+	memory_level->MpllAdFuncCntl = cpu_to_be32(memory_level->MpllAdFuncCntl);
+	memory_level->MpllDqFuncCntl = cpu_to_be32(memory_level->MpllDqFuncCntl);
+	memory_level->MclkPwrmgtCntl = cpu_to_be32(memory_level->MclkPwrmgtCntl);
+	memory_level->DllCntl = cpu_to_be32(memory_level->DllCntl);
+	memory_level->MpllSs1 = cpu_to_be32(memory_level->MpllSs1);
+	memory_level->MpllSs2 = cpu_to_be32(memory_level->MpllSs2);
+
+	return 0;
+}
+
+static int ci_populate_smc_acpi_level(struct amdgpu_device *adev,
+				      SMU7_Discrete_DpmTable *table)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct atom_clock_dividers dividers;
+	SMU7_Discrete_VoltageLevel voltage_level;
+	u32 spll_func_cntl = pi->clock_registers.cg_spll_func_cntl;
+	u32 spll_func_cntl_2 = pi->clock_registers.cg_spll_func_cntl_2;
+	u32 dll_cntl = pi->clock_registers.dll_cntl;
+	u32 mclk_pwrmgt_cntl = pi->clock_registers.mclk_pwrmgt_cntl;
+	int ret;
+
+	table->ACPILevel.Flags &= ~PPSMC_SWSTATE_FLAG_DC;
+
+	if (pi->acpi_vddc)
+		table->ACPILevel.MinVddc = cpu_to_be32(pi->acpi_vddc * VOLTAGE_SCALE);
+	else
+		table->ACPILevel.MinVddc = cpu_to_be32(pi->min_vddc_in_pp_table * VOLTAGE_SCALE);
+
+	table->ACPILevel.MinVddcPhases = pi->vddc_phase_shed_control ? 0 : 1;
+
+	table->ACPILevel.SclkFrequency = adev->clock.spll.reference_freq;
+
+	ret = amdgpu_atombios_get_clock_dividers(adev,
+						 COMPUTE_GPUCLK_INPUT_FLAG_SCLK,
+						 table->ACPILevel.SclkFrequency, false, &dividers);
+	if (ret)
+		return ret;
+
+	table->ACPILevel.SclkDid = (u8)dividers.post_divider;
+	table->ACPILevel.DisplayWatermark = PPSMC_DISPLAY_WATERMARK_LOW;
+	table->ACPILevel.DeepSleepDivId = 0;
+
+	spll_func_cntl &= ~CG_SPLL_FUNC_CNTL__SPLL_PWRON_MASK;
+	spll_func_cntl |= CG_SPLL_FUNC_CNTL__SPLL_RESET_MASK;
+
+	spll_func_cntl_2 &= ~CG_SPLL_FUNC_CNTL_2__SCLK_MUX_SEL_MASK;
+	spll_func_cntl_2 |= (4 << CG_SPLL_FUNC_CNTL_2__SCLK_MUX_SEL__SHIFT);
+
+	table->ACPILevel.CgSpllFuncCntl = spll_func_cntl;
+	table->ACPILevel.CgSpllFuncCntl2 = spll_func_cntl_2;
+	table->ACPILevel.CgSpllFuncCntl3 = pi->clock_registers.cg_spll_func_cntl_3;
+	table->ACPILevel.CgSpllFuncCntl4 = pi->clock_registers.cg_spll_func_cntl_4;
+	table->ACPILevel.SpllSpreadSpectrum = pi->clock_registers.cg_spll_spread_spectrum;
+	table->ACPILevel.SpllSpreadSpectrum2 = pi->clock_registers.cg_spll_spread_spectrum_2;
+	table->ACPILevel.CcPwrDynRm = 0;
+	table->ACPILevel.CcPwrDynRm1 = 0;
+
+	table->ACPILevel.Flags = cpu_to_be32(table->ACPILevel.Flags);
+	table->ACPILevel.MinVddcPhases = cpu_to_be32(table->ACPILevel.MinVddcPhases);
+	table->ACPILevel.SclkFrequency = cpu_to_be32(table->ACPILevel.SclkFrequency);
+	table->ACPILevel.CgSpllFuncCntl = cpu_to_be32(table->ACPILevel.CgSpllFuncCntl);
+	table->ACPILevel.CgSpllFuncCntl2 = cpu_to_be32(table->ACPILevel.CgSpllFuncCntl2);
+	table->ACPILevel.CgSpllFuncCntl3 = cpu_to_be32(table->ACPILevel.CgSpllFuncCntl3);
+	table->ACPILevel.CgSpllFuncCntl4 = cpu_to_be32(table->ACPILevel.CgSpllFuncCntl4);
+	table->ACPILevel.SpllSpreadSpectrum = cpu_to_be32(table->ACPILevel.SpllSpreadSpectrum);
+	table->ACPILevel.SpllSpreadSpectrum2 = cpu_to_be32(table->ACPILevel.SpllSpreadSpectrum2);
+	table->ACPILevel.CcPwrDynRm = cpu_to_be32(table->ACPILevel.CcPwrDynRm);
+	table->ACPILevel.CcPwrDynRm1 = cpu_to_be32(table->ACPILevel.CcPwrDynRm1);
+
+	table->MemoryACPILevel.MinVddc = table->ACPILevel.MinVddc;
+	table->MemoryACPILevel.MinVddcPhases = table->ACPILevel.MinVddcPhases;
+
+	if (pi->vddci_control != CISLANDS_VOLTAGE_CONTROL_NONE) {
+		if (pi->acpi_vddci)
+			table->MemoryACPILevel.MinVddci =
+				cpu_to_be32(pi->acpi_vddci * VOLTAGE_SCALE);
+		else
+			table->MemoryACPILevel.MinVddci =
+				cpu_to_be32(pi->min_vddci_in_pp_table * VOLTAGE_SCALE);
+	}
+
+	if (ci_populate_mvdd_value(adev, 0, &voltage_level))
+		table->MemoryACPILevel.MinMvdd = 0;
+	else
+		table->MemoryACPILevel.MinMvdd =
+			cpu_to_be32(voltage_level.Voltage * VOLTAGE_SCALE);
+
+	mclk_pwrmgt_cntl |= MCLK_PWRMGT_CNTL__MRDCK0_RESET_MASK |
+		MCLK_PWRMGT_CNTL__MRDCK1_RESET_MASK;
+	mclk_pwrmgt_cntl &= ~(MCLK_PWRMGT_CNTL__MRDCK0_PDNB_MASK |
+			MCLK_PWRMGT_CNTL__MRDCK1_PDNB_MASK);
+
+	dll_cntl &= ~(DLL_CNTL__MRDCK0_BYPASS_MASK | DLL_CNTL__MRDCK1_BYPASS_MASK);
+
+	table->MemoryACPILevel.DllCntl = cpu_to_be32(dll_cntl);
+	table->MemoryACPILevel.MclkPwrmgtCntl = cpu_to_be32(mclk_pwrmgt_cntl);
+	table->MemoryACPILevel.MpllAdFuncCntl =
+		cpu_to_be32(pi->clock_registers.mpll_ad_func_cntl);
+	table->MemoryACPILevel.MpllDqFuncCntl =
+		cpu_to_be32(pi->clock_registers.mpll_dq_func_cntl);
+	table->MemoryACPILevel.MpllFuncCntl =
+		cpu_to_be32(pi->clock_registers.mpll_func_cntl);
+	table->MemoryACPILevel.MpllFuncCntl_1 =
+		cpu_to_be32(pi->clock_registers.mpll_func_cntl_1);
+	table->MemoryACPILevel.MpllFuncCntl_2 =
+		cpu_to_be32(pi->clock_registers.mpll_func_cntl_2);
+	table->MemoryACPILevel.MpllSs1 = cpu_to_be32(pi->clock_registers.mpll_ss1);
+	table->MemoryACPILevel.MpllSs2 = cpu_to_be32(pi->clock_registers.mpll_ss2);
+
+	table->MemoryACPILevel.EnabledForThrottle = 0;
+	table->MemoryACPILevel.EnabledForActivity = 0;
+	table->MemoryACPILevel.UpH = 0;
+	table->MemoryACPILevel.DownH = 100;
+	table->MemoryACPILevel.VoltageDownH = 0;
+	table->MemoryACPILevel.ActivityLevel =
+		cpu_to_be16((u16)pi->mclk_activity_target);
+
+	table->MemoryACPILevel.StutterEnable = false;
+	table->MemoryACPILevel.StrobeEnable = false;
+	table->MemoryACPILevel.EdcReadEnable = false;
+	table->MemoryACPILevel.EdcWriteEnable = false;
+	table->MemoryACPILevel.RttEnable = false;
+
+	return 0;
+}
+
+
+static int ci_enable_ulv(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ulv_parm *ulv = &pi->ulv;
+
+	if (ulv->supported) {
+		if (enable)
+			return (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableULV) == PPSMC_Result_OK) ?
+				0 : -EINVAL;
+		else
+			return (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DisableULV) == PPSMC_Result_OK) ?
+				0 : -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_populate_ulv_level(struct amdgpu_device *adev,
+				 SMU7_Discrete_Ulv *state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u16 ulv_voltage = adev->pm.dpm.backbias_response_time;
+
+	state->CcPwrDynRm = 0;
+	state->CcPwrDynRm1 = 0;
+
+	if (ulv_voltage == 0) {
+		pi->ulv.supported = false;
+		return 0;
+	}
+
+	if (pi->voltage_control != CISLANDS_VOLTAGE_CONTROL_BY_SVID2) {
+		if (ulv_voltage > adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[0].v)
+			state->VddcOffset = 0;
+		else
+			state->VddcOffset =
+				adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[0].v - ulv_voltage;
+	} else {
+		if (ulv_voltage > adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[0].v)
+			state->VddcOffsetVid = 0;
+		else
+			state->VddcOffsetVid = (u8)
+				((adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[0].v - ulv_voltage) *
+				 VOLTAGE_VID_OFFSET_SCALE2 / VOLTAGE_VID_OFFSET_SCALE1);
+	}
+	state->VddcPhase = pi->vddc_phase_shed_control ? 0 : 1;
+
+	state->CcPwrDynRm = cpu_to_be32(state->CcPwrDynRm);
+	state->CcPwrDynRm1 = cpu_to_be32(state->CcPwrDynRm1);
+	state->VddcOffset = cpu_to_be16(state->VddcOffset);
+
+	return 0;
+}
+
+static int ci_calculate_sclk_params(struct amdgpu_device *adev,
+				    u32 engine_clock,
+				    SMU7_Discrete_GraphicsLevel *sclk)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct atom_clock_dividers dividers;
+	u32 spll_func_cntl_3 = pi->clock_registers.cg_spll_func_cntl_3;
+	u32 spll_func_cntl_4 = pi->clock_registers.cg_spll_func_cntl_4;
+	u32 cg_spll_spread_spectrum = pi->clock_registers.cg_spll_spread_spectrum;
+	u32 cg_spll_spread_spectrum_2 = pi->clock_registers.cg_spll_spread_spectrum_2;
+	u32 reference_clock = adev->clock.spll.reference_freq;
+	u32 reference_divider;
+	u32 fbdiv;
+	int ret;
+
+	ret = amdgpu_atombios_get_clock_dividers(adev,
+						 COMPUTE_GPUCLK_INPUT_FLAG_SCLK,
+						 engine_clock, false, &dividers);
+	if (ret)
+		return ret;
+
+	reference_divider = 1 + dividers.ref_div;
+	fbdiv = dividers.fb_div & 0x3FFFFFF;
+
+	spll_func_cntl_3 &= ~CG_SPLL_FUNC_CNTL_3__SPLL_FB_DIV_MASK;
+	spll_func_cntl_3 |= (fbdiv << CG_SPLL_FUNC_CNTL_3__SPLL_FB_DIV__SHIFT);
+	spll_func_cntl_3 |= CG_SPLL_FUNC_CNTL_3__SPLL_DITHEN_MASK;
+
+	if (pi->caps_sclk_ss_support) {
+		struct amdgpu_atom_ss ss;
+		u32 vco_freq = engine_clock * dividers.post_div;
+
+		if (amdgpu_atombios_get_asic_ss_info(adev, &ss,
+						     ASIC_INTERNAL_ENGINE_SS, vco_freq)) {
+			u32 clk_s = reference_clock * 5 / (reference_divider * ss.rate);
+			u32 clk_v = 4 * ss.percentage * fbdiv / (clk_s * 10000);
+
+			cg_spll_spread_spectrum &= ~(CG_SPLL_SPREAD_SPECTRUM__CLKS_MASK | CG_SPLL_SPREAD_SPECTRUM__SSEN_MASK);
+			cg_spll_spread_spectrum |= (clk_s << CG_SPLL_SPREAD_SPECTRUM__CLKS__SHIFT);
+			cg_spll_spread_spectrum |= (1 << CG_SPLL_SPREAD_SPECTRUM__SSEN__SHIFT);
+
+			cg_spll_spread_spectrum_2 &= ~CG_SPLL_SPREAD_SPECTRUM_2__CLKV_MASK;
+			cg_spll_spread_spectrum_2 |= (clk_v << CG_SPLL_SPREAD_SPECTRUM_2__CLKV__SHIFT);
+		}
+	}
+
+	sclk->SclkFrequency = engine_clock;
+	sclk->CgSpllFuncCntl3 = spll_func_cntl_3;
+	sclk->CgSpllFuncCntl4 = spll_func_cntl_4;
+	sclk->SpllSpreadSpectrum = cg_spll_spread_spectrum;
+	sclk->SpllSpreadSpectrum2  = cg_spll_spread_spectrum_2;
+	sclk->SclkDid = (u8)dividers.post_divider;
+
+	return 0;
+}
+
+static int ci_populate_single_graphic_level(struct amdgpu_device *adev,
+					    u32 engine_clock,
+					    u16 sclk_activity_level_t,
+					    SMU7_Discrete_GraphicsLevel *graphic_level)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret;
+
+	ret = ci_calculate_sclk_params(adev, engine_clock, graphic_level);
+	if (ret)
+		return ret;
+
+	ret = ci_get_dependency_volt_by_clk(adev,
+					    &adev->pm.dpm.dyn_state.vddc_dependency_on_sclk,
+					    engine_clock, &graphic_level->MinVddc);
+	if (ret)
+		return ret;
+
+	graphic_level->SclkFrequency = engine_clock;
+
+	graphic_level->Flags =  0;
+	graphic_level->MinVddcPhases = 1;
+
+	if (pi->vddc_phase_shed_control)
+		ci_populate_phase_value_based_on_sclk(adev,
+						      &adev->pm.dpm.dyn_state.phase_shedding_limits_table,
+						      engine_clock,
+						      &graphic_level->MinVddcPhases);
+
+	graphic_level->ActivityLevel = sclk_activity_level_t;
+
+	graphic_level->CcPwrDynRm = 0;
+	graphic_level->CcPwrDynRm1 = 0;
+	graphic_level->EnabledForThrottle = 1;
+	graphic_level->UpH = 0;
+	graphic_level->DownH = 0;
+	graphic_level->VoltageDownH = 0;
+	graphic_level->PowerThrottle = 0;
+
+	if (pi->caps_sclk_ds)
+		graphic_level->DeepSleepDivId = ci_get_sleep_divider_id_from_clock(engine_clock,
+										   CISLAND_MINIMUM_ENGINE_CLOCK);
+
+	graphic_level->DisplayWatermark = PPSMC_DISPLAY_WATERMARK_LOW;
+
+	graphic_level->Flags = cpu_to_be32(graphic_level->Flags);
+	graphic_level->MinVddc = cpu_to_be32(graphic_level->MinVddc * VOLTAGE_SCALE);
+	graphic_level->MinVddcPhases = cpu_to_be32(graphic_level->MinVddcPhases);
+	graphic_level->SclkFrequency = cpu_to_be32(graphic_level->SclkFrequency);
+	graphic_level->ActivityLevel = cpu_to_be16(graphic_level->ActivityLevel);
+	graphic_level->CgSpllFuncCntl3 = cpu_to_be32(graphic_level->CgSpllFuncCntl3);
+	graphic_level->CgSpllFuncCntl4 = cpu_to_be32(graphic_level->CgSpllFuncCntl4);
+	graphic_level->SpllSpreadSpectrum = cpu_to_be32(graphic_level->SpllSpreadSpectrum);
+	graphic_level->SpllSpreadSpectrum2 = cpu_to_be32(graphic_level->SpllSpreadSpectrum2);
+	graphic_level->CcPwrDynRm = cpu_to_be32(graphic_level->CcPwrDynRm);
+	graphic_level->CcPwrDynRm1 = cpu_to_be32(graphic_level->CcPwrDynRm1);
+
+	return 0;
+}
+
+static int ci_populate_all_graphic_levels(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_dpm_table *dpm_table = &pi->dpm_table;
+	u32 level_array_address = pi->dpm_table_start +
+		offsetof(SMU7_Discrete_DpmTable, GraphicsLevel);
+	u32 level_array_size = sizeof(SMU7_Discrete_GraphicsLevel) *
+		SMU7_MAX_LEVELS_GRAPHICS;
+	SMU7_Discrete_GraphicsLevel *levels = pi->smc_state_table.GraphicsLevel;
+	u32 i, ret;
+
+	memset(levels, 0, level_array_size);
+
+	for (i = 0; i < dpm_table->sclk_table.count; i++) {
+		ret = ci_populate_single_graphic_level(adev,
+						       dpm_table->sclk_table.dpm_levels[i].value,
+						       (u16)pi->activity_target[i],
+						       &pi->smc_state_table.GraphicsLevel[i]);
+		if (ret)
+			return ret;
+		if (i > 1)
+			pi->smc_state_table.GraphicsLevel[i].DeepSleepDivId = 0;
+		if (i == (dpm_table->sclk_table.count - 1))
+			pi->smc_state_table.GraphicsLevel[i].DisplayWatermark =
+				PPSMC_DISPLAY_WATERMARK_HIGH;
+	}
+	pi->smc_state_table.GraphicsLevel[0].EnabledForActivity = 1;
+
+	pi->smc_state_table.GraphicsDpmLevelCount = (u8)dpm_table->sclk_table.count;
+	pi->dpm_level_enable_mask.sclk_dpm_enable_mask =
+		ci_get_dpm_level_enable_mask_value(&dpm_table->sclk_table);
+
+	ret = amdgpu_ci_copy_bytes_to_smc(adev, level_array_address,
+				   (u8 *)levels, level_array_size,
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int ci_populate_ulv_state(struct amdgpu_device *adev,
+				 SMU7_Discrete_Ulv *ulv_level)
+{
+	return ci_populate_ulv_level(adev, ulv_level);
+}
+
+static int ci_populate_all_memory_levels(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_dpm_table *dpm_table = &pi->dpm_table;
+	u32 level_array_address = pi->dpm_table_start +
+		offsetof(SMU7_Discrete_DpmTable, MemoryLevel);
+	u32 level_array_size = sizeof(SMU7_Discrete_MemoryLevel) *
+		SMU7_MAX_LEVELS_MEMORY;
+	SMU7_Discrete_MemoryLevel *levels = pi->smc_state_table.MemoryLevel;
+	u32 i, ret;
+
+	memset(levels, 0, level_array_size);
+
+	for (i = 0; i < dpm_table->mclk_table.count; i++) {
+		if (dpm_table->mclk_table.dpm_levels[i].value == 0)
+			return -EINVAL;
+		ret = ci_populate_single_memory_level(adev,
+						      dpm_table->mclk_table.dpm_levels[i].value,
+						      &pi->smc_state_table.MemoryLevel[i]);
+		if (ret)
+			return ret;
+	}
+
+	if ((dpm_table->mclk_table.count >= 2) &&
+	    ((adev->pdev->device == 0x67B0) || (adev->pdev->device == 0x67B1))) {
+		pi->smc_state_table.MemoryLevel[1].MinVddc =
+			pi->smc_state_table.MemoryLevel[0].MinVddc;
+		pi->smc_state_table.MemoryLevel[1].MinVddcPhases =
+			pi->smc_state_table.MemoryLevel[0].MinVddcPhases;
+	}
+
+	pi->smc_state_table.MemoryLevel[0].ActivityLevel = cpu_to_be16(0x1F);
+
+	pi->smc_state_table.MemoryDpmLevelCount = (u8)dpm_table->mclk_table.count;
+	pi->dpm_level_enable_mask.mclk_dpm_enable_mask =
+		ci_get_dpm_level_enable_mask_value(&dpm_table->mclk_table);
+
+	pi->smc_state_table.MemoryLevel[dpm_table->mclk_table.count - 1].DisplayWatermark =
+		PPSMC_DISPLAY_WATERMARK_HIGH;
+
+	ret = amdgpu_ci_copy_bytes_to_smc(adev, level_array_address,
+				   (u8 *)levels, level_array_size,
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static void ci_reset_single_dpm_table(struct amdgpu_device *adev,
+				      struct ci_single_dpm_table* dpm_table,
+				      u32 count)
+{
+	u32 i;
+
+	dpm_table->count = count;
+	for (i = 0; i < MAX_REGULAR_DPM_NUMBER; i++)
+		dpm_table->dpm_levels[i].enabled = false;
+}
+
+static void ci_setup_pcie_table_entry(struct ci_single_dpm_table* dpm_table,
+				      u32 index, u32 pcie_gen, u32 pcie_lanes)
+{
+	dpm_table->dpm_levels[index].value = pcie_gen;
+	dpm_table->dpm_levels[index].param1 = pcie_lanes;
+	dpm_table->dpm_levels[index].enabled = true;
+}
+
+static int ci_setup_default_pcie_tables(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (!pi->use_pcie_performance_levels && !pi->use_pcie_powersaving_levels)
+		return -EINVAL;
+
+	if (pi->use_pcie_performance_levels && !pi->use_pcie_powersaving_levels) {
+		pi->pcie_gen_powersaving = pi->pcie_gen_performance;
+		pi->pcie_lane_powersaving = pi->pcie_lane_performance;
+	} else if (!pi->use_pcie_performance_levels && pi->use_pcie_powersaving_levels) {
+		pi->pcie_gen_performance = pi->pcie_gen_powersaving;
+		pi->pcie_lane_performance = pi->pcie_lane_powersaving;
+	}
+
+	ci_reset_single_dpm_table(adev,
+				  &pi->dpm_table.pcie_speed_table,
+				  SMU7_MAX_LEVELS_LINK);
+
+	if (adev->asic_type == CHIP_BONAIRE)
+		ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 0,
+					  pi->pcie_gen_powersaving.min,
+					  pi->pcie_lane_powersaving.max);
+	else
+		ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 0,
+					  pi->pcie_gen_powersaving.min,
+					  pi->pcie_lane_powersaving.min);
+	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 1,
+				  pi->pcie_gen_performance.min,
+				  pi->pcie_lane_performance.min);
+	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 2,
+				  pi->pcie_gen_powersaving.min,
+				  pi->pcie_lane_powersaving.max);
+	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 3,
+				  pi->pcie_gen_performance.min,
+				  pi->pcie_lane_performance.max);
+	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 4,
+				  pi->pcie_gen_powersaving.max,
+				  pi->pcie_lane_powersaving.max);
+	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 5,
+				  pi->pcie_gen_performance.max,
+				  pi->pcie_lane_performance.max);
+
+	pi->dpm_table.pcie_speed_table.count = 6;
+
+	return 0;
+}
+
+static int ci_setup_default_dpm_tables(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *allowed_sclk_vddc_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	struct amdgpu_clock_voltage_dependency_table *allowed_mclk_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk;
+	struct amdgpu_cac_leakage_table *std_voltage_table =
+		&adev->pm.dpm.dyn_state.cac_leakage_table;
+	u32 i;
+
+	if (allowed_sclk_vddc_table == NULL)
+		return -EINVAL;
+	if (allowed_sclk_vddc_table->count < 1)
+		return -EINVAL;
+	if (allowed_mclk_table == NULL)
+		return -EINVAL;
+	if (allowed_mclk_table->count < 1)
+		return -EINVAL;
+
+	memset(&pi->dpm_table, 0, sizeof(struct ci_dpm_table));
+
+	ci_reset_single_dpm_table(adev,
+				  &pi->dpm_table.sclk_table,
+				  SMU7_MAX_LEVELS_GRAPHICS);
+	ci_reset_single_dpm_table(adev,
+				  &pi->dpm_table.mclk_table,
+				  SMU7_MAX_LEVELS_MEMORY);
+	ci_reset_single_dpm_table(adev,
+				  &pi->dpm_table.vddc_table,
+				  SMU7_MAX_LEVELS_VDDC);
+	ci_reset_single_dpm_table(adev,
+				  &pi->dpm_table.vddci_table,
+				  SMU7_MAX_LEVELS_VDDCI);
+	ci_reset_single_dpm_table(adev,
+				  &pi->dpm_table.mvdd_table,
+				  SMU7_MAX_LEVELS_MVDD);
+
+	pi->dpm_table.sclk_table.count = 0;
+	for (i = 0; i < allowed_sclk_vddc_table->count; i++) {
+		if ((i == 0) ||
+		    (pi->dpm_table.sclk_table.dpm_levels[pi->dpm_table.sclk_table.count-1].value !=
+		     allowed_sclk_vddc_table->entries[i].clk)) {
+			pi->dpm_table.sclk_table.dpm_levels[pi->dpm_table.sclk_table.count].value =
+				allowed_sclk_vddc_table->entries[i].clk;
+			pi->dpm_table.sclk_table.dpm_levels[pi->dpm_table.sclk_table.count].enabled =
+				(i == 0) ? true : false;
+			pi->dpm_table.sclk_table.count++;
+		}
+	}
+
+	pi->dpm_table.mclk_table.count = 0;
+	for (i = 0; i < allowed_mclk_table->count; i++) {
+		if ((i == 0) ||
+		    (pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count-1].value !=
+		     allowed_mclk_table->entries[i].clk)) {
+			pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count].value =
+				allowed_mclk_table->entries[i].clk;
+			pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count].enabled =
+				(i == 0) ? true : false;
+			pi->dpm_table.mclk_table.count++;
+		}
+	}
+
+	for (i = 0; i < allowed_sclk_vddc_table->count; i++) {
+		pi->dpm_table.vddc_table.dpm_levels[i].value =
+			allowed_sclk_vddc_table->entries[i].v;
+		pi->dpm_table.vddc_table.dpm_levels[i].param1 =
+			std_voltage_table->entries[i].leakage;
+		pi->dpm_table.vddc_table.dpm_levels[i].enabled = true;
+	}
+	pi->dpm_table.vddc_table.count = allowed_sclk_vddc_table->count;
+
+	allowed_mclk_table = &adev->pm.dpm.dyn_state.vddci_dependency_on_mclk;
+	if (allowed_mclk_table) {
+		for (i = 0; i < allowed_mclk_table->count; i++) {
+			pi->dpm_table.vddci_table.dpm_levels[i].value =
+				allowed_mclk_table->entries[i].v;
+			pi->dpm_table.vddci_table.dpm_levels[i].enabled = true;
+		}
+		pi->dpm_table.vddci_table.count = allowed_mclk_table->count;
+	}
+
+	allowed_mclk_table = &adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk;
+	if (allowed_mclk_table) {
+		for (i = 0; i < allowed_mclk_table->count; i++) {
+			pi->dpm_table.mvdd_table.dpm_levels[i].value =
+				allowed_mclk_table->entries[i].v;
+			pi->dpm_table.mvdd_table.dpm_levels[i].enabled = true;
+		}
+		pi->dpm_table.mvdd_table.count = allowed_mclk_table->count;
+	}
+
+	ci_setup_default_pcie_tables(adev);
+
+	/* save a copy of the default DPM table */
+	memcpy(&(pi->golden_dpm_table), &(pi->dpm_table),
+			sizeof(struct ci_dpm_table));
+
+	return 0;
+}
+
+static int ci_find_boot_level(struct ci_single_dpm_table *table,
+			      u32 value, u32 *boot_level)
+{
+	u32 i;
+	int ret = -EINVAL;
+
+	for(i = 0; i < table->count; i++) {
+		if (value == table->dpm_levels[i].value) {
+			*boot_level = i;
+			ret = 0;
+		}
+	}
+
+	return ret;
+}
+
+static int ci_init_smc_table(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ulv_parm *ulv = &pi->ulv;
+	struct amdgpu_ps *amdgpu_boot_state = adev->pm.dpm.boot_ps;
+	SMU7_Discrete_DpmTable *table = &pi->smc_state_table;
+	int ret;
+
+	ret = ci_setup_default_dpm_tables(adev);
+	if (ret)
+		return ret;
+
+	if (pi->voltage_control != CISLANDS_VOLTAGE_CONTROL_NONE)
+		ci_populate_smc_voltage_tables(adev, table);
+
+	ci_init_fps_limits(adev);
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_HARDWAREDC)
+		table->SystemFlags |= PPSMC_SYSTEMFLAG_GPIO_DC;
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_STEPVDDC)
+		table->SystemFlags |= PPSMC_SYSTEMFLAG_STEPVDDC;
+
+	if (adev->gmc.vram_type == AMDGPU_VRAM_TYPE_GDDR5)
+		table->SystemFlags |= PPSMC_SYSTEMFLAG_GDDR5;
+
+	if (ulv->supported) {
+		ret = ci_populate_ulv_state(adev, &pi->smc_state_table.Ulv);
+		if (ret)
+			return ret;
+		WREG32_SMC(ixCG_ULV_PARAMETER, ulv->cg_ulv_parameter);
+	}
+
+	ret = ci_populate_all_graphic_levels(adev);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_all_memory_levels(adev);
+	if (ret)
+		return ret;
+
+	ci_populate_smc_link_level(adev, table);
+
+	ret = ci_populate_smc_acpi_level(adev, table);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_smc_vce_level(adev, table);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_smc_acp_level(adev, table);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_smc_samu_level(adev, table);
+	if (ret)
+		return ret;
+
+	ret = ci_do_program_memory_timing_parameters(adev);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_smc_uvd_level(adev, table);
+	if (ret)
+		return ret;
+
+	table->UvdBootLevel  = 0;
+	table->VceBootLevel  = 0;
+	table->AcpBootLevel  = 0;
+	table->SamuBootLevel  = 0;
+	table->GraphicsBootLevel  = 0;
+	table->MemoryBootLevel  = 0;
+
+	ret = ci_find_boot_level(&pi->dpm_table.sclk_table,
+				 pi->vbios_boot_state.sclk_bootup_value,
+				 (u32 *)&pi->smc_state_table.GraphicsBootLevel);
+
+	ret = ci_find_boot_level(&pi->dpm_table.mclk_table,
+				 pi->vbios_boot_state.mclk_bootup_value,
+				 (u32 *)&pi->smc_state_table.MemoryBootLevel);
+
+	table->BootVddc = pi->vbios_boot_state.vddc_bootup_value;
+	table->BootVddci = pi->vbios_boot_state.vddci_bootup_value;
+	table->BootMVdd = pi->vbios_boot_state.mvdd_bootup_value;
+
+	ci_populate_smc_initial_state(adev, amdgpu_boot_state);
+
+	ret = ci_populate_bapm_parameters_in_dpm_table(adev);
+	if (ret)
+		return ret;
+
+	table->UVDInterval = 1;
+	table->VCEInterval = 1;
+	table->ACPInterval = 1;
+	table->SAMUInterval = 1;
+	table->GraphicsVoltageChangeEnable = 1;
+	table->GraphicsThermThrottleEnable = 1;
+	table->GraphicsInterval = 1;
+	table->VoltageInterval = 1;
+	table->ThermalInterval = 1;
+	table->TemperatureLimitHigh = (u16)((pi->thermal_temp_setting.temperature_high *
+					     CISLANDS_Q88_FORMAT_CONVERSION_UNIT) / 1000);
+	table->TemperatureLimitLow = (u16)((pi->thermal_temp_setting.temperature_low *
+					    CISLANDS_Q88_FORMAT_CONVERSION_UNIT) / 1000);
+	table->MemoryVoltageChangeEnable = 1;
+	table->MemoryInterval = 1;
+	table->VoltageResponseTime = 0;
+	table->VddcVddciDelta = 4000;
+	table->PhaseResponseTime = 0;
+	table->MemoryThermThrottleEnable = 1;
+	table->PCIeBootLinkLevel = pi->dpm_table.pcie_speed_table.count - 1;
+	table->PCIeGenInterval = 1;
+	if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2)
+		table->SVI2Enable  = 1;
+	else
+		table->SVI2Enable  = 0;
+
+	table->ThermGpio = 17;
+	table->SclkStepSize = 0x4000;
+
+	table->SystemFlags = cpu_to_be32(table->SystemFlags);
+	table->SmioMaskVddcVid = cpu_to_be32(table->SmioMaskVddcVid);
+	table->SmioMaskVddcPhase = cpu_to_be32(table->SmioMaskVddcPhase);
+	table->SmioMaskVddciVid = cpu_to_be32(table->SmioMaskVddciVid);
+	table->SmioMaskMvddVid = cpu_to_be32(table->SmioMaskMvddVid);
+	table->SclkStepSize = cpu_to_be32(table->SclkStepSize);
+	table->TemperatureLimitHigh = cpu_to_be16(table->TemperatureLimitHigh);
+	table->TemperatureLimitLow = cpu_to_be16(table->TemperatureLimitLow);
+	table->VddcVddciDelta = cpu_to_be16(table->VddcVddciDelta);
+	table->VoltageResponseTime = cpu_to_be16(table->VoltageResponseTime);
+	table->PhaseResponseTime = cpu_to_be16(table->PhaseResponseTime);
+	table->BootVddc = cpu_to_be16(table->BootVddc * VOLTAGE_SCALE);
+	table->BootVddci = cpu_to_be16(table->BootVddci * VOLTAGE_SCALE);
+	table->BootMVdd = cpu_to_be16(table->BootMVdd * VOLTAGE_SCALE);
+
+	ret = amdgpu_ci_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Discrete_DpmTable, SystemFlags),
+				   (u8 *)&table->SystemFlags,
+				   sizeof(SMU7_Discrete_DpmTable) - 3 * sizeof(SMU7_PIDController),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static void ci_trim_single_dpm_states(struct amdgpu_device *adev,
+				      struct ci_single_dpm_table *dpm_table,
+				      u32 low_limit, u32 high_limit)
+{
+	u32 i;
+
+	for (i = 0; i < dpm_table->count; i++) {
+		if ((dpm_table->dpm_levels[i].value < low_limit) ||
+		    (dpm_table->dpm_levels[i].value > high_limit))
+			dpm_table->dpm_levels[i].enabled = false;
+		else
+			dpm_table->dpm_levels[i].enabled = true;
+	}
+}
+
+static void ci_trim_pcie_dpm_states(struct amdgpu_device *adev,
+				    u32 speed_low, u32 lanes_low,
+				    u32 speed_high, u32 lanes_high)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_single_dpm_table *pcie_table = &pi->dpm_table.pcie_speed_table;
+	u32 i, j;
+
+	for (i = 0; i < pcie_table->count; i++) {
+		if ((pcie_table->dpm_levels[i].value < speed_low) ||
+		    (pcie_table->dpm_levels[i].param1 < lanes_low) ||
+		    (pcie_table->dpm_levels[i].value > speed_high) ||
+		    (pcie_table->dpm_levels[i].param1 > lanes_high))
+			pcie_table->dpm_levels[i].enabled = false;
+		else
+			pcie_table->dpm_levels[i].enabled = true;
+	}
+
+	for (i = 0; i < pcie_table->count; i++) {
+		if (pcie_table->dpm_levels[i].enabled) {
+			for (j = i + 1; j < pcie_table->count; j++) {
+				if (pcie_table->dpm_levels[j].enabled) {
+					if ((pcie_table->dpm_levels[i].value == pcie_table->dpm_levels[j].value) &&
+					    (pcie_table->dpm_levels[i].param1 == pcie_table->dpm_levels[j].param1))
+						pcie_table->dpm_levels[j].enabled = false;
+				}
+			}
+		}
+	}
+}
+
+static int ci_trim_dpm_states(struct amdgpu_device *adev,
+			      struct amdgpu_ps *amdgpu_state)
+{
+	struct ci_ps *state = ci_get_ps(amdgpu_state);
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 high_limit_count;
+
+	if (state->performance_level_count < 1)
+		return -EINVAL;
+
+	if (state->performance_level_count == 1)
+		high_limit_count = 0;
+	else
+		high_limit_count = 1;
+
+	ci_trim_single_dpm_states(adev,
+				  &pi->dpm_table.sclk_table,
+				  state->performance_levels[0].sclk,
+				  state->performance_levels[high_limit_count].sclk);
+
+	ci_trim_single_dpm_states(adev,
+				  &pi->dpm_table.mclk_table,
+				  state->performance_levels[0].mclk,
+				  state->performance_levels[high_limit_count].mclk);
+
+	ci_trim_pcie_dpm_states(adev,
+				state->performance_levels[0].pcie_gen,
+				state->performance_levels[0].pcie_lane,
+				state->performance_levels[high_limit_count].pcie_gen,
+				state->performance_levels[high_limit_count].pcie_lane);
+
+	return 0;
+}
+
+static int ci_apply_disp_minimum_voltage_request(struct amdgpu_device *adev)
+{
+	struct amdgpu_clock_voltage_dependency_table *disp_voltage_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk;
+	struct amdgpu_clock_voltage_dependency_table *vddc_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	u32 requested_voltage = 0;
+	u32 i;
+
+	if (disp_voltage_table == NULL)
+		return -EINVAL;
+	if (!disp_voltage_table->count)
+		return -EINVAL;
+
+	for (i = 0; i < disp_voltage_table->count; i++) {
+		if (adev->clock.current_dispclk == disp_voltage_table->entries[i].clk)
+			requested_voltage = disp_voltage_table->entries[i].v;
+	}
+
+	for (i = 0; i < vddc_table->count; i++) {
+		if (requested_voltage <= vddc_table->entries[i].v) {
+			requested_voltage = vddc_table->entries[i].v;
+			return (amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+								  PPSMC_MSG_VddC_Request,
+								  requested_voltage * VOLTAGE_SCALE) == PPSMC_Result_OK) ?
+				0 : -EINVAL;
+		}
+	}
+
+	return -EINVAL;
+}
+
+static int ci_upload_dpm_level_enable_mask(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result result;
+
+	ci_apply_disp_minimum_voltage_request(adev);
+
+	if (!pi->sclk_dpm_key_disabled) {
+		if (pi->dpm_level_enable_mask.sclk_dpm_enable_mask) {
+			result = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+								   PPSMC_MSG_SCLKDPM_SetEnabledMask,
+								   pi->dpm_level_enable_mask.sclk_dpm_enable_mask);
+			if (result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+	}
+
+	if (!pi->mclk_dpm_key_disabled) {
+		if (pi->dpm_level_enable_mask.mclk_dpm_enable_mask) {
+			result = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+								   PPSMC_MSG_MCLKDPM_SetEnabledMask,
+								   pi->dpm_level_enable_mask.mclk_dpm_enable_mask);
+			if (result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+	}
+
+#if 0
+	if (!pi->pcie_dpm_key_disabled) {
+		if (pi->dpm_level_enable_mask.pcie_dpm_enable_mask) {
+			result = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+								   PPSMC_MSG_PCIeDPM_SetEnabledMask,
+								   pi->dpm_level_enable_mask.pcie_dpm_enable_mask);
+			if (result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+	}
+#endif
+
+	return 0;
+}
+
+static void ci_find_dpm_states_clocks_in_dpm_table(struct amdgpu_device *adev,
+						   struct amdgpu_ps *amdgpu_state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *state = ci_get_ps(amdgpu_state);
+	struct ci_single_dpm_table *sclk_table = &pi->dpm_table.sclk_table;
+	u32 sclk = state->performance_levels[state->performance_level_count-1].sclk;
+	struct ci_single_dpm_table *mclk_table = &pi->dpm_table.mclk_table;
+	u32 mclk = state->performance_levels[state->performance_level_count-1].mclk;
+	u32 i;
+
+	pi->need_update_smu7_dpm_table = 0;
+
+	for (i = 0; i < sclk_table->count; i++) {
+		if (sclk == sclk_table->dpm_levels[i].value)
+			break;
+	}
+
+	if (i >= sclk_table->count) {
+		pi->need_update_smu7_dpm_table |= DPMTABLE_OD_UPDATE_SCLK;
+	} else {
+		/* XXX check display min clock requirements */
+		if (CISLAND_MINIMUM_ENGINE_CLOCK != CISLAND_MINIMUM_ENGINE_CLOCK)
+			pi->need_update_smu7_dpm_table |= DPMTABLE_UPDATE_SCLK;
+	}
+
+	for (i = 0; i < mclk_table->count; i++) {
+		if (mclk == mclk_table->dpm_levels[i].value)
+			break;
+	}
+
+	if (i >= mclk_table->count)
+		pi->need_update_smu7_dpm_table |= DPMTABLE_OD_UPDATE_MCLK;
+
+	if (adev->pm.dpm.current_active_crtc_count !=
+	    adev->pm.dpm.new_active_crtc_count)
+		pi->need_update_smu7_dpm_table |= DPMTABLE_UPDATE_MCLK;
+}
+
+static int ci_populate_and_upload_sclk_mclk_dpm_levels(struct amdgpu_device *adev,
+						       struct amdgpu_ps *amdgpu_state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *state = ci_get_ps(amdgpu_state);
+	u32 sclk = state->performance_levels[state->performance_level_count-1].sclk;
+	u32 mclk = state->performance_levels[state->performance_level_count-1].mclk;
+	struct ci_dpm_table *dpm_table = &pi->dpm_table;
+	int ret;
+
+	if (!pi->need_update_smu7_dpm_table)
+		return 0;
+
+	if (pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_SCLK)
+		dpm_table->sclk_table.dpm_levels[dpm_table->sclk_table.count-1].value = sclk;
+
+	if (pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_MCLK)
+		dpm_table->mclk_table.dpm_levels[dpm_table->mclk_table.count-1].value = mclk;
+
+	if (pi->need_update_smu7_dpm_table & (DPMTABLE_OD_UPDATE_SCLK | DPMTABLE_UPDATE_SCLK)) {
+		ret = ci_populate_all_graphic_levels(adev);
+		if (ret)
+			return ret;
+	}
+
+	if (pi->need_update_smu7_dpm_table & (DPMTABLE_OD_UPDATE_MCLK | DPMTABLE_UPDATE_MCLK)) {
+		ret = ci_populate_all_memory_levels(adev);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int ci_enable_uvd_dpm(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct amdgpu_clock_and_voltage_limits *max_limits;
+	int i;
+
+	if (adev->pm.ac_power)
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	else
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
+
+	if (enable) {
+		pi->dpm_level_enable_mask.uvd_dpm_enable_mask = 0;
+
+		for (i = adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count - 1; i >= 0; i--) {
+			if (adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[i].v <= max_limits->vddc) {
+				pi->dpm_level_enable_mask.uvd_dpm_enable_mask |= 1 << i;
+
+				if (!pi->caps_uvd_dpm)
+					break;
+			}
+		}
+
+		amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+						  PPSMC_MSG_UVDDPM_SetEnabledMask,
+						  pi->dpm_level_enable_mask.uvd_dpm_enable_mask);
+
+		if (pi->last_mclk_dpm_enable_mask & 0x1) {
+			pi->uvd_enabled = true;
+			pi->dpm_level_enable_mask.mclk_dpm_enable_mask &= 0xFFFFFFFE;
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+							  PPSMC_MSG_MCLKDPM_SetEnabledMask,
+							  pi->dpm_level_enable_mask.mclk_dpm_enable_mask);
+		}
+	} else {
+		if (pi->uvd_enabled) {
+			pi->uvd_enabled = false;
+			pi->dpm_level_enable_mask.mclk_dpm_enable_mask |= 1;
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+							  PPSMC_MSG_MCLKDPM_SetEnabledMask,
+							  pi->dpm_level_enable_mask.mclk_dpm_enable_mask);
+		}
+	}
+
+	return (amdgpu_ci_send_msg_to_smc(adev, enable ?
+				   PPSMC_MSG_UVDDPM_Enable : PPSMC_MSG_UVDDPM_Disable) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+static int ci_enable_vce_dpm(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct amdgpu_clock_and_voltage_limits *max_limits;
+	int i;
+
+	if (adev->pm.ac_power)
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	else
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
+
+	if (enable) {
+		pi->dpm_level_enable_mask.vce_dpm_enable_mask = 0;
+		for (i = adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.count - 1; i >= 0; i--) {
+			if (adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].v <= max_limits->vddc) {
+				pi->dpm_level_enable_mask.vce_dpm_enable_mask |= 1 << i;
+
+				if (!pi->caps_vce_dpm)
+					break;
+			}
+		}
+
+		amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+						  PPSMC_MSG_VCEDPM_SetEnabledMask,
+						  pi->dpm_level_enable_mask.vce_dpm_enable_mask);
+	}
+
+	return (amdgpu_ci_send_msg_to_smc(adev, enable ?
+				   PPSMC_MSG_VCEDPM_Enable : PPSMC_MSG_VCEDPM_Disable) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+#if 0
+static int ci_enable_samu_dpm(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct amdgpu_clock_and_voltage_limits *max_limits;
+	int i;
+
+	if (adev->pm.ac_power)
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	else
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
+
+	if (enable) {
+		pi->dpm_level_enable_mask.samu_dpm_enable_mask = 0;
+		for (i = adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.count - 1; i >= 0; i--) {
+			if (adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[i].v <= max_limits->vddc) {
+				pi->dpm_level_enable_mask.samu_dpm_enable_mask |= 1 << i;
+
+				if (!pi->caps_samu_dpm)
+					break;
+			}
+		}
+
+		amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+						  PPSMC_MSG_SAMUDPM_SetEnabledMask,
+						  pi->dpm_level_enable_mask.samu_dpm_enable_mask);
+	}
+	return (amdgpu_ci_send_msg_to_smc(adev, enable ?
+				   PPSMC_MSG_SAMUDPM_Enable : PPSMC_MSG_SAMUDPM_Disable) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+static int ci_enable_acp_dpm(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct amdgpu_clock_and_voltage_limits *max_limits;
+	int i;
+
+	if (adev->pm.ac_power)
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	else
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
+
+	if (enable) {
+		pi->dpm_level_enable_mask.acp_dpm_enable_mask = 0;
+		for (i = adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.count - 1; i >= 0; i--) {
+			if (adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[i].v <= max_limits->vddc) {
+				pi->dpm_level_enable_mask.acp_dpm_enable_mask |= 1 << i;
+
+				if (!pi->caps_acp_dpm)
+					break;
+			}
+		}
+
+		amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+						  PPSMC_MSG_ACPDPM_SetEnabledMask,
+						  pi->dpm_level_enable_mask.acp_dpm_enable_mask);
+	}
+
+	return (amdgpu_ci_send_msg_to_smc(adev, enable ?
+				   PPSMC_MSG_ACPDPM_Enable : PPSMC_MSG_ACPDPM_Disable) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+#endif
+
+static int ci_update_uvd_dpm(struct amdgpu_device *adev, bool gate)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+	int ret = 0;
+
+	if (!gate) {
+		/* turn the clocks on when decoding */
+		if (pi->caps_uvd_dpm ||
+		    (adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count <= 0))
+			pi->smc_state_table.UvdBootLevel = 0;
+		else
+			pi->smc_state_table.UvdBootLevel =
+				adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count - 1;
+
+		tmp = RREG32_SMC(ixDPM_TABLE_475);
+		tmp &= ~DPM_TABLE_475__UvdBootLevel_MASK;
+		tmp |= (pi->smc_state_table.UvdBootLevel << DPM_TABLE_475__UvdBootLevel__SHIFT);
+		WREG32_SMC(ixDPM_TABLE_475, tmp);
+		ret = ci_enable_uvd_dpm(adev, true);
+	} else {
+		ret = ci_enable_uvd_dpm(adev, false);
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}
+
+static u8 ci_get_vce_boot_level(struct amdgpu_device *adev)
+{
+	u8 i;
+	u32 min_evclk = 30000; /* ??? */
+	struct amdgpu_vce_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+
+	for (i = 0; i < table->count; i++) {
+		if (table->entries[i].evclk >= min_evclk)
+			return i;
+	}
+
+	return table->count - 1;
+}
+
+static int ci_update_vce_dpm(struct amdgpu_device *adev,
+			     struct amdgpu_ps *amdgpu_new_state,
+			     struct amdgpu_ps *amdgpu_current_state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret = 0;
+	u32 tmp;
+
+	if (amdgpu_current_state->evclk != amdgpu_new_state->evclk) {
+		if (amdgpu_new_state->evclk) {
+			pi->smc_state_table.VceBootLevel = ci_get_vce_boot_level(adev);
+			tmp = RREG32_SMC(ixDPM_TABLE_475);
+			tmp &= ~DPM_TABLE_475__VceBootLevel_MASK;
+			tmp |= (pi->smc_state_table.VceBootLevel << DPM_TABLE_475__VceBootLevel__SHIFT);
+			WREG32_SMC(ixDPM_TABLE_475, tmp);
+
+			ret = ci_enable_vce_dpm(adev, true);
+		} else {
+			ret = ci_enable_vce_dpm(adev, false);
+			if (ret)
+				return ret;
+		}
+	}
+	return ret;
+}
+
+#if 0
+static int ci_update_samu_dpm(struct amdgpu_device *adev, bool gate)
+{
+	return ci_enable_samu_dpm(adev, gate);
+}
+
+static int ci_update_acp_dpm(struct amdgpu_device *adev, bool gate)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+
+	if (!gate) {
+		pi->smc_state_table.AcpBootLevel = 0;
+
+		tmp = RREG32_SMC(ixDPM_TABLE_475);
+		tmp &= ~AcpBootLevel_MASK;
+		tmp |= AcpBootLevel(pi->smc_state_table.AcpBootLevel);
+		WREG32_SMC(ixDPM_TABLE_475, tmp);
+	}
+
+	return ci_enable_acp_dpm(adev, !gate);
+}
+#endif
+
+static int ci_generate_dpm_level_enable_mask(struct amdgpu_device *adev,
+					     struct amdgpu_ps *amdgpu_state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret;
+
+	ret = ci_trim_dpm_states(adev, amdgpu_state);
+	if (ret)
+		return ret;
+
+	pi->dpm_level_enable_mask.sclk_dpm_enable_mask =
+		ci_get_dpm_level_enable_mask_value(&pi->dpm_table.sclk_table);
+	pi->dpm_level_enable_mask.mclk_dpm_enable_mask =
+		ci_get_dpm_level_enable_mask_value(&pi->dpm_table.mclk_table);
+	pi->last_mclk_dpm_enable_mask =
+		pi->dpm_level_enable_mask.mclk_dpm_enable_mask;
+	if (pi->uvd_enabled) {
+		if (pi->dpm_level_enable_mask.mclk_dpm_enable_mask & 1)
+			pi->dpm_level_enable_mask.mclk_dpm_enable_mask &= 0xFFFFFFFE;
+	}
+	pi->dpm_level_enable_mask.pcie_dpm_enable_mask =
+		ci_get_dpm_level_enable_mask_value(&pi->dpm_table.pcie_speed_table);
+
+	return 0;
+}
+
+static u32 ci_get_lowest_enabled_level(struct amdgpu_device *adev,
+				       u32 level_mask)
+{
+	u32 level = 0;
+
+	while ((level_mask & (1 << level)) == 0)
+		level++;
+
+	return level;
+}
+
+
+static int ci_dpm_force_performance_level(void *handle,
+					  enum amd_dpm_forced_level level)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp, levels, i;
+	int ret;
+
+	if (level == AMD_DPM_FORCED_LEVEL_HIGH) {
+		if ((!pi->pcie_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.pcie_dpm_enable_mask) {
+			levels = 0;
+			tmp = pi->dpm_level_enable_mask.pcie_dpm_enable_mask;
+			while (tmp >>= 1)
+				levels++;
+			if (levels) {
+				ret = ci_dpm_force_state_pcie(adev, level);
+				if (ret)
+					return ret;
+				for (i = 0; i < adev->usec_timeout; i++) {
+					tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX_1) &
+					TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX_MASK) >>
+					TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX__SHIFT;
+					if (tmp == levels)
+						break;
+					udelay(1);
+				}
+			}
+		}
+		if ((!pi->sclk_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.sclk_dpm_enable_mask) {
+			levels = 0;
+			tmp = pi->dpm_level_enable_mask.sclk_dpm_enable_mask;
+			while (tmp >>= 1)
+				levels++;
+			if (levels) {
+				ret = ci_dpm_force_state_sclk(adev, levels);
+				if (ret)
+					return ret;
+				for (i = 0; i < adev->usec_timeout; i++) {
+					tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
+					TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX_MASK) >>
+					TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX__SHIFT;
+					if (tmp == levels)
+						break;
+					udelay(1);
+				}
+			}
+		}
+		if ((!pi->mclk_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.mclk_dpm_enable_mask) {
+			levels = 0;
+			tmp = pi->dpm_level_enable_mask.mclk_dpm_enable_mask;
+			while (tmp >>= 1)
+				levels++;
+			if (levels) {
+				ret = ci_dpm_force_state_mclk(adev, levels);
+				if (ret)
+					return ret;
+				for (i = 0; i < adev->usec_timeout; i++) {
+					tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
+					TARGET_AND_CURRENT_PROFILE_INDEX__CURR_MCLK_INDEX_MASK) >>
+					TARGET_AND_CURRENT_PROFILE_INDEX__CURR_MCLK_INDEX__SHIFT;
+					if (tmp == levels)
+						break;
+					udelay(1);
+				}
+			}
+		}
+	} else if (level == AMD_DPM_FORCED_LEVEL_LOW) {
+		if ((!pi->sclk_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.sclk_dpm_enable_mask) {
+			levels = ci_get_lowest_enabled_level(adev,
+							     pi->dpm_level_enable_mask.sclk_dpm_enable_mask);
+			ret = ci_dpm_force_state_sclk(adev, levels);
+			if (ret)
+				return ret;
+			for (i = 0; i < adev->usec_timeout; i++) {
+				tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
+				TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX_MASK) >>
+				TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX__SHIFT;
+				if (tmp == levels)
+					break;
+				udelay(1);
+			}
+		}
+		if ((!pi->mclk_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.mclk_dpm_enable_mask) {
+			levels = ci_get_lowest_enabled_level(adev,
+							     pi->dpm_level_enable_mask.mclk_dpm_enable_mask);
+			ret = ci_dpm_force_state_mclk(adev, levels);
+			if (ret)
+				return ret;
+			for (i = 0; i < adev->usec_timeout; i++) {
+				tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
+				TARGET_AND_CURRENT_PROFILE_INDEX__CURR_MCLK_INDEX_MASK) >>
+				TARGET_AND_CURRENT_PROFILE_INDEX__CURR_MCLK_INDEX__SHIFT;
+				if (tmp == levels)
+					break;
+				udelay(1);
+			}
+		}
+		if ((!pi->pcie_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.pcie_dpm_enable_mask) {
+			levels = ci_get_lowest_enabled_level(adev,
+							     pi->dpm_level_enable_mask.pcie_dpm_enable_mask);
+			ret = ci_dpm_force_state_pcie(adev, levels);
+			if (ret)
+				return ret;
+			for (i = 0; i < adev->usec_timeout; i++) {
+				tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX_1) &
+				TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX_MASK) >>
+				TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX__SHIFT;
+				if (tmp == levels)
+					break;
+				udelay(1);
+			}
+		}
+	} else if (level == AMD_DPM_FORCED_LEVEL_AUTO) {
+		if (!pi->pcie_dpm_key_disabled) {
+			PPSMC_Result smc_result;
+
+			smc_result = amdgpu_ci_send_msg_to_smc(adev,
+							       PPSMC_MSG_PCIeDPM_UnForceLevel);
+			if (smc_result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+		ret = ci_upload_dpm_level_enable_mask(adev);
+		if (ret)
+			return ret;
+	}
+
+	adev->pm.dpm.forced_level = level;
+
+	return 0;
+}
+
+static int ci_set_mc_special_registers(struct amdgpu_device *adev,
+				       struct ci_mc_reg_table *table)
+{
+	u8 i, j, k;
+	u32 temp_reg;
+
+	for (i = 0, j = table->last; i < table->last; i++) {
+		if (j >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+			return -EINVAL;
+		switch(table->mc_reg_address[i].s1) {
+		case mmMC_SEQ_MISC1:
+			temp_reg = RREG32(mmMC_PMG_CMD_EMRS);
+			table->mc_reg_address[j].s1 = mmMC_PMG_CMD_EMRS;
+			table->mc_reg_address[j].s0 = mmMC_SEQ_PMG_CMD_EMRS_LP;
+			for (k = 0; k < table->num_entries; k++) {
+				table->mc_reg_table_entry[k].mc_data[j] =
+					((temp_reg & 0xffff0000)) | ((table->mc_reg_table_entry[k].mc_data[i] & 0xffff0000) >> 16);
+			}
+			j++;
+
+			if (j >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+				return -EINVAL;
+			temp_reg = RREG32(mmMC_PMG_CMD_MRS);
+			table->mc_reg_address[j].s1 = mmMC_PMG_CMD_MRS;
+			table->mc_reg_address[j].s0 = mmMC_SEQ_PMG_CMD_MRS_LP;
+			for (k = 0; k < table->num_entries; k++) {
+				table->mc_reg_table_entry[k].mc_data[j] =
+					(temp_reg & 0xffff0000) | (table->mc_reg_table_entry[k].mc_data[i] & 0x0000ffff);
+				if (adev->gmc.vram_type != AMDGPU_VRAM_TYPE_GDDR5)
+					table->mc_reg_table_entry[k].mc_data[j] |= 0x100;
+			}
+			j++;
+
+			if (adev->gmc.vram_type != AMDGPU_VRAM_TYPE_GDDR5) {
+				if (j >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+					return -EINVAL;
+				table->mc_reg_address[j].s1 = mmMC_PMG_AUTO_CMD;
+				table->mc_reg_address[j].s0 = mmMC_PMG_AUTO_CMD;
+				for (k = 0; k < table->num_entries; k++) {
+					table->mc_reg_table_entry[k].mc_data[j] =
+						(table->mc_reg_table_entry[k].mc_data[i] & 0xffff0000) >> 16;
+				}
+				j++;
+			}
+			break;
+		case mmMC_SEQ_RESERVE_M:
+			temp_reg = RREG32(mmMC_PMG_CMD_MRS1);
+			table->mc_reg_address[j].s1 = mmMC_PMG_CMD_MRS1;
+			table->mc_reg_address[j].s0 = mmMC_SEQ_PMG_CMD_MRS1_LP;
+			for (k = 0; k < table->num_entries; k++) {
+				table->mc_reg_table_entry[k].mc_data[j] =
+					(temp_reg & 0xffff0000) | (table->mc_reg_table_entry[k].mc_data[i] & 0x0000ffff);
+			}
+			j++;
+			break;
+		default:
+			break;
+		}
+
+	}
+
+	table->last = j;
+
+	return 0;
+}
+
+static bool ci_check_s0_mc_reg_index(u16 in_reg, u16 *out_reg)
+{
+	bool result = true;
+
+	switch(in_reg) {
+	case mmMC_SEQ_RAS_TIMING:
+		*out_reg = mmMC_SEQ_RAS_TIMING_LP;
+		break;
+	case mmMC_SEQ_DLL_STBY:
+		*out_reg = mmMC_SEQ_DLL_STBY_LP;
+		break;
+	case mmMC_SEQ_G5PDX_CMD0:
+		*out_reg = mmMC_SEQ_G5PDX_CMD0_LP;
+		break;
+	case mmMC_SEQ_G5PDX_CMD1:
+		*out_reg = mmMC_SEQ_G5PDX_CMD1_LP;
+		break;
+	case mmMC_SEQ_G5PDX_CTRL:
+		*out_reg = mmMC_SEQ_G5PDX_CTRL_LP;
+		break;
+	case mmMC_SEQ_CAS_TIMING:
+		*out_reg = mmMC_SEQ_CAS_TIMING_LP;
+	    break;
+	case mmMC_SEQ_MISC_TIMING:
+		*out_reg = mmMC_SEQ_MISC_TIMING_LP;
+		break;
+	case mmMC_SEQ_MISC_TIMING2:
+		*out_reg = mmMC_SEQ_MISC_TIMING2_LP;
+		break;
+	case mmMC_SEQ_PMG_DVS_CMD:
+		*out_reg = mmMC_SEQ_PMG_DVS_CMD_LP;
+		break;
+	case mmMC_SEQ_PMG_DVS_CTL:
+		*out_reg = mmMC_SEQ_PMG_DVS_CTL_LP;
+		break;
+	case mmMC_SEQ_RD_CTL_D0:
+		*out_reg = mmMC_SEQ_RD_CTL_D0_LP;
+		break;
+	case mmMC_SEQ_RD_CTL_D1:
+		*out_reg = mmMC_SEQ_RD_CTL_D1_LP;
+		break;
+	case mmMC_SEQ_WR_CTL_D0:
+		*out_reg = mmMC_SEQ_WR_CTL_D0_LP;
+		break;
+	case mmMC_SEQ_WR_CTL_D1:
+		*out_reg = mmMC_SEQ_WR_CTL_D1_LP;
+		break;
+	case mmMC_PMG_CMD_EMRS:
+		*out_reg = mmMC_SEQ_PMG_CMD_EMRS_LP;
+		break;
+	case mmMC_PMG_CMD_MRS:
+		*out_reg = mmMC_SEQ_PMG_CMD_MRS_LP;
+		break;
+	case mmMC_PMG_CMD_MRS1:
+		*out_reg = mmMC_SEQ_PMG_CMD_MRS1_LP;
+		break;
+	case mmMC_SEQ_PMG_TIMING:
+		*out_reg = mmMC_SEQ_PMG_TIMING_LP;
+		break;
+	case mmMC_PMG_CMD_MRS2:
+		*out_reg = mmMC_SEQ_PMG_CMD_MRS2_LP;
+		break;
+	case mmMC_SEQ_WR_CTL_2:
+		*out_reg = mmMC_SEQ_WR_CTL_2_LP;
+		break;
+	default:
+		result = false;
+		break;
+	}
+
+	return result;
+}
+
+static void ci_set_valid_flag(struct ci_mc_reg_table *table)
+{
+	u8 i, j;
+
+	for (i = 0; i < table->last; i++) {
+		for (j = 1; j < table->num_entries; j++) {
+			if (table->mc_reg_table_entry[j-1].mc_data[i] !=
+			    table->mc_reg_table_entry[j].mc_data[i]) {
+				table->valid_flag |= 1 << i;
+				break;
+			}
+		}
+	}
+}
+
+static void ci_set_s0_mc_reg_index(struct ci_mc_reg_table *table)
+{
+	u32 i;
+	u16 address;
+
+	for (i = 0; i < table->last; i++) {
+		table->mc_reg_address[i].s0 =
+			ci_check_s0_mc_reg_index(table->mc_reg_address[i].s1, &address) ?
+			address : table->mc_reg_address[i].s1;
+	}
+}
+
+static int ci_copy_vbios_mc_reg_table(const struct atom_mc_reg_table *table,
+				      struct ci_mc_reg_table *ci_table)
+{
+	u8 i, j;
+
+	if (table->last > SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+		return -EINVAL;
+	if (table->num_entries > MAX_AC_TIMING_ENTRIES)
+		return -EINVAL;
+
+	for (i = 0; i < table->last; i++)
+		ci_table->mc_reg_address[i].s1 = table->mc_reg_address[i].s1;
+
+	ci_table->last = table->last;
+
+	for (i = 0; i < table->num_entries; i++) {
+		ci_table->mc_reg_table_entry[i].mclk_max =
+			table->mc_reg_table_entry[i].mclk_max;
+		for (j = 0; j < table->last; j++)
+			ci_table->mc_reg_table_entry[i].mc_data[j] =
+				table->mc_reg_table_entry[i].mc_data[j];
+	}
+	ci_table->num_entries = table->num_entries;
+
+	return 0;
+}
+
+static int ci_register_patching_mc_seq(struct amdgpu_device *adev,
+				       struct ci_mc_reg_table *table)
+{
+	u8 i, k;
+	u32 tmp;
+	bool patch;
+
+	tmp = RREG32(mmMC_SEQ_MISC0);
+	patch = ((tmp & 0x0000f00) == 0x300) ? true : false;
+
+	if (patch &&
+	    ((adev->pdev->device == 0x67B0) ||
+	     (adev->pdev->device == 0x67B1))) {
+		for (i = 0; i < table->last; i++) {
+			if (table->last >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+				return -EINVAL;
+			switch (table->mc_reg_address[i].s1) {
+			case mmMC_SEQ_MISC1:
+				for (k = 0; k < table->num_entries; k++) {
+					if ((table->mc_reg_table_entry[k].mclk_max == 125000) ||
+					    (table->mc_reg_table_entry[k].mclk_max == 137500))
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFFFFF8) |
+							0x00000007;
+				}
+				break;
+			case mmMC_SEQ_WR_CTL_D0:
+				for (k = 0; k < table->num_entries; k++) {
+					if ((table->mc_reg_table_entry[k].mclk_max == 125000) ||
+					    (table->mc_reg_table_entry[k].mclk_max == 137500))
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFF0F00) |
+							0x0000D0DD;
+				}
+				break;
+			case mmMC_SEQ_WR_CTL_D1:
+				for (k = 0; k < table->num_entries; k++) {
+					if ((table->mc_reg_table_entry[k].mclk_max == 125000) ||
+					    (table->mc_reg_table_entry[k].mclk_max == 137500))
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFF0F00) |
+							0x0000D0DD;
+				}
+				break;
+			case mmMC_SEQ_WR_CTL_2:
+				for (k = 0; k < table->num_entries; k++) {
+					if ((table->mc_reg_table_entry[k].mclk_max == 125000) ||
+					    (table->mc_reg_table_entry[k].mclk_max == 137500))
+						table->mc_reg_table_entry[k].mc_data[i] = 0;
+				}
+				break;
+			case mmMC_SEQ_CAS_TIMING:
+				for (k = 0; k < table->num_entries; k++) {
+					if (table->mc_reg_table_entry[k].mclk_max == 125000)
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFE0FE0F) |
+							0x000C0140;
+					else if (table->mc_reg_table_entry[k].mclk_max == 137500)
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFE0FE0F) |
+							0x000C0150;
+				}
+				break;
+			case mmMC_SEQ_MISC_TIMING:
+				for (k = 0; k < table->num_entries; k++) {
+					if (table->mc_reg_table_entry[k].mclk_max == 125000)
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFFFFE0) |
+							0x00000030;
+					else if (table->mc_reg_table_entry[k].mclk_max == 137500)
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFFFFE0) |
+							0x00000035;
+				}
+				break;
+			default:
+				break;
+			}
+		}
+
+		WREG32(mmMC_SEQ_IO_DEBUG_INDEX, 3);
+		tmp = RREG32(mmMC_SEQ_IO_DEBUG_DATA);
+		tmp = (tmp & 0xFFF8FFFF) | (1 << 16);
+		WREG32(mmMC_SEQ_IO_DEBUG_INDEX, 3);
+		WREG32(mmMC_SEQ_IO_DEBUG_DATA, tmp);
+	}
+
+	return 0;
+}
+
+static int ci_initialize_mc_reg_table(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct atom_mc_reg_table *table;
+	struct ci_mc_reg_table *ci_table = &pi->mc_reg_table;
+	u8 module_index = ci_get_memory_module_index(adev);
+	int ret;
+
+	table = kzalloc(sizeof(struct atom_mc_reg_table), GFP_KERNEL);
+	if (!table)
+		return -ENOMEM;
+
+	WREG32(mmMC_SEQ_RAS_TIMING_LP, RREG32(mmMC_SEQ_RAS_TIMING));
+	WREG32(mmMC_SEQ_CAS_TIMING_LP, RREG32(mmMC_SEQ_CAS_TIMING));
+	WREG32(mmMC_SEQ_DLL_STBY_LP, RREG32(mmMC_SEQ_DLL_STBY));
+	WREG32(mmMC_SEQ_G5PDX_CMD0_LP, RREG32(mmMC_SEQ_G5PDX_CMD0));
+	WREG32(mmMC_SEQ_G5PDX_CMD1_LP, RREG32(mmMC_SEQ_G5PDX_CMD1));
+	WREG32(mmMC_SEQ_G5PDX_CTRL_LP, RREG32(mmMC_SEQ_G5PDX_CTRL));
+	WREG32(mmMC_SEQ_PMG_DVS_CMD_LP, RREG32(mmMC_SEQ_PMG_DVS_CMD));
+	WREG32(mmMC_SEQ_PMG_DVS_CTL_LP, RREG32(mmMC_SEQ_PMG_DVS_CTL));
+	WREG32(mmMC_SEQ_MISC_TIMING_LP, RREG32(mmMC_SEQ_MISC_TIMING));
+	WREG32(mmMC_SEQ_MISC_TIMING2_LP, RREG32(mmMC_SEQ_MISC_TIMING2));
+	WREG32(mmMC_SEQ_PMG_CMD_EMRS_LP, RREG32(mmMC_PMG_CMD_EMRS));
+	WREG32(mmMC_SEQ_PMG_CMD_MRS_LP, RREG32(mmMC_PMG_CMD_MRS));
+	WREG32(mmMC_SEQ_PMG_CMD_MRS1_LP, RREG32(mmMC_PMG_CMD_MRS1));
+	WREG32(mmMC_SEQ_WR_CTL_D0_LP, RREG32(mmMC_SEQ_WR_CTL_D0));
+	WREG32(mmMC_SEQ_WR_CTL_D1_LP, RREG32(mmMC_SEQ_WR_CTL_D1));
+	WREG32(mmMC_SEQ_RD_CTL_D0_LP, RREG32(mmMC_SEQ_RD_CTL_D0));
+	WREG32(mmMC_SEQ_RD_CTL_D1_LP, RREG32(mmMC_SEQ_RD_CTL_D1));
+	WREG32(mmMC_SEQ_PMG_TIMING_LP, RREG32(mmMC_SEQ_PMG_TIMING));
+	WREG32(mmMC_SEQ_PMG_CMD_MRS2_LP, RREG32(mmMC_PMG_CMD_MRS2));
+	WREG32(mmMC_SEQ_WR_CTL_2_LP, RREG32(mmMC_SEQ_WR_CTL_2));
+
+	ret = amdgpu_atombios_init_mc_reg_table(adev, module_index, table);
+	if (ret)
+		goto init_mc_done;
+
+	ret = ci_copy_vbios_mc_reg_table(table, ci_table);
+	if (ret)
+		goto init_mc_done;
+
+	ci_set_s0_mc_reg_index(ci_table);
+
+	ret = ci_register_patching_mc_seq(adev, ci_table);
+	if (ret)
+		goto init_mc_done;
+
+	ret = ci_set_mc_special_registers(adev, ci_table);
+	if (ret)
+		goto init_mc_done;
+
+	ci_set_valid_flag(ci_table);
+
+init_mc_done:
+	kfree(table);
+
+	return ret;
+}
+
+static int ci_populate_mc_reg_addresses(struct amdgpu_device *adev,
+					SMU7_Discrete_MCRegisters *mc_reg_table)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 i, j;
+
+	for (i = 0, j = 0; j < pi->mc_reg_table.last; j++) {
+		if (pi->mc_reg_table.valid_flag & (1 << j)) {
+			if (i >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+				return -EINVAL;
+			mc_reg_table->address[i].s0 = cpu_to_be16(pi->mc_reg_table.mc_reg_address[j].s0);
+			mc_reg_table->address[i].s1 = cpu_to_be16(pi->mc_reg_table.mc_reg_address[j].s1);
+			i++;
+		}
+	}
+
+	mc_reg_table->last = (u8)i;
+
+	return 0;
+}
+
+static void ci_convert_mc_registers(const struct ci_mc_reg_entry *entry,
+				    SMU7_Discrete_MCRegisterSet *data,
+				    u32 num_entries, u32 valid_flag)
+{
+	u32 i, j;
+
+	for (i = 0, j = 0; j < num_entries; j++) {
+		if (valid_flag & (1 << j)) {
+			data->value[i] = cpu_to_be32(entry->mc_data[j]);
+			i++;
+		}
+	}
+}
+
+static void ci_convert_mc_reg_table_entry_to_smc(struct amdgpu_device *adev,
+						 const u32 memory_clock,
+						 SMU7_Discrete_MCRegisterSet *mc_reg_table_data)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 i = 0;
+
+	for(i = 0; i < pi->mc_reg_table.num_entries; i++) {
+		if (memory_clock <= pi->mc_reg_table.mc_reg_table_entry[i].mclk_max)
+			break;
+	}
+
+	if ((i == pi->mc_reg_table.num_entries) && (i > 0))
+		--i;
+
+	ci_convert_mc_registers(&pi->mc_reg_table.mc_reg_table_entry[i],
+				mc_reg_table_data, pi->mc_reg_table.last,
+				pi->mc_reg_table.valid_flag);
+}
+
+static void ci_convert_mc_reg_table_to_smc(struct amdgpu_device *adev,
+					   SMU7_Discrete_MCRegisters *mc_reg_table)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 i;
+
+	for (i = 0; i < pi->dpm_table.mclk_table.count; i++)
+		ci_convert_mc_reg_table_entry_to_smc(adev,
+						     pi->dpm_table.mclk_table.dpm_levels[i].value,
+						     &mc_reg_table->data[i]);
+}
+
+static int ci_populate_initial_mc_reg_table(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret;
+
+	memset(&pi->smc_mc_reg_table, 0, sizeof(SMU7_Discrete_MCRegisters));
+
+	ret = ci_populate_mc_reg_addresses(adev, &pi->smc_mc_reg_table);
+	if (ret)
+		return ret;
+	ci_convert_mc_reg_table_to_smc(adev, &pi->smc_mc_reg_table);
+
+	return amdgpu_ci_copy_bytes_to_smc(adev,
+				    pi->mc_reg_table_start,
+				    (u8 *)&pi->smc_mc_reg_table,
+				    sizeof(SMU7_Discrete_MCRegisters),
+				    pi->sram_end);
+}
+
+static int ci_update_and_upload_mc_reg_table(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (!(pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_MCLK))
+		return 0;
+
+	memset(&pi->smc_mc_reg_table, 0, sizeof(SMU7_Discrete_MCRegisters));
+
+	ci_convert_mc_reg_table_to_smc(adev, &pi->smc_mc_reg_table);
+
+	return amdgpu_ci_copy_bytes_to_smc(adev,
+				    pi->mc_reg_table_start +
+				    offsetof(SMU7_Discrete_MCRegisters, data[0]),
+				    (u8 *)&pi->smc_mc_reg_table.data[0],
+				    sizeof(SMU7_Discrete_MCRegisterSet) *
+				    pi->dpm_table.mclk_table.count,
+				    pi->sram_end);
+}
+
+static void ci_enable_voltage_control(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+
+	tmp |= GENERAL_PWRMGT__VOLT_PWRMGT_EN_MASK;
+	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+}
+
+static enum amdgpu_pcie_gen ci_get_maximum_link_speed(struct amdgpu_device *adev,
+						      struct amdgpu_ps *amdgpu_state)
+{
+	struct ci_ps *state = ci_get_ps(amdgpu_state);
+	int i;
+	u16 pcie_speed, max_speed = 0;
+
+	for (i = 0; i < state->performance_level_count; i++) {
+		pcie_speed = state->performance_levels[i].pcie_gen;
+		if (max_speed < pcie_speed)
+			max_speed = pcie_speed;
+	}
+
+	return max_speed;
+}
+
+static u16 ci_get_current_pcie_speed(struct amdgpu_device *adev)
+{
+	u32 speed_cntl = 0;
+
+	speed_cntl = RREG32_PCIE(ixPCIE_LC_SPEED_CNTL) &
+		PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE_MASK;
+	speed_cntl >>= PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE__SHIFT;
+
+	return (u16)speed_cntl;
+}
+
+static int ci_get_current_pcie_lane_number(struct amdgpu_device *adev)
+{
+	u32 link_width = 0;
+
+	link_width = RREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL) &
+		PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD_MASK;
+	link_width >>= PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD__SHIFT;
+
+	switch (link_width) {
+	case 1:
+		return 1;
+	case 2:
+		return 2;
+	case 3:
+		return 4;
+	case 4:
+		return 8;
+	case 0:
+	case 6:
+	default:
+		return 16;
+	}
+}
+
+static void ci_request_link_speed_change_before_state_change(struct amdgpu_device *adev,
+							     struct amdgpu_ps *amdgpu_new_state,
+							     struct amdgpu_ps *amdgpu_current_state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	enum amdgpu_pcie_gen target_link_speed =
+		ci_get_maximum_link_speed(adev, amdgpu_new_state);
+	enum amdgpu_pcie_gen current_link_speed;
+
+	if (pi->force_pcie_gen == AMDGPU_PCIE_GEN_INVALID)
+		current_link_speed = ci_get_maximum_link_speed(adev, amdgpu_current_state);
+	else
+		current_link_speed = pi->force_pcie_gen;
+
+	pi->force_pcie_gen = AMDGPU_PCIE_GEN_INVALID;
+	pi->pspp_notify_required = false;
+	if (target_link_speed > current_link_speed) {
+		switch (target_link_speed) {
+#ifdef CONFIG_ACPI
+		case AMDGPU_PCIE_GEN3:
+			if (amdgpu_acpi_pcie_performance_request(adev, PCIE_PERF_REQ_PECI_GEN3, false) == 0)
+				break;
+			pi->force_pcie_gen = AMDGPU_PCIE_GEN2;
+			if (current_link_speed == AMDGPU_PCIE_GEN2)
+				break;
+		case AMDGPU_PCIE_GEN2:
+			if (amdgpu_acpi_pcie_performance_request(adev, PCIE_PERF_REQ_PECI_GEN2, false) == 0)
+				break;
+#endif
+		default:
+			pi->force_pcie_gen = ci_get_current_pcie_speed(adev);
+			break;
+		}
+	} else {
+		if (target_link_speed < current_link_speed)
+			pi->pspp_notify_required = true;
+	}
+}
+
+static void ci_notify_link_speed_change_after_state_change(struct amdgpu_device *adev,
+							   struct amdgpu_ps *amdgpu_new_state,
+							   struct amdgpu_ps *amdgpu_current_state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	enum amdgpu_pcie_gen target_link_speed =
+		ci_get_maximum_link_speed(adev, amdgpu_new_state);
+	u8 request;
+
+	if (pi->pspp_notify_required) {
+		if (target_link_speed == AMDGPU_PCIE_GEN3)
+			request = PCIE_PERF_REQ_PECI_GEN3;
+		else if (target_link_speed == AMDGPU_PCIE_GEN2)
+			request = PCIE_PERF_REQ_PECI_GEN2;
+		else
+			request = PCIE_PERF_REQ_PECI_GEN1;
+
+		if ((request == PCIE_PERF_REQ_PECI_GEN1) &&
+		    (ci_get_current_pcie_speed(adev) > 0))
+			return;
+
+#ifdef CONFIG_ACPI
+		amdgpu_acpi_pcie_performance_request(adev, request, false);
+#endif
+	}
+}
+
+static int ci_set_private_data_variables_based_on_pptable(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *allowed_sclk_vddc_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	struct amdgpu_clock_voltage_dependency_table *allowed_mclk_vddc_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk;
+	struct amdgpu_clock_voltage_dependency_table *allowed_mclk_vddci_table =
+		&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk;
+
+	if (allowed_sclk_vddc_table == NULL)
+		return -EINVAL;
+	if (allowed_sclk_vddc_table->count < 1)
+		return -EINVAL;
+	if (allowed_mclk_vddc_table == NULL)
+		return -EINVAL;
+	if (allowed_mclk_vddc_table->count < 1)
+		return -EINVAL;
+	if (allowed_mclk_vddci_table == NULL)
+		return -EINVAL;
+	if (allowed_mclk_vddci_table->count < 1)
+		return -EINVAL;
+
+	pi->min_vddc_in_pp_table = allowed_sclk_vddc_table->entries[0].v;
+	pi->max_vddc_in_pp_table =
+		allowed_sclk_vddc_table->entries[allowed_sclk_vddc_table->count - 1].v;
+
+	pi->min_vddci_in_pp_table = allowed_mclk_vddci_table->entries[0].v;
+	pi->max_vddci_in_pp_table =
+		allowed_mclk_vddci_table->entries[allowed_mclk_vddci_table->count - 1].v;
+
+	adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.sclk =
+		allowed_sclk_vddc_table->entries[allowed_sclk_vddc_table->count - 1].clk;
+	adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.mclk =
+		allowed_mclk_vddc_table->entries[allowed_sclk_vddc_table->count - 1].clk;
+	adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.vddc =
+		allowed_sclk_vddc_table->entries[allowed_sclk_vddc_table->count - 1].v;
+	adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.vddci =
+		allowed_mclk_vddci_table->entries[allowed_mclk_vddci_table->count - 1].v;
+
+	return 0;
+}
+
+static void ci_patch_with_vddc_leakage(struct amdgpu_device *adev, u16 *vddc)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_leakage_voltage *leakage_table = &pi->vddc_leakage;
+	u32 leakage_index;
+
+	for (leakage_index = 0; leakage_index < leakage_table->count; leakage_index++) {
+		if (leakage_table->leakage_id[leakage_index] == *vddc) {
+			*vddc = leakage_table->actual_voltage[leakage_index];
+			break;
+		}
+	}
+}
+
+static void ci_patch_with_vddci_leakage(struct amdgpu_device *adev, u16 *vddci)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_leakage_voltage *leakage_table = &pi->vddci_leakage;
+	u32 leakage_index;
+
+	for (leakage_index = 0; leakage_index < leakage_table->count; leakage_index++) {
+		if (leakage_table->leakage_id[leakage_index] == *vddci) {
+			*vddci = leakage_table->actual_voltage[leakage_index];
+			break;
+		}
+	}
+}
+
+static void ci_patch_clock_voltage_dependency_table_with_vddc_leakage(struct amdgpu_device *adev,
+								      struct amdgpu_clock_voltage_dependency_table *table)
+{
+	u32 i;
+
+	if (table) {
+		for (i = 0; i < table->count; i++)
+			ci_patch_with_vddc_leakage(adev, &table->entries[i].v);
+	}
+}
+
+static void ci_patch_clock_voltage_dependency_table_with_vddci_leakage(struct amdgpu_device *adev,
+								       struct amdgpu_clock_voltage_dependency_table *table)
+{
+	u32 i;
+
+	if (table) {
+		for (i = 0; i < table->count; i++)
+			ci_patch_with_vddci_leakage(adev, &table->entries[i].v);
+	}
+}
+
+static void ci_patch_vce_clock_voltage_dependency_table_with_vddc_leakage(struct amdgpu_device *adev,
+									  struct amdgpu_vce_clock_voltage_dependency_table *table)
+{
+	u32 i;
+
+	if (table) {
+		for (i = 0; i < table->count; i++)
+			ci_patch_with_vddc_leakage(adev, &table->entries[i].v);
+	}
+}
+
+static void ci_patch_uvd_clock_voltage_dependency_table_with_vddc_leakage(struct amdgpu_device *adev,
+									  struct amdgpu_uvd_clock_voltage_dependency_table *table)
+{
+	u32 i;
+
+	if (table) {
+		for (i = 0; i < table->count; i++)
+			ci_patch_with_vddc_leakage(adev, &table->entries[i].v);
+	}
+}
+
+static void ci_patch_vddc_phase_shed_limit_table_with_vddc_leakage(struct amdgpu_device *adev,
+								   struct amdgpu_phase_shedding_limits_table *table)
+{
+	u32 i;
+
+	if (table) {
+		for (i = 0; i < table->count; i++)
+			ci_patch_with_vddc_leakage(adev, &table->entries[i].voltage);
+	}
+}
+
+static void ci_patch_clock_voltage_limits_with_vddc_leakage(struct amdgpu_device *adev,
+							    struct amdgpu_clock_and_voltage_limits *table)
+{
+	if (table) {
+		ci_patch_with_vddc_leakage(adev, (u16 *)&table->vddc);
+		ci_patch_with_vddci_leakage(adev, (u16 *)&table->vddci);
+	}
+}
+
+static void ci_patch_cac_leakage_table_with_vddc_leakage(struct amdgpu_device *adev,
+							 struct amdgpu_cac_leakage_table *table)
+{
+	u32 i;
+
+	if (table) {
+		for (i = 0; i < table->count; i++)
+			ci_patch_with_vddc_leakage(adev, &table->entries[i].vddc);
+	}
+}
+
+static void ci_patch_dependency_tables_with_leakage(struct amdgpu_device *adev)
+{
+
+	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								  &adev->pm.dpm.dyn_state.vddc_dependency_on_sclk);
+	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								  &adev->pm.dpm.dyn_state.vddc_dependency_on_mclk);
+	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								  &adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk);
+	ci_patch_clock_voltage_dependency_table_with_vddci_leakage(adev,
+								   &adev->pm.dpm.dyn_state.vddci_dependency_on_mclk);
+	ci_patch_vce_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								      &adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table);
+	ci_patch_uvd_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								      &adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table);
+	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								  &adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table);
+	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								  &adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table);
+	ci_patch_vddc_phase_shed_limit_table_with_vddc_leakage(adev,
+							       &adev->pm.dpm.dyn_state.phase_shedding_limits_table);
+	ci_patch_clock_voltage_limits_with_vddc_leakage(adev,
+							&adev->pm.dpm.dyn_state.max_clock_voltage_on_ac);
+	ci_patch_clock_voltage_limits_with_vddc_leakage(adev,
+							&adev->pm.dpm.dyn_state.max_clock_voltage_on_dc);
+	ci_patch_cac_leakage_table_with_vddc_leakage(adev,
+						     &adev->pm.dpm.dyn_state.cac_leakage_table);
+
+}
+
+static void ci_update_current_ps(struct amdgpu_device *adev,
+				 struct amdgpu_ps *rps)
+{
+	struct ci_ps *new_ps = ci_get_ps(rps);
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	pi->current_rps = *rps;
+	pi->current_ps = *new_ps;
+	pi->current_rps.ps_priv = &pi->current_ps;
+	adev->pm.dpm.current_ps = &pi->current_rps;
+}
+
+static void ci_update_requested_ps(struct amdgpu_device *adev,
+				   struct amdgpu_ps *rps)
+{
+	struct ci_ps *new_ps = ci_get_ps(rps);
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	pi->requested_rps = *rps;
+	pi->requested_ps = *new_ps;
+	pi->requested_rps.ps_priv = &pi->requested_ps;
+	adev->pm.dpm.requested_ps = &pi->requested_rps;
+}
+
+static int ci_dpm_pre_set_power_state(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_ps requested_ps = *adev->pm.dpm.requested_ps;
+	struct amdgpu_ps *new_ps = &requested_ps;
+
+	ci_update_requested_ps(adev, new_ps);
+
+	ci_apply_state_adjust_rules(adev, &pi->requested_rps);
+
+	return 0;
+}
+
+static void ci_dpm_post_set_power_state(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_ps *new_ps = &pi->requested_rps;
+
+	ci_update_current_ps(adev, new_ps);
+}
+
+
+static void ci_dpm_setup_asic(struct amdgpu_device *adev)
+{
+	ci_read_clock_registers(adev);
+	ci_enable_acpi_power_management(adev);
+	ci_init_sclk_t(adev);
+}
+
+static int ci_dpm_enable(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_ps *boot_ps = adev->pm.dpm.boot_ps;
+	int ret;
+
+	if (pi->voltage_control != CISLANDS_VOLTAGE_CONTROL_NONE) {
+		ci_enable_voltage_control(adev);
+		ret = ci_construct_voltage_tables(adev);
+		if (ret) {
+			DRM_ERROR("ci_construct_voltage_tables failed\n");
+			return ret;
+		}
+	}
+	if (pi->caps_dynamic_ac_timing) {
+		ret = ci_initialize_mc_reg_table(adev);
+		if (ret)
+			pi->caps_dynamic_ac_timing = false;
+	}
+	if (pi->dynamic_ss)
+		ci_enable_spread_spectrum(adev, true);
+	if (pi->thermal_protection)
+		ci_enable_thermal_protection(adev, true);
+	ci_program_sstp(adev);
+	ci_enable_display_gap(adev);
+	ci_program_vc(adev);
+	ret = ci_upload_firmware(adev);
+	if (ret) {
+		DRM_ERROR("ci_upload_firmware failed\n");
+		return ret;
+	}
+	ret = ci_process_firmware_header(adev);
+	if (ret) {
+		DRM_ERROR("ci_process_firmware_header failed\n");
+		return ret;
+	}
+	ret = ci_initial_switch_from_arb_f0_to_f1(adev);
+	if (ret) {
+		DRM_ERROR("ci_initial_switch_from_arb_f0_to_f1 failed\n");
+		return ret;
+	}
+	ret = ci_init_smc_table(adev);
+	if (ret) {
+		DRM_ERROR("ci_init_smc_table failed\n");
+		return ret;
+	}
+	ret = ci_init_arb_table_index(adev);
+	if (ret) {
+		DRM_ERROR("ci_init_arb_table_index failed\n");
+		return ret;
+	}
+	if (pi->caps_dynamic_ac_timing) {
+		ret = ci_populate_initial_mc_reg_table(adev);
+		if (ret) {
+			DRM_ERROR("ci_populate_initial_mc_reg_table failed\n");
+			return ret;
+		}
+	}
+	ret = ci_populate_pm_base(adev);
+	if (ret) {
+		DRM_ERROR("ci_populate_pm_base failed\n");
+		return ret;
+	}
+	ci_dpm_start_smc(adev);
+	ci_enable_vr_hot_gpio_interrupt(adev);
+	ret = ci_notify_smc_display_change(adev, false);
+	if (ret) {
+		DRM_ERROR("ci_notify_smc_display_change failed\n");
+		return ret;
+	}
+	ci_enable_sclk_control(adev, true);
+	ret = ci_enable_ulv(adev, true);
+	if (ret) {
+		DRM_ERROR("ci_enable_ulv failed\n");
+		return ret;
+	}
+	ret = ci_enable_ds_master_switch(adev, true);
+	if (ret) {
+		DRM_ERROR("ci_enable_ds_master_switch failed\n");
+		return ret;
+	}
+	ret = ci_start_dpm(adev);
+	if (ret) {
+		DRM_ERROR("ci_start_dpm failed\n");
+		return ret;
+	}
+	ret = ci_enable_didt(adev, true);
+	if (ret) {
+		DRM_ERROR("ci_enable_didt failed\n");
+		return ret;
+	}
+	ret = ci_enable_smc_cac(adev, true);
+	if (ret) {
+		DRM_ERROR("ci_enable_smc_cac failed\n");
+		return ret;
+	}
+	ret = ci_enable_power_containment(adev, true);
+	if (ret) {
+		DRM_ERROR("ci_enable_power_containment failed\n");
+		return ret;
+	}
+
+	ret = ci_power_control_set_level(adev);
+	if (ret) {
+		DRM_ERROR("ci_power_control_set_level failed\n");
+		return ret;
+	}
+
+	ci_enable_auto_throttle_source(adev, AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL, true);
+
+	ret = ci_enable_thermal_based_sclk_dpm(adev, true);
+	if (ret) {
+		DRM_ERROR("ci_enable_thermal_based_sclk_dpm failed\n");
+		return ret;
+	}
+
+	ci_thermal_start_thermal_controller(adev);
+
+	ci_update_current_ps(adev, boot_ps);
+
+	return 0;
+}
+
+static void ci_dpm_disable(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_ps *boot_ps = adev->pm.dpm.boot_ps;
+
+	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
+		       AMDGPU_THERMAL_IRQ_LOW_TO_HIGH);
+	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
+		       AMDGPU_THERMAL_IRQ_HIGH_TO_LOW);
+
+	ci_dpm_powergate_uvd(adev, true);
+
+	if (!amdgpu_ci_is_smc_running(adev))
+		return;
+
+	ci_thermal_stop_thermal_controller(adev);
+
+	if (pi->thermal_protection)
+		ci_enable_thermal_protection(adev, false);
+	ci_enable_power_containment(adev, false);
+	ci_enable_smc_cac(adev, false);
+	ci_enable_didt(adev, false);
+	ci_enable_spread_spectrum(adev, false);
+	ci_enable_auto_throttle_source(adev, AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL, false);
+	ci_stop_dpm(adev);
+	ci_enable_ds_master_switch(adev, false);
+	ci_enable_ulv(adev, false);
+	ci_clear_vc(adev);
+	ci_reset_to_default(adev);
+	ci_dpm_stop_smc(adev);
+	ci_force_switch_to_arb_f0(adev);
+	ci_enable_thermal_based_sclk_dpm(adev, false);
+
+	ci_update_current_ps(adev, boot_ps);
+}
+
+static int ci_dpm_set_power_state(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_ps *new_ps = &pi->requested_rps;
+	struct amdgpu_ps *old_ps = &pi->current_rps;
+	int ret;
+
+	ci_find_dpm_states_clocks_in_dpm_table(adev, new_ps);
+	if (pi->pcie_performance_request)
+		ci_request_link_speed_change_before_state_change(adev, new_ps, old_ps);
+	ret = ci_freeze_sclk_mclk_dpm(adev);
+	if (ret) {
+		DRM_ERROR("ci_freeze_sclk_mclk_dpm failed\n");
+		return ret;
+	}
+	ret = ci_populate_and_upload_sclk_mclk_dpm_levels(adev, new_ps);
+	if (ret) {
+		DRM_ERROR("ci_populate_and_upload_sclk_mclk_dpm_levels failed\n");
+		return ret;
+	}
+	ret = ci_generate_dpm_level_enable_mask(adev, new_ps);
+	if (ret) {
+		DRM_ERROR("ci_generate_dpm_level_enable_mask failed\n");
+		return ret;
+	}
+
+	ret = ci_update_vce_dpm(adev, new_ps, old_ps);
+	if (ret) {
+		DRM_ERROR("ci_update_vce_dpm failed\n");
+		return ret;
+	}
+
+	ret = ci_update_sclk_t(adev);
+	if (ret) {
+		DRM_ERROR("ci_update_sclk_t failed\n");
+		return ret;
+	}
+	if (pi->caps_dynamic_ac_timing) {
+		ret = ci_update_and_upload_mc_reg_table(adev);
+		if (ret) {
+			DRM_ERROR("ci_update_and_upload_mc_reg_table failed\n");
+			return ret;
+		}
+	}
+	ret = ci_program_memory_timing_parameters(adev);
+	if (ret) {
+		DRM_ERROR("ci_program_memory_timing_parameters failed\n");
+		return ret;
+	}
+	ret = ci_unfreeze_sclk_mclk_dpm(adev);
+	if (ret) {
+		DRM_ERROR("ci_unfreeze_sclk_mclk_dpm failed\n");
+		return ret;
+	}
+	ret = ci_upload_dpm_level_enable_mask(adev);
+	if (ret) {
+		DRM_ERROR("ci_upload_dpm_level_enable_mask failed\n");
+		return ret;
+	}
+	if (pi->pcie_performance_request)
+		ci_notify_link_speed_change_after_state_change(adev, new_ps, old_ps);
+
+	return 0;
+}
+
+#if 0
+static void ci_dpm_reset_asic(struct amdgpu_device *adev)
+{
+	ci_set_boot_state(adev);
+}
+#endif
+
+static void ci_dpm_display_configuration_changed(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	ci_program_display_gap(adev);
+}
+
+union power_info {
+	struct _ATOM_POWERPLAY_INFO info;
+	struct _ATOM_POWERPLAY_INFO_V2 info_2;
+	struct _ATOM_POWERPLAY_INFO_V3 info_3;
+	struct _ATOM_PPLIB_POWERPLAYTABLE pplib;
+	struct _ATOM_PPLIB_POWERPLAYTABLE2 pplib2;
+	struct _ATOM_PPLIB_POWERPLAYTABLE3 pplib3;
+};
+
+union pplib_clock_info {
+	struct _ATOM_PPLIB_R600_CLOCK_INFO r600;
+	struct _ATOM_PPLIB_RS780_CLOCK_INFO rs780;
+	struct _ATOM_PPLIB_EVERGREEN_CLOCK_INFO evergreen;
+	struct _ATOM_PPLIB_SUMO_CLOCK_INFO sumo;
+	struct _ATOM_PPLIB_SI_CLOCK_INFO si;
+	struct _ATOM_PPLIB_CI_CLOCK_INFO ci;
+};
+
+union pplib_power_state {
+	struct _ATOM_PPLIB_STATE v1;
+	struct _ATOM_PPLIB_STATE_V2 v2;
+};
+
+static void ci_parse_pplib_non_clock_info(struct amdgpu_device *adev,
+					  struct amdgpu_ps *rps,
+					  struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info,
+					  u8 table_rev)
+{
+	rps->caps = le32_to_cpu(non_clock_info->ulCapsAndSettings);
+	rps->class = le16_to_cpu(non_clock_info->usClassification);
+	rps->class2 = le16_to_cpu(non_clock_info->usClassification2);
+
+	if (ATOM_PPLIB_NONCLOCKINFO_VER1 < table_rev) {
+		rps->vclk = le32_to_cpu(non_clock_info->ulVCLK);
+		rps->dclk = le32_to_cpu(non_clock_info->ulDCLK);
+	} else {
+		rps->vclk = 0;
+		rps->dclk = 0;
+	}
+
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_BOOT)
+		adev->pm.dpm.boot_ps = rps;
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
+		adev->pm.dpm.uvd_ps = rps;
+}
+
+static void ci_parse_pplib_clock_info(struct amdgpu_device *adev,
+				      struct amdgpu_ps *rps, int index,
+				      union pplib_clock_info *clock_info)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *ps = ci_get_ps(rps);
+	struct ci_pl *pl = &ps->performance_levels[index];
+
+	ps->performance_level_count = index + 1;
+
+	pl->sclk = le16_to_cpu(clock_info->ci.usEngineClockLow);
+	pl->sclk |= clock_info->ci.ucEngineClockHigh << 16;
+	pl->mclk = le16_to_cpu(clock_info->ci.usMemoryClockLow);
+	pl->mclk |= clock_info->ci.ucMemoryClockHigh << 16;
+
+	pl->pcie_gen = amdgpu_get_pcie_gen_support(adev,
+						   pi->sys_pcie_mask,
+						   pi->vbios_boot_state.pcie_gen_bootup_value,
+						   clock_info->ci.ucPCIEGen);
+	pl->pcie_lane = amdgpu_get_pcie_lane_support(adev,
+						     pi->vbios_boot_state.pcie_lane_bootup_value,
+						     le16_to_cpu(clock_info->ci.usPCIELane));
+
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_ACPI) {
+		pi->acpi_pcie_gen = pl->pcie_gen;
+	}
+
+	if (rps->class2 & ATOM_PPLIB_CLASSIFICATION2_ULV) {
+		pi->ulv.supported = true;
+		pi->ulv.pl = *pl;
+		pi->ulv.cg_ulv_parameter = CISLANDS_CGULVPARAMETER_DFLT;
+	}
+
+	/* patch up boot state */
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_BOOT) {
+		pl->mclk = pi->vbios_boot_state.mclk_bootup_value;
+		pl->sclk = pi->vbios_boot_state.sclk_bootup_value;
+		pl->pcie_gen = pi->vbios_boot_state.pcie_gen_bootup_value;
+		pl->pcie_lane = pi->vbios_boot_state.pcie_lane_bootup_value;
+	}
+
+	switch (rps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) {
+	case ATOM_PPLIB_CLASSIFICATION_UI_BATTERY:
+		pi->use_pcie_powersaving_levels = true;
+		if (pi->pcie_gen_powersaving.max < pl->pcie_gen)
+			pi->pcie_gen_powersaving.max = pl->pcie_gen;
+		if (pi->pcie_gen_powersaving.min > pl->pcie_gen)
+			pi->pcie_gen_powersaving.min = pl->pcie_gen;
+		if (pi->pcie_lane_powersaving.max < pl->pcie_lane)
+			pi->pcie_lane_powersaving.max = pl->pcie_lane;
+		if (pi->pcie_lane_powersaving.min > pl->pcie_lane)
+			pi->pcie_lane_powersaving.min = pl->pcie_lane;
+		break;
+	case ATOM_PPLIB_CLASSIFICATION_UI_PERFORMANCE:
+		pi->use_pcie_performance_levels = true;
+		if (pi->pcie_gen_performance.max < pl->pcie_gen)
+			pi->pcie_gen_performance.max = pl->pcie_gen;
+		if (pi->pcie_gen_performance.min > pl->pcie_gen)
+			pi->pcie_gen_performance.min = pl->pcie_gen;
+		if (pi->pcie_lane_performance.max < pl->pcie_lane)
+			pi->pcie_lane_performance.max = pl->pcie_lane;
+		if (pi->pcie_lane_performance.min > pl->pcie_lane)
+			pi->pcie_lane_performance.min = pl->pcie_lane;
+		break;
+	default:
+		break;
+	}
+}
+
+static int ci_parse_power_table(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info;
+	union pplib_power_state *power_state;
+	int i, j, k, non_clock_array_index, clock_array_index;
+	union pplib_clock_info *clock_info;
+	struct _StateArray *state_array;
+	struct _ClockInfoArray *clock_info_array;
+	struct _NonClockInfoArray *non_clock_info_array;
+	union power_info *power_info;
+	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
+	u16 data_offset;
+	u8 frev, crev;
+	u8 *power_state_offset;
+	struct ci_ps *ps;
+
+	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset))
+		return -EINVAL;
+	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
+
+	amdgpu_add_thermal_controller(adev);
+
+	state_array = (struct _StateArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usStateArrayOffset));
+	clock_info_array = (struct _ClockInfoArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usClockInfoArrayOffset));
+	non_clock_info_array = (struct _NonClockInfoArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usNonClockInfoArrayOffset));
+
+	adev->pm.dpm.ps = kcalloc(state_array->ucNumEntries,
+				  sizeof(struct amdgpu_ps),
+				  GFP_KERNEL);
+	if (!adev->pm.dpm.ps)
+		return -ENOMEM;
+	power_state_offset = (u8 *)state_array->states;
+	for (i = 0; i < state_array->ucNumEntries; i++) {
+		u8 *idx;
+		power_state = (union pplib_power_state *)power_state_offset;
+		non_clock_array_index = power_state->v2.nonClockInfoIndex;
+		non_clock_info = (struct _ATOM_PPLIB_NONCLOCK_INFO *)
+			&non_clock_info_array->nonClockInfo[non_clock_array_index];
+		ps = kzalloc(sizeof(struct ci_ps), GFP_KERNEL);
+		if (ps == NULL) {
+			kfree(adev->pm.dpm.ps);
+			return -ENOMEM;
+		}
+		adev->pm.dpm.ps[i].ps_priv = ps;
+		ci_parse_pplib_non_clock_info(adev, &adev->pm.dpm.ps[i],
+					      non_clock_info,
+					      non_clock_info_array->ucEntrySize);
+		k = 0;
+		idx = (u8 *)&power_state->v2.clockInfoIndex[0];
+		for (j = 0; j < power_state->v2.ucNumDPMLevels; j++) {
+			clock_array_index = idx[j];
+			if (clock_array_index >= clock_info_array->ucNumEntries)
+				continue;
+			if (k >= CISLANDS_MAX_HARDWARE_POWERLEVELS)
+				break;
+			clock_info = (union pplib_clock_info *)
+				((u8 *)&clock_info_array->clockInfo[0] +
+				 (clock_array_index * clock_info_array->ucEntrySize));
+			ci_parse_pplib_clock_info(adev,
+						  &adev->pm.dpm.ps[i], k,
+						  clock_info);
+			k++;
+		}
+		power_state_offset += 2 + power_state->v2.ucNumDPMLevels;
+	}
+	adev->pm.dpm.num_ps = state_array->ucNumEntries;
+
+	/* fill in the vce power states */
+	for (i = 0; i < adev->pm.dpm.num_of_vce_states; i++) {
+		u32 sclk, mclk;
+		clock_array_index = adev->pm.dpm.vce_states[i].clk_idx;
+		clock_info = (union pplib_clock_info *)
+			&clock_info_array->clockInfo[clock_array_index * clock_info_array->ucEntrySize];
+		sclk = le16_to_cpu(clock_info->ci.usEngineClockLow);
+		sclk |= clock_info->ci.ucEngineClockHigh << 16;
+		mclk = le16_to_cpu(clock_info->ci.usMemoryClockLow);
+		mclk |= clock_info->ci.ucMemoryClockHigh << 16;
+		adev->pm.dpm.vce_states[i].sclk = sclk;
+		adev->pm.dpm.vce_states[i].mclk = mclk;
+	}
+
+	return 0;
+}
+
+static int ci_get_vbios_boot_values(struct amdgpu_device *adev,
+				    struct ci_vbios_boot_state *boot_state)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, FirmwareInfo);
+	ATOM_FIRMWARE_INFO_V2_2 *firmware_info;
+	u8 frev, crev;
+	u16 data_offset;
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		firmware_info =
+			(ATOM_FIRMWARE_INFO_V2_2 *)(mode_info->atom_context->bios +
+						    data_offset);
+		boot_state->mvdd_bootup_value = le16_to_cpu(firmware_info->usBootUpMVDDCVoltage);
+		boot_state->vddc_bootup_value = le16_to_cpu(firmware_info->usBootUpVDDCVoltage);
+		boot_state->vddci_bootup_value = le16_to_cpu(firmware_info->usBootUpVDDCIVoltage);
+		boot_state->pcie_gen_bootup_value = ci_get_current_pcie_speed(adev);
+		boot_state->pcie_lane_bootup_value = ci_get_current_pcie_lane_number(adev);
+		boot_state->sclk_bootup_value = le32_to_cpu(firmware_info->ulDefaultEngineClock);
+		boot_state->mclk_bootup_value = le32_to_cpu(firmware_info->ulDefaultMemoryClock);
+
+		return 0;
+	}
+	return -EINVAL;
+}
+
+static void ci_dpm_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->pm.dpm.num_ps; i++) {
+		kfree(adev->pm.dpm.ps[i].ps_priv);
+	}
+	kfree(adev->pm.dpm.ps);
+	kfree(adev->pm.dpm.priv);
+	kfree(adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries);
+	amdgpu_free_extended_power_table(adev);
+}
+
+/**
+ * ci_dpm_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int ci_dpm_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		if ((adev->pdev->revision == 0x80) ||
+		    (adev->pdev->revision == 0x81) ||
+		    (adev->pdev->device == 0x665f))
+			chip_name = "bonaire_k";
+		else
+			chip_name = "bonaire";
+		break;
+	case CHIP_HAWAII:
+		if (adev->pdev->revision == 0x80)
+			chip_name = "hawaii_k";
+		else
+			chip_name = "hawaii";
+		break;
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+	default: BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_smc.bin", chip_name);
+	err = request_firmware(&adev->pm.fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->pm.fw);
+
+out:
+	if (err) {
+		pr_err("cik_smc: Failed to load firmware \"%s\"\n", fw_name);
+		release_firmware(adev->pm.fw);
+		adev->pm.fw = NULL;
+	}
+	return err;
+}
+
+static int ci_dpm_init(struct amdgpu_device *adev)
+{
+	int index = GetIndexIntoMasterTable(DATA, ASIC_InternalSS_Info);
+	SMU7_Discrete_DpmTable *dpm_table;
+	struct amdgpu_gpio_rec gpio;
+	u16 data_offset, size;
+	u8 frev, crev;
+	struct ci_power_info *pi;
+	int ret;
+
+	pi = kzalloc(sizeof(struct ci_power_info), GFP_KERNEL);
+	if (pi == NULL)
+		return -ENOMEM;
+	adev->pm.dpm.priv = pi;
+
+	pi->sys_pcie_mask =
+		adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_MASK;
+
+	pi->force_pcie_gen = AMDGPU_PCIE_GEN_INVALID;
+
+	pi->pcie_gen_performance.max = AMDGPU_PCIE_GEN1;
+	pi->pcie_gen_performance.min = AMDGPU_PCIE_GEN3;
+	pi->pcie_gen_powersaving.max = AMDGPU_PCIE_GEN1;
+	pi->pcie_gen_powersaving.min = AMDGPU_PCIE_GEN3;
+
+	pi->pcie_lane_performance.max = 0;
+	pi->pcie_lane_performance.min = 16;
+	pi->pcie_lane_powersaving.max = 0;
+	pi->pcie_lane_powersaving.min = 16;
+
+	ret = ci_get_vbios_boot_values(adev, &pi->vbios_boot_state);
+	if (ret) {
+		ci_dpm_fini(adev);
+		return ret;
+	}
+
+	ret = amdgpu_get_platform_caps(adev);
+	if (ret) {
+		ci_dpm_fini(adev);
+		return ret;
+	}
+
+	ret = amdgpu_parse_extended_power_table(adev);
+	if (ret) {
+		ci_dpm_fini(adev);
+		return ret;
+	}
+
+	ret = ci_parse_power_table(adev);
+	if (ret) {
+		ci_dpm_fini(adev);
+		return ret;
+	}
+
+	pi->dll_default_on = false;
+	pi->sram_end = SMC_RAM_END;
+
+	pi->activity_target[0] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[1] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[2] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[3] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[4] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[5] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[6] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[7] = CISLAND_TARGETACTIVITY_DFLT;
+
+	pi->mclk_activity_target = CISLAND_MCLK_TARGETACTIVITY_DFLT;
+
+	pi->sclk_dpm_key_disabled = 0;
+	pi->mclk_dpm_key_disabled = 0;
+	pi->pcie_dpm_key_disabled = 0;
+	pi->thermal_sclk_dpm_enabled = 0;
+
+	if (adev->powerplay.pp_feature & PP_SCLK_DEEP_SLEEP_MASK)
+		pi->caps_sclk_ds = true;
+	else
+		pi->caps_sclk_ds = false;
+
+	pi->mclk_strobe_mode_threshold = 40000;
+	pi->mclk_stutter_mode_threshold = 40000;
+	pi->mclk_edc_enable_threshold = 40000;
+	pi->mclk_edc_wr_enable_threshold = 40000;
+
+	ci_initialize_powertune_defaults(adev);
+
+	pi->caps_fps = false;
+
+	pi->caps_sclk_throttle_low_notification = false;
+
+	pi->caps_uvd_dpm = true;
+	pi->caps_vce_dpm = true;
+
+	ci_get_leakage_voltages(adev);
+	ci_patch_dependency_tables_with_leakage(adev);
+	ci_set_private_data_variables_based_on_pptable(adev);
+
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries =
+		kcalloc(4,
+			sizeof(struct amdgpu_clock_voltage_dependency_entry),
+			GFP_KERNEL);
+	if (!adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries) {
+		ci_dpm_fini(adev);
+		return -ENOMEM;
+	}
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.count = 4;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[0].clk = 0;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[0].v = 0;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[1].clk = 36000;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[1].v = 720;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[2].clk = 54000;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[2].v = 810;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[3].clk = 72000;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[3].v = 900;
+
+	adev->pm.dpm.dyn_state.mclk_sclk_ratio = 4;
+	adev->pm.dpm.dyn_state.sclk_mclk_delta = 15000;
+	adev->pm.dpm.dyn_state.vddc_vddci_delta = 200;
+
+	adev->pm.dpm.dyn_state.valid_sclk_values.count = 0;
+	adev->pm.dpm.dyn_state.valid_sclk_values.values = NULL;
+	adev->pm.dpm.dyn_state.valid_mclk_values.count = 0;
+	adev->pm.dpm.dyn_state.valid_mclk_values.values = NULL;
+
+	if (adev->asic_type == CHIP_HAWAII) {
+		pi->thermal_temp_setting.temperature_low = 94500;
+		pi->thermal_temp_setting.temperature_high = 95000;
+		pi->thermal_temp_setting.temperature_shutdown = 104000;
+	} else {
+		pi->thermal_temp_setting.temperature_low = 99500;
+		pi->thermal_temp_setting.temperature_high = 100000;
+		pi->thermal_temp_setting.temperature_shutdown = 104000;
+	}
+
+	pi->uvd_enabled = false;
+
+	dpm_table = &pi->smc_state_table;
+
+	gpio = amdgpu_atombios_lookup_gpio(adev, VDDC_VRHOT_GPIO_PINID);
+	if (gpio.valid) {
+		dpm_table->VRHotGpio = gpio.shift;
+		adev->pm.dpm.platform_caps |= ATOM_PP_PLATFORM_CAP_REGULATOR_HOT;
+	} else {
+		dpm_table->VRHotGpio = CISLANDS_UNUSED_GPIO_PIN;
+		adev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_REGULATOR_HOT;
+	}
+
+	gpio = amdgpu_atombios_lookup_gpio(adev, PP_AC_DC_SWITCH_GPIO_PINID);
+	if (gpio.valid) {
+		dpm_table->AcDcGpio = gpio.shift;
+		adev->pm.dpm.platform_caps |= ATOM_PP_PLATFORM_CAP_HARDWAREDC;
+	} else {
+		dpm_table->AcDcGpio = CISLANDS_UNUSED_GPIO_PIN;
+		adev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_HARDWAREDC;
+	}
+
+	gpio = amdgpu_atombios_lookup_gpio(adev, VDDC_PCC_GPIO_PINID);
+	if (gpio.valid) {
+		u32 tmp = RREG32_SMC(ixCNB_PWRMGT_CNTL);
+
+		switch (gpio.shift) {
+		case 0:
+			tmp &= ~CNB_PWRMGT_CNTL__GNB_SLOW_MODE_MASK;
+			tmp |= 1 << CNB_PWRMGT_CNTL__GNB_SLOW_MODE__SHIFT;
+			break;
+		case 1:
+			tmp &= ~CNB_PWRMGT_CNTL__GNB_SLOW_MODE_MASK;
+			tmp |= 2 << CNB_PWRMGT_CNTL__GNB_SLOW_MODE__SHIFT;
+			break;
+		case 2:
+			tmp |= CNB_PWRMGT_CNTL__GNB_SLOW_MASK;
+			break;
+		case 3:
+			tmp |= CNB_PWRMGT_CNTL__FORCE_NB_PS1_MASK;
+			break;
+		case 4:
+			tmp |= CNB_PWRMGT_CNTL__DPM_ENABLED_MASK;
+			break;
+		default:
+			DRM_INFO("Invalid PCC GPIO: %u!\n", gpio.shift);
+			break;
+		}
+		WREG32_SMC(ixCNB_PWRMGT_CNTL, tmp);
+	}
+
+	pi->voltage_control = CISLANDS_VOLTAGE_CONTROL_NONE;
+	pi->vddci_control = CISLANDS_VOLTAGE_CONTROL_NONE;
+	pi->mvdd_control = CISLANDS_VOLTAGE_CONTROL_NONE;
+	if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_VDDC, VOLTAGE_OBJ_GPIO_LUT))
+		pi->voltage_control = CISLANDS_VOLTAGE_CONTROL_BY_GPIO;
+	else if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_VDDC, VOLTAGE_OBJ_SVID2))
+		pi->voltage_control = CISLANDS_VOLTAGE_CONTROL_BY_SVID2;
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_VDDCI_CONTROL) {
+		if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_VDDCI, VOLTAGE_OBJ_GPIO_LUT))
+			pi->vddci_control = CISLANDS_VOLTAGE_CONTROL_BY_GPIO;
+		else if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_VDDCI, VOLTAGE_OBJ_SVID2))
+			pi->vddci_control = CISLANDS_VOLTAGE_CONTROL_BY_SVID2;
+		else
+			adev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_VDDCI_CONTROL;
+	}
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_MVDDCONTROL) {
+		if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_MVDDC, VOLTAGE_OBJ_GPIO_LUT))
+			pi->mvdd_control = CISLANDS_VOLTAGE_CONTROL_BY_GPIO;
+		else if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_MVDDC, VOLTAGE_OBJ_SVID2))
+			pi->mvdd_control = CISLANDS_VOLTAGE_CONTROL_BY_SVID2;
+		else
+			adev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_MVDDCONTROL;
+	}
+
+	pi->vddc_phase_shed_control = true;
+
+#if defined(CONFIG_ACPI)
+	pi->pcie_performance_request =
+		amdgpu_acpi_is_pcie_performance_request_supported(adev);
+#else
+	pi->pcie_performance_request = false;
+#endif
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		pi->caps_sclk_ss_support = true;
+		pi->caps_mclk_ss_support = true;
+		pi->dynamic_ss = true;
+	} else {
+		pi->caps_sclk_ss_support = false;
+		pi->caps_mclk_ss_support = false;
+		pi->dynamic_ss = true;
+	}
+
+	if (adev->pm.int_thermal_type != THERMAL_TYPE_NONE)
+		pi->thermal_protection = true;
+	else
+		pi->thermal_protection = false;
+
+	pi->caps_dynamic_ac_timing = true;
+
+	pi->uvd_power_gated = true;
+
+	/* make sure dc limits are valid */
+	if ((adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.sclk == 0) ||
+	    (adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.mclk == 0))
+		adev->pm.dpm.dyn_state.max_clock_voltage_on_dc =
+			adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+
+	pi->fan_ctrl_is_in_default_mode = true;
+
+	return 0;
+}
+
+static void
+ci_dpm_debugfs_print_current_performance_level(void *handle,
+					       struct seq_file *m)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_ps *rps = &pi->current_rps;
+	u32 sclk = ci_get_average_sclk_freq(adev);
+	u32 mclk = ci_get_average_mclk_freq(adev);
+	u32 activity_percent = 50;
+	int ret;
+
+	ret = ci_read_smc_soft_register(adev, offsetof(SMU7_SoftRegisters, AverageGraphicsA),
+					&activity_percent);
+
+	if (ret == 0) {
+		activity_percent += 0x80;
+		activity_percent >>= 8;
+		activity_percent = activity_percent > 100 ? 100 : activity_percent;
+	}
+
+	seq_printf(m, "uvd %sabled\n", pi->uvd_power_gated ? "dis" : "en");
+	seq_printf(m, "vce %sabled\n", rps->vce_active ? "en" : "dis");
+	seq_printf(m, "power level avg    sclk: %u mclk: %u\n",
+		   sclk, mclk);
+	seq_printf(m, "GPU load: %u %%\n", activity_percent);
+}
+
+static void ci_dpm_print_power_state(void *handle, void *current_ps)
+{
+	struct amdgpu_ps *rps = (struct amdgpu_ps *)current_ps;
+	struct ci_ps *ps = ci_get_ps(rps);
+	struct ci_pl *pl;
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_dpm_print_class_info(rps->class, rps->class2);
+	amdgpu_dpm_print_cap_info(rps->caps);
+	printk("\tuvd    vclk: %d dclk: %d\n", rps->vclk, rps->dclk);
+	for (i = 0; i < ps->performance_level_count; i++) {
+		pl = &ps->performance_levels[i];
+		printk("\t\tpower level %d    sclk: %u mclk: %u pcie gen: %u pcie lanes: %u\n",
+		       i, pl->sclk, pl->mclk, pl->pcie_gen + 1, pl->pcie_lane);
+	}
+	amdgpu_dpm_print_ps_status(adev, rps);
+}
+
+static inline bool ci_are_power_levels_equal(const struct ci_pl *ci_cpl1,
+						const struct ci_pl *ci_cpl2)
+{
+	return ((ci_cpl1->mclk == ci_cpl2->mclk) &&
+		  (ci_cpl1->sclk == ci_cpl2->sclk) &&
+		  (ci_cpl1->pcie_gen == ci_cpl2->pcie_gen) &&
+		  (ci_cpl1->pcie_lane == ci_cpl2->pcie_lane));
+}
+
+static int ci_check_state_equal(void *handle,
+				void *current_ps,
+				void *request_ps,
+				bool *equal)
+{
+	struct ci_ps *ci_cps;
+	struct ci_ps *ci_rps;
+	int i;
+	struct amdgpu_ps *cps = (struct amdgpu_ps *)current_ps;
+	struct amdgpu_ps *rps = (struct amdgpu_ps *)request_ps;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev == NULL || cps == NULL || rps == NULL || equal == NULL)
+		return -EINVAL;
+
+	ci_cps = ci_get_ps((struct amdgpu_ps *)cps);
+	ci_rps = ci_get_ps((struct amdgpu_ps *)rps);
+
+	if (ci_cps == NULL) {
+		*equal = false;
+		return 0;
+	}
+
+	if (ci_cps->performance_level_count != ci_rps->performance_level_count) {
+
+		*equal = false;
+		return 0;
+	}
+
+	for (i = 0; i < ci_cps->performance_level_count; i++) {
+		if (!ci_are_power_levels_equal(&(ci_cps->performance_levels[i]),
+					&(ci_rps->performance_levels[i]))) {
+			*equal = false;
+			return 0;
+		}
+	}
+
+	/* If all performance levels are the same try to use the UVD clocks to break the tie.*/
+	*equal = ((cps->vclk == rps->vclk) && (cps->dclk == rps->dclk));
+	*equal &= ((cps->evclk == rps->evclk) && (cps->ecclk == rps->ecclk));
+
+	return 0;
+}
+
+static u32 ci_dpm_get_sclk(void *handle, bool low)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *requested_state = ci_get_ps(&pi->requested_rps);
+
+	if (low)
+		return requested_state->performance_levels[0].sclk;
+	else
+		return requested_state->performance_levels[requested_state->performance_level_count - 1].sclk;
+}
+
+static u32 ci_dpm_get_mclk(void *handle, bool low)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *requested_state = ci_get_ps(&pi->requested_rps);
+
+	if (low)
+		return requested_state->performance_levels[0].mclk;
+	else
+		return requested_state->performance_levels[requested_state->performance_level_count - 1].mclk;
+}
+
+/* get temperature in millidegrees */
+static int ci_dpm_get_temp(void *handle)
+{
+	u32 temp;
+	int actual_temp = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	temp = (RREG32_SMC(ixCG_MULT_THERMAL_STATUS) & CG_MULT_THERMAL_STATUS__CTF_TEMP_MASK) >>
+		CG_MULT_THERMAL_STATUS__CTF_TEMP__SHIFT;
+
+	if (temp & 0x200)
+		actual_temp = 255;
+	else
+		actual_temp = temp & 0x1ff;
+
+	actual_temp = actual_temp * 1000;
+
+	return actual_temp;
+}
+
+static int ci_set_temperature_range(struct amdgpu_device *adev)
+{
+	int ret;
+
+	ret = ci_thermal_enable_alert(adev, false);
+	if (ret)
+		return ret;
+	ret = ci_thermal_set_temperature_range(adev, CISLANDS_TEMP_RANGE_MIN,
+					       CISLANDS_TEMP_RANGE_MAX);
+	if (ret)
+		return ret;
+	ret = ci_thermal_enable_alert(adev, true);
+	if (ret)
+		return ret;
+	return ret;
+}
+
+static int ci_dpm_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->powerplay.pp_funcs = &ci_dpm_funcs;
+	adev->powerplay.pp_handle = adev;
+	ci_dpm_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int ci_dpm_late_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->pm.dpm_enabled)
+		return 0;
+
+	/* init the sysfs and debugfs files late */
+	ret = amdgpu_pm_sysfs_init(adev);
+	if (ret)
+		return ret;
+
+	ret = ci_set_temperature_range(adev);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int ci_dpm_sw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 230,
+				&adev->pm.dpm.thermal.irq);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 231,
+				&adev->pm.dpm.thermal.irq);
+	if (ret)
+		return ret;
+
+	/* default to balanced state */
+	adev->pm.dpm.state = POWER_STATE_TYPE_BALANCED;
+	adev->pm.dpm.user_state = POWER_STATE_TYPE_BALANCED;
+	adev->pm.dpm.forced_level = AMD_DPM_FORCED_LEVEL_AUTO;
+	adev->pm.default_sclk = adev->clock.default_sclk;
+	adev->pm.default_mclk = adev->clock.default_mclk;
+	adev->pm.current_sclk = adev->clock.default_sclk;
+	adev->pm.current_mclk = adev->clock.default_mclk;
+	adev->pm.int_thermal_type = THERMAL_TYPE_NONE;
+
+	ret = ci_dpm_init_microcode(adev);
+	if (ret)
+		return ret;
+
+	if (amdgpu_dpm == 0)
+		return 0;
+
+	INIT_WORK(&adev->pm.dpm.thermal.work, amdgpu_dpm_thermal_work_handler);
+	mutex_lock(&adev->pm.mutex);
+	ret = ci_dpm_init(adev);
+	if (ret)
+		goto dpm_failed;
+	adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
+	if (amdgpu_dpm == 1)
+		amdgpu_pm_print_power_states(adev);
+	mutex_unlock(&adev->pm.mutex);
+	DRM_INFO("amdgpu: dpm initialized\n");
+
+	return 0;
+
+dpm_failed:
+	ci_dpm_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+	DRM_ERROR("amdgpu: dpm initialization failed\n");
+	return ret;
+}
+
+static int ci_dpm_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	flush_work(&adev->pm.dpm.thermal.work);
+
+	mutex_lock(&adev->pm.mutex);
+	ci_dpm_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+
+	release_firmware(adev->pm.fw);
+	adev->pm.fw = NULL;
+
+	return 0;
+}
+
+static int ci_dpm_hw_init(void *handle)
+{
+	int ret;
+
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!amdgpu_dpm) {
+		ret = ci_upload_firmware(adev);
+		if (ret) {
+			DRM_ERROR("ci_upload_firmware failed\n");
+			return ret;
+		}
+		ci_dpm_start_smc(adev);
+		return 0;
+	}
+
+	mutex_lock(&adev->pm.mutex);
+	ci_dpm_setup_asic(adev);
+	ret = ci_dpm_enable(adev);
+	if (ret)
+		adev->pm.dpm_enabled = false;
+	else
+		adev->pm.dpm_enabled = true;
+	mutex_unlock(&adev->pm.mutex);
+
+	return ret;
+}
+
+static int ci_dpm_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		ci_dpm_disable(adev);
+		mutex_unlock(&adev->pm.mutex);
+	} else {
+		ci_dpm_stop_smc(adev);
+	}
+
+	return 0;
+}
+
+static int ci_dpm_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
+			       AMDGPU_THERMAL_IRQ_LOW_TO_HIGH);
+		amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
+			       AMDGPU_THERMAL_IRQ_HIGH_TO_LOW);
+		adev->pm.dpm.last_user_state = adev->pm.dpm.user_state;
+		adev->pm.dpm.last_state = adev->pm.dpm.state;
+		adev->pm.dpm.user_state = POWER_STATE_TYPE_INTERNAL_BOOT;
+		adev->pm.dpm.state = POWER_STATE_TYPE_INTERNAL_BOOT;
+		mutex_unlock(&adev->pm.mutex);
+		amdgpu_pm_compute_clocks(adev);
+
+	}
+
+	return 0;
+}
+
+static int ci_dpm_resume(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		/* asic init will reset to the boot state */
+		mutex_lock(&adev->pm.mutex);
+		ci_dpm_setup_asic(adev);
+		ret = ci_dpm_enable(adev);
+		if (ret)
+			adev->pm.dpm_enabled = false;
+		else
+			adev->pm.dpm_enabled = true;
+		adev->pm.dpm.user_state = adev->pm.dpm.last_user_state;
+		adev->pm.dpm.state = adev->pm.dpm.last_state;
+		mutex_unlock(&adev->pm.mutex);
+		if (adev->pm.dpm_enabled)
+			amdgpu_pm_compute_clocks(adev);
+	}
+	return 0;
+}
+
+static bool ci_dpm_is_idle(void *handle)
+{
+	/* XXX */
+	return true;
+}
+
+static int ci_dpm_wait_for_idle(void *handle)
+{
+	/* XXX */
+	return 0;
+}
+
+static int ci_dpm_soft_reset(void *handle)
+{
+	return 0;
+}
+
+static int ci_dpm_set_interrupt_state(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      unsigned type,
+				      enum amdgpu_interrupt_state state)
+{
+	u32 cg_thermal_int;
+
+	switch (type) {
+	case AMDGPU_THERMAL_IRQ_LOW_TO_HIGH:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
+			cg_thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT, cg_thermal_int);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
+			cg_thermal_int &= ~CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT, cg_thermal_int);
+			break;
+		default:
+			break;
+		}
+		break;
+
+	case AMDGPU_THERMAL_IRQ_HIGH_TO_LOW:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
+			cg_thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT, cg_thermal_int);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
+			cg_thermal_int &= ~CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT, cg_thermal_int);
+			break;
+		default:
+			break;
+		}
+		break;
+
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int ci_dpm_process_interrupt(struct amdgpu_device *adev,
+				    struct amdgpu_irq_src *source,
+				    struct amdgpu_iv_entry *entry)
+{
+	bool queue_thermal = false;
+
+	if (entry == NULL)
+		return -EINVAL;
+
+	switch (entry->src_id) {
+	case 230: /* thermal low to high */
+		DRM_DEBUG("IH: thermal low to high\n");
+		adev->pm.dpm.thermal.high_to_low = false;
+		queue_thermal = true;
+		break;
+	case 231: /* thermal high to low */
+		DRM_DEBUG("IH: thermal high to low\n");
+		adev->pm.dpm.thermal.high_to_low = true;
+		queue_thermal = true;
+		break;
+	default:
+		break;
+	}
+
+	if (queue_thermal)
+		schedule_work(&adev->pm.dpm.thermal.work);
+
+	return 0;
+}
+
+static int ci_dpm_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int ci_dpm_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static int ci_dpm_print_clock_levels(void *handle,
+		enum pp_clock_type type, char *buf)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_single_dpm_table *sclk_table = &pi->dpm_table.sclk_table;
+	struct ci_single_dpm_table *mclk_table = &pi->dpm_table.mclk_table;
+	struct ci_single_dpm_table *pcie_table = &pi->dpm_table.pcie_speed_table;
+
+	int i, now, size = 0;
+	uint32_t clock, pcie_speed;
+
+	switch (type) {
+	case PP_SCLK:
+		amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_API_GetSclkFrequency);
+		clock = RREG32(mmSMC_MSG_ARG_0);
+
+		for (i = 0; i < sclk_table->count; i++) {
+			if (clock > sclk_table->dpm_levels[i].value)
+				continue;
+			break;
+		}
+		now = i;
+
+		for (i = 0; i < sclk_table->count; i++)
+			size += snprintf(buf + size, PAGE_SIZE - size, "%d: %uMhz %s\n",
+					i, sclk_table->dpm_levels[i].value / 100,
+					(i == now) ? "*" : "");
+		break;
+	case PP_MCLK:
+		amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_API_GetMclkFrequency);
+		clock = RREG32(mmSMC_MSG_ARG_0);
+
+		for (i = 0; i < mclk_table->count; i++) {
+			if (clock > mclk_table->dpm_levels[i].value)
+				continue;
+			break;
+		}
+		now = i;
+
+		for (i = 0; i < mclk_table->count; i++)
+			size += snprintf(buf + size, PAGE_SIZE - size, "%d: %uMhz %s\n",
+					i, mclk_table->dpm_levels[i].value / 100,
+					(i == now) ? "*" : "");
+		break;
+	case PP_PCIE:
+		pcie_speed = ci_get_current_pcie_speed(adev);
+		for (i = 0; i < pcie_table->count; i++) {
+			if (pcie_speed != pcie_table->dpm_levels[i].value)
+				continue;
+			break;
+		}
+		now = i;
+
+		for (i = 0; i < pcie_table->count; i++)
+			size += snprintf(buf + size, PAGE_SIZE - size, "%d: %s %s\n", i,
+					(pcie_table->dpm_levels[i].value == 0) ? "2.5GT/s, x1" :
+					(pcie_table->dpm_levels[i].value == 1) ? "5.0GT/s, x16" :
+					(pcie_table->dpm_levels[i].value == 2) ? "8.0GT/s, x16" : "",
+					(i == now) ? "*" : "");
+		break;
+	default:
+		break;
+	}
+
+	return size;
+}
+
+static int ci_dpm_force_clock_level(void *handle,
+		enum pp_clock_type type, uint32_t mask)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (adev->pm.dpm.forced_level != AMD_DPM_FORCED_LEVEL_MANUAL)
+		return -EINVAL;
+
+	if (mask == 0)
+		return -EINVAL;
+
+	switch (type) {
+	case PP_SCLK:
+		if (!pi->sclk_dpm_key_disabled)
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+					PPSMC_MSG_SCLKDPM_SetEnabledMask,
+					pi->dpm_level_enable_mask.sclk_dpm_enable_mask & mask);
+		break;
+
+	case PP_MCLK:
+		if (!pi->mclk_dpm_key_disabled)
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+					PPSMC_MSG_MCLKDPM_SetEnabledMask,
+					pi->dpm_level_enable_mask.mclk_dpm_enable_mask & mask);
+		break;
+
+	case PP_PCIE:
+	{
+		uint32_t tmp = mask & pi->dpm_level_enable_mask.pcie_dpm_enable_mask;
+
+		if (!pi->pcie_dpm_key_disabled) {
+			if (fls(tmp) != ffs(tmp))
+				amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PCIeDPM_UnForceLevel);
+			else
+				amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+					PPSMC_MSG_PCIeDPM_ForceLevel,
+					fls(tmp) - 1);
+		}
+		break;
+	}
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int ci_dpm_get_sclk_od(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_single_dpm_table *sclk_table = &(pi->dpm_table.sclk_table);
+	struct ci_single_dpm_table *golden_sclk_table =
+			&(pi->golden_dpm_table.sclk_table);
+	int value;
+
+	value = (sclk_table->dpm_levels[sclk_table->count - 1].value -
+			golden_sclk_table->dpm_levels[golden_sclk_table->count - 1].value) *
+			100 /
+			golden_sclk_table->dpm_levels[golden_sclk_table->count - 1].value;
+
+	return value;
+}
+
+static int ci_dpm_set_sclk_od(void *handle, uint32_t value)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *ps = ci_get_ps(adev->pm.dpm.requested_ps);
+	struct ci_single_dpm_table *golden_sclk_table =
+			&(pi->golden_dpm_table.sclk_table);
+
+	if (value > 20)
+		value = 20;
+
+	ps->performance_levels[ps->performance_level_count - 1].sclk =
+			golden_sclk_table->dpm_levels[golden_sclk_table->count - 1].value *
+			value / 100 +
+			golden_sclk_table->dpm_levels[golden_sclk_table->count - 1].value;
+
+	return 0;
+}
+
+static int ci_dpm_get_mclk_od(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_single_dpm_table *mclk_table = &(pi->dpm_table.mclk_table);
+	struct ci_single_dpm_table *golden_mclk_table =
+			&(pi->golden_dpm_table.mclk_table);
+	int value;
+
+	value = (mclk_table->dpm_levels[mclk_table->count - 1].value -
+			golden_mclk_table->dpm_levels[golden_mclk_table->count - 1].value) *
+			100 /
+			golden_mclk_table->dpm_levels[golden_mclk_table->count - 1].value;
+
+	return value;
+}
+
+static int ci_dpm_set_mclk_od(void *handle, uint32_t value)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *ps = ci_get_ps(adev->pm.dpm.requested_ps);
+	struct ci_single_dpm_table *golden_mclk_table =
+			&(pi->golden_dpm_table.mclk_table);
+
+	if (value > 20)
+		value = 20;
+
+	ps->performance_levels[ps->performance_level_count - 1].mclk =
+			golden_mclk_table->dpm_levels[golden_mclk_table->count - 1].value *
+			value / 100 +
+			golden_mclk_table->dpm_levels[golden_mclk_table->count - 1].value;
+
+	return 0;
+}
+
+static int ci_dpm_read_sensor(void *handle, int idx,
+			      void *value, int *size)
+{
+	u32 activity_percent = 50;
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* size must be at least 4 bytes for all sensors */
+	if (*size < 4)
+		return -EINVAL;
+
+	switch (idx) {
+	case AMDGPU_PP_SENSOR_GFX_SCLK:
+		*((uint32_t *)value) = ci_get_average_sclk_freq(adev);
+		*size = 4;
+		return 0;
+	case AMDGPU_PP_SENSOR_GFX_MCLK:
+		*((uint32_t *)value) = ci_get_average_mclk_freq(adev);
+		*size = 4;
+		return 0;
+	case AMDGPU_PP_SENSOR_GPU_TEMP:
+		*((uint32_t *)value) = ci_dpm_get_temp(adev);
+		*size = 4;
+		return 0;
+	case AMDGPU_PP_SENSOR_GPU_LOAD:
+		ret = ci_read_smc_soft_register(adev,
+						offsetof(SMU7_SoftRegisters,
+							 AverageGraphicsA),
+						&activity_percent);
+		if (ret == 0) {
+			activity_percent += 0x80;
+			activity_percent >>= 8;
+			activity_percent =
+				activity_percent > 100 ? 100 : activity_percent;
+		}
+		*((uint32_t *)value) = activity_percent;
+		*size = 4;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
+static int ci_set_powergating_by_smu(void *handle,
+				uint32_t block_type, bool gate)
+{
+	switch (block_type) {
+	case AMD_IP_BLOCK_TYPE_UVD:
+		ci_dpm_powergate_uvd(handle, gate);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static const struct amd_ip_funcs ci_dpm_ip_funcs = {
+	.name = "ci_dpm",
+	.early_init = ci_dpm_early_init,
+	.late_init = ci_dpm_late_init,
+	.sw_init = ci_dpm_sw_init,
+	.sw_fini = ci_dpm_sw_fini,
+	.hw_init = ci_dpm_hw_init,
+	.hw_fini = ci_dpm_hw_fini,
+	.suspend = ci_dpm_suspend,
+	.resume = ci_dpm_resume,
+	.is_idle = ci_dpm_is_idle,
+	.wait_for_idle = ci_dpm_wait_for_idle,
+	.soft_reset = ci_dpm_soft_reset,
+	.set_clockgating_state = ci_dpm_set_clockgating_state,
+	.set_powergating_state = ci_dpm_set_powergating_state,
+};
+
+const struct amdgpu_ip_block_version ci_smu_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_SMC,
+	.major = 7,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &ci_dpm_ip_funcs,
+};
+
+static const struct amd_pm_funcs ci_dpm_funcs = {
+	.pre_set_power_state = &ci_dpm_pre_set_power_state,
+	.set_power_state = &ci_dpm_set_power_state,
+	.post_set_power_state = &ci_dpm_post_set_power_state,
+	.display_configuration_changed = &ci_dpm_display_configuration_changed,
+	.get_sclk = &ci_dpm_get_sclk,
+	.get_mclk = &ci_dpm_get_mclk,
+	.print_power_state = &ci_dpm_print_power_state,
+	.debugfs_print_current_performance_level = &ci_dpm_debugfs_print_current_performance_level,
+	.force_performance_level = &ci_dpm_force_performance_level,
+	.vblank_too_short = &ci_dpm_vblank_too_short,
+	.set_powergating_by_smu = &ci_set_powergating_by_smu,
+	.set_fan_control_mode = &ci_dpm_set_fan_control_mode,
+	.get_fan_control_mode = &ci_dpm_get_fan_control_mode,
+	.set_fan_speed_percent = &ci_dpm_set_fan_speed_percent,
+	.get_fan_speed_percent = &ci_dpm_get_fan_speed_percent,
+	.print_clock_levels = ci_dpm_print_clock_levels,
+	.force_clock_level = ci_dpm_force_clock_level,
+	.get_sclk_od = ci_dpm_get_sclk_od,
+	.set_sclk_od = ci_dpm_set_sclk_od,
+	.get_mclk_od = ci_dpm_get_mclk_od,
+	.set_mclk_od = ci_dpm_set_mclk_od,
+	.check_state_equal = ci_check_state_equal,
+	.get_vce_clock_state = amdgpu_get_vce_clock_state,
+	.read_sensor = ci_dpm_read_sensor,
+};
+
+static const struct amdgpu_irq_src_funcs ci_dpm_irq_funcs = {
+	.set = ci_dpm_set_interrupt_state,
+	.process = ci_dpm_process_interrupt,
+};
+
+static void ci_dpm_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->pm.dpm.thermal.irq.num_types = AMDGPU_THERMAL_IRQ_LAST;
+	adev->pm.dpm.thermal.irq.funcs = &ci_dpm_irq_funcs;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ci_smc.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ci_smc.c
new file mode 100644
index 00000000000..b8ba51e045b
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ci_smc.c
@@ -0,0 +1,279 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "cikd.h"
+#include "ppsmc.h"
+#include "amdgpu_ucode.h"
+#include "ci_dpm.h"
+
+#include "smu/smu_7_0_1_d.h"
+#include "smu/smu_7_0_1_sh_mask.h"
+
+static int ci_set_smc_sram_address(struct amdgpu_device *adev,
+				   u32 smc_address, u32 limit)
+{
+	if (smc_address & 3)
+		return -EINVAL;
+	if ((smc_address + 3) > limit)
+		return -EINVAL;
+
+	WREG32(mmSMC_IND_INDEX_0, smc_address);
+	WREG32_P(mmSMC_IND_ACCESS_CNTL, 0, ~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
+
+	return 0;
+}
+
+int amdgpu_ci_copy_bytes_to_smc(struct amdgpu_device *adev,
+			 u32 smc_start_address,
+			 const u8 *src, u32 byte_count, u32 limit)
+{
+	unsigned long flags;
+	u32 data, original_data;
+	u32 addr;
+	u32 extra_shift;
+	int ret = 0;
+
+	if (smc_start_address & 3)
+		return -EINVAL;
+	if ((smc_start_address + byte_count) > limit)
+		return -EINVAL;
+
+	addr = smc_start_address;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	while (byte_count >= 4) {
+		/* SMC address space is BE */
+		data = (src[0] << 24) | (src[1] << 16) | (src[2] << 8) | src[3];
+
+		ret = ci_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			goto done;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+
+		src += 4;
+		byte_count -= 4;
+		addr += 4;
+	}
+
+	/* RMW for the final bytes */
+	if (byte_count > 0) {
+		data = 0;
+
+		ret = ci_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			goto done;
+
+		original_data = RREG32(mmSMC_IND_DATA_0);
+
+		extra_shift = 8 * (4 - byte_count);
+
+		while (byte_count > 0) {
+			data = (data << 8) + *src++;
+			byte_count--;
+		}
+
+		data <<= extra_shift;
+
+		data |= (original_data & ~((~0UL) << extra_shift));
+
+		ret = ci_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			goto done;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+	}
+
+done:
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return ret;
+}
+
+void amdgpu_ci_start_smc(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+
+	tmp &= ~SMC_SYSCON_RESET_CNTL__rst_reg_MASK;
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, tmp);
+}
+
+void amdgpu_ci_reset_smc(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+
+	tmp |= SMC_SYSCON_RESET_CNTL__rst_reg_MASK;
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, tmp);
+}
+
+int amdgpu_ci_program_jump_on_start(struct amdgpu_device *adev)
+{
+	static u8 data[] = { 0xE0, 0x00, 0x80, 0x40 };
+
+	return amdgpu_ci_copy_bytes_to_smc(adev, 0x0, data, 4, sizeof(data)+1);
+}
+
+void amdgpu_ci_stop_smc_clock(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+
+	tmp |= SMC_SYSCON_CLOCK_CNTL_0__ck_disable_MASK;
+
+	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, tmp);
+}
+
+void amdgpu_ci_start_smc_clock(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+
+	tmp &= ~SMC_SYSCON_CLOCK_CNTL_0__ck_disable_MASK;
+
+	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, tmp);
+}
+
+bool amdgpu_ci_is_smc_running(struct amdgpu_device *adev)
+{
+	u32 clk = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	u32 pc_c = RREG32_SMC(ixSMC_PC_C);
+
+	if (!(clk & SMC_SYSCON_CLOCK_CNTL_0__ck_disable_MASK) && (0x20100 <= pc_c))
+		return true;
+
+	return false;
+}
+
+PPSMC_Result amdgpu_ci_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg)
+{
+	u32 tmp;
+	int i;
+
+	if (!amdgpu_ci_is_smc_running(adev))
+		return PPSMC_Result_Failed;
+
+	WREG32(mmSMC_MESSAGE_0, msg);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmSMC_RESP_0);
+		if (tmp != 0)
+			break;
+		udelay(1);
+	}
+	tmp = RREG32(mmSMC_RESP_0);
+
+	return (PPSMC_Result)tmp;
+}
+
+PPSMC_Result amdgpu_ci_wait_for_smc_inactive(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	int i;
+
+	if (!amdgpu_ci_is_smc_running(adev))
+		return PPSMC_Result_OK;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+		if ((tmp & SMC_SYSCON_CLOCK_CNTL_0__cken_MASK) == 0)
+			break;
+		udelay(1);
+	}
+
+	return PPSMC_Result_OK;
+}
+
+int amdgpu_ci_load_smc_ucode(struct amdgpu_device *adev, u32 limit)
+{
+	const struct smc_firmware_header_v1_0 *hdr;
+	unsigned long flags;
+	u32 ucode_start_address;
+	u32 ucode_size;
+	const u8 *src;
+	u32 data;
+
+	if (!adev->pm.fw)
+		return -EINVAL;
+
+	hdr = (const struct smc_firmware_header_v1_0 *)adev->pm.fw->data;
+	amdgpu_ucode_print_smc_hdr(&hdr->header);
+
+	adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	ucode_start_address = le32_to_cpu(hdr->ucode_start_addr);
+	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
+	src = (const u8 *)
+		(adev->pm.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	if (ucode_size & 3)
+		return -EINVAL;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmSMC_IND_INDEX_0, ucode_start_address);
+	WREG32_P(mmSMC_IND_ACCESS_CNTL, SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK,
+		~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
+	while (ucode_size >= 4) {
+		/* SMC address space is BE */
+		data = (src[0] << 24) | (src[1] << 16) | (src[2] << 8) | src[3];
+
+		WREG32(mmSMC_IND_DATA_0, data);
+
+		src += 4;
+		ucode_size -= 4;
+	}
+	WREG32_P(mmSMC_IND_ACCESS_CNTL, 0, ~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return 0;
+}
+
+int amdgpu_ci_read_smc_sram_dword(struct amdgpu_device *adev,
+			   u32 smc_address, u32 *value, u32 limit)
+{
+	unsigned long flags;
+	int ret;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	ret = ci_set_smc_sram_address(adev, smc_address, limit);
+	if (ret == 0)
+		*value = RREG32(mmSMC_IND_DATA_0);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return ret;
+}
+
+int amdgpu_ci_write_smc_sram_dword(struct amdgpu_device *adev,
+			    u32 smc_address, u32 value, u32 limit)
+{
+	unsigned long flags;
+	int ret;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	ret = ci_set_smc_sram_address(adev, smc_address, limit);
+	if (ret == 0)
+		WREG32(mmSMC_IND_DATA_0, value);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return ret;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_cik.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_cik.c
new file mode 100644
index 00000000000..e9ffcaf5c2a
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_cik.c
@@ -0,0 +1,2085 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+#include <linux/firmware.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+#include "cikd.h"
+#include "atom.h"
+#include "amd_pcie.h"
+
+#include "cik.h"
+#include "gmc_v7_0.h"
+#include "cik_ih.h"
+#include "dce_v8_0.h"
+#include "gfx_v7_0.h"
+#include "cik_sdma.h"
+#include "uvd_v4_2.h"
+#include "vce_v2_0.h"
+#include "cik_dpm.h"
+
+#include "uvd/uvd_4_2_d.h"
+
+#include "smu/smu_7_0_1_d.h"
+#include "smu/smu_7_0_1_sh_mask.h"
+
+#include "dce/dce_8_0_d.h"
+#include "dce/dce_8_0_sh_mask.h"
+
+#include "bif/bif_4_1_d.h"
+#include "bif/bif_4_1_sh_mask.h"
+
+#include "gca/gfx_7_2_d.h"
+#include "gca/gfx_7_2_enum.h"
+#include "gca/gfx_7_2_sh_mask.h"
+
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+#include "amdgpu_dm.h"
+#include "amdgpu_amdkfd.h"
+#include "dce_virtual.h"
+
+/*
+ * Indirect registers accessor
+ */
+static u32 cik_pcie_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(mmPCIE_INDEX, reg);
+	(void)RREG32(mmPCIE_INDEX);
+	r = RREG32(mmPCIE_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+	return r;
+}
+
+static void cik_pcie_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(mmPCIE_INDEX, reg);
+	(void)RREG32(mmPCIE_INDEX);
+	WREG32(mmPCIE_DATA, v);
+	(void)RREG32(mmPCIE_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+static u32 cik_smc_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmSMC_IND_INDEX_0, (reg));
+	r = RREG32(mmSMC_IND_DATA_0);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return r;
+}
+
+static void cik_smc_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmSMC_IND_INDEX_0, (reg));
+	WREG32(mmSMC_IND_DATA_0, (v));
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+}
+
+static u32 cik_uvd_ctx_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->uvd_ctx_idx_lock, flags);
+	WREG32(mmUVD_CTX_INDEX, ((reg) & 0x1ff));
+	r = RREG32(mmUVD_CTX_DATA);
+	spin_unlock_irqrestore(&adev->uvd_ctx_idx_lock, flags);
+	return r;
+}
+
+static void cik_uvd_ctx_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->uvd_ctx_idx_lock, flags);
+	WREG32(mmUVD_CTX_INDEX, ((reg) & 0x1ff));
+	WREG32(mmUVD_CTX_DATA, (v));
+	spin_unlock_irqrestore(&adev->uvd_ctx_idx_lock, flags);
+}
+
+static u32 cik_didt_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(mmDIDT_IND_INDEX, (reg));
+	r = RREG32(mmDIDT_IND_DATA);
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+	return r;
+}
+
+static void cik_didt_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(mmDIDT_IND_INDEX, (reg));
+	WREG32(mmDIDT_IND_DATA, (v));
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+}
+
+static const u32 bonaire_golden_spm_registers[] =
+{
+	0xc200, 0xe0ffffff, 0xe0000000
+};
+
+static const u32 bonaire_golden_common_registers[] =
+{
+	0x31dc, 0xffffffff, 0x00000800,
+	0x31dd, 0xffffffff, 0x00000800,
+	0x31e6, 0xffffffff, 0x00007fbf,
+	0x31e7, 0xffffffff, 0x00007faf
+};
+
+static const u32 bonaire_golden_registers[] =
+{
+	0xcd5, 0x00000333, 0x00000333,
+	0xcd4, 0x000c0fc0, 0x00040200,
+	0x2684, 0x00010000, 0x00058208,
+	0xf000, 0xffff1fff, 0x00140000,
+	0xf080, 0xfdfc0fff, 0x00000100,
+	0xf08d, 0x40000000, 0x40000200,
+	0x260c, 0xffffffff, 0x00000000,
+	0x260d, 0xf00fffff, 0x00000400,
+	0x260e, 0x0002021c, 0x00020200,
+	0x31e, 0x00000080, 0x00000000,
+	0x16ec, 0x000000f0, 0x00000070,
+	0x16f0, 0xf0311fff, 0x80300000,
+	0x263e, 0x73773777, 0x12010001,
+	0xd43, 0x00810000, 0x408af000,
+	0x1c0c, 0x31000111, 0x00000011,
+	0xbd2, 0x73773777, 0x12010001,
+	0x883, 0x00007fb6, 0x0021a1b1,
+	0x884, 0x00007fb6, 0x002021b1,
+	0x860, 0x00007fb6, 0x00002191,
+	0x886, 0x00007fb6, 0x002121b1,
+	0x887, 0x00007fb6, 0x002021b1,
+	0x877, 0x00007fb6, 0x00002191,
+	0x878, 0x00007fb6, 0x00002191,
+	0xd8a, 0x0000003f, 0x0000000a,
+	0xd8b, 0x0000003f, 0x0000000a,
+	0xab9, 0x00073ffe, 0x000022a2,
+	0x903, 0x000007ff, 0x00000000,
+	0x2285, 0xf000003f, 0x00000007,
+	0x22fc, 0x00002001, 0x00000001,
+	0x22c9, 0xffffffff, 0x00ffffff,
+	0xc281, 0x0000ff0f, 0x00000000,
+	0xa293, 0x07ffffff, 0x06000000,
+	0x136, 0x00000fff, 0x00000100,
+	0xf9e, 0x00000001, 0x00000002,
+	0x2440, 0x03000000, 0x0362c688,
+	0x2300, 0x000000ff, 0x00000001,
+	0x390, 0x00001fff, 0x00001fff,
+	0x2418, 0x0000007f, 0x00000020,
+	0x2542, 0x00010000, 0x00010000,
+	0x2b05, 0x000003ff, 0x000000f3,
+	0x2b03, 0xffffffff, 0x00001032
+};
+
+static const u32 bonaire_mgcg_cgcg_init[] =
+{
+	0x3108, 0xffffffff, 0xfffffffc,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf0a8, 0xffffffff, 0x00000100,
+	0xf082, 0xffffffff, 0x00000100,
+	0xf0b0, 0xffffffff, 0xc0000100,
+	0xf0b2, 0xffffffff, 0xc0000100,
+	0xf0b1, 0xffffffff, 0xc0000100,
+	0x1579, 0xffffffff, 0x00600100,
+	0xf0a0, 0xffffffff, 0x00000100,
+	0xf085, 0xffffffff, 0x06000100,
+	0xf088, 0xffffffff, 0x00000100,
+	0xf086, 0xffffffff, 0x06000100,
+	0xf081, 0xffffffff, 0x00000100,
+	0xf0b8, 0xffffffff, 0x00000100,
+	0xf089, 0xffffffff, 0x00000100,
+	0xf080, 0xffffffff, 0x00000100,
+	0xf08c, 0xffffffff, 0x00000100,
+	0xf08d, 0xffffffff, 0x00000100,
+	0xf094, 0xffffffff, 0x00000100,
+	0xf095, 0xffffffff, 0x00000100,
+	0xf096, 0xffffffff, 0x00000100,
+	0xf097, 0xffffffff, 0x00000100,
+	0xf098, 0xffffffff, 0x00000100,
+	0xf09f, 0xffffffff, 0x00000100,
+	0xf09e, 0xffffffff, 0x00000100,
+	0xf084, 0xffffffff, 0x06000100,
+	0xf0a4, 0xffffffff, 0x00000100,
+	0xf09d, 0xffffffff, 0x00000100,
+	0xf0ad, 0xffffffff, 0x00000100,
+	0xf0ac, 0xffffffff, 0x00000100,
+	0xf09c, 0xffffffff, 0x00000100,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf008, 0xffffffff, 0x00010000,
+	0xf009, 0xffffffff, 0x00030002,
+	0xf00a, 0xffffffff, 0x00040007,
+	0xf00b, 0xffffffff, 0x00060005,
+	0xf00c, 0xffffffff, 0x00090008,
+	0xf00d, 0xffffffff, 0x00010000,
+	0xf00e, 0xffffffff, 0x00030002,
+	0xf00f, 0xffffffff, 0x00040007,
+	0xf010, 0xffffffff, 0x00060005,
+	0xf011, 0xffffffff, 0x00090008,
+	0xf012, 0xffffffff, 0x00010000,
+	0xf013, 0xffffffff, 0x00030002,
+	0xf014, 0xffffffff, 0x00040007,
+	0xf015, 0xffffffff, 0x00060005,
+	0xf016, 0xffffffff, 0x00090008,
+	0xf017, 0xffffffff, 0x00010000,
+	0xf018, 0xffffffff, 0x00030002,
+	0xf019, 0xffffffff, 0x00040007,
+	0xf01a, 0xffffffff, 0x00060005,
+	0xf01b, 0xffffffff, 0x00090008,
+	0xf01c, 0xffffffff, 0x00010000,
+	0xf01d, 0xffffffff, 0x00030002,
+	0xf01e, 0xffffffff, 0x00040007,
+	0xf01f, 0xffffffff, 0x00060005,
+	0xf020, 0xffffffff, 0x00090008,
+	0xf021, 0xffffffff, 0x00010000,
+	0xf022, 0xffffffff, 0x00030002,
+	0xf023, 0xffffffff, 0x00040007,
+	0xf024, 0xffffffff, 0x00060005,
+	0xf025, 0xffffffff, 0x00090008,
+	0xf026, 0xffffffff, 0x00010000,
+	0xf027, 0xffffffff, 0x00030002,
+	0xf028, 0xffffffff, 0x00040007,
+	0xf029, 0xffffffff, 0x00060005,
+	0xf02a, 0xffffffff, 0x00090008,
+	0xf000, 0xffffffff, 0x96e00200,
+	0x21c2, 0xffffffff, 0x00900100,
+	0x3109, 0xffffffff, 0x0020003f,
+	0xe, 0xffffffff, 0x0140001c,
+	0xf, 0x000f0000, 0x000f0000,
+	0x88, 0xffffffff, 0xc060000c,
+	0x89, 0xc0000fff, 0x00000100,
+	0x3e4, 0xffffffff, 0x00000100,
+	0x3e6, 0x00000101, 0x00000000,
+	0x82a, 0xffffffff, 0x00000104,
+	0x1579, 0xff000fff, 0x00000100,
+	0xc33, 0xc0000fff, 0x00000104,
+	0x3079, 0x00000001, 0x00000001,
+	0x3403, 0xff000ff0, 0x00000100,
+	0x3603, 0xff000ff0, 0x00000100
+};
+
+static const u32 spectre_golden_spm_registers[] =
+{
+	0xc200, 0xe0ffffff, 0xe0000000
+};
+
+static const u32 spectre_golden_common_registers[] =
+{
+	0x31dc, 0xffffffff, 0x00000800,
+	0x31dd, 0xffffffff, 0x00000800,
+	0x31e6, 0xffffffff, 0x00007fbf,
+	0x31e7, 0xffffffff, 0x00007faf
+};
+
+static const u32 spectre_golden_registers[] =
+{
+	0xf000, 0xffff1fff, 0x96940200,
+	0xf003, 0xffff0001, 0xff000000,
+	0xf080, 0xfffc0fff, 0x00000100,
+	0x1bb6, 0x00010101, 0x00010000,
+	0x260d, 0xf00fffff, 0x00000400,
+	0x260e, 0xfffffffc, 0x00020200,
+	0x16ec, 0x000000f0, 0x00000070,
+	0x16f0, 0xf0311fff, 0x80300000,
+	0x263e, 0x73773777, 0x12010001,
+	0x26df, 0x00ff0000, 0x00fc0000,
+	0xbd2, 0x73773777, 0x12010001,
+	0x2285, 0xf000003f, 0x00000007,
+	0x22c9, 0xffffffff, 0x00ffffff,
+	0xa0d4, 0x3f3f3fff, 0x00000082,
+	0xa0d5, 0x0000003f, 0x00000000,
+	0xf9e, 0x00000001, 0x00000002,
+	0x244f, 0xffff03df, 0x00000004,
+	0x31da, 0x00000008, 0x00000008,
+	0x2300, 0x000008ff, 0x00000800,
+	0x2542, 0x00010000, 0x00010000,
+	0x2b03, 0xffffffff, 0x54763210,
+	0x853e, 0x01ff01ff, 0x00000002,
+	0x8526, 0x007ff800, 0x00200000,
+	0x8057, 0xffffffff, 0x00000f40,
+	0xc24d, 0xffffffff, 0x00000001
+};
+
+static const u32 spectre_mgcg_cgcg_init[] =
+{
+	0x3108, 0xffffffff, 0xfffffffc,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf0a8, 0xffffffff, 0x00000100,
+	0xf082, 0xffffffff, 0x00000100,
+	0xf0b0, 0xffffffff, 0x00000100,
+	0xf0b2, 0xffffffff, 0x00000100,
+	0xf0b1, 0xffffffff, 0x00000100,
+	0x1579, 0xffffffff, 0x00600100,
+	0xf0a0, 0xffffffff, 0x00000100,
+	0xf085, 0xffffffff, 0x06000100,
+	0xf088, 0xffffffff, 0x00000100,
+	0xf086, 0xffffffff, 0x06000100,
+	0xf081, 0xffffffff, 0x00000100,
+	0xf0b8, 0xffffffff, 0x00000100,
+	0xf089, 0xffffffff, 0x00000100,
+	0xf080, 0xffffffff, 0x00000100,
+	0xf08c, 0xffffffff, 0x00000100,
+	0xf08d, 0xffffffff, 0x00000100,
+	0xf094, 0xffffffff, 0x00000100,
+	0xf095, 0xffffffff, 0x00000100,
+	0xf096, 0xffffffff, 0x00000100,
+	0xf097, 0xffffffff, 0x00000100,
+	0xf098, 0xffffffff, 0x00000100,
+	0xf09f, 0xffffffff, 0x00000100,
+	0xf09e, 0xffffffff, 0x00000100,
+	0xf084, 0xffffffff, 0x06000100,
+	0xf0a4, 0xffffffff, 0x00000100,
+	0xf09d, 0xffffffff, 0x00000100,
+	0xf0ad, 0xffffffff, 0x00000100,
+	0xf0ac, 0xffffffff, 0x00000100,
+	0xf09c, 0xffffffff, 0x00000100,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf008, 0xffffffff, 0x00010000,
+	0xf009, 0xffffffff, 0x00030002,
+	0xf00a, 0xffffffff, 0x00040007,
+	0xf00b, 0xffffffff, 0x00060005,
+	0xf00c, 0xffffffff, 0x00090008,
+	0xf00d, 0xffffffff, 0x00010000,
+	0xf00e, 0xffffffff, 0x00030002,
+	0xf00f, 0xffffffff, 0x00040007,
+	0xf010, 0xffffffff, 0x00060005,
+	0xf011, 0xffffffff, 0x00090008,
+	0xf012, 0xffffffff, 0x00010000,
+	0xf013, 0xffffffff, 0x00030002,
+	0xf014, 0xffffffff, 0x00040007,
+	0xf015, 0xffffffff, 0x00060005,
+	0xf016, 0xffffffff, 0x00090008,
+	0xf017, 0xffffffff, 0x00010000,
+	0xf018, 0xffffffff, 0x00030002,
+	0xf019, 0xffffffff, 0x00040007,
+	0xf01a, 0xffffffff, 0x00060005,
+	0xf01b, 0xffffffff, 0x00090008,
+	0xf01c, 0xffffffff, 0x00010000,
+	0xf01d, 0xffffffff, 0x00030002,
+	0xf01e, 0xffffffff, 0x00040007,
+	0xf01f, 0xffffffff, 0x00060005,
+	0xf020, 0xffffffff, 0x00090008,
+	0xf021, 0xffffffff, 0x00010000,
+	0xf022, 0xffffffff, 0x00030002,
+	0xf023, 0xffffffff, 0x00040007,
+	0xf024, 0xffffffff, 0x00060005,
+	0xf025, 0xffffffff, 0x00090008,
+	0xf026, 0xffffffff, 0x00010000,
+	0xf027, 0xffffffff, 0x00030002,
+	0xf028, 0xffffffff, 0x00040007,
+	0xf029, 0xffffffff, 0x00060005,
+	0xf02a, 0xffffffff, 0x00090008,
+	0xf02b, 0xffffffff, 0x00010000,
+	0xf02c, 0xffffffff, 0x00030002,
+	0xf02d, 0xffffffff, 0x00040007,
+	0xf02e, 0xffffffff, 0x00060005,
+	0xf02f, 0xffffffff, 0x00090008,
+	0xf000, 0xffffffff, 0x96e00200,
+	0x21c2, 0xffffffff, 0x00900100,
+	0x3109, 0xffffffff, 0x0020003f,
+	0xe, 0xffffffff, 0x0140001c,
+	0xf, 0x000f0000, 0x000f0000,
+	0x88, 0xffffffff, 0xc060000c,
+	0x89, 0xc0000fff, 0x00000100,
+	0x3e4, 0xffffffff, 0x00000100,
+	0x3e6, 0x00000101, 0x00000000,
+	0x82a, 0xffffffff, 0x00000104,
+	0x1579, 0xff000fff, 0x00000100,
+	0xc33, 0xc0000fff, 0x00000104,
+	0x3079, 0x00000001, 0x00000001,
+	0x3403, 0xff000ff0, 0x00000100,
+	0x3603, 0xff000ff0, 0x00000100
+};
+
+static const u32 kalindi_golden_spm_registers[] =
+{
+	0xc200, 0xe0ffffff, 0xe0000000
+};
+
+static const u32 kalindi_golden_common_registers[] =
+{
+	0x31dc, 0xffffffff, 0x00000800,
+	0x31dd, 0xffffffff, 0x00000800,
+	0x31e6, 0xffffffff, 0x00007fbf,
+	0x31e7, 0xffffffff, 0x00007faf
+};
+
+static const u32 kalindi_golden_registers[] =
+{
+	0xf000, 0xffffdfff, 0x6e944040,
+	0x1579, 0xff607fff, 0xfc000100,
+	0xf088, 0xff000fff, 0x00000100,
+	0xf089, 0xff000fff, 0x00000100,
+	0xf080, 0xfffc0fff, 0x00000100,
+	0x1bb6, 0x00010101, 0x00010000,
+	0x260c, 0xffffffff, 0x00000000,
+	0x260d, 0xf00fffff, 0x00000400,
+	0x16ec, 0x000000f0, 0x00000070,
+	0x16f0, 0xf0311fff, 0x80300000,
+	0x263e, 0x73773777, 0x12010001,
+	0x263f, 0xffffffff, 0x00000010,
+	0x26df, 0x00ff0000, 0x00fc0000,
+	0x200c, 0x00001f0f, 0x0000100a,
+	0xbd2, 0x73773777, 0x12010001,
+	0x902, 0x000fffff, 0x000c007f,
+	0x2285, 0xf000003f, 0x00000007,
+	0x22c9, 0x3fff3fff, 0x00ffcfff,
+	0xc281, 0x0000ff0f, 0x00000000,
+	0xa293, 0x07ffffff, 0x06000000,
+	0x136, 0x00000fff, 0x00000100,
+	0xf9e, 0x00000001, 0x00000002,
+	0x31da, 0x00000008, 0x00000008,
+	0x2300, 0x000000ff, 0x00000003,
+	0x853e, 0x01ff01ff, 0x00000002,
+	0x8526, 0x007ff800, 0x00200000,
+	0x8057, 0xffffffff, 0x00000f40,
+	0x2231, 0x001f3ae3, 0x00000082,
+	0x2235, 0x0000001f, 0x00000010,
+	0xc24d, 0xffffffff, 0x00000000
+};
+
+static const u32 kalindi_mgcg_cgcg_init[] =
+{
+	0x3108, 0xffffffff, 0xfffffffc,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf0a8, 0xffffffff, 0x00000100,
+	0xf082, 0xffffffff, 0x00000100,
+	0xf0b0, 0xffffffff, 0x00000100,
+	0xf0b2, 0xffffffff, 0x00000100,
+	0xf0b1, 0xffffffff, 0x00000100,
+	0x1579, 0xffffffff, 0x00600100,
+	0xf0a0, 0xffffffff, 0x00000100,
+	0xf085, 0xffffffff, 0x06000100,
+	0xf088, 0xffffffff, 0x00000100,
+	0xf086, 0xffffffff, 0x06000100,
+	0xf081, 0xffffffff, 0x00000100,
+	0xf0b8, 0xffffffff, 0x00000100,
+	0xf089, 0xffffffff, 0x00000100,
+	0xf080, 0xffffffff, 0x00000100,
+	0xf08c, 0xffffffff, 0x00000100,
+	0xf08d, 0xffffffff, 0x00000100,
+	0xf094, 0xffffffff, 0x00000100,
+	0xf095, 0xffffffff, 0x00000100,
+	0xf096, 0xffffffff, 0x00000100,
+	0xf097, 0xffffffff, 0x00000100,
+	0xf098, 0xffffffff, 0x00000100,
+	0xf09f, 0xffffffff, 0x00000100,
+	0xf09e, 0xffffffff, 0x00000100,
+	0xf084, 0xffffffff, 0x06000100,
+	0xf0a4, 0xffffffff, 0x00000100,
+	0xf09d, 0xffffffff, 0x00000100,
+	0xf0ad, 0xffffffff, 0x00000100,
+	0xf0ac, 0xffffffff, 0x00000100,
+	0xf09c, 0xffffffff, 0x00000100,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf008, 0xffffffff, 0x00010000,
+	0xf009, 0xffffffff, 0x00030002,
+	0xf00a, 0xffffffff, 0x00040007,
+	0xf00b, 0xffffffff, 0x00060005,
+	0xf00c, 0xffffffff, 0x00090008,
+	0xf00d, 0xffffffff, 0x00010000,
+	0xf00e, 0xffffffff, 0x00030002,
+	0xf00f, 0xffffffff, 0x00040007,
+	0xf010, 0xffffffff, 0x00060005,
+	0xf011, 0xffffffff, 0x00090008,
+	0xf000, 0xffffffff, 0x96e00200,
+	0x21c2, 0xffffffff, 0x00900100,
+	0x3109, 0xffffffff, 0x0020003f,
+	0xe, 0xffffffff, 0x0140001c,
+	0xf, 0x000f0000, 0x000f0000,
+	0x88, 0xffffffff, 0xc060000c,
+	0x89, 0xc0000fff, 0x00000100,
+	0x82a, 0xffffffff, 0x00000104,
+	0x1579, 0xff000fff, 0x00000100,
+	0xc33, 0xc0000fff, 0x00000104,
+	0x3079, 0x00000001, 0x00000001,
+	0x3403, 0xff000ff0, 0x00000100,
+	0x3603, 0xff000ff0, 0x00000100
+};
+
+static const u32 hawaii_golden_spm_registers[] =
+{
+	0xc200, 0xe0ffffff, 0xe0000000
+};
+
+static const u32 hawaii_golden_common_registers[] =
+{
+	0xc200, 0xffffffff, 0xe0000000,
+	0xa0d4, 0xffffffff, 0x3a00161a,
+	0xa0d5, 0xffffffff, 0x0000002e,
+	0x2684, 0xffffffff, 0x00018208,
+	0x263e, 0xffffffff, 0x12011003
+};
+
+static const u32 hawaii_golden_registers[] =
+{
+	0xcd5, 0x00000333, 0x00000333,
+	0x2684, 0x00010000, 0x00058208,
+	0x260c, 0xffffffff, 0x00000000,
+	0x260d, 0xf00fffff, 0x00000400,
+	0x260e, 0x0002021c, 0x00020200,
+	0x31e, 0x00000080, 0x00000000,
+	0x16ec, 0x000000f0, 0x00000070,
+	0x16f0, 0xf0311fff, 0x80300000,
+	0xd43, 0x00810000, 0x408af000,
+	0x1c0c, 0x31000111, 0x00000011,
+	0xbd2, 0x73773777, 0x12010001,
+	0x848, 0x0000007f, 0x0000001b,
+	0x877, 0x00007fb6, 0x00002191,
+	0xd8a, 0x0000003f, 0x0000000a,
+	0xd8b, 0x0000003f, 0x0000000a,
+	0xab9, 0x00073ffe, 0x000022a2,
+	0x903, 0x000007ff, 0x00000000,
+	0x22fc, 0x00002001, 0x00000001,
+	0x22c9, 0xffffffff, 0x00ffffff,
+	0xc281, 0x0000ff0f, 0x00000000,
+	0xa293, 0x07ffffff, 0x06000000,
+	0xf9e, 0x00000001, 0x00000002,
+	0x31da, 0x00000008, 0x00000008,
+	0x31dc, 0x00000f00, 0x00000800,
+	0x31dd, 0x00000f00, 0x00000800,
+	0x31e6, 0x00ffffff, 0x00ff7fbf,
+	0x31e7, 0x00ffffff, 0x00ff7faf,
+	0x2300, 0x000000ff, 0x00000800,
+	0x390, 0x00001fff, 0x00001fff,
+	0x2418, 0x0000007f, 0x00000020,
+	0x2542, 0x00010000, 0x00010000,
+	0x2b80, 0x00100000, 0x000ff07c,
+	0x2b05, 0x000003ff, 0x0000000f,
+	0x2b04, 0xffffffff, 0x7564fdec,
+	0x2b03, 0xffffffff, 0x3120b9a8,
+	0x2b02, 0x20000000, 0x0f9c0000
+};
+
+static const u32 hawaii_mgcg_cgcg_init[] =
+{
+	0x3108, 0xffffffff, 0xfffffffd,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf0a8, 0xffffffff, 0x00000100,
+	0xf082, 0xffffffff, 0x00000100,
+	0xf0b0, 0xffffffff, 0x00000100,
+	0xf0b2, 0xffffffff, 0x00000100,
+	0xf0b1, 0xffffffff, 0x00000100,
+	0x1579, 0xffffffff, 0x00200100,
+	0xf0a0, 0xffffffff, 0x00000100,
+	0xf085, 0xffffffff, 0x06000100,
+	0xf088, 0xffffffff, 0x00000100,
+	0xf086, 0xffffffff, 0x06000100,
+	0xf081, 0xffffffff, 0x00000100,
+	0xf0b8, 0xffffffff, 0x00000100,
+	0xf089, 0xffffffff, 0x00000100,
+	0xf080, 0xffffffff, 0x00000100,
+	0xf08c, 0xffffffff, 0x00000100,
+	0xf08d, 0xffffffff, 0x00000100,
+	0xf094, 0xffffffff, 0x00000100,
+	0xf095, 0xffffffff, 0x00000100,
+	0xf096, 0xffffffff, 0x00000100,
+	0xf097, 0xffffffff, 0x00000100,
+	0xf098, 0xffffffff, 0x00000100,
+	0xf09f, 0xffffffff, 0x00000100,
+	0xf09e, 0xffffffff, 0x00000100,
+	0xf084, 0xffffffff, 0x06000100,
+	0xf0a4, 0xffffffff, 0x00000100,
+	0xf09d, 0xffffffff, 0x00000100,
+	0xf0ad, 0xffffffff, 0x00000100,
+	0xf0ac, 0xffffffff, 0x00000100,
+	0xf09c, 0xffffffff, 0x00000100,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf008, 0xffffffff, 0x00010000,
+	0xf009, 0xffffffff, 0x00030002,
+	0xf00a, 0xffffffff, 0x00040007,
+	0xf00b, 0xffffffff, 0x00060005,
+	0xf00c, 0xffffffff, 0x00090008,
+	0xf00d, 0xffffffff, 0x00010000,
+	0xf00e, 0xffffffff, 0x00030002,
+	0xf00f, 0xffffffff, 0x00040007,
+	0xf010, 0xffffffff, 0x00060005,
+	0xf011, 0xffffffff, 0x00090008,
+	0xf012, 0xffffffff, 0x00010000,
+	0xf013, 0xffffffff, 0x00030002,
+	0xf014, 0xffffffff, 0x00040007,
+	0xf015, 0xffffffff, 0x00060005,
+	0xf016, 0xffffffff, 0x00090008,
+	0xf017, 0xffffffff, 0x00010000,
+	0xf018, 0xffffffff, 0x00030002,
+	0xf019, 0xffffffff, 0x00040007,
+	0xf01a, 0xffffffff, 0x00060005,
+	0xf01b, 0xffffffff, 0x00090008,
+	0xf01c, 0xffffffff, 0x00010000,
+	0xf01d, 0xffffffff, 0x00030002,
+	0xf01e, 0xffffffff, 0x00040007,
+	0xf01f, 0xffffffff, 0x00060005,
+	0xf020, 0xffffffff, 0x00090008,
+	0xf021, 0xffffffff, 0x00010000,
+	0xf022, 0xffffffff, 0x00030002,
+	0xf023, 0xffffffff, 0x00040007,
+	0xf024, 0xffffffff, 0x00060005,
+	0xf025, 0xffffffff, 0x00090008,
+	0xf026, 0xffffffff, 0x00010000,
+	0xf027, 0xffffffff, 0x00030002,
+	0xf028, 0xffffffff, 0x00040007,
+	0xf029, 0xffffffff, 0x00060005,
+	0xf02a, 0xffffffff, 0x00090008,
+	0xf02b, 0xffffffff, 0x00010000,
+	0xf02c, 0xffffffff, 0x00030002,
+	0xf02d, 0xffffffff, 0x00040007,
+	0xf02e, 0xffffffff, 0x00060005,
+	0xf02f, 0xffffffff, 0x00090008,
+	0xf030, 0xffffffff, 0x00010000,
+	0xf031, 0xffffffff, 0x00030002,
+	0xf032, 0xffffffff, 0x00040007,
+	0xf033, 0xffffffff, 0x00060005,
+	0xf034, 0xffffffff, 0x00090008,
+	0xf035, 0xffffffff, 0x00010000,
+	0xf036, 0xffffffff, 0x00030002,
+	0xf037, 0xffffffff, 0x00040007,
+	0xf038, 0xffffffff, 0x00060005,
+	0xf039, 0xffffffff, 0x00090008,
+	0xf03a, 0xffffffff, 0x00010000,
+	0xf03b, 0xffffffff, 0x00030002,
+	0xf03c, 0xffffffff, 0x00040007,
+	0xf03d, 0xffffffff, 0x00060005,
+	0xf03e, 0xffffffff, 0x00090008,
+	0x30c6, 0xffffffff, 0x00020200,
+	0xcd4, 0xffffffff, 0x00000200,
+	0x570, 0xffffffff, 0x00000400,
+	0x157a, 0xffffffff, 0x00000000,
+	0xbd4, 0xffffffff, 0x00000902,
+	0xf000, 0xffffffff, 0x96940200,
+	0x21c2, 0xffffffff, 0x00900100,
+	0x3109, 0xffffffff, 0x0020003f,
+	0xe, 0xffffffff, 0x0140001c,
+	0xf, 0x000f0000, 0x000f0000,
+	0x88, 0xffffffff, 0xc060000c,
+	0x89, 0xc0000fff, 0x00000100,
+	0x3e4, 0xffffffff, 0x00000100,
+	0x3e6, 0x00000101, 0x00000000,
+	0x82a, 0xffffffff, 0x00000104,
+	0x1579, 0xff000fff, 0x00000100,
+	0xc33, 0xc0000fff, 0x00000104,
+	0x3079, 0x00000001, 0x00000001,
+	0x3403, 0xff000ff0, 0x00000100,
+	0x3603, 0xff000ff0, 0x00000100
+};
+
+static const u32 godavari_golden_registers[] =
+{
+	0x1579, 0xff607fff, 0xfc000100,
+	0x1bb6, 0x00010101, 0x00010000,
+	0x260c, 0xffffffff, 0x00000000,
+	0x260c0, 0xf00fffff, 0x00000400,
+	0x184c, 0xffffffff, 0x00010000,
+	0x16ec, 0x000000f0, 0x00000070,
+	0x16f0, 0xf0311fff, 0x80300000,
+	0x263e, 0x73773777, 0x12010001,
+	0x263f, 0xffffffff, 0x00000010,
+	0x200c, 0x00001f0f, 0x0000100a,
+	0xbd2, 0x73773777, 0x12010001,
+	0x902, 0x000fffff, 0x000c007f,
+	0x2285, 0xf000003f, 0x00000007,
+	0x22c9, 0xffffffff, 0x00ff0fff,
+	0xc281, 0x0000ff0f, 0x00000000,
+	0xa293, 0x07ffffff, 0x06000000,
+	0x136, 0x00000fff, 0x00000100,
+	0x3405, 0x00010000, 0x00810001,
+	0x3605, 0x00010000, 0x00810001,
+	0xf9e, 0x00000001, 0x00000002,
+	0x31da, 0x00000008, 0x00000008,
+	0x31dc, 0x00000f00, 0x00000800,
+	0x31dd, 0x00000f00, 0x00000800,
+	0x31e6, 0x00ffffff, 0x00ff7fbf,
+	0x31e7, 0x00ffffff, 0x00ff7faf,
+	0x2300, 0x000000ff, 0x00000001,
+	0x853e, 0x01ff01ff, 0x00000002,
+	0x8526, 0x007ff800, 0x00200000,
+	0x8057, 0xffffffff, 0x00000f40,
+	0x2231, 0x001f3ae3, 0x00000082,
+	0x2235, 0x0000001f, 0x00000010,
+	0xc24d, 0xffffffff, 0x00000000
+};
+
+static void cik_init_golden_registers(struct amdgpu_device *adev)
+{
+	/* Some of the registers might be dependent on GRBM_GFX_INDEX */
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		amdgpu_device_program_register_sequence(adev,
+							bonaire_mgcg_cgcg_init,
+							ARRAY_SIZE(bonaire_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							bonaire_golden_registers,
+							ARRAY_SIZE(bonaire_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							bonaire_golden_common_registers,
+							ARRAY_SIZE(bonaire_golden_common_registers));
+		amdgpu_device_program_register_sequence(adev,
+							bonaire_golden_spm_registers,
+							ARRAY_SIZE(bonaire_golden_spm_registers));
+		break;
+	case CHIP_KABINI:
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_mgcg_cgcg_init,
+							ARRAY_SIZE(kalindi_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_golden_registers,
+							ARRAY_SIZE(kalindi_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_golden_common_registers,
+							ARRAY_SIZE(kalindi_golden_common_registers));
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_golden_spm_registers,
+							ARRAY_SIZE(kalindi_golden_spm_registers));
+		break;
+	case CHIP_MULLINS:
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_mgcg_cgcg_init,
+							ARRAY_SIZE(kalindi_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							godavari_golden_registers,
+							ARRAY_SIZE(godavari_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_golden_common_registers,
+							ARRAY_SIZE(kalindi_golden_common_registers));
+		amdgpu_device_program_register_sequence(adev,
+							kalindi_golden_spm_registers,
+							ARRAY_SIZE(kalindi_golden_spm_registers));
+		break;
+	case CHIP_KAVERI:
+		amdgpu_device_program_register_sequence(adev,
+							spectre_mgcg_cgcg_init,
+							ARRAY_SIZE(spectre_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							spectre_golden_registers,
+							ARRAY_SIZE(spectre_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							spectre_golden_common_registers,
+							ARRAY_SIZE(spectre_golden_common_registers));
+		amdgpu_device_program_register_sequence(adev,
+							spectre_golden_spm_registers,
+							ARRAY_SIZE(spectre_golden_spm_registers));
+		break;
+	case CHIP_HAWAII:
+		amdgpu_device_program_register_sequence(adev,
+							hawaii_mgcg_cgcg_init,
+							ARRAY_SIZE(hawaii_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							hawaii_golden_registers,
+							ARRAY_SIZE(hawaii_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							hawaii_golden_common_registers,
+							ARRAY_SIZE(hawaii_golden_common_registers));
+		amdgpu_device_program_register_sequence(adev,
+							hawaii_golden_spm_registers,
+							ARRAY_SIZE(hawaii_golden_spm_registers));
+		break;
+	default:
+		break;
+	}
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+/**
+ * cik_get_xclk - get the xclk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns the reference clock used by the gfx engine
+ * (CIK).
+ */
+static u32 cik_get_xclk(struct amdgpu_device *adev)
+{
+	u32 reference_clock = adev->clock.spll.reference_freq;
+
+	if (adev->flags & AMD_IS_APU) {
+		if (RREG32_SMC(ixGENERAL_PWRMGT) & GENERAL_PWRMGT__GPU_COUNTER_CLK_MASK)
+			return reference_clock / 2;
+	} else {
+		if (RREG32_SMC(ixCG_CLKPIN_CNTL) & CG_CLKPIN_CNTL__XTALIN_DIVIDE_MASK)
+			return reference_clock / 4;
+	}
+	return reference_clock;
+}
+
+/**
+ * cik_srbm_select - select specific register instances
+ *
+ * @adev: amdgpu_device pointer
+ * @me: selected ME (micro engine)
+ * @pipe: pipe
+ * @queue: queue
+ * @vmid: VMID
+ *
+ * Switches the currently active registers instances.  Some
+ * registers are instanced per VMID, others are instanced per
+ * me/pipe/queue combination.
+ */
+void cik_srbm_select(struct amdgpu_device *adev,
+		     u32 me, u32 pipe, u32 queue, u32 vmid)
+{
+	u32 srbm_gfx_cntl =
+		(((pipe << SRBM_GFX_CNTL__PIPEID__SHIFT) & SRBM_GFX_CNTL__PIPEID_MASK)|
+		((me << SRBM_GFX_CNTL__MEID__SHIFT) & SRBM_GFX_CNTL__MEID_MASK)|
+		((vmid << SRBM_GFX_CNTL__VMID__SHIFT) & SRBM_GFX_CNTL__VMID_MASK)|
+		((queue << SRBM_GFX_CNTL__QUEUEID__SHIFT) & SRBM_GFX_CNTL__QUEUEID_MASK));
+	WREG32(mmSRBM_GFX_CNTL, srbm_gfx_cntl);
+}
+
+static void cik_vga_set_state(struct amdgpu_device *adev, bool state)
+{
+	uint32_t tmp;
+
+	tmp = RREG32(mmCONFIG_CNTL);
+	if (!state)
+		tmp |= CONFIG_CNTL__VGA_DIS_MASK;
+	else
+		tmp &= ~CONFIG_CNTL__VGA_DIS_MASK;
+	WREG32(mmCONFIG_CNTL, tmp);
+}
+
+static bool cik_read_disabled_bios(struct amdgpu_device *adev)
+{
+	u32 bus_cntl;
+	u32 d1vga_control = 0;
+	u32 d2vga_control = 0;
+	u32 vga_render_control = 0;
+	u32 rom_cntl;
+	bool r;
+
+	bus_cntl = RREG32(mmBUS_CNTL);
+	if (adev->mode_info.num_crtc) {
+		d1vga_control = RREG32(mmD1VGA_CONTROL);
+		d2vga_control = RREG32(mmD2VGA_CONTROL);
+		vga_render_control = RREG32(mmVGA_RENDER_CONTROL);
+	}
+	rom_cntl = RREG32_SMC(ixROM_CNTL);
+
+	/* enable the rom */
+	WREG32(mmBUS_CNTL, (bus_cntl & ~BUS_CNTL__BIOS_ROM_DIS_MASK));
+	if (adev->mode_info.num_crtc) {
+		/* Disable VGA mode */
+		WREG32(mmD1VGA_CONTROL,
+		       (d1vga_control & ~(D1VGA_CONTROL__D1VGA_MODE_ENABLE_MASK |
+					  D1VGA_CONTROL__D1VGA_TIMING_SELECT_MASK)));
+		WREG32(mmD2VGA_CONTROL,
+		       (d2vga_control & ~(D1VGA_CONTROL__D1VGA_MODE_ENABLE_MASK |
+					  D1VGA_CONTROL__D1VGA_TIMING_SELECT_MASK)));
+		WREG32(mmVGA_RENDER_CONTROL,
+		       (vga_render_control & ~VGA_RENDER_CONTROL__VGA_VSTATUS_CNTL_MASK));
+	}
+	WREG32_SMC(ixROM_CNTL, rom_cntl | ROM_CNTL__SCK_OVERWRITE_MASK);
+
+	r = amdgpu_read_bios(adev);
+
+	/* restore regs */
+	WREG32(mmBUS_CNTL, bus_cntl);
+	if (adev->mode_info.num_crtc) {
+		WREG32(mmD1VGA_CONTROL, d1vga_control);
+		WREG32(mmD2VGA_CONTROL, d2vga_control);
+		WREG32(mmVGA_RENDER_CONTROL, vga_render_control);
+	}
+	WREG32_SMC(ixROM_CNTL, rom_cntl);
+	return r;
+}
+
+static bool cik_read_bios_from_rom(struct amdgpu_device *adev,
+				   u8 *bios, u32 length_bytes)
+{
+	u32 *dw_ptr;
+	unsigned long flags;
+	u32 i, length_dw;
+
+	if (bios == NULL)
+		return false;
+	if (length_bytes == 0)
+		return false;
+	/* APU vbios image is part of sbios image */
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	dw_ptr = (u32 *)bios;
+	length_dw = roundup2(length_bytes, 4) / 4;
+	/* take the smc lock since we are using the smc index */
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	/* set rom index to 0 */
+	WREG32(mmSMC_IND_INDEX_0, ixROM_INDEX);
+	WREG32(mmSMC_IND_DATA_0, 0);
+	/* set index to data for continous read */
+	WREG32(mmSMC_IND_INDEX_0, ixROM_DATA);
+	for (i = 0; i < length_dw; i++)
+		dw_ptr[i] = RREG32(mmSMC_IND_DATA_0);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return true;
+}
+
+static const struct amdgpu_allowed_register_entry cik_allowed_read_registers[] = {
+	{mmGRBM_STATUS},
+	{mmGB_ADDR_CONFIG},
+	{mmMC_ARB_RAMCFG},
+	{mmGB_TILE_MODE0},
+	{mmGB_TILE_MODE1},
+	{mmGB_TILE_MODE2},
+	{mmGB_TILE_MODE3},
+	{mmGB_TILE_MODE4},
+	{mmGB_TILE_MODE5},
+	{mmGB_TILE_MODE6},
+	{mmGB_TILE_MODE7},
+	{mmGB_TILE_MODE8},
+	{mmGB_TILE_MODE9},
+	{mmGB_TILE_MODE10},
+	{mmGB_TILE_MODE11},
+	{mmGB_TILE_MODE12},
+	{mmGB_TILE_MODE13},
+	{mmGB_TILE_MODE14},
+	{mmGB_TILE_MODE15},
+	{mmGB_TILE_MODE16},
+	{mmGB_TILE_MODE17},
+	{mmGB_TILE_MODE18},
+	{mmGB_TILE_MODE19},
+	{mmGB_TILE_MODE20},
+	{mmGB_TILE_MODE21},
+	{mmGB_TILE_MODE22},
+	{mmGB_TILE_MODE23},
+	{mmGB_TILE_MODE24},
+	{mmGB_TILE_MODE25},
+	{mmGB_TILE_MODE26},
+	{mmGB_TILE_MODE27},
+	{mmGB_TILE_MODE28},
+	{mmGB_TILE_MODE29},
+	{mmGB_TILE_MODE30},
+	{mmGB_TILE_MODE31},
+	{mmGB_MACROTILE_MODE0},
+	{mmGB_MACROTILE_MODE1},
+	{mmGB_MACROTILE_MODE2},
+	{mmGB_MACROTILE_MODE3},
+	{mmGB_MACROTILE_MODE4},
+	{mmGB_MACROTILE_MODE5},
+	{mmGB_MACROTILE_MODE6},
+	{mmGB_MACROTILE_MODE7},
+	{mmGB_MACROTILE_MODE8},
+	{mmGB_MACROTILE_MODE9},
+	{mmGB_MACROTILE_MODE10},
+	{mmGB_MACROTILE_MODE11},
+	{mmGB_MACROTILE_MODE12},
+	{mmGB_MACROTILE_MODE13},
+	{mmGB_MACROTILE_MODE14},
+	{mmGB_MACROTILE_MODE15},
+	{mmCC_RB_BACKEND_DISABLE, true},
+	{mmGC_USER_RB_BACKEND_DISABLE, true},
+	{mmGB_BACKEND_MAP, false},
+	{mmPA_SC_RASTER_CONFIG, true},
+	{mmPA_SC_RASTER_CONFIG_1, true},
+};
+
+
+static uint32_t cik_get_register_value(struct amdgpu_device *adev,
+				       bool indexed, u32 se_num,
+				       u32 sh_num, u32 reg_offset)
+{
+	if (indexed) {
+		uint32_t val;
+		unsigned se_idx = (se_num == 0xffffffff) ? 0 : se_num;
+		unsigned sh_idx = (sh_num == 0xffffffff) ? 0 : sh_num;
+
+		switch (reg_offset) {
+		case mmCC_RB_BACKEND_DISABLE:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].rb_backend_disable;
+		case mmGC_USER_RB_BACKEND_DISABLE:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].user_rb_backend_disable;
+		case mmPA_SC_RASTER_CONFIG:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].raster_config;
+		case mmPA_SC_RASTER_CONFIG_1:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].raster_config_1;
+		}
+
+		mutex_lock(&adev->grbm_idx_mutex);
+		if (se_num != 0xffffffff || sh_num != 0xffffffff)
+			amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+
+		val = RREG32(reg_offset);
+
+		if (se_num != 0xffffffff || sh_num != 0xffffffff)
+			amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		mutex_unlock(&adev->grbm_idx_mutex);
+		return val;
+	} else {
+		unsigned idx;
+
+		switch (reg_offset) {
+		case mmGB_ADDR_CONFIG:
+			return adev->gfx.config.gb_addr_config;
+		case mmMC_ARB_RAMCFG:
+			return adev->gfx.config.mc_arb_ramcfg;
+		case mmGB_TILE_MODE0:
+		case mmGB_TILE_MODE1:
+		case mmGB_TILE_MODE2:
+		case mmGB_TILE_MODE3:
+		case mmGB_TILE_MODE4:
+		case mmGB_TILE_MODE5:
+		case mmGB_TILE_MODE6:
+		case mmGB_TILE_MODE7:
+		case mmGB_TILE_MODE8:
+		case mmGB_TILE_MODE9:
+		case mmGB_TILE_MODE10:
+		case mmGB_TILE_MODE11:
+		case mmGB_TILE_MODE12:
+		case mmGB_TILE_MODE13:
+		case mmGB_TILE_MODE14:
+		case mmGB_TILE_MODE15:
+		case mmGB_TILE_MODE16:
+		case mmGB_TILE_MODE17:
+		case mmGB_TILE_MODE18:
+		case mmGB_TILE_MODE19:
+		case mmGB_TILE_MODE20:
+		case mmGB_TILE_MODE21:
+		case mmGB_TILE_MODE22:
+		case mmGB_TILE_MODE23:
+		case mmGB_TILE_MODE24:
+		case mmGB_TILE_MODE25:
+		case mmGB_TILE_MODE26:
+		case mmGB_TILE_MODE27:
+		case mmGB_TILE_MODE28:
+		case mmGB_TILE_MODE29:
+		case mmGB_TILE_MODE30:
+		case mmGB_TILE_MODE31:
+			idx = (reg_offset - mmGB_TILE_MODE0);
+			return adev->gfx.config.tile_mode_array[idx];
+		case mmGB_MACROTILE_MODE0:
+		case mmGB_MACROTILE_MODE1:
+		case mmGB_MACROTILE_MODE2:
+		case mmGB_MACROTILE_MODE3:
+		case mmGB_MACROTILE_MODE4:
+		case mmGB_MACROTILE_MODE5:
+		case mmGB_MACROTILE_MODE6:
+		case mmGB_MACROTILE_MODE7:
+		case mmGB_MACROTILE_MODE8:
+		case mmGB_MACROTILE_MODE9:
+		case mmGB_MACROTILE_MODE10:
+		case mmGB_MACROTILE_MODE11:
+		case mmGB_MACROTILE_MODE12:
+		case mmGB_MACROTILE_MODE13:
+		case mmGB_MACROTILE_MODE14:
+		case mmGB_MACROTILE_MODE15:
+			idx = (reg_offset - mmGB_MACROTILE_MODE0);
+			return adev->gfx.config.macrotile_mode_array[idx];
+		default:
+			return RREG32(reg_offset);
+		}
+	}
+}
+
+static int cik_read_register(struct amdgpu_device *adev, u32 se_num,
+			     u32 sh_num, u32 reg_offset, u32 *value)
+{
+	uint32_t i;
+
+	*value = 0;
+	for (i = 0; i < ARRAY_SIZE(cik_allowed_read_registers); i++) {
+		bool indexed = cik_allowed_read_registers[i].grbm_indexed;
+
+		if (reg_offset != cik_allowed_read_registers[i].reg_offset)
+			continue;
+
+		*value = cik_get_register_value(adev, indexed, se_num, sh_num,
+						reg_offset);
+		return 0;
+	}
+	return -EINVAL;
+}
+
+struct kv_reset_save_regs {
+	u32 gmcon_reng_execute;
+	u32 gmcon_misc;
+	u32 gmcon_misc3;
+};
+
+static void kv_save_regs_for_reset(struct amdgpu_device *adev,
+				   struct kv_reset_save_regs *save)
+{
+	save->gmcon_reng_execute = RREG32(mmGMCON_RENG_EXECUTE);
+	save->gmcon_misc = RREG32(mmGMCON_MISC);
+	save->gmcon_misc3 = RREG32(mmGMCON_MISC3);
+
+	WREG32(mmGMCON_RENG_EXECUTE, save->gmcon_reng_execute &
+		~GMCON_RENG_EXECUTE__RENG_EXECUTE_ON_PWR_UP_MASK);
+	WREG32(mmGMCON_MISC, save->gmcon_misc &
+		~(GMCON_MISC__RENG_EXECUTE_ON_REG_UPDATE_MASK |
+			GMCON_MISC__STCTRL_STUTTER_EN_MASK));
+}
+
+static void kv_restore_regs_for_reset(struct amdgpu_device *adev,
+				      struct kv_reset_save_regs *save)
+{
+	int i;
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0x200010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0x300010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x210000);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0xa00010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x21003);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0xb00010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x2b00);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0xc00010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0xd00010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x420000);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0x100010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x120202);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0x500010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x3e3e36);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0x600010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x373f3e);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0x700010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x3e1332);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0xe00010ff);
+
+	WREG32(mmGMCON_MISC3, save->gmcon_misc3);
+	WREG32(mmGMCON_MISC, save->gmcon_misc);
+	WREG32(mmGMCON_RENG_EXECUTE, save->gmcon_reng_execute);
+}
+
+static int cik_gpu_pci_config_reset(struct amdgpu_device *adev)
+{
+	struct kv_reset_save_regs kv_save = { 0 };
+	u32 i;
+	int r = -EINVAL;
+
+	dev_info(adev->dev, "GPU pci config reset\n");
+
+	if (adev->flags & AMD_IS_APU)
+		kv_save_regs_for_reset(adev, &kv_save);
+
+	/* disable BM */
+	pci_clear_master(adev->pdev);
+	/* reset */
+	amdgpu_device_pci_config_reset(adev);
+
+	udelay(100);
+
+	/* wait for asic to come out of reset */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmCONFIG_MEMSIZE) != 0xffffffff) {
+			/* enable BM */
+			pci_set_master(adev->pdev);
+			adev->has_hw_reset = true;
+			r = 0;
+			break;
+		}
+		udelay(1);
+	}
+
+	/* does asic init need to be run first??? */
+	if (adev->flags & AMD_IS_APU)
+		kv_restore_regs_for_reset(adev, &kv_save);
+
+	return r;
+}
+
+/**
+ * cik_asic_reset - soft reset GPU
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up which blocks are hung and attempt
+ * to reset them.
+ * Returns 0 for success.
+ */
+static int cik_asic_reset(struct amdgpu_device *adev)
+{
+	int r;
+
+	amdgpu_atombios_scratch_regs_engine_hung(adev, true);
+
+	r = cik_gpu_pci_config_reset(adev);
+
+	amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+
+	return r;
+}
+
+static u32 cik_get_config_memsize(struct amdgpu_device *adev)
+{
+	return RREG32(mmCONFIG_MEMSIZE);
+}
+
+static int cik_set_uvd_clock(struct amdgpu_device *adev, u32 clock,
+			      u32 cntl_reg, u32 status_reg)
+{
+	int r, i;
+	struct atom_clock_dividers dividers;
+	uint32_t tmp;
+
+	r = amdgpu_atombios_get_clock_dividers(adev,
+					       COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+					       clock, false, &dividers);
+	if (r)
+		return r;
+
+	tmp = RREG32_SMC(cntl_reg);
+	tmp &= ~(CG_DCLK_CNTL__DCLK_DIR_CNTL_EN_MASK |
+		CG_DCLK_CNTL__DCLK_DIVIDER_MASK);
+	tmp |= dividers.post_divider;
+	WREG32_SMC(cntl_reg, tmp);
+
+	for (i = 0; i < 100; i++) {
+		if (RREG32_SMC(status_reg) & CG_DCLK_STATUS__DCLK_STATUS_MASK)
+			break;
+		mdelay(10);
+	}
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	return 0;
+}
+
+static int cik_set_uvd_clocks(struct amdgpu_device *adev, u32 vclk, u32 dclk)
+{
+	int r = 0;
+
+	r = cik_set_uvd_clock(adev, vclk, ixCG_VCLK_CNTL, ixCG_VCLK_STATUS);
+	if (r)
+		return r;
+
+	r = cik_set_uvd_clock(adev, dclk, ixCG_DCLK_CNTL, ixCG_DCLK_STATUS);
+	return r;
+}
+
+static int cik_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk)
+{
+	int r, i;
+	struct atom_clock_dividers dividers;
+	u32 tmp;
+
+	r = amdgpu_atombios_get_clock_dividers(adev,
+					       COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+					       ecclk, false, &dividers);
+	if (r)
+		return r;
+
+	for (i = 0; i < 100; i++) {
+		if (RREG32_SMC(ixCG_ECLK_STATUS) & CG_ECLK_STATUS__ECLK_STATUS_MASK)
+			break;
+		mdelay(10);
+	}
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	tmp = RREG32_SMC(ixCG_ECLK_CNTL);
+	tmp &= ~(CG_ECLK_CNTL__ECLK_DIR_CNTL_EN_MASK |
+		CG_ECLK_CNTL__ECLK_DIVIDER_MASK);
+	tmp |= dividers.post_divider;
+	WREG32_SMC(ixCG_ECLK_CNTL, tmp);
+
+	for (i = 0; i < 100; i++) {
+		if (RREG32_SMC(ixCG_ECLK_STATUS) & CG_ECLK_STATUS__ECLK_STATUS_MASK)
+			break;
+		mdelay(10);
+	}
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	return 0;
+}
+
+static void cik_pcie_gen3_enable(struct amdgpu_device *adev)
+{
+	struct pci_dev *root = adev->pdev->bus->self;
+	int bridge_pos, gpu_pos;
+	u32 speed_cntl, current_data_rate;
+	int i;
+	u16 tmp16;
+
+	if (pci_is_root_bus(adev->pdev->bus))
+		return;
+
+	if (amdgpu_pcie_gen2 == 0)
+		return;
+
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	if (!(adev->pm.pcie_gen_mask & (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+					CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)))
+		return;
+
+	speed_cntl = RREG32_PCIE(ixPCIE_LC_SPEED_CNTL);
+	current_data_rate = (speed_cntl & PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE_MASK) >>
+		PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE__SHIFT;
+	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3) {
+		if (current_data_rate == 2) {
+			DRM_INFO("PCIE gen 3 link speeds already enabled\n");
+			return;
+		}
+		DRM_INFO("enabling PCIE gen 3 link speeds, disable with amdgpu.pcie_gen2=0\n");
+	} else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2) {
+		if (current_data_rate == 1) {
+			DRM_INFO("PCIE gen 2 link speeds already enabled\n");
+			return;
+		}
+		DRM_INFO("enabling PCIE gen 2 link speeds, disable with amdgpu.pcie_gen2=0\n");
+	}
+
+	bridge_pos = pci_pcie_cap(root);
+	if (!bridge_pos)
+		return;
+
+	gpu_pos = pci_pcie_cap(adev->pdev);
+	if (!gpu_pos)
+		return;
+
+	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3) {
+		/* re-try equalization if gen3 is not already enabled */
+		if (current_data_rate != 2) {
+			u16 bridge_cfg, gpu_cfg;
+			u16 bridge_cfg2, gpu_cfg2;
+			u32 max_lw, current_lw, tmp;
+
+			pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &bridge_cfg);
+			pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &gpu_cfg);
+
+			tmp16 = bridge_cfg | PCI_EXP_LNKCTL_HAWD;
+			pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL, tmp16);
+
+			tmp16 = gpu_cfg | PCI_EXP_LNKCTL_HAWD;
+			pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, tmp16);
+
+			tmp = RREG32_PCIE(ixPCIE_LC_STATUS1);
+			max_lw = (tmp & PCIE_LC_STATUS1__LC_DETECTED_LINK_WIDTH_MASK) >>
+				PCIE_LC_STATUS1__LC_DETECTED_LINK_WIDTH__SHIFT;
+			current_lw = (tmp & PCIE_LC_STATUS1__LC_OPERATING_LINK_WIDTH_MASK)
+				>> PCIE_LC_STATUS1__LC_OPERATING_LINK_WIDTH__SHIFT;
+
+			if (current_lw < max_lw) {
+				tmp = RREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL);
+				if (tmp & PCIE_LC_LINK_WIDTH_CNTL__LC_RENEGOTIATION_SUPPORT_MASK) {
+					tmp &= ~(PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_MASK |
+						PCIE_LC_LINK_WIDTH_CNTL__LC_UPCONFIGURE_DIS_MASK);
+					tmp |= (max_lw <<
+						PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH__SHIFT);
+					tmp |= PCIE_LC_LINK_WIDTH_CNTL__LC_UPCONFIGURE_SUPPORT_MASK |
+					PCIE_LC_LINK_WIDTH_CNTL__LC_RENEGOTIATE_EN_MASK |
+					PCIE_LC_LINK_WIDTH_CNTL__LC_RECONFIG_NOW_MASK;
+					WREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL, tmp);
+				}
+			}
+
+			for (i = 0; i < 10; i++) {
+				/* check status */
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_DEVSTA, &tmp16);
+				if (tmp16 & PCI_EXP_DEVSTA_TRPND)
+					break;
+
+				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &bridge_cfg);
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &gpu_cfg);
+
+				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, &bridge_cfg2);
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &gpu_cfg2);
+
+				tmp = RREG32_PCIE(ixPCIE_LC_CNTL4);
+				tmp |= PCIE_LC_CNTL4__LC_SET_QUIESCE_MASK;
+				WREG32_PCIE(ixPCIE_LC_CNTL4, tmp);
+
+				tmp = RREG32_PCIE(ixPCIE_LC_CNTL4);
+				tmp |= PCIE_LC_CNTL4__LC_REDO_EQ_MASK;
+				WREG32_PCIE(ixPCIE_LC_CNTL4, tmp);
+
+				drm_msleep(100);
+
+				/* linkctl */
+				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &tmp16);
+				tmp16 &= ~PCI_EXP_LNKCTL_HAWD;
+				tmp16 |= (bridge_cfg & PCI_EXP_LNKCTL_HAWD);
+				pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL, tmp16);
+
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &tmp16);
+				tmp16 &= ~PCI_EXP_LNKCTL_HAWD;
+				tmp16 |= (gpu_cfg & PCI_EXP_LNKCTL_HAWD);
+				pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, tmp16);
+
+				/* linkctl2 */
+				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, &tmp16);
+				tmp16 &= ~((1 << 4) | (7 << 9));
+				tmp16 |= (bridge_cfg2 & ((1 << 4) | (7 << 9)));
+				pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, tmp16);
+
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &tmp16);
+				tmp16 &= ~((1 << 4) | (7 << 9));
+				tmp16 |= (gpu_cfg2 & ((1 << 4) | (7 << 9)));
+				pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, tmp16);
+
+				tmp = RREG32_PCIE(ixPCIE_LC_CNTL4);
+				tmp &= ~PCIE_LC_CNTL4__LC_SET_QUIESCE_MASK;
+				WREG32_PCIE(ixPCIE_LC_CNTL4, tmp);
+			}
+		}
+	}
+
+	/* set the link speed */
+	speed_cntl |= PCIE_LC_SPEED_CNTL__LC_FORCE_EN_SW_SPEED_CHANGE_MASK |
+		PCIE_LC_SPEED_CNTL__LC_FORCE_DIS_HW_SPEED_CHANGE_MASK;
+	speed_cntl &= ~PCIE_LC_SPEED_CNTL__LC_FORCE_DIS_SW_SPEED_CHANGE_MASK;
+	WREG32_PCIE(ixPCIE_LC_SPEED_CNTL, speed_cntl);
+
+	pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &tmp16);
+	tmp16 &= ~0xf;
+	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)
+		tmp16 |= 3; /* gen3 */
+	else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2)
+		tmp16 |= 2; /* gen2 */
+	else
+		tmp16 |= 1; /* gen1 */
+	pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, tmp16);
+
+	speed_cntl = RREG32_PCIE(ixPCIE_LC_SPEED_CNTL);
+	speed_cntl |= PCIE_LC_SPEED_CNTL__LC_INITIATE_LINK_SPEED_CHANGE_MASK;
+	WREG32_PCIE(ixPCIE_LC_SPEED_CNTL, speed_cntl);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		speed_cntl = RREG32_PCIE(ixPCIE_LC_SPEED_CNTL);
+		if ((speed_cntl & PCIE_LC_SPEED_CNTL__LC_INITIATE_LINK_SPEED_CHANGE_MASK) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+static void cik_program_aspm(struct amdgpu_device *adev)
+{
+	u32 data, orig;
+	bool disable_l0s = false, disable_l1 = false, disable_plloff_in_l1 = false;
+	bool disable_clkreq = false;
+
+	if (amdgpu_aspm == 0)
+		return;
+
+	if (pci_is_root_bus(adev->pdev->bus))
+		return;
+
+	/* XXX double check APUs */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	orig = data = RREG32_PCIE(ixPCIE_LC_N_FTS_CNTL);
+	data &= ~PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS_MASK;
+	data |= (0x24 << PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS__SHIFT) |
+		PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS_OVERRIDE_EN_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_LC_N_FTS_CNTL, data);
+
+	orig = data = RREG32_PCIE(ixPCIE_LC_CNTL3);
+	data |= PCIE_LC_CNTL3__LC_GO_TO_RECOVERY_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_LC_CNTL3, data);
+
+	orig = data = RREG32_PCIE(ixPCIE_P_CNTL);
+	data |= PCIE_P_CNTL__P_IGNORE_EDB_ERR_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_P_CNTL, data);
+
+	orig = data = RREG32_PCIE(ixPCIE_LC_CNTL);
+	data &= ~(PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK |
+		PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK);
+	data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	if (!disable_l0s)
+		data |= (7 << PCIE_LC_CNTL__LC_L0S_INACTIVITY__SHIFT);
+
+	if (!disable_l1) {
+		data |= (7 << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT);
+		data &= ~PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+		if (orig != data)
+			WREG32_PCIE(ixPCIE_LC_CNTL, data);
+
+		if (!disable_plloff_in_l1) {
+			bool clk_req_support;
+
+			orig = data = RREG32_PCIE(ixPB0_PIF_PWRDOWN_0);
+			data &= ~(PB0_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_OFF_0_MASK |
+				PB0_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_TXS2_0_MASK);
+			data |= (7 << PB0_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_OFF_0__SHIFT) |
+				(7 << PB0_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_TXS2_0__SHIFT);
+			if (orig != data)
+				WREG32_PCIE(ixPB0_PIF_PWRDOWN_0, data);
+
+			orig = data = RREG32_PCIE(ixPB0_PIF_PWRDOWN_1);
+			data &= ~(PB0_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_OFF_1_MASK |
+				PB0_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_TXS2_1_MASK);
+			data |= (7 << PB0_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_OFF_1__SHIFT) |
+				(7 << PB0_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_TXS2_1__SHIFT);
+			if (orig != data)
+				WREG32_PCIE(ixPB0_PIF_PWRDOWN_1, data);
+
+			orig = data = RREG32_PCIE(ixPB1_PIF_PWRDOWN_0);
+			data &= ~(PB1_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_OFF_0_MASK |
+				PB1_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_TXS2_0_MASK);
+			data |= (7 << PB1_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_OFF_0__SHIFT) |
+				(7 << PB1_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_TXS2_0__SHIFT);
+			if (orig != data)
+				WREG32_PCIE(ixPB1_PIF_PWRDOWN_0, data);
+
+			orig = data = RREG32_PCIE(ixPB1_PIF_PWRDOWN_1);
+			data &= ~(PB1_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_OFF_1_MASK |
+				PB1_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_TXS2_1_MASK);
+			data |= (7 << PB1_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_OFF_1__SHIFT) |
+				(7 << PB1_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_TXS2_1__SHIFT);
+			if (orig != data)
+				WREG32_PCIE(ixPB1_PIF_PWRDOWN_1, data);
+
+			orig = data = RREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL);
+			data &= ~PCIE_LC_LINK_WIDTH_CNTL__LC_DYN_LANES_PWR_STATE_MASK;
+			data |= ~(3 << PCIE_LC_LINK_WIDTH_CNTL__LC_DYN_LANES_PWR_STATE__SHIFT);
+			if (orig != data)
+				WREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL, data);
+
+			if (!disable_clkreq) {
+				struct pci_dev *root = adev->pdev->bus->self;
+				u32 lnkcap;
+
+				clk_req_support = false;
+				pcie_capability_read_dword(root, PCI_EXP_LNKCAP, &lnkcap);
+				if (lnkcap & PCI_EXP_LNKCAP_CLKPM)
+					clk_req_support = true;
+			} else {
+				clk_req_support = false;
+			}
+
+			if (clk_req_support) {
+				orig = data = RREG32_PCIE(ixPCIE_LC_CNTL2);
+				data |= PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK |
+					PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L23_MASK;
+				if (orig != data)
+					WREG32_PCIE(ixPCIE_LC_CNTL2, data);
+
+				orig = data = RREG32_SMC(ixTHM_CLK_CNTL);
+				data &= ~(THM_CLK_CNTL__CMON_CLK_SEL_MASK |
+					THM_CLK_CNTL__TMON_CLK_SEL_MASK);
+				data |= (1 << THM_CLK_CNTL__CMON_CLK_SEL__SHIFT) |
+					(1 << THM_CLK_CNTL__TMON_CLK_SEL__SHIFT);
+				if (orig != data)
+					WREG32_SMC(ixTHM_CLK_CNTL, data);
+
+				orig = data = RREG32_SMC(ixMISC_CLK_CTRL);
+				data &= ~(MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL_MASK |
+					MISC_CLK_CTRL__ZCLK_SEL_MASK);
+				data |= (1 << MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL__SHIFT) |
+					(1 << MISC_CLK_CTRL__ZCLK_SEL__SHIFT);
+				if (orig != data)
+					WREG32_SMC(ixMISC_CLK_CTRL, data);
+
+				orig = data = RREG32_SMC(ixCG_CLKPIN_CNTL);
+				data &= ~CG_CLKPIN_CNTL__BCLK_AS_XCLK_MASK;
+				if (orig != data)
+					WREG32_SMC(ixCG_CLKPIN_CNTL, data);
+
+				orig = data = RREG32_SMC(ixCG_CLKPIN_CNTL_2);
+				data &= ~CG_CLKPIN_CNTL_2__FORCE_BIF_REFCLK_EN_MASK;
+				if (orig != data)
+					WREG32_SMC(ixCG_CLKPIN_CNTL_2, data);
+
+				orig = data = RREG32_SMC(ixMPLL_BYPASSCLK_SEL);
+				data &= ~MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK;
+				data |= (4 << MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT);
+				if (orig != data)
+					WREG32_SMC(ixMPLL_BYPASSCLK_SEL, data);
+			}
+		}
+	} else {
+		if (orig != data)
+			WREG32_PCIE(ixPCIE_LC_CNTL, data);
+	}
+
+	orig = data = RREG32_PCIE(ixPCIE_CNTL2);
+	data |= PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+		PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+		PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_CNTL2, data);
+
+	if (!disable_l0s) {
+		data = RREG32_PCIE(ixPCIE_LC_N_FTS_CNTL);
+		if ((data & PCIE_LC_N_FTS_CNTL__LC_N_FTS_MASK) ==
+				PCIE_LC_N_FTS_CNTL__LC_N_FTS_MASK) {
+			data = RREG32_PCIE(ixPCIE_LC_STATUS1);
+			if ((data & PCIE_LC_STATUS1__LC_REVERSE_XMIT_MASK) &&
+			(data & PCIE_LC_STATUS1__LC_REVERSE_RCVR_MASK)) {
+				orig = data = RREG32_PCIE(ixPCIE_LC_CNTL);
+				data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
+				if (orig != data)
+					WREG32_PCIE(ixPCIE_LC_CNTL, data);
+			}
+		}
+	}
+}
+
+static uint32_t cik_get_rev_id(struct amdgpu_device *adev)
+{
+	return (RREG32(mmCC_DRM_ID_STRAPS) & CC_DRM_ID_STRAPS__ATI_REV_ID_MASK)
+		>> CC_DRM_ID_STRAPS__ATI_REV_ID__SHIFT;
+}
+
+static void cik_detect_hw_virtualization(struct amdgpu_device *adev)
+{
+	if (is_virtual_machine()) /* passthrough mode */
+		adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
+}
+
+static void cik_flush_hdp(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 1);
+		RREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL);
+	} else {
+		amdgpu_ring_emit_wreg(ring, mmHDP_MEM_COHERENCY_FLUSH_CNTL, 1);
+	}
+}
+
+static void cik_invalidate_hdp(struct amdgpu_device *adev,
+			       struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32(mmHDP_DEBUG0, 1);
+		RREG32(mmHDP_DEBUG0);
+	} else {
+		amdgpu_ring_emit_wreg(ring, mmHDP_DEBUG0, 1);
+	}
+}
+
+static bool cik_need_full_reset(struct amdgpu_device *adev)
+{
+	/* change this when we support soft reset */
+	return true;
+}
+
+static const struct amdgpu_asic_funcs cik_asic_funcs =
+{
+	.read_disabled_bios = &cik_read_disabled_bios,
+	.read_bios_from_rom = &cik_read_bios_from_rom,
+	.read_register = &cik_read_register,
+	.reset = &cik_asic_reset,
+	.set_vga_state = &cik_vga_set_state,
+	.get_xclk = &cik_get_xclk,
+	.set_uvd_clocks = &cik_set_uvd_clocks,
+	.set_vce_clocks = &cik_set_vce_clocks,
+	.get_config_memsize = &cik_get_config_memsize,
+	.flush_hdp = &cik_flush_hdp,
+	.invalidate_hdp = &cik_invalidate_hdp,
+	.need_full_reset = &cik_need_full_reset,
+};
+
+static int cik_common_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->smc_rreg = &cik_smc_rreg;
+	adev->smc_wreg = &cik_smc_wreg;
+	adev->pcie_rreg = &cik_pcie_rreg;
+	adev->pcie_wreg = &cik_pcie_wreg;
+	adev->uvd_ctx_rreg = &cik_uvd_ctx_rreg;
+	adev->uvd_ctx_wreg = &cik_uvd_ctx_wreg;
+	adev->didt_rreg = &cik_didt_rreg;
+	adev->didt_wreg = &cik_didt_wreg;
+
+	adev->asic_funcs = &cik_asic_funcs;
+
+	adev->rev_id = cik_get_rev_id(adev);
+	adev->external_rev_id = 0xFF;
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			/*AMD_CG_SUPPORT_GFX_CGCG |*/
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_CGTS_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_VCE_MGCG |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x14;
+		break;
+	case CHIP_HAWAII:
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			/*AMD_CG_SUPPORT_GFX_CGCG |*/
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_VCE_MGCG |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = 0x28;
+		break;
+	case CHIP_KAVERI:
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			/*AMD_CG_SUPPORT_GFX_CGCG |*/
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_CGTS_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_VCE_MGCG |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_MGCG;
+		adev->pg_flags =
+			/*AMD_PG_SUPPORT_GFX_PG |
+			  AMD_PG_SUPPORT_GFX_SMG |
+			  AMD_PG_SUPPORT_GFX_DMG |*/
+			AMD_PG_SUPPORT_UVD |
+			AMD_PG_SUPPORT_VCE |
+			/*  AMD_PG_SUPPORT_CP |
+			  AMD_PG_SUPPORT_GDS |
+			  AMD_PG_SUPPORT_RLC_SMU_HS |
+			  AMD_PG_SUPPORT_ACP |
+			  AMD_PG_SUPPORT_SAMU |*/
+			0;
+		if (adev->pdev->device == 0x1312 ||
+			adev->pdev->device == 0x1316 ||
+			adev->pdev->device == 0x1317)
+			adev->external_rev_id = 0x41;
+		else
+			adev->external_rev_id = 0x1;
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			/*AMD_CG_SUPPORT_GFX_CGCG |*/
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_CGTS_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_VCE_MGCG |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_MGCG;
+		adev->pg_flags =
+			/*AMD_PG_SUPPORT_GFX_PG |
+			  AMD_PG_SUPPORT_GFX_SMG | */
+			AMD_PG_SUPPORT_UVD |
+			/*AMD_PG_SUPPORT_VCE |
+			  AMD_PG_SUPPORT_CP |
+			  AMD_PG_SUPPORT_GDS |
+			  AMD_PG_SUPPORT_RLC_SMU_HS |
+			  AMD_PG_SUPPORT_SAMU |*/
+			0;
+		if (adev->asic_type == CHIP_KABINI) {
+			if (adev->rev_id == 0)
+				adev->external_rev_id = 0x81;
+			else if (adev->rev_id == 1)
+				adev->external_rev_id = 0x82;
+			else if (adev->rev_id == 2)
+				adev->external_rev_id = 0x85;
+		} else
+			adev->external_rev_id = adev->rev_id + 0xa1;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int cik_common_sw_init(void *handle)
+{
+	return 0;
+}
+
+static int cik_common_sw_fini(void *handle)
+{
+	return 0;
+}
+
+static int cik_common_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* move the golden regs per IP block */
+	cik_init_golden_registers(adev);
+	/* enable pcie gen2/3 link */
+	cik_pcie_gen3_enable(adev);
+	/* enable aspm */
+	cik_program_aspm(adev);
+
+	return 0;
+}
+
+static int cik_common_hw_fini(void *handle)
+{
+	return 0;
+}
+
+static int cik_common_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cik_common_hw_fini(adev);
+}
+
+static int cik_common_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cik_common_hw_init(adev);
+}
+
+static bool cik_common_is_idle(void *handle)
+{
+	return true;
+}
+
+static int cik_common_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static int cik_common_soft_reset(void *handle)
+{
+	/* XXX hard reset?? */
+	return 0;
+}
+
+static int cik_common_set_clockgating_state(void *handle,
+					    enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int cik_common_set_powergating_state(void *handle,
+					    enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs cik_common_ip_funcs = {
+	.name = "cik_common",
+	.early_init = cik_common_early_init,
+	.late_init = NULL,
+	.sw_init = cik_common_sw_init,
+	.sw_fini = cik_common_sw_fini,
+	.hw_init = cik_common_hw_init,
+	.hw_fini = cik_common_hw_fini,
+	.suspend = cik_common_suspend,
+	.resume = cik_common_resume,
+	.is_idle = cik_common_is_idle,
+	.wait_for_idle = cik_common_wait_for_idle,
+	.soft_reset = cik_common_soft_reset,
+	.set_clockgating_state = cik_common_set_clockgating_state,
+	.set_powergating_state = cik_common_set_powergating_state,
+};
+
+static const struct amdgpu_ip_block_version cik_common_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_COMMON,
+	.major = 1,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &cik_common_ip_funcs,
+};
+
+int cik_set_ip_blocks(struct amdgpu_device *adev)
+{
+	cik_detect_hw_virtualization(adev);
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		amdgpu_device_ip_block_add(adev, &cik_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v7_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_ih_ip_block);
+		if (amdgpu_dpm == -1)
+			amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		else
+			amdgpu_device_ip_block_add(adev, &ci_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
+		else
+			amdgpu_device_ip_block_add(adev, &dce_v8_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v7_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_sdma_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v4_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v2_0_ip_block);
+		break;
+	case CHIP_HAWAII:
+		amdgpu_device_ip_block_add(adev, &cik_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v7_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_ih_ip_block);
+		if (amdgpu_dpm == -1)
+			amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		else
+			amdgpu_device_ip_block_add(adev, &ci_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
+		else
+			amdgpu_device_ip_block_add(adev, &dce_v8_5_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v7_3_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_sdma_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v4_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v2_0_ip_block);
+		break;
+	case CHIP_KAVERI:
+		amdgpu_device_ip_block_add(adev, &cik_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v7_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &kv_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
+		else
+			amdgpu_device_ip_block_add(adev, &dce_v8_1_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v7_1_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_sdma_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v4_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v2_0_ip_block);
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		amdgpu_device_ip_block_add(adev, &cik_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v7_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &kv_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
+		else
+			amdgpu_device_ip_block_add(adev, &dce_v8_3_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v7_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &cik_sdma_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v4_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v2_0_ip_block);
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+	return 0;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_cik_sdma.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_cik_sdma.c
new file mode 100644
index 00000000000..d61e18decab
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_cik_sdma.c
@@ -0,0 +1,1408 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_trace.h"
+#include "cikd.h"
+#include "cik.h"
+
+#include "bif/bif_4_1_d.h"
+#include "bif/bif_4_1_sh_mask.h"
+
+#include "gca/gfx_7_2_d.h"
+#include "gca/gfx_7_2_enum.h"
+#include "gca/gfx_7_2_sh_mask.h"
+
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+static const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
+{
+	SDMA0_REGISTER_OFFSET,
+	SDMA1_REGISTER_OFFSET
+};
+
+static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev);
+static void cik_sdma_set_irq_funcs(struct amdgpu_device *adev);
+static void cik_sdma_set_buffer_funcs(struct amdgpu_device *adev);
+static void cik_sdma_set_vm_pte_funcs(struct amdgpu_device *adev);
+static int cik_sdma_soft_reset(void *handle);
+
+MODULE_FIRMWARE("amdgpu/bonaire_sdma.bin");
+MODULE_FIRMWARE("amdgpu/bonaire_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_sdma.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_sdma.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/kabini_sdma.bin");
+MODULE_FIRMWARE("amdgpu/kabini_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/mullins_sdma.bin");
+MODULE_FIRMWARE("amdgpu/mullins_sdma1.bin");
+
+u32 amdgpu_cik_gpu_check_soft_reset(struct amdgpu_device *adev);
+
+
+static void cik_sdma_free_microcode(struct amdgpu_device *adev)
+{
+	int i;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+			release_firmware(adev->sdma.instance[i].fw);
+			adev->sdma.instance[i].fw = NULL;
+	}
+}
+
+/*
+ * sDMA - System DMA
+ * Starting with CIK, the GPU has new asynchronous
+ * DMA engines.  These engines are used for compute
+ * and gfx.  There are two DMA engines (SDMA0, SDMA1)
+ * and each one supports 1 ring buffer used for gfx
+ * and 2 queues used for compute.
+ *
+ * The programming model is very similar to the CP
+ * (ring buffer, IBs, etc.), but sDMA has it's own
+ * packet format that is different from the PM4 format
+ * used by the CP. sDMA supports copying data, writing
+ * embedded data, solid fills, and a number of other
+ * things.  It also has support for tiling/detiling of
+ * buffers.
+ */
+
+/**
+ * cik_sdma_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int cik_sdma_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err = 0, i;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		chip_name = "bonaire";
+		break;
+	case CHIP_HAWAII:
+		chip_name = "hawaii";
+		break;
+	case CHIP_KAVERI:
+		chip_name = "kaveri";
+		break;
+	case CHIP_KABINI:
+		chip_name = "kabini";
+		break;
+	case CHIP_MULLINS:
+		chip_name = "mullins";
+		break;
+	default: BUG();
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (i == 0)
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma.bin", chip_name);
+		else
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma1.bin", chip_name);
+		err = request_firmware(&adev->sdma.instance[i].fw, fw_name, adev->dev);
+		if (err)
+			goto out;
+		err = amdgpu_ucode_validate(adev->sdma.instance[i].fw);
+	}
+out:
+	if (err) {
+		pr_err("cik_sdma: Failed to load firmware \"%s\"\n", fw_name);
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			release_firmware(adev->sdma.instance[i].fw);
+			adev->sdma.instance[i].fw = NULL;
+		}
+	}
+	return err;
+}
+
+/**
+ * cik_sdma_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware (CIK+).
+ */
+static uint64_t cik_sdma_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	u32 rptr;
+
+	rptr = ring->adev->wb.wb[ring->rptr_offs];
+
+	return (rptr & 0x3fffc) >> 2;
+}
+
+/**
+ * cik_sdma_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware (CIK+).
+ */
+static uint64_t cik_sdma_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return (RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) & 0x3fffc) >> 2;
+}
+
+/**
+ * cik_sdma_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware (CIK+).
+ */
+static void cik_sdma_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],
+		       	(lower_32_bits(ring->wptr) << 2) & 0x3fffc);
+}
+
+static void cik_sdma_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			amdgpu_ring_write(ring, ring->funcs->nop |
+					  SDMA_NOP_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->funcs->nop);
+}
+
+/**
+ * cik_sdma_ring_emit_ib - Schedule an IB on the DMA engine
+ *
+ * @ring: amdgpu ring pointer
+ * @ib: IB object to schedule
+ *
+ * Schedule an IB in the DMA ring (CIK).
+ */
+static void cik_sdma_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib,
+				  unsigned vmid, bool ctx_switch)
+{
+	u32 extra_bits = vmid & 0xf;
+
+	/* IB packet must end on a 8 DW boundary */
+	cik_sdma_ring_insert_nop(ring, (12 - (lower_32_bits(ring->wptr) & 7)) % 8);
+
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_INDIRECT_BUFFER, 0, extra_bits));
+	amdgpu_ring_write(ring, ib->gpu_addr & 0xffffffe0); /* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xffffffff);
+	amdgpu_ring_write(ring, ib->length_dw);
+
+}
+
+/**
+ * cik_sdma_ring_emit_hdp_flush - emit an hdp flush on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Emit an hdp flush packet on the requested DMA ring.
+ */
+static void cik_sdma_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	u32 extra_bits = (SDMA_POLL_REG_MEM_EXTRA_OP(1) |
+			  SDMA_POLL_REG_MEM_EXTRA_FUNC(3)); /* == */
+	u32 ref_and_mask;
+
+	if (ring->me == 0)
+		ref_and_mask = GPU_HDP_FLUSH_DONE__SDMA0_MASK;
+	else
+		ref_and_mask = GPU_HDP_FLUSH_DONE__SDMA1_MASK;
+
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_POLL_REG_MEM, 0, extra_bits));
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_DONE << 2);
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_REQ << 2);
+	amdgpu_ring_write(ring, ref_and_mask); /* reference */
+	amdgpu_ring_write(ring, ref_and_mask); /* mask */
+	amdgpu_ring_write(ring, (0xfff << 16) | 10); /* retry count, poll interval */
+}
+
+/**
+ * cik_sdma_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @fence: amdgpu fence object
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed (CIK).
+ */
+static void cik_sdma_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				     unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_FENCE, 0, 0));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_FENCE, 0, 0));
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+
+	/* generate an interrupt */
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_TRAP, 0, 0));
+}
+
+/**
+ * cik_sdma_gfx_stop - stop the gfx async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the gfx async dma ring buffers (CIK).
+ */
+static void cik_sdma_gfx_stop(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *sdma0 = &adev->sdma.instance[0].ring;
+	struct amdgpu_ring *sdma1 = &adev->sdma.instance[1].ring;
+	u32 rb_cntl;
+	int i;
+
+	if ((adev->mman.buffer_funcs_ring == sdma0) ||
+	    (adev->mman.buffer_funcs_ring == sdma1))
+			amdgpu_ttm_set_buffer_funcs_status(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
+		rb_cntl &= ~SDMA0_GFX_RB_CNTL__RB_ENABLE_MASK;
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], 0);
+	}
+	sdma0->ready = false;
+	sdma1->ready = false;
+}
+
+/**
+ * cik_sdma_rlc_stop - stop the compute async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute async dma queues (CIK).
+ */
+static void cik_sdma_rlc_stop(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+}
+
+/**
+ * cik_ctx_switch_enable - stop the async dma engines context switch
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs context switch.
+ *
+ * Halt or unhalt the async dma engines context switch (VI).
+ */
+static void cik_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl, phase_quantum = 0;
+	int i;
+
+	if (amdgpu_sdma_phase_quantum) {
+		unsigned value = amdgpu_sdma_phase_quantum;
+		unsigned unit = 0;
+
+		while (value > (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
+				SDMA0_PHASE0_QUANTUM__VALUE__SHIFT)) {
+			value = (value + 1) >> 1;
+			unit++;
+		}
+		if (unit > (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
+			    SDMA0_PHASE0_QUANTUM__UNIT__SHIFT)) {
+			value = (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
+				 SDMA0_PHASE0_QUANTUM__VALUE__SHIFT);
+			unit = (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
+				SDMA0_PHASE0_QUANTUM__UNIT__SHIFT);
+			WARN_ONCE(1,
+			"clamping sdma_phase_quantum to %uK clock cycles\n",
+				  value << unit);
+		}
+		phase_quantum =
+			value << SDMA0_PHASE0_QUANTUM__VALUE__SHIFT |
+			unit  << SDMA0_PHASE0_QUANTUM__UNIT__SHIFT;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		f32_cntl = RREG32(mmSDMA0_CNTL + sdma_offsets[i]);
+		if (enable) {
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+					AUTO_CTXSW_ENABLE, 1);
+			if (amdgpu_sdma_phase_quantum) {
+				WREG32(mmSDMA0_PHASE0_QUANTUM + sdma_offsets[i],
+				       phase_quantum);
+				WREG32(mmSDMA0_PHASE1_QUANTUM + sdma_offsets[i],
+				       phase_quantum);
+			}
+		} else {
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+					AUTO_CTXSW_ENABLE, 0);
+		}
+
+		WREG32(mmSDMA0_CNTL + sdma_offsets[i], f32_cntl);
+	}
+}
+
+/**
+ * cik_sdma_enable - stop the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs.
+ *
+ * Halt or unhalt the async dma engines (CIK).
+ */
+static void cik_sdma_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 me_cntl;
+	int i;
+
+	if (!enable) {
+		cik_sdma_gfx_stop(adev);
+		cik_sdma_rlc_stop(adev);
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		me_cntl = RREG32(mmSDMA0_F32_CNTL + sdma_offsets[i]);
+		if (enable)
+			me_cntl &= ~SDMA0_F32_CNTL__HALT_MASK;
+		else
+			me_cntl |= SDMA0_F32_CNTL__HALT_MASK;
+		WREG32(mmSDMA0_F32_CNTL + sdma_offsets[i], me_cntl);
+	}
+}
+
+/**
+ * cik_sdma_gfx_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the gfx DMA ring buffers and enable them (CIK).
+ * Returns 0 for success, error for failure.
+ */
+static int cik_sdma_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl, ib_cntl;
+	u32 rb_bufsz;
+	u32 wb_offset;
+	int i, j, r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		wb_offset = (ring->rptr_offs * 4);
+
+		mutex_lock(&adev->srbm_mutex);
+		for (j = 0; j < 16; j++) {
+			cik_srbm_select(adev, 0, 0, 0, j);
+			/* SDMA GFX */
+			WREG32(mmSDMA0_GFX_VIRTUAL_ADDR + sdma_offsets[i], 0);
+			WREG32(mmSDMA0_GFX_APE1_CNTL + sdma_offsets[i], 0);
+			/* XXX SDMA RLC - todo */
+		}
+		cik_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		WREG32(mmSDMA0_TILING_CONFIG + sdma_offsets[i],
+		       adev->gfx.config.gb_addr_config & 0x70);
+
+		WREG32(mmSDMA0_SEM_INCOMPLETE_TIMER_CNTL + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i], 0);
+
+		/* Set ring buffer size in dwords */
+		rb_bufsz = order_base_2(ring->ring_size / 4);
+		rb_cntl = rb_bufsz << 1;
+#ifdef __BIG_ENDIAN
+		rb_cntl |= SDMA0_GFX_RB_CNTL__RB_SWAP_ENABLE_MASK |
+			SDMA0_GFX_RB_CNTL__RPTR_WRITEBACK_SWAP_ENABLE_MASK;
+#endif
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+
+		/* Initialize the ring buffer's read and write pointers */
+		WREG32(mmSDMA0_GFX_RB_RPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_IB_RPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_IB_OFFSET + sdma_offsets[i], 0);
+
+		/* set the wb address whether it's enabled or not */
+		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i],
+		       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
+		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i],
+		       ((adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC));
+
+		rb_cntl |= SDMA0_GFX_RB_CNTL__RPTR_WRITEBACK_ENABLE_MASK;
+
+		WREG32(mmSDMA0_GFX_RB_BASE + sdma_offsets[i], ring->gpu_addr >> 8);
+		WREG32(mmSDMA0_GFX_RB_BASE_HI + sdma_offsets[i], ring->gpu_addr >> 40);
+
+		ring->wptr = 0;
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], lower_32_bits(ring->wptr) << 2);
+
+		/* enable DMA RB */
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i],
+		       rb_cntl | SDMA0_GFX_RB_CNTL__RB_ENABLE_MASK);
+
+		ib_cntl = SDMA0_GFX_IB_CNTL__IB_ENABLE_MASK;
+#ifdef __BIG_ENDIAN
+		ib_cntl |= SDMA0_GFX_IB_CNTL__IB_SWAP_ENABLE_MASK;
+#endif
+		/* enable DMA IBs */
+		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
+
+		ring->ready = true;
+	}
+
+	cik_sdma_enable(adev, true);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		r = amdgpu_ring_test_ring(ring);
+		if (r) {
+			ring->ready = false;
+			return r;
+		}
+
+		if (adev->mman.buffer_funcs_ring == ring)
+			amdgpu_ttm_set_buffer_funcs_status(adev, true);
+	}
+
+	return 0;
+}
+
+/**
+ * cik_sdma_rlc_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the compute DMA queues and enable them (CIK).
+ * Returns 0 for success, error for failure.
+ */
+static int cik_sdma_rlc_resume(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+	return 0;
+}
+
+/**
+ * cik_sdma_load_microcode - load the sDMA ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the sDMA0/1 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int cik_sdma_load_microcode(struct amdgpu_device *adev)
+{
+	const struct sdma_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+	int i, j;
+
+	/* halt the MEs */
+	cik_sdma_enable(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (!adev->sdma.instance[i].fw)
+			return -EINVAL;
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		amdgpu_ucode_print_sdma_hdr(&hdr->header);
+		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+		adev->sdma.instance[i].fw_version = le32_to_cpu(hdr->header.ucode_version);
+		adev->sdma.instance[i].feature_version = le32_to_cpu(hdr->ucode_feature_version);
+		if (adev->sdma.instance[i].feature_version >= 20)
+			adev->sdma.instance[i].burst_nop = true;
+		fw_data = (const __le32 *)
+			(adev->sdma.instance[i].fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], 0);
+		for (j = 0; j < fw_size; j++)
+			WREG32(mmSDMA0_UCODE_DATA + sdma_offsets[i], le32_to_cpup(fw_data++));
+		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], adev->sdma.instance[i].fw_version);
+	}
+
+	return 0;
+}
+
+/**
+ * cik_sdma_start - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the DMA engines and enable them (CIK).
+ * Returns 0 for success, error for failure.
+ */
+static int cik_sdma_start(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = cik_sdma_load_microcode(adev);
+	if (r)
+		return r;
+
+	/* halt the engine before programing */
+	cik_sdma_enable(adev, false);
+	/* enable sdma ring preemption */
+	cik_ctx_switch_enable(adev, true);
+
+	/* start the gfx rings and rlc compute queues */
+	r = cik_sdma_gfx_resume(adev);
+	if (r)
+		return r;
+	r = cik_sdma_rlc_resume(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * cik_sdma_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory. (CIK).
+ * Returns 0 for success, error for failure.
+ */
+static int cik_sdma_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_alloc(ring, 5);
+	if (r) {
+		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_device_wb_free(adev, index);
+		return r;
+	}
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, 1); /* number of DWs to follow */
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_device_wb_free(adev, index);
+
+	return r;
+}
+
+/**
+ * cik_sdma_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test a simple IB in the DMA ring (CIK).
+ * Returns 0 on success, error on failure.
+ */
+static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned index;
+	u32 tmp = 0;
+	u64 gpu_addr;
+	long r;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, 256, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err0;
+	}
+
+	ib.ptr[0] = SDMA_PACKET(SDMA_OPCODE_WRITE,
+				SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr);
+	ib.ptr[3] = 1;
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.length_dw = 5;
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err1;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
+		goto err1;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto err1;
+	}
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+	if (tmp == 0xDEADBEEF) {
+		DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
+	} else {
+		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+		r = -EINVAL;
+	}
+
+err1:
+	amdgpu_ib_free(adev, &ib, NULL);
+	dma_fence_put(f);
+err0:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+/**
+ * cik_sdma_vm_copy_pages - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using sDMA (CIK).
+ */
+static void cik_sdma_vm_copy_pte(struct amdgpu_ib *ib,
+				 uint64_t pe, uint64_t src,
+				 unsigned count)
+{
+	unsigned bytes = count * 8;
+
+	ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_COPY,
+		SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
+	ib->ptr[ib->length_dw++] = bytes;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+}
+
+/**
+ * cik_sdma_vm_write_pages - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @value: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ *
+ * Update PTEs by writing them manually using sDMA (CIK).
+ */
+static void cik_sdma_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
+				  uint64_t value, unsigned count,
+				  uint32_t incr)
+{
+	unsigned ndw = count * 2;
+
+	ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_WRITE,
+		SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = ndw;
+	for (; ndw > 0; ndw -= 2) {
+		ib->ptr[ib->length_dw++] = lower_32_bits(value);
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		value += incr;
+	}
+}
+
+/**
+ * cik_sdma_vm_set_pages - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA (CIK).
+ */
+static void cik_sdma_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe,
+				    uint64_t addr, unsigned count,
+				    uint32_t incr, uint64_t flags)
+{
+	/* for physically contiguous pages (vram) */
+	ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_GENERATE_PTE_PDE, 0, 0);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = lower_32_bits(flags); /* mask */
+	ib->ptr[ib->length_dw++] = upper_32_bits(flags);
+	ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = incr; /* increment size */
+	ib->ptr[ib->length_dw++] = 0;
+	ib->ptr[ib->length_dw++] = count; /* number of entries */
+}
+
+/**
+ * cik_sdma_vm_pad_ib - pad the IB to the required number of dw
+ *
+ * @ib: indirect buffer to fill with padding
+ *
+ */
+static void cik_sdma_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	u32 pad_count;
+	int i;
+
+	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	for (i = 0; i < pad_count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			ib->ptr[ib->length_dw++] =
+					SDMA_PACKET(SDMA_OPCODE_NOP, 0, 0) |
+					SDMA_NOP_COUNT(pad_count - 1);
+		else
+			ib->ptr[ib->length_dw++] =
+					SDMA_PACKET(SDMA_OPCODE_NOP, 0, 0);
+}
+
+/**
+ * cik_sdma_ring_emit_pipeline_sync - sync the pipeline
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Make sure all previous operations are completed (CIK).
+ */
+static void cik_sdma_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	/* wait for idle */
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_POLL_REG_MEM, 0,
+					    SDMA_POLL_REG_MEM_EXTRA_OP(0) |
+					    SDMA_POLL_REG_MEM_EXTRA_FUNC(3) | /* equal */
+					    SDMA_POLL_REG_MEM_EXTRA_M));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+	amdgpu_ring_write(ring, seq); /* reference */
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
+	amdgpu_ring_write(ring, (0xfff << 16) | 4); /* retry count, poll interval */
+}
+
+/**
+ * cik_sdma_ring_emit_vm_flush - cik vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vm: amdgpu_vm pointer
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA (CIK).
+ */
+static void cik_sdma_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned vmid, uint64_t pd_addr)
+{
+	u32 extra_bits = (SDMA_POLL_REG_MEM_EXTRA_OP(0) |
+			  SDMA_POLL_REG_MEM_EXTRA_FUNC(0)); /* always */
+
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_POLL_REG_MEM, 0, extra_bits));
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST << 2);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0); /* reference */
+	amdgpu_ring_write(ring, 0); /* mask */
+	amdgpu_ring_write(ring, (0xfff << 16) | 10); /* retry count, poll interval */
+}
+
+static void cik_sdma_ring_emit_wreg(struct amdgpu_ring *ring,
+				    uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, val);
+}
+
+static void cik_enable_sdma_mgcg(struct amdgpu_device *adev,
+				 bool enable)
+{
+	u32 orig, data;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_MGCG)) {
+		WREG32(mmSDMA0_CLK_CTRL + SDMA0_REGISTER_OFFSET, 0x00000100);
+		WREG32(mmSDMA0_CLK_CTRL + SDMA1_REGISTER_OFFSET, 0x00000100);
+	} else {
+		orig = data = RREG32(mmSDMA0_CLK_CTRL + SDMA0_REGISTER_OFFSET);
+		data |= 0xff000000;
+		if (data != orig)
+			WREG32(mmSDMA0_CLK_CTRL + SDMA0_REGISTER_OFFSET, data);
+
+		orig = data = RREG32(mmSDMA0_CLK_CTRL + SDMA1_REGISTER_OFFSET);
+		data |= 0xff000000;
+		if (data != orig)
+			WREG32(mmSDMA0_CLK_CTRL + SDMA1_REGISTER_OFFSET, data);
+	}
+}
+
+static void cik_enable_sdma_mgls(struct amdgpu_device *adev,
+				 bool enable)
+{
+	u32 orig, data;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_LS)) {
+		orig = data = RREG32(mmSDMA0_POWER_CNTL + SDMA0_REGISTER_OFFSET);
+		data |= 0x100;
+		if (orig != data)
+			WREG32(mmSDMA0_POWER_CNTL + SDMA0_REGISTER_OFFSET, data);
+
+		orig = data = RREG32(mmSDMA0_POWER_CNTL + SDMA1_REGISTER_OFFSET);
+		data |= 0x100;
+		if (orig != data)
+			WREG32(mmSDMA0_POWER_CNTL + SDMA1_REGISTER_OFFSET, data);
+	} else {
+		orig = data = RREG32(mmSDMA0_POWER_CNTL + SDMA0_REGISTER_OFFSET);
+		data &= ~0x100;
+		if (orig != data)
+			WREG32(mmSDMA0_POWER_CNTL + SDMA0_REGISTER_OFFSET, data);
+
+		orig = data = RREG32(mmSDMA0_POWER_CNTL + SDMA1_REGISTER_OFFSET);
+		data &= ~0x100;
+		if (orig != data)
+			WREG32(mmSDMA0_POWER_CNTL + SDMA1_REGISTER_OFFSET, data);
+	}
+}
+
+static int cik_sdma_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->sdma.num_instances = SDMA_MAX_INSTANCE;
+
+	cik_sdma_set_ring_funcs(adev);
+	cik_sdma_set_irq_funcs(adev);
+	cik_sdma_set_buffer_funcs(adev);
+	cik_sdma_set_vm_pte_funcs(adev);
+
+	return 0;
+}
+
+static int cik_sdma_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r, i;
+
+	r = cik_sdma_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load sdma firmware!\n");
+		return r;
+	}
+
+	/* SDMA trap event */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 224,
+			      &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	/* SDMA Privileged inst */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 241,
+			      &adev->sdma.illegal_inst_irq);
+	if (r)
+		return r;
+
+	/* SDMA Privileged inst */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 247,
+			      &adev->sdma.illegal_inst_irq);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		snprintf(ring->name, sizeof(ring->name), "sdma%d", i);
+		r = amdgpu_ring_init(adev, ring, 1024,
+				     &adev->sdma.trap_irq,
+				     (i == 0) ?
+				     AMDGPU_SDMA_IRQ_TRAP0 :
+				     AMDGPU_SDMA_IRQ_TRAP1);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+static int cik_sdma_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+
+	cik_sdma_free_microcode(adev);
+	return 0;
+}
+
+static int cik_sdma_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = cik_sdma_start(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int cik_sdma_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	cik_ctx_switch_enable(adev, false);
+	cik_sdma_enable(adev, false);
+
+	return 0;
+}
+
+static int cik_sdma_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cik_sdma_hw_fini(adev);
+}
+
+static int cik_sdma_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	cik_sdma_soft_reset(handle);
+
+	return cik_sdma_hw_init(adev);
+}
+
+static bool cik_sdma_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS2);
+
+	if (tmp & (SRBM_STATUS2__SDMA_BUSY_MASK |
+				SRBM_STATUS2__SDMA1_BUSY_MASK))
+	    return false;
+
+	return true;
+}
+
+static int cik_sdma_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmSRBM_STATUS2) & (SRBM_STATUS2__SDMA_BUSY_MASK |
+				SRBM_STATUS2__SDMA1_BUSY_MASK);
+
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int cik_sdma_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS2);
+
+	if (tmp & SRBM_STATUS2__SDMA_BUSY_MASK) {
+		/* sdma0 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET);
+		tmp |= SDMA0_F32_CNTL__HALT_MASK;
+		WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp);
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK;
+	}
+	if (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK) {
+		/* sdma1 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET);
+		tmp |= SDMA0_F32_CNTL__HALT_MASK;
+		WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp);
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK;
+	}
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int cik_sdma_set_trap_irq_state(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *src,
+				       unsigned type,
+				       enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	switch (type) {
+	case AMDGPU_SDMA_IRQ_TRAP0:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
+			sdma_cntl &= ~SDMA0_CNTL__TRAP_ENABLE_MASK;
+			WREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
+			sdma_cntl |= SDMA0_CNTL__TRAP_ENABLE_MASK;
+			WREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	case AMDGPU_SDMA_IRQ_TRAP1:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
+			sdma_cntl &= ~SDMA0_CNTL__TRAP_ENABLE_MASK;
+			WREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
+			sdma_cntl |= SDMA0_CNTL__TRAP_ENABLE_MASK;
+			WREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int cik_sdma_process_trap_irq(struct amdgpu_device *adev,
+				     struct amdgpu_irq_src *source,
+				     struct amdgpu_iv_entry *entry)
+{
+	u8 instance_id, queue_id;
+
+	instance_id = (entry->ring_id & 0x3) >> 0;
+	queue_id = (entry->ring_id & 0xc) >> 2;
+	DRM_DEBUG("IH: SDMA trap\n");
+	switch (instance_id) {
+	case 0:
+		switch (queue_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[0].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		}
+		break;
+	case 1:
+		switch (queue_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[1].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		}
+		break;
+	}
+
+	return 0;
+}
+
+static int cik_sdma_process_illegal_inst_irq(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *source,
+					     struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in SDMA command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int cik_sdma_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	bool gate = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_CG_STATE_GATE)
+		gate = true;
+
+	cik_enable_sdma_mgcg(adev, gate);
+	cik_enable_sdma_mgls(adev, gate);
+
+	return 0;
+}
+
+static int cik_sdma_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs cik_sdma_ip_funcs = {
+	.name = "cik_sdma",
+	.early_init = cik_sdma_early_init,
+	.late_init = NULL,
+	.sw_init = cik_sdma_sw_init,
+	.sw_fini = cik_sdma_sw_fini,
+	.hw_init = cik_sdma_hw_init,
+	.hw_fini = cik_sdma_hw_fini,
+	.suspend = cik_sdma_suspend,
+	.resume = cik_sdma_resume,
+	.is_idle = cik_sdma_is_idle,
+	.wait_for_idle = cik_sdma_wait_for_idle,
+	.soft_reset = cik_sdma_soft_reset,
+	.set_clockgating_state = cik_sdma_set_clockgating_state,
+	.set_powergating_state = cik_sdma_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs cik_sdma_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xf,
+	.nop = SDMA_PACKET(SDMA_OPCODE_NOP, 0, 0),
+	.support_64bit_ptrs = false,
+	.get_rptr = cik_sdma_ring_get_rptr,
+	.get_wptr = cik_sdma_ring_get_wptr,
+	.set_wptr = cik_sdma_ring_set_wptr,
+	.emit_frame_size =
+		6 + /* cik_sdma_ring_emit_hdp_flush */
+		3 + /* hdp invalidate */
+		6 + /* cik_sdma_ring_emit_pipeline_sync */
+		CIK_FLUSH_GPU_TLB_NUM_WREG * 3 + 6 + /* cik_sdma_ring_emit_vm_flush */
+		9 + 9 + 9, /* cik_sdma_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 7 + 4, /* cik_sdma_ring_emit_ib */
+	.emit_ib = cik_sdma_ring_emit_ib,
+	.emit_fence = cik_sdma_ring_emit_fence,
+	.emit_pipeline_sync = cik_sdma_ring_emit_pipeline_sync,
+	.emit_vm_flush = cik_sdma_ring_emit_vm_flush,
+	.emit_hdp_flush = cik_sdma_ring_emit_hdp_flush,
+	.test_ring = cik_sdma_ring_test_ring,
+	.test_ib = cik_sdma_ring_test_ib,
+	.insert_nop = cik_sdma_ring_insert_nop,
+	.pad_ib = cik_sdma_ring_pad_ib,
+	.emit_wreg = cik_sdma_ring_emit_wreg,
+};
+
+static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->sdma.instance[i].ring.funcs = &cik_sdma_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs cik_sdma_trap_irq_funcs = {
+	.set = cik_sdma_set_trap_irq_state,
+	.process = cik_sdma_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs cik_sdma_illegal_inst_irq_funcs = {
+	.process = cik_sdma_process_illegal_inst_irq,
+};
+
+static void cik_sdma_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_LAST;
+	adev->sdma.trap_irq.funcs = &cik_sdma_trap_irq_funcs;
+	adev->sdma.illegal_inst_irq.funcs = &cik_sdma_illegal_inst_irq_funcs;
+}
+
+/**
+ * cik_sdma_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Copy GPU buffers using the DMA engine (CIK).
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void cik_sdma_emit_copy_buffer(struct amdgpu_ib *ib,
+				      uint64_t src_offset,
+				      uint64_t dst_offset,
+				      uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_COPY, SDMA_COPY_SUB_OPCODE_LINEAR, 0);
+	ib->ptr[ib->length_dw++] = byte_count;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+}
+
+/**
+ * cik_sdma_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine (CIK).
+ */
+static void cik_sdma_emit_fill_buffer(struct amdgpu_ib *ib,
+				      uint32_t src_data,
+				      uint64_t dst_offset,
+				      uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_CONSTANT_FILL, 0, 0);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = byte_count;
+}
+
+static const struct amdgpu_buffer_funcs cik_sdma_buffer_funcs = {
+	.copy_max_bytes = 0x1fffff,
+	.copy_num_dw = 7,
+	.emit_copy_buffer = cik_sdma_emit_copy_buffer,
+
+	.fill_max_bytes = 0x1fffff,
+	.fill_num_dw = 5,
+	.emit_fill_buffer = cik_sdma_emit_fill_buffer,
+};
+
+static void cik_sdma_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mman.buffer_funcs == NULL) {
+		adev->mman.buffer_funcs = &cik_sdma_buffer_funcs;
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+	}
+}
+
+static const struct amdgpu_vm_pte_funcs cik_sdma_vm_pte_funcs = {
+	.copy_pte_num_dw = 7,
+	.copy_pte = cik_sdma_vm_copy_pte,
+
+	.write_pte = cik_sdma_vm_write_pte,
+	.set_pte_pde = cik_sdma_vm_set_pte_pde,
+};
+
+static void cik_sdma_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	if (adev->vm_manager.vm_pte_funcs == NULL) {
+		adev->vm_manager.vm_pte_funcs = &cik_sdma_vm_pte_funcs;
+		for (i = 0; i < adev->sdma.num_instances; i++)
+			adev->vm_manager.vm_pte_rings[i] =
+				&adev->sdma.instance[i].ring;
+
+		adev->vm_manager.vm_pte_num_rings = adev->sdma.num_instances;
+	}
+}
+
+const struct amdgpu_ip_block_version cik_sdma_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_SDMA,
+	.major = 2,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &cik_sdma_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_connectors.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_connectors.c
new file mode 100644
index 00000000000..914b7be8463
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_connectors.c
@@ -0,0 +1,1881 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <drm/drmP.h>
+#include <drm/drm_edid.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/drm_fb_helper.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "atom.h"
+#include "atombios_encoders.h"
+#include "atombios_dp.h"
+#include "amdgpu_connectors.h"
+#include "amdgpu_i2c.h"
+
+#include <linux/pm_runtime.h>
+
+void amdgpu_connector_hotplug(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	/* bail if the connector does not have hpd pin, e.g.,
+	 * VGA, TV, etc.
+	 */
+	if (amdgpu_connector->hpd.hpd == AMDGPU_HPD_NONE)
+		return;
+
+	amdgpu_display_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
+
+	/* if the connector is already off, don't turn it back on */
+	if (connector->dpms != DRM_MODE_DPMS_ON)
+		return;
+
+	/* just deal with DP (not eDP) here. */
+	if (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort) {
+		struct amdgpu_connector_atom_dig *dig_connector =
+			amdgpu_connector->con_priv;
+
+		/* if existing sink type was not DP no need to retrain */
+		if (dig_connector->dp_sink_type != CONNECTOR_OBJECT_ID_DISPLAYPORT)
+			return;
+
+		/* first get sink type as it may be reset after (un)plug */
+		dig_connector->dp_sink_type = amdgpu_atombios_dp_get_sinktype(amdgpu_connector);
+		/* don't do anything if sink is not display port, i.e.,
+		 * passive dp->(dvi|hdmi) adaptor
+		 */
+		if (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT &&
+		    amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd) &&
+		    amdgpu_atombios_dp_needs_link_train(amdgpu_connector)) {
+			/* Don't start link training before we have the DPCD */
+			if (amdgpu_atombios_dp_get_dpcd(amdgpu_connector))
+				return;
+
+			/* Turn the connector off and back on immediately, which
+			 * will trigger link training
+			 */
+			drm_helper_connector_dpms(connector, DRM_MODE_DPMS_OFF);
+			drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON);
+		}
+	}
+}
+
+static void amdgpu_connector_property_change_mode(struct drm_encoder *encoder)
+{
+	struct drm_crtc *crtc = encoder->crtc;
+
+	if (crtc && crtc->enabled) {
+		drm_crtc_helper_set_mode(crtc, &crtc->mode,
+					 crtc->x, crtc->y, crtc->primary->fb);
+	}
+}
+
+int amdgpu_connector_get_monitor_bpc(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *dig_connector;
+	int bpc = 8;
+	unsigned mode_clock, max_tmds_clock;
+
+	switch (connector->connector_type) {
+	case DRM_MODE_CONNECTOR_DVII:
+	case DRM_MODE_CONNECTOR_HDMIB:
+		if (amdgpu_connector->use_digital) {
+			if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+				if (connector->display_info.bpc)
+					bpc = connector->display_info.bpc;
+			}
+		}
+		break;
+	case DRM_MODE_CONNECTOR_DVID:
+	case DRM_MODE_CONNECTOR_HDMIA:
+		if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+			if (connector->display_info.bpc)
+				bpc = connector->display_info.bpc;
+		}
+		break;
+	case DRM_MODE_CONNECTOR_DisplayPort:
+		dig_connector = amdgpu_connector->con_priv;
+		if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) ||
+		    (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP) ||
+		    drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+			if (connector->display_info.bpc)
+				bpc = connector->display_info.bpc;
+		}
+		break;
+	case DRM_MODE_CONNECTOR_eDP:
+	case DRM_MODE_CONNECTOR_LVDS:
+		if (connector->display_info.bpc)
+			bpc = connector->display_info.bpc;
+		else {
+			const struct drm_connector_helper_funcs *connector_funcs =
+				connector->helper_private;
+			struct drm_encoder *encoder = connector_funcs->best_encoder(connector);
+			struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+			struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+			if (dig->lcd_misc & ATOM_PANEL_MISC_V13_6BIT_PER_COLOR)
+				bpc = 6;
+			else if (dig->lcd_misc & ATOM_PANEL_MISC_V13_8BIT_PER_COLOR)
+				bpc = 8;
+		}
+		break;
+	}
+
+	if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+		/*
+		 * Pre DCE-8 hw can't handle > 12 bpc, and more than 12 bpc doesn't make
+		 * much sense without support for > 12 bpc framebuffers. RGB 4:4:4 at
+		 * 12 bpc is always supported on hdmi deep color sinks, as this is
+		 * required by the HDMI-1.3 spec. Clamp to a safe 12 bpc maximum.
+		 */
+		if (bpc > 12) {
+			DRM_DEBUG("%s: HDMI deep color %d bpc unsupported. Using 12 bpc.\n",
+				  connector->name, bpc);
+			bpc = 12;
+		}
+
+		/* Any defined maximum tmds clock limit we must not exceed? */
+		if (connector->display_info.max_tmds_clock > 0) {
+			/* mode_clock is clock in kHz for mode to be modeset on this connector */
+			mode_clock = amdgpu_connector->pixelclock_for_modeset;
+
+			/* Maximum allowable input clock in kHz */
+			max_tmds_clock = connector->display_info.max_tmds_clock;
+
+			DRM_DEBUG("%s: hdmi mode dotclock %d kHz, max tmds input clock %d kHz.\n",
+				  connector->name, mode_clock, max_tmds_clock);
+
+			/* Check if bpc is within clock limit. Try to degrade gracefully otherwise */
+			if ((bpc == 12) && (mode_clock * 3/2 > max_tmds_clock)) {
+				if ((connector->display_info.edid_hdmi_dc_modes & DRM_EDID_HDMI_DC_30) &&
+				    (mode_clock * 5/4 <= max_tmds_clock))
+					bpc = 10;
+				else
+					bpc = 8;
+
+				DRM_DEBUG("%s: HDMI deep color 12 bpc exceeds max tmds clock. Using %d bpc.\n",
+					  connector->name, bpc);
+			}
+
+			if ((bpc == 10) && (mode_clock * 5/4 > max_tmds_clock)) {
+				bpc = 8;
+				DRM_DEBUG("%s: HDMI deep color 10 bpc exceeds max tmds clock. Using %d bpc.\n",
+					  connector->name, bpc);
+			}
+		} else if (bpc > 8) {
+			/* max_tmds_clock missing, but hdmi spec mandates it for deep color. */
+			DRM_DEBUG("%s: Required max tmds clock for HDMI deep color missing. Using 8 bpc.\n",
+				  connector->name);
+			bpc = 8;
+		}
+	}
+
+	if ((amdgpu_deep_color == 0) && (bpc > 8)) {
+		DRM_DEBUG("%s: Deep color disabled. Set amdgpu module param deep_color=1 to enable.\n",
+			  connector->name);
+		bpc = 8;
+	}
+
+	DRM_DEBUG("%s: Display bpc=%d, returned bpc=%d\n",
+		  connector->name, connector->display_info.bpc, bpc);
+
+	return bpc;
+}
+
+static void
+amdgpu_connector_update_scratch_regs(struct drm_connector *connector,
+				      enum drm_connector_status status)
+{
+	struct drm_encoder *best_encoder;
+	struct drm_encoder *encoder;
+	const struct drm_connector_helper_funcs *connector_funcs = connector->helper_private;
+	bool connected;
+	int i;
+
+	best_encoder = connector_funcs->best_encoder(connector);
+
+	drm_connector_for_each_possible_encoder(connector, encoder, i) {
+		if ((encoder == best_encoder) && (status == connector_status_connected))
+			connected = true;
+		else
+			connected = false;
+
+		amdgpu_atombios_encoder_set_bios_scratch_regs(connector, encoder, connected);
+	}
+}
+
+static struct drm_encoder *
+amdgpu_connector_find_encoder(struct drm_connector *connector,
+			       int encoder_type)
+{
+	struct drm_encoder *encoder;
+	int i;
+
+	drm_connector_for_each_possible_encoder(connector, encoder, i) {
+		if (encoder->encoder_type == encoder_type)
+			return encoder;
+	}
+
+	return NULL;
+}
+
+struct edid *amdgpu_connector_edid(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_property_blob *edid_blob = connector->edid_blob_ptr;
+
+	if (amdgpu_connector->edid) {
+		return amdgpu_connector->edid;
+	} else if (edid_blob) {
+		struct edid *edid = kmemdup(edid_blob->data, edid_blob->length, GFP_KERNEL);
+		if (edid)
+			amdgpu_connector->edid = edid;
+	}
+	return amdgpu_connector->edid;
+}
+
+static struct edid *
+amdgpu_connector_get_hardcoded_edid(struct amdgpu_device *adev)
+{
+	struct edid *edid;
+
+	if (adev->mode_info.bios_hardcoded_edid) {
+		edid = kmalloc(adev->mode_info.bios_hardcoded_edid_size, GFP_KERNEL);
+		if (edid) {
+			memcpy((unsigned char *)edid,
+			       (unsigned char *)adev->mode_info.bios_hardcoded_edid,
+			       adev->mode_info.bios_hardcoded_edid_size);
+			return edid;
+		}
+	}
+	return NULL;
+}
+
+static void amdgpu_connector_get_edid(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	if (amdgpu_connector->edid)
+		return;
+
+	/* on hw with routers, select right port */
+	if (amdgpu_connector->router.ddc_valid)
+		amdgpu_i2c_router_select_ddc_port(amdgpu_connector);
+
+	if ((amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector) !=
+	     ENCODER_OBJECT_ID_NONE) &&
+	    amdgpu_connector->ddc_bus->has_aux) {
+		amdgpu_connector->edid = drm_get_edid(connector,
+						      &amdgpu_connector->ddc_bus->aux.ddc);
+	} else if ((connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort) ||
+		   (connector->connector_type == DRM_MODE_CONNECTOR_eDP)) {
+		struct amdgpu_connector_atom_dig *dig = amdgpu_connector->con_priv;
+
+		if ((dig->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT ||
+		     dig->dp_sink_type == CONNECTOR_OBJECT_ID_eDP) &&
+		    amdgpu_connector->ddc_bus->has_aux)
+			amdgpu_connector->edid = drm_get_edid(connector,
+							      &amdgpu_connector->ddc_bus->aux.ddc);
+		else if (amdgpu_connector->ddc_bus)
+			amdgpu_connector->edid = drm_get_edid(connector,
+							      &amdgpu_connector->ddc_bus->adapter);
+	} else if (amdgpu_connector->ddc_bus) {
+		amdgpu_connector->edid = drm_get_edid(connector,
+						      &amdgpu_connector->ddc_bus->adapter);
+	}
+
+	if (!amdgpu_connector->edid) {
+		/* some laptops provide a hardcoded edid in rom for LCDs */
+		if (((connector->connector_type == DRM_MODE_CONNECTOR_LVDS) ||
+		     (connector->connector_type == DRM_MODE_CONNECTOR_eDP)))
+			amdgpu_connector->edid = amdgpu_connector_get_hardcoded_edid(adev);
+	}
+}
+
+static void amdgpu_connector_free_edid(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	kfree(amdgpu_connector->edid);
+	amdgpu_connector->edid = NULL;
+}
+
+static int amdgpu_connector_ddc_get_modes(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	int ret;
+
+	if (amdgpu_connector->edid) {
+		drm_connector_update_edid_property(connector, amdgpu_connector->edid);
+		ret = drm_add_edid_modes(connector, amdgpu_connector->edid);
+		return ret;
+	}
+	drm_connector_update_edid_property(connector, NULL);
+	return 0;
+}
+
+static struct drm_encoder *
+amdgpu_connector_best_single_encoder(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder;
+	int i;
+
+	/* pick the first one */
+	drm_connector_for_each_possible_encoder(connector, encoder, i)
+		return encoder;
+
+	return NULL;
+}
+
+static void amdgpu_get_native_mode(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	if (encoder == NULL)
+		return;
+
+	amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	if (!list_empty(&connector->probed_modes)) {
+		struct drm_display_mode *preferred_mode =
+			list_first_entry(&connector->probed_modes,
+					 struct drm_display_mode, head);
+
+		amdgpu_encoder->native_mode = *preferred_mode;
+	} else {
+		amdgpu_encoder->native_mode.clock = 0;
+	}
+}
+
+static struct drm_display_mode *
+amdgpu_connector_lcd_native_mode(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_display_mode *mode = NULL;
+	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+
+	if (native_mode->hdisplay != 0 &&
+	    native_mode->vdisplay != 0 &&
+	    native_mode->clock != 0) {
+		mode = drm_mode_duplicate(dev, native_mode);
+		mode->type = DRM_MODE_TYPE_PREFERRED | DRM_MODE_TYPE_DRIVER;
+		drm_mode_set_name(mode);
+
+		DRM_DEBUG_KMS("Adding native panel mode %s\n", mode->name);
+	} else if (native_mode->hdisplay != 0 &&
+		   native_mode->vdisplay != 0) {
+		/* mac laptops without an edid */
+		/* Note that this is not necessarily the exact panel mode,
+		 * but an approximation based on the cvt formula.  For these
+		 * systems we should ideally read the mode info out of the
+		 * registers or add a mode table, but this works and is much
+		 * simpler.
+		 */
+		mode = drm_cvt_mode(dev, native_mode->hdisplay, native_mode->vdisplay, 60, true, false, false);
+		mode->type = DRM_MODE_TYPE_PREFERRED | DRM_MODE_TYPE_DRIVER;
+		DRM_DEBUG_KMS("Adding cvt approximation of native panel mode %s\n", mode->name);
+	}
+	return mode;
+}
+
+static void amdgpu_connector_add_common_modes(struct drm_encoder *encoder,
+					       struct drm_connector *connector)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_display_mode *mode = NULL;
+	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+	int i;
+	static const struct mode_size {
+		int w;
+		int h;
+	} common_modes[17] = {
+		{ 640,  480},
+		{ 720,  480},
+		{ 800,  600},
+		{ 848,  480},
+		{1024,  768},
+		{1152,  768},
+		{1280,  720},
+		{1280,  800},
+		{1280,  854},
+		{1280,  960},
+		{1280, 1024},
+		{1440,  900},
+		{1400, 1050},
+		{1680, 1050},
+		{1600, 1200},
+		{1920, 1080},
+		{1920, 1200}
+	};
+
+	for (i = 0; i < 17; i++) {
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_TV_SUPPORT)) {
+			if (common_modes[i].w > 1024 ||
+			    common_modes[i].h > 768)
+				continue;
+		}
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+			if (common_modes[i].w > native_mode->hdisplay ||
+			    common_modes[i].h > native_mode->vdisplay ||
+			    (common_modes[i].w == native_mode->hdisplay &&
+			     common_modes[i].h == native_mode->vdisplay))
+				continue;
+		}
+		if (common_modes[i].w < 320 || common_modes[i].h < 200)
+			continue;
+
+		mode = drm_cvt_mode(dev, common_modes[i].w, common_modes[i].h, 60, false, false, false);
+		drm_mode_probed_add(connector, mode);
+	}
+}
+
+static int amdgpu_connector_set_property(struct drm_connector *connector,
+					  struct drm_property *property,
+					  uint64_t val)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	if (property == adev->mode_info.coherent_mode_property) {
+		struct amdgpu_encoder_atom_dig *dig;
+		bool new_coherent_mode;
+
+		/* need to find digital encoder on connector */
+		encoder = amdgpu_connector_find_encoder(connector, DRM_MODE_ENCODER_TMDS);
+		if (!encoder)
+			return 0;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		if (!amdgpu_encoder->enc_priv)
+			return 0;
+
+		dig = amdgpu_encoder->enc_priv;
+		new_coherent_mode = val ? true : false;
+		if (dig->coherent_mode != new_coherent_mode) {
+			dig->coherent_mode = new_coherent_mode;
+			amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+		}
+	}
+
+	if (property == adev->mode_info.audio_property) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		/* need to find digital encoder on connector */
+		encoder = amdgpu_connector_find_encoder(connector, DRM_MODE_ENCODER_TMDS);
+		if (!encoder)
+			return 0;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		if (amdgpu_connector->audio != val) {
+			amdgpu_connector->audio = val;
+			amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+		}
+	}
+
+	if (property == adev->mode_info.dither_property) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		/* need to find digital encoder on connector */
+		encoder = amdgpu_connector_find_encoder(connector, DRM_MODE_ENCODER_TMDS);
+		if (!encoder)
+			return 0;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		if (amdgpu_connector->dither != val) {
+			amdgpu_connector->dither = val;
+			amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+		}
+	}
+
+	if (property == adev->mode_info.underscan_property) {
+		/* need to find digital encoder on connector */
+		encoder = amdgpu_connector_find_encoder(connector, DRM_MODE_ENCODER_TMDS);
+		if (!encoder)
+			return 0;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		if (amdgpu_encoder->underscan_type != val) {
+			amdgpu_encoder->underscan_type = val;
+			amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+		}
+	}
+
+	if (property == adev->mode_info.underscan_hborder_property) {
+		/* need to find digital encoder on connector */
+		encoder = amdgpu_connector_find_encoder(connector, DRM_MODE_ENCODER_TMDS);
+		if (!encoder)
+			return 0;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		if (amdgpu_encoder->underscan_hborder != val) {
+			amdgpu_encoder->underscan_hborder = val;
+			amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+		}
+	}
+
+	if (property == adev->mode_info.underscan_vborder_property) {
+		/* need to find digital encoder on connector */
+		encoder = amdgpu_connector_find_encoder(connector, DRM_MODE_ENCODER_TMDS);
+		if (!encoder)
+			return 0;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		if (amdgpu_encoder->underscan_vborder != val) {
+			amdgpu_encoder->underscan_vborder = val;
+			amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+		}
+	}
+
+	if (property == adev->mode_info.load_detect_property) {
+		struct amdgpu_connector *amdgpu_connector =
+			to_amdgpu_connector(connector);
+
+		if (val == 0)
+			amdgpu_connector->dac_load_detect = false;
+		else
+			amdgpu_connector->dac_load_detect = true;
+	}
+
+	if (property == dev->mode_config.scaling_mode_property) {
+		enum amdgpu_rmx_type rmx_type;
+
+		if (connector->encoder) {
+			amdgpu_encoder = to_amdgpu_encoder(connector->encoder);
+		} else {
+			const struct drm_connector_helper_funcs *connector_funcs = connector->helper_private;
+			amdgpu_encoder = to_amdgpu_encoder(connector_funcs->best_encoder(connector));
+		}
+
+		switch (val) {
+		default:
+		case DRM_MODE_SCALE_NONE: rmx_type = RMX_OFF; break;
+		case DRM_MODE_SCALE_CENTER: rmx_type = RMX_CENTER; break;
+		case DRM_MODE_SCALE_ASPECT: rmx_type = RMX_ASPECT; break;
+		case DRM_MODE_SCALE_FULLSCREEN: rmx_type = RMX_FULL; break;
+		}
+		if (amdgpu_encoder->rmx_type == rmx_type)
+			return 0;
+
+		if ((rmx_type != DRM_MODE_SCALE_NONE) &&
+		    (amdgpu_encoder->native_mode.clock == 0))
+			return 0;
+
+		amdgpu_encoder->rmx_type = rmx_type;
+
+		amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+	}
+
+	return 0;
+}
+
+static void
+amdgpu_connector_fixup_lcd_native_mode(struct drm_encoder *encoder,
+					struct drm_connector *connector)
+{
+	struct amdgpu_encoder *amdgpu_encoder =	to_amdgpu_encoder(encoder);
+	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+	struct drm_display_mode *t, *mode;
+
+	/* If the EDID preferred mode doesn't match the native mode, use it */
+	list_for_each_entry_safe(mode, t, &connector->probed_modes, head) {
+		if (mode->type & DRM_MODE_TYPE_PREFERRED) {
+			if (mode->hdisplay != native_mode->hdisplay ||
+			    mode->vdisplay != native_mode->vdisplay)
+				memcpy(native_mode, mode, sizeof(*mode));
+		}
+	}
+
+	/* Try to get native mode details from EDID if necessary */
+	if (!native_mode->clock) {
+		list_for_each_entry_safe(mode, t, &connector->probed_modes, head) {
+			if (mode->hdisplay == native_mode->hdisplay &&
+			    mode->vdisplay == native_mode->vdisplay) {
+				*native_mode = *mode;
+				drm_mode_set_crtcinfo(native_mode, CRTC_INTERLACE_HALVE_V);
+				DRM_DEBUG_KMS("Determined LVDS native mode details from EDID\n");
+				break;
+			}
+		}
+	}
+
+	if (!native_mode->clock) {
+		DRM_DEBUG_KMS("No LVDS native mode details, disabling RMX\n");
+		amdgpu_encoder->rmx_type = RMX_OFF;
+	}
+}
+
+static int amdgpu_connector_lvds_get_modes(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder;
+	int ret = 0;
+	struct drm_display_mode *mode;
+
+	amdgpu_connector_get_edid(connector);
+	ret = amdgpu_connector_ddc_get_modes(connector);
+	if (ret > 0) {
+		encoder = amdgpu_connector_best_single_encoder(connector);
+		if (encoder) {
+			amdgpu_connector_fixup_lcd_native_mode(encoder, connector);
+			/* add scaled modes */
+			amdgpu_connector_add_common_modes(encoder, connector);
+		}
+		return ret;
+	}
+
+	encoder = amdgpu_connector_best_single_encoder(connector);
+	if (!encoder)
+		return 0;
+
+	/* we have no EDID modes */
+	mode = amdgpu_connector_lcd_native_mode(encoder);
+	if (mode) {
+		ret = 1;
+		drm_mode_probed_add(connector, mode);
+		/* add the width/height from vbios tables if available */
+		connector->display_info.width_mm = mode->width_mm;
+		connector->display_info.height_mm = mode->height_mm;
+		/* add scaled modes */
+		amdgpu_connector_add_common_modes(encoder, connector);
+	}
+
+	return ret;
+}
+
+static enum drm_mode_status amdgpu_connector_lvds_mode_valid(struct drm_connector *connector,
+					     struct drm_display_mode *mode)
+{
+	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
+
+	if ((mode->hdisplay < 320) || (mode->vdisplay < 240))
+		return MODE_PANEL;
+
+	if (encoder) {
+		struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+		struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+
+		/* AVIVO hardware supports downscaling modes larger than the panel
+		 * to the panel size, but I'm not sure this is desirable.
+		 */
+		if ((mode->hdisplay > native_mode->hdisplay) ||
+		    (mode->vdisplay > native_mode->vdisplay))
+			return MODE_PANEL;
+
+		/* if scaling is disabled, block non-native modes */
+		if (amdgpu_encoder->rmx_type == RMX_OFF) {
+			if ((mode->hdisplay != native_mode->hdisplay) ||
+			    (mode->vdisplay != native_mode->vdisplay))
+				return MODE_PANEL;
+		}
+	}
+
+	return MODE_OK;
+}
+
+static enum drm_connector_status
+amdgpu_connector_lvds_detect(struct drm_connector *connector, bool force)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
+	enum drm_connector_status ret = connector_status_disconnected;
+	int r;
+
+	r = pm_runtime_get_sync(connector->dev->dev);
+	if (r < 0)
+		return connector_status_disconnected;
+
+	if (encoder) {
+		struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+		struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+
+		/* check if panel is valid */
+		if (native_mode->hdisplay >= 320 && native_mode->vdisplay >= 240)
+			ret = connector_status_connected;
+
+	}
+
+	/* check for edid as well */
+	amdgpu_connector_get_edid(connector);
+	if (amdgpu_connector->edid)
+		ret = connector_status_connected;
+	/* check acpi lid status ??? */
+
+	amdgpu_connector_update_scratch_regs(connector, ret);
+
+	pm_runtime_mark_last_busy(connector->dev->dev);
+	pm_runtime_put_autosuspend(connector->dev->dev);
+
+	return ret;
+}
+
+static void amdgpu_connector_unregister(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	if (amdgpu_connector->ddc_bus && amdgpu_connector->ddc_bus->has_aux) {
+		drm_dp_aux_unregister(&amdgpu_connector->ddc_bus->aux);
+		amdgpu_connector->ddc_bus->has_aux = false;
+	}
+}
+
+static void amdgpu_connector_destroy(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	amdgpu_connector_free_edid(connector);
+	kfree(amdgpu_connector->con_priv);
+	drm_connector_unregister(connector);
+	drm_connector_cleanup(connector);
+	kfree(connector);
+}
+
+static int amdgpu_connector_set_lcd_property(struct drm_connector *connector,
+					      struct drm_property *property,
+					      uint64_t value)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_encoder *amdgpu_encoder;
+	enum amdgpu_rmx_type rmx_type;
+
+	DRM_DEBUG_KMS("\n");
+	if (property != dev->mode_config.scaling_mode_property)
+		return 0;
+
+	if (connector->encoder)
+		amdgpu_encoder = to_amdgpu_encoder(connector->encoder);
+	else {
+		const struct drm_connector_helper_funcs *connector_funcs = connector->helper_private;
+		amdgpu_encoder = to_amdgpu_encoder(connector_funcs->best_encoder(connector));
+	}
+
+	switch (value) {
+	case DRM_MODE_SCALE_NONE: rmx_type = RMX_OFF; break;
+	case DRM_MODE_SCALE_CENTER: rmx_type = RMX_CENTER; break;
+	case DRM_MODE_SCALE_ASPECT: rmx_type = RMX_ASPECT; break;
+	default:
+	case DRM_MODE_SCALE_FULLSCREEN: rmx_type = RMX_FULL; break;
+	}
+	if (amdgpu_encoder->rmx_type == rmx_type)
+		return 0;
+
+	amdgpu_encoder->rmx_type = rmx_type;
+
+	amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+	return 0;
+}
+
+
+static const struct drm_connector_helper_funcs amdgpu_connector_lvds_helper_funcs = {
+	.get_modes = amdgpu_connector_lvds_get_modes,
+	.mode_valid = amdgpu_connector_lvds_mode_valid,
+	.best_encoder = amdgpu_connector_best_single_encoder,
+};
+
+static const struct drm_connector_funcs amdgpu_connector_lvds_funcs = {
+	.dpms = drm_helper_connector_dpms,
+	.detect = amdgpu_connector_lvds_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.early_unregister = amdgpu_connector_unregister,
+	.destroy = amdgpu_connector_destroy,
+	.set_property = amdgpu_connector_set_lcd_property,
+};
+
+static int amdgpu_connector_vga_get_modes(struct drm_connector *connector)
+{
+	int ret;
+
+	amdgpu_connector_get_edid(connector);
+	ret = amdgpu_connector_ddc_get_modes(connector);
+
+	return ret;
+}
+
+static enum drm_mode_status amdgpu_connector_vga_mode_valid(struct drm_connector *connector,
+					    struct drm_display_mode *mode)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* XXX check mode bandwidth */
+
+	if ((mode->clock / 10) > adev->clock.max_pixel_clock)
+		return MODE_CLOCK_HIGH;
+
+	return MODE_OK;
+}
+
+static enum drm_connector_status
+amdgpu_connector_vga_detect(struct drm_connector *connector, bool force)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_encoder *encoder;
+	const struct drm_encoder_helper_funcs *encoder_funcs;
+	bool dret = false;
+	enum drm_connector_status ret = connector_status_disconnected;
+	int r;
+
+	r = pm_runtime_get_sync(connector->dev->dev);
+	if (r < 0)
+		return connector_status_disconnected;
+
+	encoder = amdgpu_connector_best_single_encoder(connector);
+	if (!encoder)
+		ret = connector_status_disconnected;
+
+	if (amdgpu_connector->ddc_bus)
+		dret = amdgpu_display_ddc_probe(amdgpu_connector, false);
+	if (dret) {
+		amdgpu_connector->detected_by_load = false;
+		amdgpu_connector_free_edid(connector);
+		amdgpu_connector_get_edid(connector);
+
+		if (!amdgpu_connector->edid) {
+			DRM_ERROR("%s: probed a monitor but no|invalid EDID\n",
+					connector->name);
+			ret = connector_status_connected;
+		} else {
+			amdgpu_connector->use_digital =
+				!!(amdgpu_connector->edid->input & DRM_EDID_INPUT_DIGITAL);
+
+			/* some oems have boards with separate digital and analog connectors
+			 * with a shared ddc line (often vga + hdmi)
+			 */
+			if (amdgpu_connector->use_digital && amdgpu_connector->shared_ddc) {
+				amdgpu_connector_free_edid(connector);
+				ret = connector_status_disconnected;
+			} else {
+				ret = connector_status_connected;
+			}
+		}
+	} else {
+
+		/* if we aren't forcing don't do destructive polling */
+		if (!force) {
+			/* only return the previous status if we last
+			 * detected a monitor via load.
+			 */
+			if (amdgpu_connector->detected_by_load)
+				ret = connector->status;
+			goto out;
+		}
+
+		if (amdgpu_connector->dac_load_detect && encoder) {
+			encoder_funcs = encoder->helper_private;
+			ret = encoder_funcs->detect(encoder, connector);
+			if (ret != connector_status_disconnected)
+				amdgpu_connector->detected_by_load = true;
+		}
+	}
+
+	amdgpu_connector_update_scratch_regs(connector, ret);
+
+out:
+	pm_runtime_mark_last_busy(connector->dev->dev);
+	pm_runtime_put_autosuspend(connector->dev->dev);
+
+	return ret;
+}
+
+static const struct drm_connector_helper_funcs amdgpu_connector_vga_helper_funcs = {
+	.get_modes = amdgpu_connector_vga_get_modes,
+	.mode_valid = amdgpu_connector_vga_mode_valid,
+	.best_encoder = amdgpu_connector_best_single_encoder,
+};
+
+static const struct drm_connector_funcs amdgpu_connector_vga_funcs = {
+	.dpms = drm_helper_connector_dpms,
+	.detect = amdgpu_connector_vga_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.early_unregister = amdgpu_connector_unregister,
+	.destroy = amdgpu_connector_destroy,
+	.set_property = amdgpu_connector_set_property,
+};
+
+static bool
+amdgpu_connector_check_hpd_status_unchanged(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	enum drm_connector_status status;
+
+	if (amdgpu_connector->hpd.hpd != AMDGPU_HPD_NONE) {
+		if (amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd))
+			status = connector_status_connected;
+		else
+			status = connector_status_disconnected;
+		if (connector->status == status)
+			return true;
+	}
+
+	return false;
+}
+
+/*
+ * DVI is complicated
+ * Do a DDC probe, if DDC probe passes, get the full EDID so
+ * we can do analog/digital monitor detection at this point.
+ * If the monitor is an analog monitor or we got no DDC,
+ * we need to find the DAC encoder object for this connector.
+ * If we got no DDC, we do load detection on the DAC encoder object.
+ * If we got analog DDC or load detection passes on the DAC encoder
+ * we have to check if this analog encoder is shared with anyone else (TV)
+ * if its shared we have to set the other connector to disconnected.
+ */
+static enum drm_connector_status
+amdgpu_connector_dvi_detect(struct drm_connector *connector, bool force)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	const struct drm_encoder_helper_funcs *encoder_funcs;
+	int r;
+	enum drm_connector_status ret = connector_status_disconnected;
+	bool dret = false, broken_edid = false;
+
+	r = pm_runtime_get_sync(connector->dev->dev);
+	if (r < 0)
+		return connector_status_disconnected;
+
+	if (!force && amdgpu_connector_check_hpd_status_unchanged(connector)) {
+		ret = connector->status;
+		goto exit;
+	}
+
+	if (amdgpu_connector->ddc_bus)
+		dret = amdgpu_display_ddc_probe(amdgpu_connector, false);
+	if (dret) {
+		amdgpu_connector->detected_by_load = false;
+		amdgpu_connector_free_edid(connector);
+		amdgpu_connector_get_edid(connector);
+
+		if (!amdgpu_connector->edid) {
+			DRM_ERROR("%s: probed a monitor but no|invalid EDID\n",
+					connector->name);
+			ret = connector_status_connected;
+			broken_edid = true; /* defer use_digital to later */
+		} else {
+			amdgpu_connector->use_digital =
+				!!(amdgpu_connector->edid->input & DRM_EDID_INPUT_DIGITAL);
+
+			/* some oems have boards with separate digital and analog connectors
+			 * with a shared ddc line (often vga + hdmi)
+			 */
+			if ((!amdgpu_connector->use_digital) && amdgpu_connector->shared_ddc) {
+				amdgpu_connector_free_edid(connector);
+				ret = connector_status_disconnected;
+			} else {
+				ret = connector_status_connected;
+			}
+
+			/* This gets complicated.  We have boards with VGA + HDMI with a
+			 * shared DDC line and we have boards with DVI-D + HDMI with a shared
+			 * DDC line.  The latter is more complex because with DVI<->HDMI adapters
+			 * you don't really know what's connected to which port as both are digital.
+			 */
+			if (amdgpu_connector->shared_ddc && (ret == connector_status_connected)) {
+				struct drm_connector *list_connector;
+				struct amdgpu_connector *list_amdgpu_connector;
+				list_for_each_entry(list_connector, &dev->mode_config.connector_list, head) {
+					if (connector == list_connector)
+						continue;
+					list_amdgpu_connector = to_amdgpu_connector(list_connector);
+					if (list_amdgpu_connector->shared_ddc &&
+					    (list_amdgpu_connector->ddc_bus->rec.i2c_id ==
+					     amdgpu_connector->ddc_bus->rec.i2c_id)) {
+						/* cases where both connectors are digital */
+						if (list_connector->connector_type != DRM_MODE_CONNECTOR_VGA) {
+							/* hpd is our only option in this case */
+							if (!amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd)) {
+								amdgpu_connector_free_edid(connector);
+								ret = connector_status_disconnected;
+							}
+						}
+					}
+				}
+			}
+		}
+	}
+
+	if ((ret == connector_status_connected) && (amdgpu_connector->use_digital == true))
+		goto out;
+
+	/* DVI-D and HDMI-A are digital only */
+	if ((connector->connector_type == DRM_MODE_CONNECTOR_DVID) ||
+	    (connector->connector_type == DRM_MODE_CONNECTOR_HDMIA))
+		goto out;
+
+	/* if we aren't forcing don't do destructive polling */
+	if (!force) {
+		/* only return the previous status if we last
+		 * detected a monitor via load.
+		 */
+		if (amdgpu_connector->detected_by_load)
+			ret = connector->status;
+		goto out;
+	}
+
+	/* find analog encoder */
+	if (amdgpu_connector->dac_load_detect) {
+		struct drm_encoder *encoder;
+		int i;
+
+		drm_connector_for_each_possible_encoder(connector, encoder, i) {
+			if (encoder->encoder_type != DRM_MODE_ENCODER_DAC &&
+			    encoder->encoder_type != DRM_MODE_ENCODER_TVDAC)
+				continue;
+
+			encoder_funcs = encoder->helper_private;
+			if (encoder_funcs->detect) {
+				if (!broken_edid) {
+					if (ret != connector_status_connected) {
+						/* deal with analog monitors without DDC */
+						ret = encoder_funcs->detect(encoder, connector);
+						if (ret == connector_status_connected) {
+							amdgpu_connector->use_digital = false;
+						}
+						if (ret != connector_status_disconnected)
+							amdgpu_connector->detected_by_load = true;
+					}
+				} else {
+					enum drm_connector_status lret;
+					/* assume digital unless load detected otherwise */
+					amdgpu_connector->use_digital = true;
+					lret = encoder_funcs->detect(encoder, connector);
+					DRM_DEBUG_KMS("load_detect %x returned: %x\n",encoder->encoder_type,lret);
+					if (lret == connector_status_connected)
+						amdgpu_connector->use_digital = false;
+				}
+				break;
+			}
+		}
+	}
+
+out:
+	/* updated in get modes as well since we need to know if it's analog or digital */
+	amdgpu_connector_update_scratch_regs(connector, ret);
+
+exit:
+	pm_runtime_mark_last_busy(connector->dev->dev);
+	pm_runtime_put_autosuspend(connector->dev->dev);
+
+	return ret;
+}
+
+/* okay need to be smart in here about which encoder to pick */
+static struct drm_encoder *
+amdgpu_connector_dvi_encoder(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_encoder *encoder;
+	int i;
+
+	drm_connector_for_each_possible_encoder(connector, encoder, i) {
+		if (amdgpu_connector->use_digital == true) {
+			if (encoder->encoder_type == DRM_MODE_ENCODER_TMDS)
+				return encoder;
+		} else {
+			if (encoder->encoder_type == DRM_MODE_ENCODER_DAC ||
+			    encoder->encoder_type == DRM_MODE_ENCODER_TVDAC)
+				return encoder;
+		}
+	}
+
+	/* see if we have a default encoder  TODO */
+
+	/* then check use digitial */
+	/* pick the first one */
+	drm_connector_for_each_possible_encoder(connector, encoder, i)
+		return encoder;
+
+	return NULL;
+}
+
+static void amdgpu_connector_dvi_force(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	if (connector->force == DRM_FORCE_ON)
+		amdgpu_connector->use_digital = false;
+	if (connector->force == DRM_FORCE_ON_DIGITAL)
+		amdgpu_connector->use_digital = true;
+}
+
+static enum drm_mode_status amdgpu_connector_dvi_mode_valid(struct drm_connector *connector,
+					    struct drm_display_mode *mode)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	/* XXX check mode bandwidth */
+
+	if (amdgpu_connector->use_digital && (mode->clock > 165000)) {
+		if ((amdgpu_connector->connector_object_id == CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I) ||
+		    (amdgpu_connector->connector_object_id == CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D) ||
+		    (amdgpu_connector->connector_object_id == CONNECTOR_OBJECT_ID_HDMI_TYPE_B)) {
+			return MODE_OK;
+		} else if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+			/* HDMI 1.3+ supports max clock of 340 Mhz */
+			if (mode->clock > 340000)
+				return MODE_CLOCK_HIGH;
+			else
+				return MODE_OK;
+		} else {
+			return MODE_CLOCK_HIGH;
+		}
+	}
+
+	/* check against the max pixel clock */
+	if ((mode->clock / 10) > adev->clock.max_pixel_clock)
+		return MODE_CLOCK_HIGH;
+
+	return MODE_OK;
+}
+
+static const struct drm_connector_helper_funcs amdgpu_connector_dvi_helper_funcs = {
+	.get_modes = amdgpu_connector_vga_get_modes,
+	.mode_valid = amdgpu_connector_dvi_mode_valid,
+	.best_encoder = amdgpu_connector_dvi_encoder,
+};
+
+static const struct drm_connector_funcs amdgpu_connector_dvi_funcs = {
+	.dpms = drm_helper_connector_dpms,
+	.detect = amdgpu_connector_dvi_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.set_property = amdgpu_connector_set_property,
+	.early_unregister = amdgpu_connector_unregister,
+	.destroy = amdgpu_connector_destroy,
+	.force = amdgpu_connector_dvi_force,
+};
+
+static int amdgpu_connector_dp_get_modes(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *amdgpu_dig_connector = amdgpu_connector->con_priv;
+	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
+	int ret;
+
+	if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) ||
+	    (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) {
+		struct drm_display_mode *mode;
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) {
+			if (!amdgpu_dig_connector->edp_on)
+				amdgpu_atombios_encoder_set_edp_panel_power(connector,
+								     ATOM_TRANSMITTER_ACTION_POWER_ON);
+			amdgpu_connector_get_edid(connector);
+			ret = amdgpu_connector_ddc_get_modes(connector);
+			if (!amdgpu_dig_connector->edp_on)
+				amdgpu_atombios_encoder_set_edp_panel_power(connector,
+								     ATOM_TRANSMITTER_ACTION_POWER_OFF);
+		} else {
+			/* need to setup ddc on the bridge */
+			if (amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector) !=
+			    ENCODER_OBJECT_ID_NONE) {
+				if (encoder)
+					amdgpu_atombios_encoder_setup_ext_encoder_ddc(encoder);
+			}
+			amdgpu_connector_get_edid(connector);
+			ret = amdgpu_connector_ddc_get_modes(connector);
+		}
+
+		if (ret > 0) {
+			if (encoder) {
+				amdgpu_connector_fixup_lcd_native_mode(encoder, connector);
+				/* add scaled modes */
+				amdgpu_connector_add_common_modes(encoder, connector);
+			}
+			return ret;
+		}
+
+		if (!encoder)
+			return 0;
+
+		/* we have no EDID modes */
+		mode = amdgpu_connector_lcd_native_mode(encoder);
+		if (mode) {
+			ret = 1;
+			drm_mode_probed_add(connector, mode);
+			/* add the width/height from vbios tables if available */
+			connector->display_info.width_mm = mode->width_mm;
+			connector->display_info.height_mm = mode->height_mm;
+			/* add scaled modes */
+			amdgpu_connector_add_common_modes(encoder, connector);
+		}
+	} else {
+		/* need to setup ddc on the bridge */
+		if (amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector) !=
+			ENCODER_OBJECT_ID_NONE) {
+			if (encoder)
+				amdgpu_atombios_encoder_setup_ext_encoder_ddc(encoder);
+		}
+		amdgpu_connector_get_edid(connector);
+		ret = amdgpu_connector_ddc_get_modes(connector);
+
+		amdgpu_get_native_mode(connector);
+	}
+
+	return ret;
+}
+
+u16 amdgpu_connector_encoder_get_dp_bridge_encoder_id(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+	int i;
+
+	drm_connector_for_each_possible_encoder(connector, encoder, i) {
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		switch (amdgpu_encoder->encoder_id) {
+		case ENCODER_OBJECT_ID_TRAVIS:
+		case ENCODER_OBJECT_ID_NUTMEG:
+			return amdgpu_encoder->encoder_id;
+		default:
+			break;
+		}
+	}
+
+	return ENCODER_OBJECT_ID_NONE;
+}
+
+static bool amdgpu_connector_encoder_is_hbr2(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+	int i;
+	bool found = false;
+
+	drm_connector_for_each_possible_encoder(connector, encoder, i) {
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		if (amdgpu_encoder->caps & ATOM_ENCODER_CAP_RECORD_HBR2)
+			found = true;
+	}
+
+	return found;
+}
+
+bool amdgpu_connector_is_dp12_capable(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if ((adev->clock.default_dispclk >= 53900) &&
+	    amdgpu_connector_encoder_is_hbr2(connector)) {
+		return true;
+	}
+
+	return false;
+}
+
+static enum drm_connector_status
+amdgpu_connector_dp_detect(struct drm_connector *connector, bool force)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	enum drm_connector_status ret = connector_status_disconnected;
+	struct amdgpu_connector_atom_dig *amdgpu_dig_connector = amdgpu_connector->con_priv;
+	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
+	int r;
+
+	r = pm_runtime_get_sync(connector->dev->dev);
+	if (r < 0)
+		return connector_status_disconnected;
+
+	if (!force && amdgpu_connector_check_hpd_status_unchanged(connector)) {
+		ret = connector->status;
+		goto out;
+	}
+
+	amdgpu_connector_free_edid(connector);
+
+	if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) ||
+	    (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) {
+		if (encoder) {
+			struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+			struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+
+			/* check if panel is valid */
+			if (native_mode->hdisplay >= 320 && native_mode->vdisplay >= 240)
+				ret = connector_status_connected;
+		}
+		/* eDP is always DP */
+		amdgpu_dig_connector->dp_sink_type = CONNECTOR_OBJECT_ID_DISPLAYPORT;
+		if (!amdgpu_dig_connector->edp_on)
+			amdgpu_atombios_encoder_set_edp_panel_power(connector,
+							     ATOM_TRANSMITTER_ACTION_POWER_ON);
+		if (!amdgpu_atombios_dp_get_dpcd(amdgpu_connector))
+			ret = connector_status_connected;
+		if (!amdgpu_dig_connector->edp_on)
+			amdgpu_atombios_encoder_set_edp_panel_power(connector,
+							     ATOM_TRANSMITTER_ACTION_POWER_OFF);
+	} else if (amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector) !=
+		   ENCODER_OBJECT_ID_NONE) {
+		/* DP bridges are always DP */
+		amdgpu_dig_connector->dp_sink_type = CONNECTOR_OBJECT_ID_DISPLAYPORT;
+		/* get the DPCD from the bridge */
+		amdgpu_atombios_dp_get_dpcd(amdgpu_connector);
+
+		if (encoder) {
+			/* setup ddc on the bridge */
+			amdgpu_atombios_encoder_setup_ext_encoder_ddc(encoder);
+			/* bridge chips are always aux */
+			/* try DDC */
+			if (amdgpu_display_ddc_probe(amdgpu_connector, true))
+				ret = connector_status_connected;
+			else if (amdgpu_connector->dac_load_detect) { /* try load detection */
+				const struct drm_encoder_helper_funcs *encoder_funcs = encoder->helper_private;
+				ret = encoder_funcs->detect(encoder, connector);
+			}
+		}
+	} else {
+		amdgpu_dig_connector->dp_sink_type =
+			amdgpu_atombios_dp_get_sinktype(amdgpu_connector);
+		if (amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd)) {
+			ret = connector_status_connected;
+			if (amdgpu_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT)
+				amdgpu_atombios_dp_get_dpcd(amdgpu_connector);
+		} else {
+			if (amdgpu_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) {
+				if (!amdgpu_atombios_dp_get_dpcd(amdgpu_connector))
+					ret = connector_status_connected;
+			} else {
+				/* try non-aux ddc (DP to DVI/HDMI/etc. adapter) */
+				if (amdgpu_display_ddc_probe(amdgpu_connector,
+							     false))
+					ret = connector_status_connected;
+			}
+		}
+	}
+
+	amdgpu_connector_update_scratch_regs(connector, ret);
+out:
+	pm_runtime_mark_last_busy(connector->dev->dev);
+	pm_runtime_put_autosuspend(connector->dev->dev);
+
+	return ret;
+}
+
+static enum drm_mode_status amdgpu_connector_dp_mode_valid(struct drm_connector *connector,
+					   struct drm_display_mode *mode)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *amdgpu_dig_connector = amdgpu_connector->con_priv;
+
+	/* XXX check mode bandwidth */
+
+	if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) ||
+	    (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) {
+		struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
+
+		if ((mode->hdisplay < 320) || (mode->vdisplay < 240))
+			return MODE_PANEL;
+
+		if (encoder) {
+			struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+			struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+
+			/* AVIVO hardware supports downscaling modes larger than the panel
+			 * to the panel size, but I'm not sure this is desirable.
+			 */
+			if ((mode->hdisplay > native_mode->hdisplay) ||
+			    (mode->vdisplay > native_mode->vdisplay))
+				return MODE_PANEL;
+
+			/* if scaling is disabled, block non-native modes */
+			if (amdgpu_encoder->rmx_type == RMX_OFF) {
+				if ((mode->hdisplay != native_mode->hdisplay) ||
+				    (mode->vdisplay != native_mode->vdisplay))
+					return MODE_PANEL;
+			}
+		}
+		return MODE_OK;
+	} else {
+		if ((amdgpu_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) ||
+		    (amdgpu_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) {
+			return amdgpu_atombios_dp_mode_valid_helper(connector, mode);
+		} else {
+			if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+				/* HDMI 1.3+ supports max clock of 340 Mhz */
+				if (mode->clock > 340000)
+					return MODE_CLOCK_HIGH;
+			} else {
+				if (mode->clock > 165000)
+					return MODE_CLOCK_HIGH;
+			}
+		}
+	}
+
+	return MODE_OK;
+}
+
+static const struct drm_connector_helper_funcs amdgpu_connector_dp_helper_funcs = {
+	.get_modes = amdgpu_connector_dp_get_modes,
+	.mode_valid = amdgpu_connector_dp_mode_valid,
+	.best_encoder = amdgpu_connector_dvi_encoder,
+};
+
+static const struct drm_connector_funcs amdgpu_connector_dp_funcs = {
+	.dpms = drm_helper_connector_dpms,
+	.detect = amdgpu_connector_dp_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.set_property = amdgpu_connector_set_property,
+	.early_unregister = amdgpu_connector_unregister,
+	.destroy = amdgpu_connector_destroy,
+	.force = amdgpu_connector_dvi_force,
+};
+
+static const struct drm_connector_funcs amdgpu_connector_edp_funcs = {
+	.dpms = drm_helper_connector_dpms,
+	.detect = amdgpu_connector_dp_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.set_property = amdgpu_connector_set_lcd_property,
+	.early_unregister = amdgpu_connector_unregister,
+	.destroy = amdgpu_connector_destroy,
+	.force = amdgpu_connector_dvi_force,
+};
+
+void
+amdgpu_connector_add(struct amdgpu_device *adev,
+		      uint32_t connector_id,
+		      uint32_t supported_device,
+		      int connector_type,
+		      struct amdgpu_i2c_bus_rec *i2c_bus,
+		      uint16_t connector_object_id,
+		      struct amdgpu_hpd *hpd,
+		      struct amdgpu_router *router)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+	struct amdgpu_connector_atom_dig *amdgpu_dig_connector;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+	uint32_t subpixel_order = SubPixelNone;
+	bool shared_ddc = false;
+	bool is_dp_bridge = false;
+	bool has_aux = false;
+
+	if (connector_type == DRM_MODE_CONNECTOR_Unknown)
+		return;
+
+	/* see if we already added it */
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		amdgpu_connector = to_amdgpu_connector(connector);
+		if (amdgpu_connector->connector_id == connector_id) {
+			amdgpu_connector->devices |= supported_device;
+			return;
+		}
+		if (amdgpu_connector->ddc_bus && i2c_bus->valid) {
+			if (amdgpu_connector->ddc_bus->rec.i2c_id == i2c_bus->i2c_id) {
+				amdgpu_connector->shared_ddc = true;
+				shared_ddc = true;
+			}
+			if (amdgpu_connector->router_bus && router->ddc_valid &&
+			    (amdgpu_connector->router.router_id == router->router_id)) {
+				amdgpu_connector->shared_ddc = false;
+				shared_ddc = false;
+			}
+		}
+	}
+
+	/* check if it's a dp bridge */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		if (amdgpu_encoder->devices & supported_device) {
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_TRAVIS:
+			case ENCODER_OBJECT_ID_NUTMEG:
+				is_dp_bridge = true;
+				break;
+			default:
+				break;
+			}
+		}
+	}
+
+	amdgpu_connector = kzalloc(sizeof(struct amdgpu_connector), GFP_KERNEL);
+	if (!amdgpu_connector)
+		return;
+
+	connector = &amdgpu_connector->base;
+
+	amdgpu_connector->connector_id = connector_id;
+	amdgpu_connector->devices = supported_device;
+	amdgpu_connector->shared_ddc = shared_ddc;
+	amdgpu_connector->connector_object_id = connector_object_id;
+	amdgpu_connector->hpd = *hpd;
+
+	amdgpu_connector->router = *router;
+	if (router->ddc_valid || router->cd_valid) {
+		amdgpu_connector->router_bus = amdgpu_i2c_lookup(adev, &router->i2c_info);
+		if (!amdgpu_connector->router_bus)
+			DRM_ERROR("Failed to assign router i2c bus! Check dmesg for i2c errors.\n");
+	}
+
+	if (is_dp_bridge) {
+		amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL);
+		if (!amdgpu_dig_connector)
+			goto failed;
+		amdgpu_connector->con_priv = amdgpu_dig_connector;
+		if (i2c_bus->valid) {
+			amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+			if (amdgpu_connector->ddc_bus)
+				has_aux = true;
+			else
+				DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+		}
+		switch (connector_type) {
+		case DRM_MODE_CONNECTOR_VGA:
+		case DRM_MODE_CONNECTOR_DVIA:
+		default:
+			drm_connector_init(dev, &amdgpu_connector->base,
+					   &amdgpu_connector_dp_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base,
+						 &amdgpu_connector_dp_helper_funcs);
+			connector->interlace_allowed = true;
+			connector->doublescan_allowed = true;
+			amdgpu_connector->dac_load_detect = true;
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.load_detect_property,
+						      1);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+			break;
+		case DRM_MODE_CONNECTOR_DVII:
+		case DRM_MODE_CONNECTOR_DVID:
+		case DRM_MODE_CONNECTOR_HDMIA:
+		case DRM_MODE_CONNECTOR_HDMIB:
+		case DRM_MODE_CONNECTOR_DisplayPort:
+			drm_connector_init(dev, &amdgpu_connector->base,
+					   &amdgpu_connector_dp_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base,
+						 &amdgpu_connector_dp_helper_funcs);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.underscan_property,
+						      UNDERSCAN_OFF);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.underscan_hborder_property,
+						      0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.underscan_vborder_property,
+						      0);
+
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.dither_property,
+						   AMDGPU_FMT_DITHER_DISABLE);
+
+			if (amdgpu_audio != 0)
+				drm_object_attach_property(&amdgpu_connector->base.base,
+							   adev->mode_info.audio_property,
+							   AMDGPU_AUDIO_AUTO);
+
+			subpixel_order = SubPixelHorizontalRGB;
+			connector->interlace_allowed = true;
+			if (connector_type == DRM_MODE_CONNECTOR_HDMIB)
+				connector->doublescan_allowed = true;
+			else
+				connector->doublescan_allowed = false;
+			if (connector_type == DRM_MODE_CONNECTOR_DVII) {
+				amdgpu_connector->dac_load_detect = true;
+				drm_object_attach_property(&amdgpu_connector->base.base,
+							      adev->mode_info.load_detect_property,
+							      1);
+			}
+			break;
+		case DRM_MODE_CONNECTOR_LVDS:
+		case DRM_MODE_CONNECTOR_eDP:
+			drm_connector_init(dev, &amdgpu_connector->base,
+					   &amdgpu_connector_edp_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base,
+						 &amdgpu_connector_dp_helper_funcs);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      dev->mode_config.scaling_mode_property,
+						      DRM_MODE_SCALE_FULLSCREEN);
+			subpixel_order = SubPixelHorizontalRGB;
+			connector->interlace_allowed = false;
+			connector->doublescan_allowed = false;
+			break;
+		}
+	} else {
+		switch (connector_type) {
+		case DRM_MODE_CONNECTOR_VGA:
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_vga_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_vga_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (!amdgpu_connector->ddc_bus)
+					DRM_ERROR("VGA: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			amdgpu_connector->dac_load_detect = true;
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.load_detect_property,
+						      1);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+			/* no HPD on analog connectors */
+			amdgpu_connector->hpd.hpd = AMDGPU_HPD_NONE;
+			connector->interlace_allowed = true;
+			connector->doublescan_allowed = true;
+			break;
+		case DRM_MODE_CONNECTOR_DVIA:
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_vga_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_vga_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (!amdgpu_connector->ddc_bus)
+					DRM_ERROR("DVIA: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			amdgpu_connector->dac_load_detect = true;
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.load_detect_property,
+						      1);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+			/* no HPD on analog connectors */
+			amdgpu_connector->hpd.hpd = AMDGPU_HPD_NONE;
+			connector->interlace_allowed = true;
+			connector->doublescan_allowed = true;
+			break;
+		case DRM_MODE_CONNECTOR_DVII:
+		case DRM_MODE_CONNECTOR_DVID:
+			amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL);
+			if (!amdgpu_dig_connector)
+				goto failed;
+			amdgpu_connector->con_priv = amdgpu_dig_connector;
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_dvi_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dvi_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (!amdgpu_connector->ddc_bus)
+					DRM_ERROR("DVI: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			subpixel_order = SubPixelHorizontalRGB;
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.coherent_mode_property,
+						      1);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_property,
+						   UNDERSCAN_OFF);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_hborder_property,
+						   0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_vborder_property,
+						   0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+
+			if (amdgpu_audio != 0) {
+				drm_object_attach_property(&amdgpu_connector->base.base,
+							   adev->mode_info.audio_property,
+							   AMDGPU_AUDIO_AUTO);
+			}
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.dither_property,
+						   AMDGPU_FMT_DITHER_DISABLE);
+			if (connector_type == DRM_MODE_CONNECTOR_DVII) {
+				amdgpu_connector->dac_load_detect = true;
+				drm_object_attach_property(&amdgpu_connector->base.base,
+							   adev->mode_info.load_detect_property,
+							   1);
+			}
+			connector->interlace_allowed = true;
+			if (connector_type == DRM_MODE_CONNECTOR_DVII)
+				connector->doublescan_allowed = true;
+			else
+				connector->doublescan_allowed = false;
+			break;
+		case DRM_MODE_CONNECTOR_HDMIA:
+		case DRM_MODE_CONNECTOR_HDMIB:
+			amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL);
+			if (!amdgpu_dig_connector)
+				goto failed;
+			amdgpu_connector->con_priv = amdgpu_dig_connector;
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_dvi_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dvi_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (!amdgpu_connector->ddc_bus)
+					DRM_ERROR("HDMI: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.coherent_mode_property,
+						      1);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_property,
+						   UNDERSCAN_OFF);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_hborder_property,
+						   0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_vborder_property,
+						   0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+			if (amdgpu_audio != 0) {
+				drm_object_attach_property(&amdgpu_connector->base.base,
+							   adev->mode_info.audio_property,
+							   AMDGPU_AUDIO_AUTO);
+			}
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.dither_property,
+						   AMDGPU_FMT_DITHER_DISABLE);
+			subpixel_order = SubPixelHorizontalRGB;
+			connector->interlace_allowed = true;
+			if (connector_type == DRM_MODE_CONNECTOR_HDMIB)
+				connector->doublescan_allowed = true;
+			else
+				connector->doublescan_allowed = false;
+			break;
+		case DRM_MODE_CONNECTOR_DisplayPort:
+			amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL);
+			if (!amdgpu_dig_connector)
+				goto failed;
+			amdgpu_connector->con_priv = amdgpu_dig_connector;
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_dp_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dp_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (amdgpu_connector->ddc_bus)
+					has_aux = true;
+				else
+					DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			subpixel_order = SubPixelHorizontalRGB;
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.coherent_mode_property,
+						      1);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_property,
+						   UNDERSCAN_OFF);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_hborder_property,
+						   0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_vborder_property,
+						   0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+			if (amdgpu_audio != 0) {
+				drm_object_attach_property(&amdgpu_connector->base.base,
+							   adev->mode_info.audio_property,
+							   AMDGPU_AUDIO_AUTO);
+			}
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.dither_property,
+						   AMDGPU_FMT_DITHER_DISABLE);
+			connector->interlace_allowed = true;
+			/* in theory with a DP to VGA converter... */
+			connector->doublescan_allowed = false;
+			break;
+		case DRM_MODE_CONNECTOR_eDP:
+			amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL);
+			if (!amdgpu_dig_connector)
+				goto failed;
+			amdgpu_connector->con_priv = amdgpu_dig_connector;
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_edp_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dp_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (amdgpu_connector->ddc_bus)
+					has_aux = true;
+				else
+					DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      dev->mode_config.scaling_mode_property,
+						      DRM_MODE_SCALE_FULLSCREEN);
+			subpixel_order = SubPixelHorizontalRGB;
+			connector->interlace_allowed = false;
+			connector->doublescan_allowed = false;
+			break;
+		case DRM_MODE_CONNECTOR_LVDS:
+			amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL);
+			if (!amdgpu_dig_connector)
+				goto failed;
+			amdgpu_connector->con_priv = amdgpu_dig_connector;
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_lvds_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_lvds_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (!amdgpu_connector->ddc_bus)
+					DRM_ERROR("LVDS: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      dev->mode_config.scaling_mode_property,
+						      DRM_MODE_SCALE_FULLSCREEN);
+			subpixel_order = SubPixelHorizontalRGB;
+			connector->interlace_allowed = false;
+			connector->doublescan_allowed = false;
+			break;
+		}
+	}
+
+	if (amdgpu_connector->hpd.hpd == AMDGPU_HPD_NONE) {
+		if (i2c_bus->valid) {
+			connector->polled = DRM_CONNECTOR_POLL_CONNECT |
+			                    DRM_CONNECTOR_POLL_DISCONNECT;
+		}
+	} else
+		connector->polled = DRM_CONNECTOR_POLL_HPD;
+
+	connector->display_info.subpixel_order = subpixel_order;
+	drm_connector_register(connector);
+
+	if (has_aux)
+		amdgpu_atombios_dp_aux_init(amdgpu_connector);
+
+	return;
+
+failed:
+	drm_connector_cleanup(connector);
+	kfree(connector);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_connectors.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_connectors.h
new file mode 100644
index 00000000000..61fcef15ad7
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_connectors.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_CONNECTORS_H__
+#define __AMDGPU_CONNECTORS_H__
+
+struct edid *amdgpu_connector_edid(struct drm_connector *connector);
+void amdgpu_connector_hotplug(struct drm_connector *connector);
+int amdgpu_connector_get_monitor_bpc(struct drm_connector *connector);
+u16 amdgpu_connector_encoder_get_dp_bridge_encoder_id(struct drm_connector *connector);
+bool amdgpu_connector_is_dp12_capable(struct drm_connector *connector);
+void
+amdgpu_connector_add(struct amdgpu_device *adev,
+		      uint32_t connector_id,
+		      uint32_t supported_device,
+		      int connector_type,
+		      struct amdgpu_i2c_bus_rec *i2c_bus,
+		      uint16_t connector_object_id,
+		      struct amdgpu_hpd *hpd,
+		      struct amdgpu_router *router);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_cs.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_cs.c
new file mode 100644
index 00000000000..dc44fb36b2f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_cs.c
@@ -0,0 +1,1681 @@
+/*
+ * Copyright 2008 Jerome Glisse.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Jerome Glisse <glisse@freedesktop.org>
+ */
+#include <linux/pagemap.h>
+#include <linux/sync_file.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include <drm/drm_syncobj.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+#include "amdgpu_gmc.h"
+
+static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
+				      struct drm_amdgpu_cs_chunk_fence *data,
+				      uint32_t *offset)
+{
+	struct drm_gem_object *gobj;
+	unsigned long size;
+	int r;
+
+	gobj = drm_gem_object_lookup(p->filp, data->handle);
+	if (gobj == NULL)
+		return -EINVAL;
+
+	p->uf_entry.robj = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
+	p->uf_entry.priority = 0;
+	p->uf_entry.tv.bo = &p->uf_entry.robj->tbo;
+	p->uf_entry.tv.shared = true;
+	p->uf_entry.user_pages = NULL;
+
+	drm_gem_object_put_unlocked(gobj);
+
+	size = amdgpu_bo_size(p->uf_entry.robj);
+	if (size != PAGE_SIZE || (data->offset + 8) > size) {
+		r = -EINVAL;
+		goto error_unref;
+	}
+
+	if (amdgpu_ttm_tt_get_usermm(p->uf_entry.robj->tbo.ttm)) {
+		r = -EINVAL;
+		goto error_unref;
+	}
+
+	*offset = data->offset;
+
+	return 0;
+
+error_unref:
+	amdgpu_bo_unref(&p->uf_entry.robj);
+	return r;
+}
+
+static int amdgpu_cs_bo_handles_chunk(struct amdgpu_cs_parser *p,
+				      struct drm_amdgpu_bo_list_in *data)
+{
+	int r;
+	struct drm_amdgpu_bo_list_entry *info = NULL;
+
+	r = amdgpu_bo_create_list_entry_array(data, &info);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_list_create(p->adev, p->filp, info, data->bo_number,
+				  &p->bo_list);
+	if (r)
+		goto error_free;
+
+	kvfree(info);
+	return 0;
+
+error_free:
+	if (info)
+		kvfree(info);
+
+	return r;
+}
+
+static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, union drm_amdgpu_cs *cs)
+{
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	uint64_t *chunk_array_user;
+	uint64_t *chunk_array;
+	unsigned size, num_ibs = 0;
+	uint32_t uf_offset = 0;
+	int i;
+	int ret;
+
+	if (cs->in.num_chunks == 0)
+		return 0;
+
+	chunk_array = kmalloc_array(cs->in.num_chunks, sizeof(uint64_t), GFP_KERNEL);
+	if (!chunk_array)
+		return -ENOMEM;
+
+	p->ctx = amdgpu_ctx_get(fpriv, cs->in.ctx_id);
+	if (!p->ctx) {
+		ret = -EINVAL;
+		goto free_chunk;
+	}
+
+	mutex_lock(&p->ctx->lock);
+
+	/* skip guilty context job */
+	if (atomic_read(&p->ctx->guilty) == 1) {
+		ret = -ECANCELED;
+		goto free_chunk;
+	}
+
+	/* get chunks */
+	chunk_array_user = u64_to_user_ptr(cs->in.chunks);
+	if (copy_from_user(chunk_array, chunk_array_user,
+			   sizeof(uint64_t)*cs->in.num_chunks)) {
+		ret = -EFAULT;
+		goto free_chunk;
+	}
+
+	p->nchunks = cs->in.num_chunks;
+	p->chunks = kmalloc_array(p->nchunks, sizeof(struct amdgpu_cs_chunk),
+			    GFP_KERNEL);
+	if (!p->chunks) {
+		ret = -ENOMEM;
+		goto free_chunk;
+	}
+
+	for (i = 0; i < p->nchunks; i++) {
+		struct drm_amdgpu_cs_chunk __user **chunk_ptr = NULL;
+		struct drm_amdgpu_cs_chunk user_chunk;
+		uint32_t __user *cdata;
+
+		chunk_ptr = u64_to_user_ptr(chunk_array[i]);
+		if (copy_from_user(&user_chunk, chunk_ptr,
+				       sizeof(struct drm_amdgpu_cs_chunk))) {
+			ret = -EFAULT;
+			i--;
+			goto free_partial_kdata;
+		}
+		p->chunks[i].chunk_id = user_chunk.chunk_id;
+		p->chunks[i].length_dw = user_chunk.length_dw;
+
+		size = p->chunks[i].length_dw;
+		cdata = u64_to_user_ptr(user_chunk.chunk_data);
+
+		p->chunks[i].kdata = kvmalloc_array(size, sizeof(uint32_t), GFP_KERNEL);
+		if (p->chunks[i].kdata == NULL) {
+			ret = -ENOMEM;
+			i--;
+			goto free_partial_kdata;
+		}
+		size *= sizeof(uint32_t);
+		if (copy_from_user(p->chunks[i].kdata, cdata, size)) {
+			ret = -EFAULT;
+			goto free_partial_kdata;
+		}
+
+		switch (p->chunks[i].chunk_id) {
+		case AMDGPU_CHUNK_ID_IB:
+			++num_ibs;
+			break;
+
+		case AMDGPU_CHUNK_ID_FENCE:
+			size = sizeof(struct drm_amdgpu_cs_chunk_fence);
+			if (p->chunks[i].length_dw * sizeof(uint32_t) < size) {
+				ret = -EINVAL;
+				goto free_partial_kdata;
+			}
+
+			ret = amdgpu_cs_user_fence_chunk(p, p->chunks[i].kdata,
+							 &uf_offset);
+			if (ret)
+				goto free_partial_kdata;
+
+			break;
+
+		case AMDGPU_CHUNK_ID_BO_HANDLES:
+			size = sizeof(struct drm_amdgpu_bo_list_in);
+			if (p->chunks[i].length_dw * sizeof(uint32_t) < size) {
+				ret = -EINVAL;
+				goto free_partial_kdata;
+			}
+
+			ret = amdgpu_cs_bo_handles_chunk(p, p->chunks[i].kdata);
+			if (ret)
+				goto free_partial_kdata;
+
+			break;
+
+		case AMDGPU_CHUNK_ID_DEPENDENCIES:
+		case AMDGPU_CHUNK_ID_SYNCOBJ_IN:
+		case AMDGPU_CHUNK_ID_SYNCOBJ_OUT:
+			break;
+
+		default:
+			ret = -EINVAL;
+			goto free_partial_kdata;
+		}
+	}
+
+	ret = amdgpu_job_alloc(p->adev, num_ibs, &p->job, vm);
+	if (ret)
+		goto free_all_kdata;
+
+	if (p->ctx->vram_lost_counter != p->job->vram_lost_counter) {
+		ret = -ECANCELED;
+		goto free_all_kdata;
+	}
+
+	if (p->uf_entry.robj)
+		p->job->uf_addr = uf_offset;
+	kfree(chunk_array);
+
+	/* Use this opportunity to fill in task info for the vm */
+	amdgpu_vm_set_task_info(vm);
+
+	return 0;
+
+free_all_kdata:
+	i = p->nchunks - 1;
+free_partial_kdata:
+	for (; i >= 0; i--)
+		kvfree(p->chunks[i].kdata);
+	kfree(p->chunks);
+	p->chunks = NULL;
+	p->nchunks = 0;
+free_chunk:
+	kfree(chunk_array);
+
+	return ret;
+}
+
+/* Convert microseconds to bytes. */
+static u64 us_to_bytes(struct amdgpu_device *adev, s64 us)
+{
+	if (us <= 0 || !adev->mm_stats.log2_max_MBps)
+		return 0;
+
+	/* Since accum_us is incremented by a million per second, just
+	 * multiply it by the number of MB/s to get the number of bytes.
+	 */
+	return us << adev->mm_stats.log2_max_MBps;
+}
+
+static s64 bytes_to_us(struct amdgpu_device *adev, u64 bytes)
+{
+	if (!adev->mm_stats.log2_max_MBps)
+		return 0;
+
+	return bytes >> adev->mm_stats.log2_max_MBps;
+}
+
+/* Returns how many bytes TTM can move right now. If no bytes can be moved,
+ * it returns 0. If it returns non-zero, it's OK to move at least one buffer,
+ * which means it can go over the threshold once. If that happens, the driver
+ * will be in debt and no other buffer migrations can be done until that debt
+ * is repaid.
+ *
+ * This approach allows moving a buffer of any size (it's important to allow
+ * that).
+ *
+ * The currency is simply time in microseconds and it increases as the clock
+ * ticks. The accumulated microseconds (us) are converted to bytes and
+ * returned.
+ */
+static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
+					      u64 *max_bytes,
+					      u64 *max_vis_bytes)
+{
+	s64 time_us, increment_us;
+	u64 free_vram, total_vram, used_vram;
+
+	/* Allow a maximum of 200 accumulated ms. This is basically per-IB
+	 * throttling.
+	 *
+	 * It means that in order to get full max MBps, at least 5 IBs per
+	 * second must be submitted and not more than 200ms apart from each
+	 * other.
+	 */
+	const s64 us_upper_bound = 200000;
+
+	if (!adev->mm_stats.log2_max_MBps) {
+		*max_bytes = 0;
+		*max_vis_bytes = 0;
+		return;
+	}
+
+	total_vram = adev->gmc.real_vram_size - atomic64_read(&adev->vram_pin_size);
+	used_vram = amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+	free_vram = used_vram >= total_vram ? 0 : total_vram - used_vram;
+
+	spin_lock(&adev->mm_stats.lock);
+
+	/* Increase the amount of accumulated us. */
+	time_us = ktime_to_us(ktime_get());
+	increment_us = time_us - adev->mm_stats.last_update_us;
+	adev->mm_stats.last_update_us = time_us;
+	adev->mm_stats.accum_us = min(adev->mm_stats.accum_us + increment_us,
+                                      us_upper_bound);
+
+	/* This prevents the short period of low performance when the VRAM
+	 * usage is low and the driver is in debt or doesn't have enough
+	 * accumulated us to fill VRAM quickly.
+	 *
+	 * The situation can occur in these cases:
+	 * - a lot of VRAM is freed by userspace
+	 * - the presence of a big buffer causes a lot of evictions
+	 *   (solution: split buffers into smaller ones)
+	 *
+	 * If 128 MB or 1/8th of VRAM is free, start filling it now by setting
+	 * accum_us to a positive number.
+	 */
+	if (free_vram >= 128 * 1024 * 1024 || free_vram >= total_vram / 8) {
+		s64 min_us;
+
+		/* Be more aggresive on dGPUs. Try to fill a portion of free
+		 * VRAM now.
+		 */
+		if (!(adev->flags & AMD_IS_APU))
+			min_us = bytes_to_us(adev, free_vram / 4);
+		else
+			min_us = 0; /* Reset accum_us on APUs. */
+
+		adev->mm_stats.accum_us = max(min_us, adev->mm_stats.accum_us);
+	}
+
+	/* This is set to 0 if the driver is in debt to disallow (optional)
+	 * buffer moves.
+	 */
+	*max_bytes = us_to_bytes(adev, adev->mm_stats.accum_us);
+
+	/* Do the same for visible VRAM if half of it is free */
+	if (!amdgpu_gmc_vram_full_visible(&adev->gmc)) {
+		u64 total_vis_vram = adev->gmc.visible_vram_size;
+		u64 used_vis_vram =
+			amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+
+		if (used_vis_vram < total_vis_vram) {
+			u64 free_vis_vram = total_vis_vram - used_vis_vram;
+			adev->mm_stats.accum_us_vis = min(adev->mm_stats.accum_us_vis +
+							  increment_us, us_upper_bound);
+
+			if (free_vis_vram >= total_vis_vram / 2)
+				adev->mm_stats.accum_us_vis =
+					max(bytes_to_us(adev, free_vis_vram / 2),
+					    adev->mm_stats.accum_us_vis);
+		}
+
+		*max_vis_bytes = us_to_bytes(adev, adev->mm_stats.accum_us_vis);
+	} else {
+		*max_vis_bytes = 0;
+	}
+
+	spin_unlock(&adev->mm_stats.lock);
+}
+
+/* Report how many bytes have really been moved for the last command
+ * submission. This can result in a debt that can stop buffer migrations
+ * temporarily.
+ */
+void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes,
+				  u64 num_vis_bytes)
+{
+	spin_lock(&adev->mm_stats.lock);
+	adev->mm_stats.accum_us -= bytes_to_us(adev, num_bytes);
+	adev->mm_stats.accum_us_vis -= bytes_to_us(adev, num_vis_bytes);
+	spin_unlock(&adev->mm_stats.lock);
+}
+
+static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
+				 struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct ttm_operation_ctx ctx = {
+		.interruptible = true,
+		.no_wait_gpu = false,
+		.resv = bo->tbo.resv,
+		.flags = 0
+	};
+	uint32_t domain;
+	int r;
+
+	if (bo->pin_count)
+		return 0;
+
+	/* Don't move this buffer if we have depleted our allowance
+	 * to move it. Don't move anything if the threshold is zero.
+	 */
+	if (p->bytes_moved < p->bytes_moved_threshold) {
+		if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+		    (bo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)) {
+			/* And don't move a CPU_ACCESS_REQUIRED BO to limited
+			 * visible VRAM if we've depleted our allowance to do
+			 * that.
+			 */
+			if (p->bytes_moved_vis < p->bytes_moved_vis_threshold)
+				domain = bo->preferred_domains;
+			else
+				domain = bo->allowed_domains;
+		} else {
+			domain = bo->preferred_domains;
+		}
+	} else {
+		domain = bo->allowed_domains;
+	}
+
+retry:
+	amdgpu_bo_placement_from_domain(bo, domain);
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+
+	p->bytes_moved += ctx.bytes_moved;
+	if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+	    amdgpu_bo_in_cpu_visible_vram(bo))
+		p->bytes_moved_vis += ctx.bytes_moved;
+
+	if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains) {
+		domain = bo->allowed_domains;
+		goto retry;
+	}
+
+	return r;
+}
+
+/* Last resort, try to evict something from the current working set */
+static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
+				struct amdgpu_bo *validated)
+{
+	uint32_t domain = validated->allowed_domains;
+	struct ttm_operation_ctx ctx = { true, false };
+	int r;
+
+	if (!p->evictable)
+		return false;
+
+	for (;&p->evictable->tv.head != &p->validated;
+	     p->evictable = list_prev_entry(p->evictable, tv.head)) {
+
+		struct amdgpu_bo_list_entry *candidate = p->evictable;
+		struct amdgpu_bo *bo = candidate->robj;
+		struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+		bool update_bytes_moved_vis;
+		uint32_t other;
+
+		/* If we reached our current BO we can forget it */
+		if (candidate->robj == validated)
+			break;
+
+		/* We can't move pinned BOs here */
+		if (bo->pin_count)
+			continue;
+
+		other = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type);
+
+		/* Check if this BO is in one of the domains we need space for */
+		if (!(other & domain))
+			continue;
+
+		/* Check if we can move this BO somewhere else */
+		other = bo->allowed_domains & ~domain;
+		if (!other)
+			continue;
+
+		/* Good we can try to move this BO somewhere else */
+		update_bytes_moved_vis =
+				!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+				amdgpu_bo_in_cpu_visible_vram(bo);
+		amdgpu_bo_placement_from_domain(bo, other);
+		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+		p->bytes_moved += ctx.bytes_moved;
+		if (update_bytes_moved_vis)
+			p->bytes_moved_vis += ctx.bytes_moved;
+
+		if (unlikely(r))
+			break;
+
+		p->evictable = list_prev_entry(p->evictable, tv.head);
+		list_move(&candidate->tv.head, &p->validated);
+
+		return true;
+	}
+
+	return false;
+}
+
+static int amdgpu_cs_validate(void *param, struct amdgpu_bo *bo)
+{
+	struct amdgpu_cs_parser *p = param;
+	int r;
+
+	do {
+		r = amdgpu_cs_bo_validate(p, bo);
+	} while (r == -ENOMEM && amdgpu_cs_try_evict(p, bo));
+	if (r)
+		return r;
+
+	if (bo->shadow)
+		r = amdgpu_cs_bo_validate(p, bo->shadow);
+
+	return r;
+}
+
+static int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
+			    struct list_head *validated)
+{
+	struct ttm_operation_ctx ctx = { true, false };
+	struct amdgpu_bo_list_entry *lobj;
+	int r;
+
+	list_for_each_entry(lobj, validated, tv.head) {
+		struct amdgpu_bo *bo = lobj->robj;
+		bool binding_userptr = false;
+		struct mm_struct *usermm;
+
+#ifdef notyet
+		usermm = amdgpu_ttm_tt_get_usermm(bo->tbo.ttm);
+		if (usermm && usermm != current->mm)
+			return -EPERM;
+
+		/* Check if we have user pages and nobody bound the BO already */
+		if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm) &&
+		    lobj->user_pages) {
+			amdgpu_bo_placement_from_domain(bo,
+							AMDGPU_GEM_DOMAIN_CPU);
+			r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+			if (r)
+				return r;
+			amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm,
+						     lobj->user_pages);
+			binding_userptr = true;
+		}
+#endif
+
+		if (p->evictable == lobj)
+			p->evictable = NULL;
+
+		r = amdgpu_cs_validate(p, bo);
+		if (r)
+			return r;
+
+		if (binding_userptr) {
+			kvfree(lobj->user_pages);
+			lobj->user_pages = NULL;
+		}
+	}
+	return 0;
+}
+
+static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
+				union drm_amdgpu_cs *cs)
+{
+	STUB();
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_bo_list_entry *e;
+	struct list_head duplicates;
+	struct amdgpu_bo *gds;
+	struct amdgpu_bo *gws;
+	struct amdgpu_bo *oa;
+	unsigned tries = 10;
+	int r;
+
+	INIT_LIST_HEAD(&p->validated);
+
+	/* p->bo_list could already be assigned if AMDGPU_CHUNK_ID_BO_HANDLES is present */
+	if (cs->in.bo_list_handle) {
+		if (p->bo_list)
+			return -EINVAL;
+
+		r = amdgpu_bo_list_get(fpriv, cs->in.bo_list_handle,
+				       &p->bo_list);
+		if (r)
+			return r;
+	} else if (!p->bo_list) {
+		/* Create a empty bo_list when no handle is provided */
+		r = amdgpu_bo_list_create(p->adev, p->filp, NULL, 0,
+					  &p->bo_list);
+		if (r)
+			return r;
+	}
+
+	amdgpu_bo_list_get_list(p->bo_list, &p->validated);
+	if (p->bo_list->first_userptr != p->bo_list->num_entries)
+		p->mn = amdgpu_mn_get(p->adev, AMDGPU_MN_TYPE_GFX);
+
+	INIT_LIST_HEAD(&duplicates);
+	amdgpu_vm_get_pd_bo(&fpriv->vm, &p->validated, &p->vm_pd);
+
+	if (p->uf_entry.robj && !p->uf_entry.robj->parent)
+		list_add(&p->uf_entry.tv.head, &p->validated);
+
+	while (1) {
+		struct list_head need_pages;
+
+		r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true,
+					   &duplicates);
+		if (unlikely(r != 0)) {
+			if (r != -ERESTARTSYS)
+				DRM_ERROR("ttm_eu_reserve_buffers failed.\n");
+			goto error_free_pages;
+		}
+
+		INIT_LIST_HEAD(&need_pages);
+		amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
+			struct amdgpu_bo *bo = e->robj;
+
+			if (amdgpu_ttm_tt_userptr_invalidated(bo->tbo.ttm,
+				 &e->user_invalidated) && e->user_pages) {
+
+				/* We acquired a page array, but somebody
+				 * invalidated it. Free it and try again
+				 */
+#ifdef notyet
+				release_pages(e->user_pages,
+					      bo->tbo.ttm->num_pages);
+#endif
+				kvfree(e->user_pages);
+				e->user_pages = NULL;
+			}
+
+			if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm) &&
+			    !e->user_pages) {
+				list_del(&e->tv.head);
+				list_add(&e->tv.head, &need_pages);
+
+				amdgpu_bo_unreserve(e->robj);
+			}
+		}
+
+		if (list_empty(&need_pages))
+			break;
+
+		/* Unreserve everything again. */
+		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
+
+		/* We tried too many times, just abort */
+		if (!--tries) {
+			r = -EDEADLK;
+			DRM_ERROR("deadlock in %s\n", __func__);
+			goto error_free_pages;
+		}
+
+		/* Fill the page arrays for all userptrs. */
+		list_for_each_entry(e, &need_pages, tv.head) {
+			struct ttm_tt *ttm = e->robj->tbo.ttm;
+
+			e->user_pages = kvmalloc_array(ttm->num_pages,
+							 sizeof(struct vm_page*),
+							 GFP_KERNEL | __GFP_ZERO);
+			if (!e->user_pages) {
+				r = -ENOMEM;
+				DRM_ERROR("calloc failure in %s\n", __func__);
+				goto error_free_pages;
+			}
+
+			r = amdgpu_ttm_tt_get_user_pages(ttm, e->user_pages);
+			if (r) {
+				DRM_ERROR("amdgpu_ttm_tt_get_user_pages failed.\n");
+				kvfree(e->user_pages);
+				e->user_pages = NULL;
+				goto error_free_pages;
+			}
+		}
+
+		/* And try again. */
+		list_splice(&need_pages, &p->validated);
+	}
+
+	amdgpu_cs_get_threshold_for_moves(p->adev, &p->bytes_moved_threshold,
+					  &p->bytes_moved_vis_threshold);
+	p->bytes_moved = 0;
+	p->bytes_moved_vis = 0;
+	p->evictable = list_last_entry(&p->validated,
+				       struct amdgpu_bo_list_entry,
+				       tv.head);
+
+	r = amdgpu_vm_validate_pt_bos(p->adev, &fpriv->vm,
+				      amdgpu_cs_validate, p);
+	if (r) {
+		DRM_ERROR("amdgpu_vm_validate_pt_bos() failed.\n");
+		goto error_validate;
+	}
+
+	r = amdgpu_cs_list_validate(p, &duplicates);
+	if (r) {
+		DRM_ERROR("amdgpu_cs_list_validate(duplicates) failed.\n");
+		goto error_validate;
+	}
+
+	r = amdgpu_cs_list_validate(p, &p->validated);
+	if (r) {
+		DRM_ERROR("amdgpu_cs_list_validate(validated) failed.\n");
+		goto error_validate;
+	}
+
+	amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved,
+				     p->bytes_moved_vis);
+
+	gds = p->bo_list->gds_obj;
+	gws = p->bo_list->gws_obj;
+	oa = p->bo_list->oa_obj;
+
+	amdgpu_bo_list_for_each_entry(e, p->bo_list)
+		e->bo_va = amdgpu_vm_bo_find(vm, e->robj);
+
+	if (gds) {
+		p->job->gds_base = amdgpu_bo_gpu_offset(gds);
+		p->job->gds_size = amdgpu_bo_size(gds);
+	}
+	if (gws) {
+		p->job->gws_base = amdgpu_bo_gpu_offset(gws);
+		p->job->gws_size = amdgpu_bo_size(gws);
+	}
+	if (oa) {
+		p->job->oa_base = amdgpu_bo_gpu_offset(oa);
+		p->job->oa_size = amdgpu_bo_size(oa);
+	}
+
+	if (!r && p->uf_entry.robj) {
+		struct amdgpu_bo *uf = p->uf_entry.robj;
+
+		r = amdgpu_ttm_alloc_gart(&uf->tbo);
+		p->job->uf_addr += amdgpu_bo_gpu_offset(uf);
+	}
+
+error_validate:
+	if (r)
+		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
+
+error_free_pages:
+
+	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
+		if (!e->user_pages)
+			continue;
+
+#ifdef notyet
+		release_pages(e->user_pages,
+			      e->robj->tbo.ttm->num_pages);
+#endif
+		kvfree(e->user_pages);
+	}
+
+	return r;
+}
+
+static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
+{
+	struct amdgpu_bo_list_entry *e;
+	int r;
+
+	list_for_each_entry(e, &p->validated, tv.head) {
+		struct reservation_object *resv = e->robj->tbo.resv;
+		r = amdgpu_sync_resv(p->adev, &p->job->sync, resv, p->filp,
+				     amdgpu_bo_explicit_sync(e->robj));
+
+		if (r)
+			return r;
+	}
+	return 0;
+}
+
+/**
+ * cs_parser_fini() - clean parser states
+ * @parser:	parser structure holding parsing context.
+ * @error:	error number
+ *
+ * If error is set than unvalidate buffer, otherwise just free memory
+ * used by parsing context.
+ **/
+static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error,
+				  bool backoff)
+{
+	unsigned i;
+
+	if (error && backoff)
+		ttm_eu_backoff_reservation(&parser->ticket,
+					   &parser->validated);
+
+	for (i = 0; i < parser->num_post_dep_syncobjs; i++)
+		drm_syncobj_put(parser->post_dep_syncobjs[i]);
+	kfree(parser->post_dep_syncobjs);
+
+	dma_fence_put(parser->fence);
+
+	if (parser->ctx) {
+		mutex_unlock(&parser->ctx->lock);
+		amdgpu_ctx_put(parser->ctx);
+	}
+	if (parser->bo_list)
+		amdgpu_bo_list_put(parser->bo_list);
+
+	for (i = 0; i < parser->nchunks; i++)
+		kvfree(parser->chunks[i].kdata);
+	kfree(parser->chunks);
+	if (parser->job)
+		amdgpu_job_free(parser->job);
+	amdgpu_bo_unref(&parser->uf_entry.robj);
+}
+
+static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
+{
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_device *adev = p->adev;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_bo_list_entry *e;
+	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_bo *bo;
+	int r;
+
+	r = amdgpu_vm_clear_freed(adev, vm, NULL);
+	if (r)
+		return r;
+
+	r = amdgpu_vm_bo_update(adev, fpriv->prt_va, false);
+	if (r)
+		return r;
+
+	r = amdgpu_sync_fence(adev, &p->job->sync,
+			      fpriv->prt_va->last_pt_update, false);
+	if (r)
+		return r;
+
+	if (amdgpu_sriov_vf(adev)) {
+		struct dma_fence *f;
+
+		bo_va = fpriv->csa_va;
+		BUG_ON(!bo_va);
+		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		if (r)
+			return r;
+
+		f = bo_va->last_pt_update;
+		r = amdgpu_sync_fence(adev, &p->job->sync, f, false);
+		if (r)
+			return r;
+	}
+
+	amdgpu_bo_list_for_each_entry(e, p->bo_list) {
+		struct dma_fence *f;
+
+		/* ignore duplicates */
+		bo = e->robj;
+		if (!bo)
+			continue;
+
+		bo_va = e->bo_va;
+		if (bo_va == NULL)
+			continue;
+
+		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		if (r)
+			return r;
+
+		f = bo_va->last_pt_update;
+		r = amdgpu_sync_fence(adev, &p->job->sync, f, false);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_vm_handle_moved(adev, vm);
+	if (r)
+		return r;
+
+	r = amdgpu_vm_update_directories(adev, vm);
+	if (r)
+		return r;
+
+	r = amdgpu_sync_fence(adev, &p->job->sync, vm->last_update, false);
+	if (r)
+		return r;
+
+	if (amdgpu_vm_debug) {
+		/* Invalidate all BOs to test for userspace bugs */
+		amdgpu_bo_list_for_each_entry(e, p->bo_list) {
+			/* ignore duplicates */
+			if (!e->robj)
+				continue;
+
+			amdgpu_vm_bo_invalidate(adev, e->robj, false);
+		}
+	}
+
+	return r;
+}
+
+static int amdgpu_cs_ib_vm_chunk(struct amdgpu_device *adev,
+				 struct amdgpu_cs_parser *p)
+{
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_ring *ring = p->ring;
+	int r;
+
+	/* Only for UVD/VCE VM emulation */
+	if (p->ring->funcs->parse_cs || p->ring->funcs->patch_cs_in_place) {
+		unsigned i, j;
+
+		for (i = 0, j = 0; i < p->nchunks && j < p->job->num_ibs; i++) {
+			struct drm_amdgpu_cs_chunk_ib *chunk_ib;
+			struct amdgpu_bo_va_mapping *m;
+			struct amdgpu_bo *aobj = NULL;
+			struct amdgpu_cs_chunk *chunk;
+			uint64_t offset, va_start;
+			struct amdgpu_ib *ib;
+			uint8_t *kptr;
+
+			chunk = &p->chunks[i];
+			ib = &p->job->ibs[j];
+			chunk_ib = chunk->kdata;
+
+			if (chunk->chunk_id != AMDGPU_CHUNK_ID_IB)
+				continue;
+
+			va_start = chunk_ib->va_start & AMDGPU_VA_HOLE_MASK;
+			r = amdgpu_cs_find_mapping(p, va_start, &aobj, &m);
+			if (r) {
+				DRM_ERROR("IB va_start is invalid\n");
+				return r;
+			}
+
+			if ((va_start + chunk_ib->ib_bytes) >
+			    (m->last + 1) * AMDGPU_GPU_PAGE_SIZE) {
+				DRM_ERROR("IB va_start+ib_bytes is invalid\n");
+				return -EINVAL;
+			}
+
+			/* the IB should be reserved at this point */
+			r = amdgpu_bo_kmap(aobj, (void **)&kptr);
+			if (r) {
+				return r;
+			}
+
+			offset = m->start * AMDGPU_GPU_PAGE_SIZE;
+			kptr += va_start - offset;
+
+			if (p->ring->funcs->parse_cs) {
+				memcpy(ib->ptr, kptr, chunk_ib->ib_bytes);
+				amdgpu_bo_kunmap(aobj);
+
+				r = amdgpu_ring_parse_cs(ring, p, j);
+				if (r)
+					return r;
+			} else {
+				ib->ptr = (uint32_t *)kptr;
+				r = amdgpu_ring_patch_cs_in_place(ring, p, j);
+				amdgpu_bo_kunmap(aobj);
+				if (r)
+					return r;
+			}
+
+			j++;
+		}
+	}
+
+	if (p->job->vm) {
+		p->job->vm_pd_addr = amdgpu_bo_gpu_offset(vm->root.base.bo);
+
+		r = amdgpu_bo_vm_update_pte(p);
+		if (r)
+			return r;
+
+		r = reservation_object_reserve_shared(vm->root.base.bo->tbo.resv);
+		if (r)
+			return r;
+	}
+
+	return amdgpu_cs_sync_rings(p);
+}
+
+static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
+			     struct amdgpu_cs_parser *parser)
+{
+	struct amdgpu_fpriv *fpriv = parser->filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	int i, j;
+	int r, ce_preempt = 0, de_preempt = 0;
+
+	for (i = 0, j = 0; i < parser->nchunks && j < parser->job->num_ibs; i++) {
+		struct amdgpu_cs_chunk *chunk;
+		struct amdgpu_ib *ib;
+		struct drm_amdgpu_cs_chunk_ib *chunk_ib;
+		struct amdgpu_ring *ring;
+
+		chunk = &parser->chunks[i];
+		ib = &parser->job->ibs[j];
+		chunk_ib = (struct drm_amdgpu_cs_chunk_ib *)chunk->kdata;
+
+		if (chunk->chunk_id != AMDGPU_CHUNK_ID_IB)
+			continue;
+
+		if (chunk_ib->ip_type == AMDGPU_HW_IP_GFX && amdgpu_sriov_vf(adev)) {
+			if (chunk_ib->flags & AMDGPU_IB_FLAG_PREEMPT) {
+				if (chunk_ib->flags & AMDGPU_IB_FLAG_CE)
+					ce_preempt++;
+				else
+					de_preempt++;
+			}
+
+			/* each GFX command submit allows 0 or 1 IB preemptible for CE & DE */
+			if (ce_preempt > 1 || de_preempt > 1)
+				return -EINVAL;
+		}
+
+		r = amdgpu_queue_mgr_map(adev, &parser->ctx->queue_mgr, chunk_ib->ip_type,
+					 chunk_ib->ip_instance, chunk_ib->ring, &ring);
+		if (r)
+			return r;
+
+		if (chunk_ib->flags & AMDGPU_IB_FLAG_PREAMBLE)
+			parser->job->preamble_status |=
+				AMDGPU_PREAMBLE_IB_PRESENT;
+
+		if (parser->ring && parser->ring != ring)
+			return -EINVAL;
+
+		parser->ring = ring;
+
+		r =  amdgpu_ib_get(adev, vm,
+					ring->funcs->parse_cs ? chunk_ib->ib_bytes : 0,
+					ib);
+		if (r) {
+			DRM_ERROR("Failed to get ib !\n");
+			return r;
+		}
+
+		ib->gpu_addr = chunk_ib->va_start;
+		ib->length_dw = chunk_ib->ib_bytes / 4;
+		ib->flags = chunk_ib->flags;
+
+		j++;
+	}
+
+	/* UVD & VCE fw doesn't support user fences */
+	if (parser->job->uf_addr && (
+	    parser->ring->funcs->type == AMDGPU_RING_TYPE_UVD ||
+	    parser->ring->funcs->type == AMDGPU_RING_TYPE_VCE))
+		return -EINVAL;
+
+	return amdgpu_ctx_wait_prev_fence(parser->ctx, parser->ring->idx);
+}
+
+static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p,
+				       struct amdgpu_cs_chunk *chunk)
+{
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	unsigned num_deps;
+	int i, r;
+	struct drm_amdgpu_cs_chunk_dep *deps;
+
+	deps = (struct drm_amdgpu_cs_chunk_dep *)chunk->kdata;
+	num_deps = chunk->length_dw * 4 /
+		sizeof(struct drm_amdgpu_cs_chunk_dep);
+
+	for (i = 0; i < num_deps; ++i) {
+		struct amdgpu_ring *ring;
+		struct amdgpu_ctx *ctx;
+		struct dma_fence *fence;
+
+		ctx = amdgpu_ctx_get(fpriv, deps[i].ctx_id);
+		if (ctx == NULL)
+			return -EINVAL;
+
+		r = amdgpu_queue_mgr_map(p->adev, &ctx->queue_mgr,
+					 deps[i].ip_type,
+					 deps[i].ip_instance,
+					 deps[i].ring, &ring);
+		if (r) {
+			amdgpu_ctx_put(ctx);
+			return r;
+		}
+
+		fence = amdgpu_ctx_get_fence(ctx, ring,
+					     deps[i].handle);
+		if (IS_ERR(fence)) {
+			r = PTR_ERR(fence);
+			amdgpu_ctx_put(ctx);
+			return r;
+		} else if (fence) {
+			r = amdgpu_sync_fence(p->adev, &p->job->sync, fence,
+					true);
+			dma_fence_put(fence);
+			amdgpu_ctx_put(ctx);
+			if (r)
+				return r;
+		}
+	}
+	return 0;
+}
+
+static int amdgpu_syncobj_lookup_and_add_to_sync(struct amdgpu_cs_parser *p,
+						 uint32_t handle)
+{
+	int r;
+	struct dma_fence *fence;
+	r = drm_syncobj_find_fence(p->filp, handle, &fence);
+	if (r)
+		return r;
+
+	r = amdgpu_sync_fence(p->adev, &p->job->sync, fence, true);
+	dma_fence_put(fence);
+
+	return r;
+}
+
+static int amdgpu_cs_process_syncobj_in_dep(struct amdgpu_cs_parser *p,
+					    struct amdgpu_cs_chunk *chunk)
+{
+	unsigned num_deps;
+	int i, r;
+	struct drm_amdgpu_cs_chunk_sem *deps;
+
+	deps = (struct drm_amdgpu_cs_chunk_sem *)chunk->kdata;
+	num_deps = chunk->length_dw * 4 /
+		sizeof(struct drm_amdgpu_cs_chunk_sem);
+
+	for (i = 0; i < num_deps; ++i) {
+		r = amdgpu_syncobj_lookup_and_add_to_sync(p, deps[i].handle);
+		if (r)
+			return r;
+	}
+	return 0;
+}
+
+static int amdgpu_cs_process_syncobj_out_dep(struct amdgpu_cs_parser *p,
+					     struct amdgpu_cs_chunk *chunk)
+{
+	unsigned num_deps;
+	int i;
+	struct drm_amdgpu_cs_chunk_sem *deps;
+	deps = (struct drm_amdgpu_cs_chunk_sem *)chunk->kdata;
+	num_deps = chunk->length_dw * 4 /
+		sizeof(struct drm_amdgpu_cs_chunk_sem);
+
+	p->post_dep_syncobjs = kmalloc_array(num_deps,
+					     sizeof(struct drm_syncobj *),
+					     GFP_KERNEL);
+	p->num_post_dep_syncobjs = 0;
+
+	if (!p->post_dep_syncobjs)
+		return -ENOMEM;
+
+	for (i = 0; i < num_deps; ++i) {
+		p->post_dep_syncobjs[i] = drm_syncobj_find(p->filp, deps[i].handle);
+		if (!p->post_dep_syncobjs[i])
+			return -EINVAL;
+		p->num_post_dep_syncobjs++;
+	}
+	return 0;
+}
+
+static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
+				  struct amdgpu_cs_parser *p)
+{
+	int i, r;
+
+	for (i = 0; i < p->nchunks; ++i) {
+		struct amdgpu_cs_chunk *chunk;
+
+		chunk = &p->chunks[i];
+
+		if (chunk->chunk_id == AMDGPU_CHUNK_ID_DEPENDENCIES) {
+			r = amdgpu_cs_process_fence_dep(p, chunk);
+			if (r)
+				return r;
+		} else if (chunk->chunk_id == AMDGPU_CHUNK_ID_SYNCOBJ_IN) {
+			r = amdgpu_cs_process_syncobj_in_dep(p, chunk);
+			if (r)
+				return r;
+		} else if (chunk->chunk_id == AMDGPU_CHUNK_ID_SYNCOBJ_OUT) {
+			r = amdgpu_cs_process_syncobj_out_dep(p, chunk);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+static void amdgpu_cs_post_dependencies(struct amdgpu_cs_parser *p)
+{
+	int i;
+
+	for (i = 0; i < p->num_post_dep_syncobjs; ++i)
+		drm_syncobj_replace_fence(p->post_dep_syncobjs[i], p->fence);
+}
+
+static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
+			    union drm_amdgpu_cs *cs)
+{
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_ring *ring = p->ring;
+	struct drm_sched_entity *entity = &p->ctx->rings[ring->idx].entity;
+	enum drm_sched_priority priority;
+	struct amdgpu_bo_list_entry *e;
+	struct amdgpu_job *job;
+	uint64_t seq;
+
+	int r;
+
+	job = p->job;
+	p->job = NULL;
+
+	r = drm_sched_job_init(&job->base, entity, p->filp);
+	if (r)
+		goto error_unlock;
+
+	/* No memory allocation is allowed while holding the mn lock */
+	amdgpu_mn_lock(p->mn);
+	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
+		struct amdgpu_bo *bo = e->robj;
+
+		if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm)) {
+			r = -ERESTARTSYS;
+			goto error_abort;
+		}
+	}
+
+	job->owner = p->filp;
+	p->fence = dma_fence_get(&job->base.s_fence->finished);
+
+	r = amdgpu_ctx_add_fence(p->ctx, ring, p->fence, &seq);
+	if (r) {
+		dma_fence_put(p->fence);
+		dma_fence_put(&job->base.s_fence->finished);
+		amdgpu_job_free(job);
+		amdgpu_mn_unlock(p->mn);
+		return r;
+	}
+
+	amdgpu_cs_post_dependencies(p);
+
+	if ((job->preamble_status & AMDGPU_PREAMBLE_IB_PRESENT) &&
+	    !p->ctx->preamble_presented) {
+		job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT_FIRST;
+		p->ctx->preamble_presented = true;
+	}
+
+	cs->out.handle = seq;
+	job->uf_sequence = seq;
+
+	amdgpu_job_free_resources(job);
+
+	trace_amdgpu_cs_ioctl(job);
+	amdgpu_vm_bo_trace_cs(&fpriv->vm, &p->ticket);
+	priority = job->base.s_priority;
+	drm_sched_entity_push_job(&job->base, entity);
+
+	ring = to_amdgpu_ring(entity->rq->sched);
+	amdgpu_ring_priority_get(ring, priority);
+
+	ttm_eu_fence_buffer_objects(&p->ticket, &p->validated, p->fence);
+	amdgpu_mn_unlock(p->mn);
+
+	return 0;
+
+error_abort:
+	dma_fence_put(&job->base.s_fence->finished);
+	job->base.s_fence = NULL;
+	amdgpu_mn_unlock(p->mn);
+
+error_unlock:
+	amdgpu_job_free(job);
+	return r;
+}
+
+int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	union drm_amdgpu_cs *cs = data;
+	struct amdgpu_cs_parser parser = {};
+	bool reserved_buffers = false;
+	int i, r;
+
+	if (!adev->accel_working)
+		return -EBUSY;
+
+	parser.adev = adev;
+	parser.filp = filp;
+
+	r = amdgpu_cs_parser_init(&parser, data);
+	if (r) {
+		DRM_ERROR("Failed to initialize parser !\n");
+		goto out;
+	}
+
+	r = amdgpu_cs_ib_fill(adev, &parser);
+	if (r)
+		goto out;
+
+	r = amdgpu_cs_parser_bos(&parser, data);
+	if (r) {
+		if (r == -ENOMEM)
+			DRM_ERROR("Not enough memory for command submission!\n");
+		else if (r != -ERESTARTSYS)
+			DRM_ERROR("Failed to process the buffer list %d!\n", r);
+		goto out;
+	}
+
+	reserved_buffers = true;
+
+	r = amdgpu_cs_dependencies(adev, &parser);
+	if (r) {
+		DRM_ERROR("Failed in the dependencies handling %d!\n", r);
+		goto out;
+	}
+
+	for (i = 0; i < parser.job->num_ibs; i++)
+		trace_amdgpu_cs(&parser, i);
+
+	r = amdgpu_cs_ib_vm_chunk(adev, &parser);
+	if (r)
+		goto out;
+
+	r = amdgpu_cs_submit(&parser, cs);
+
+out:
+	amdgpu_cs_parser_fini(&parser, r, reserved_buffers);
+	return r;
+}
+
+/**
+ * amdgpu_cs_wait_ioctl - wait for a command submission to finish
+ *
+ * @dev: drm device
+ * @data: data from userspace
+ * @filp: file private
+ *
+ * Wait for the command submission identified by handle to finish.
+ */
+int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data,
+			 struct drm_file *filp)
+{
+	union drm_amdgpu_wait_cs *wait = data;
+	struct amdgpu_device *adev = dev->dev_private;
+	unsigned long timeout = amdgpu_gem_timeout(wait->in.timeout);
+	struct amdgpu_ring *ring = NULL;
+	struct amdgpu_ctx *ctx;
+	struct dma_fence *fence;
+	long r;
+
+	ctx = amdgpu_ctx_get(filp->driver_priv, wait->in.ctx_id);
+	if (ctx == NULL)
+		return -EINVAL;
+
+	r = amdgpu_queue_mgr_map(adev, &ctx->queue_mgr,
+				 wait->in.ip_type, wait->in.ip_instance,
+				 wait->in.ring, &ring);
+	if (r) {
+		amdgpu_ctx_put(ctx);
+		return r;
+	}
+
+	fence = amdgpu_ctx_get_fence(ctx, ring, wait->in.handle);
+	if (IS_ERR(fence))
+		r = PTR_ERR(fence);
+	else if (fence) {
+		r = dma_fence_wait_timeout(fence, true, timeout);
+		if (r > 0 && fence->error)
+			r = fence->error;
+		dma_fence_put(fence);
+	} else
+		r = 1;
+
+	amdgpu_ctx_put(ctx);
+	if (r < 0)
+		return r;
+
+	memset(wait, 0, sizeof(*wait));
+	wait->out.status = (r == 0);
+
+	return 0;
+}
+
+/**
+ * amdgpu_cs_get_fence - helper to get fence from drm_amdgpu_fence
+ *
+ * @adev: amdgpu device
+ * @filp: file private
+ * @user: drm_amdgpu_fence copied from user space
+ */
+static struct dma_fence *amdgpu_cs_get_fence(struct amdgpu_device *adev,
+					     struct drm_file *filp,
+					     struct drm_amdgpu_fence *user)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_ctx *ctx;
+	struct dma_fence *fence;
+	int r;
+
+	ctx = amdgpu_ctx_get(filp->driver_priv, user->ctx_id);
+	if (ctx == NULL)
+		return ERR_PTR(-EINVAL);
+
+	r = amdgpu_queue_mgr_map(adev, &ctx->queue_mgr, user->ip_type,
+				 user->ip_instance, user->ring, &ring);
+	if (r) {
+		amdgpu_ctx_put(ctx);
+		return ERR_PTR(r);
+	}
+
+	fence = amdgpu_ctx_get_fence(ctx, ring, user->seq_no);
+	amdgpu_ctx_put(ctx);
+
+	return fence;
+}
+
+int amdgpu_cs_fence_to_handle_ioctl(struct drm_device *dev, void *data,
+				    struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	union drm_amdgpu_fence_to_handle *info = data;
+	struct dma_fence *fence;
+	struct drm_syncobj *syncobj;
+	struct sync_file *sync_file;
+	int fd, r;
+
+	fence = amdgpu_cs_get_fence(adev, filp, &info->in.fence);
+	if (IS_ERR(fence))
+		return PTR_ERR(fence);
+
+	switch (info->in.what) {
+	case AMDGPU_FENCE_TO_HANDLE_GET_SYNCOBJ:
+		r = drm_syncobj_create(&syncobj, 0, fence);
+		dma_fence_put(fence);
+		if (r)
+			return r;
+		r = drm_syncobj_get_handle(filp, syncobj, &info->out.handle);
+		drm_syncobj_put(syncobj);
+		return r;
+
+	case AMDGPU_FENCE_TO_HANDLE_GET_SYNCOBJ_FD:
+		r = drm_syncobj_create(&syncobj, 0, fence);
+		dma_fence_put(fence);
+		if (r)
+			return r;
+		r = drm_syncobj_get_fd(syncobj, (int*)&info->out.handle);
+		drm_syncobj_put(syncobj);
+		return r;
+
+	case AMDGPU_FENCE_TO_HANDLE_GET_SYNC_FILE_FD:
+		fd = get_unused_fd_flags(O_CLOEXEC);
+		if (fd < 0) {
+			dma_fence_put(fence);
+			return fd;
+		}
+
+		sync_file = sync_file_create(fence);
+		dma_fence_put(fence);
+		if (!sync_file) {
+			put_unused_fd(fd);
+			return -ENOMEM;
+		}
+
+		fd_install(fd, sync_file->file);
+		info->out.handle = fd;
+		return 0;
+
+	default:
+		return -EINVAL;
+	}
+}
+
+/**
+ * amdgpu_cs_wait_all_fence - wait on all fences to signal
+ *
+ * @adev: amdgpu device
+ * @filp: file private
+ * @wait: wait parameters
+ * @fences: array of drm_amdgpu_fence
+ */
+static int amdgpu_cs_wait_all_fences(struct amdgpu_device *adev,
+				     struct drm_file *filp,
+				     union drm_amdgpu_wait_fences *wait,
+				     struct drm_amdgpu_fence *fences)
+{
+	uint32_t fence_count = wait->in.fence_count;
+	unsigned int i;
+	long r = 1;
+
+	for (i = 0; i < fence_count; i++) {
+		struct dma_fence *fence;
+		unsigned long timeout = amdgpu_gem_timeout(wait->in.timeout_ns);
+
+		fence = amdgpu_cs_get_fence(adev, filp, &fences[i]);
+		if (IS_ERR(fence))
+			return PTR_ERR(fence);
+		else if (!fence)
+			continue;
+
+		r = dma_fence_wait_timeout(fence, true, timeout);
+		dma_fence_put(fence);
+		if (r < 0)
+			return r;
+
+		if (r == 0)
+			break;
+
+		if (fence->error)
+			return fence->error;
+	}
+
+	memset(wait, 0, sizeof(*wait));
+	wait->out.status = (r > 0);
+
+	return 0;
+}
+
+/**
+ * amdgpu_cs_wait_any_fence - wait on any fence to signal
+ *
+ * @adev: amdgpu device
+ * @filp: file private
+ * @wait: wait parameters
+ * @fences: array of drm_amdgpu_fence
+ */
+static int amdgpu_cs_wait_any_fence(struct amdgpu_device *adev,
+				    struct drm_file *filp,
+				    union drm_amdgpu_wait_fences *wait,
+				    struct drm_amdgpu_fence *fences)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	unsigned long timeout = amdgpu_gem_timeout(wait->in.timeout_ns);
+	uint32_t fence_count = wait->in.fence_count;
+	uint32_t first = ~0;
+	struct dma_fence **array;
+	unsigned int i;
+	long r;
+
+	/* Prepare the fence array */
+	array = kcalloc(fence_count, sizeof(struct dma_fence *), GFP_KERNEL);
+
+	if (array == NULL)
+		return -ENOMEM;
+
+	for (i = 0; i < fence_count; i++) {
+		struct dma_fence *fence;
+
+		fence = amdgpu_cs_get_fence(adev, filp, &fences[i]);
+		if (IS_ERR(fence)) {
+			r = PTR_ERR(fence);
+			goto err_free_fence_array;
+		} else if (fence) {
+			array[i] = fence;
+		} else { /* NULL, the fence has been already signaled */
+			r = 1;
+			first = i;
+			goto out;
+		}
+	}
+
+	r = dma_fence_wait_any_timeout(array, fence_count, true, timeout,
+				       &first);
+	if (r < 0)
+		goto err_free_fence_array;
+
+out:
+	memset(wait, 0, sizeof(*wait));
+	wait->out.status = (r > 0);
+	wait->out.first_signaled = first;
+
+	if (first < fence_count && array[first])
+		r = array[first]->error;
+	else
+		r = 0;
+
+err_free_fence_array:
+	for (i = 0; i < fence_count; i++)
+		dma_fence_put(array[i]);
+	kfree(array);
+
+	return r;
+#endif
+}
+
+/**
+ * amdgpu_cs_wait_fences_ioctl - wait for multiple command submissions to finish
+ *
+ * @dev: drm device
+ * @data: data from userspace
+ * @filp: file private
+ */
+int amdgpu_cs_wait_fences_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	union drm_amdgpu_wait_fences *wait = data;
+	uint32_t fence_count = wait->in.fence_count;
+	struct drm_amdgpu_fence *fences_user;
+	struct drm_amdgpu_fence *fences;
+	int r;
+
+	/* Get the fences from userspace */
+	fences = kmalloc_array(fence_count, sizeof(struct drm_amdgpu_fence),
+			GFP_KERNEL);
+	if (fences == NULL)
+		return -ENOMEM;
+
+	fences_user = u64_to_user_ptr(wait->in.fences);
+	if (copy_from_user(fences, fences_user,
+		sizeof(struct drm_amdgpu_fence) * fence_count)) {
+		r = -EFAULT;
+		goto err_free_fences;
+	}
+
+	if (wait->in.wait_all)
+		r = amdgpu_cs_wait_all_fences(adev, filp, wait, fences);
+	else
+		r = amdgpu_cs_wait_any_fence(adev, filp, wait, fences);
+
+err_free_fences:
+	kfree(fences);
+
+	return r;
+}
+
+/**
+ * amdgpu_cs_find_bo_va - find bo_va for VM address
+ *
+ * @parser: command submission parser context
+ * @addr: VM address
+ * @bo: resulting BO of the mapping found
+ *
+ * Search the buffer objects in the command submission context for a certain
+ * virtual memory address. Returns allocation structure when found, NULL
+ * otherwise.
+ */
+int amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
+			   uint64_t addr, struct amdgpu_bo **bo,
+			   struct amdgpu_bo_va_mapping **map)
+{
+	struct amdgpu_fpriv *fpriv = parser->filp->driver_priv;
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_bo_va_mapping *mapping;
+	int r;
+
+	addr /= AMDGPU_GPU_PAGE_SIZE;
+
+	mapping = amdgpu_vm_bo_lookup_mapping(vm, addr);
+	if (!mapping || !mapping->bo_va || !mapping->bo_va->base.bo)
+		return -EINVAL;
+
+	*bo = mapping->bo_va->base.bo;
+	*map = mapping;
+
+	/* Double check that the BO is reserved by this CS */
+	if (READ_ONCE((*bo)->tbo.resv->lock.ctx) != &parser->ticket)
+		return -EINVAL;
+
+	if (!((*bo)->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS)) {
+		(*bo)->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+		amdgpu_bo_placement_from_domain(*bo, (*bo)->allowed_domains);
+		r = ttm_bo_validate(&(*bo)->tbo, &(*bo)->placement, &ctx);
+		if (r)
+			return r;
+	}
+
+	return amdgpu_ttm_alloc_gart(&(*bo)->tbo);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ctx.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ctx.c
new file mode 100644
index 00000000000..ec09efecbc6
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ctx.c
@@ -0,0 +1,521 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: monk liu <monk.liu@amd.com>
+ */
+
+#include <drm/drmP.h>
+#ifdef notyet
+#include <drm/drm_auth.h>
+#endif
+#include "amdgpu.h"
+#include "amdgpu_sched.h"
+
+static int amdgpu_ctx_priority_permit(struct drm_file *filp,
+				      enum drm_sched_priority priority)
+{
+	/* NORMAL and below are accessible by everyone */
+	if (priority <= DRM_SCHED_PRIORITY_NORMAL)
+		return 0;
+
+#ifdef notyet
+	if (capable(CAP_SYS_NICE))
+		return 0;
+#endif
+
+	if (drm_is_current_master(filp))
+		return 0;
+
+	return -EACCES;
+}
+
+static int amdgpu_ctx_init(struct amdgpu_device *adev,
+			   enum drm_sched_priority priority,
+			   struct drm_file *filp,
+			   struct amdgpu_ctx *ctx)
+{
+	unsigned i, j;
+	int r;
+
+	if (priority < 0 || priority >= DRM_SCHED_PRIORITY_MAX)
+		return -EINVAL;
+
+	r = amdgpu_ctx_priority_permit(filp, priority);
+	if (r)
+		return r;
+
+	memset(ctx, 0, sizeof(*ctx));
+	ctx->adev = adev;
+	kref_init(&ctx->refcount);
+	mtx_init(&ctx->ring_lock, IPL_TTY);
+	ctx->fences = kcalloc(amdgpu_sched_jobs * AMDGPU_MAX_RINGS,
+			      sizeof(struct dma_fence*), GFP_KERNEL);
+	if (!ctx->fences)
+		return -ENOMEM;
+
+	rw_init(&ctx->lock, "amctxlk");
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		ctx->rings[i].sequence = 1;
+		ctx->rings[i].fences = &ctx->fences[amdgpu_sched_jobs * i];
+	}
+
+	ctx->reset_counter = atomic_read(&adev->gpu_reset_counter);
+	ctx->reset_counter_query = ctx->reset_counter;
+	ctx->vram_lost_counter = atomic_read(&adev->vram_lost_counter);
+	ctx->init_priority = priority;
+	ctx->override_priority = DRM_SCHED_PRIORITY_UNSET;
+
+	/* create context entity for each ring */
+	for (i = 0; i < adev->num_rings; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		struct drm_sched_rq *rq;
+
+		rq = &ring->sched.sched_rq[priority];
+
+		if (ring == &adev->gfx.kiq.ring)
+			continue;
+
+		r = drm_sched_entity_init(&ctx->rings[i].entity,
+					  &rq, 1, &ctx->guilty);
+		if (r)
+			goto failed;
+	}
+
+	r = amdgpu_queue_mgr_init(adev, &ctx->queue_mgr);
+	if (r)
+		goto failed;
+
+	return 0;
+
+failed:
+	for (j = 0; j < i; j++)
+		drm_sched_entity_destroy(&ctx->rings[j].entity);
+	kfree(ctx->fences);
+	ctx->fences = NULL;
+	return r;
+}
+
+static void amdgpu_ctx_fini(struct kref *ref)
+{
+	struct amdgpu_ctx *ctx = container_of(ref, struct amdgpu_ctx, refcount);
+	struct amdgpu_device *adev = ctx->adev;
+	unsigned i, j;
+
+	if (!adev)
+		return;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
+		for (j = 0; j < amdgpu_sched_jobs; ++j)
+			dma_fence_put(ctx->rings[i].fences[j]);
+	kfree(ctx->fences);
+	ctx->fences = NULL;
+
+	amdgpu_queue_mgr_fini(adev, &ctx->queue_mgr);
+
+	mutex_destroy(&ctx->lock);
+
+	kfree(ctx);
+}
+
+static int amdgpu_ctx_alloc(struct amdgpu_device *adev,
+			    struct amdgpu_fpriv *fpriv,
+			    struct drm_file *filp,
+			    enum drm_sched_priority priority,
+			    uint32_t *id)
+{
+	struct amdgpu_ctx_mgr *mgr = &fpriv->ctx_mgr;
+	struct amdgpu_ctx *ctx;
+	int r;
+
+	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+
+	mutex_lock(&mgr->lock);
+	r = idr_alloc(&mgr->ctx_handles, ctx, 1, 0, GFP_KERNEL);
+	if (r < 0) {
+		mutex_unlock(&mgr->lock);
+		kfree(ctx);
+		return r;
+	}
+
+	*id = (uint32_t)r;
+	r = amdgpu_ctx_init(adev, priority, filp, ctx);
+	if (r) {
+		idr_remove(&mgr->ctx_handles, *id);
+		*id = 0;
+		kfree(ctx);
+	}
+	mutex_unlock(&mgr->lock);
+	return r;
+}
+
+static void amdgpu_ctx_do_release(struct kref *ref)
+{
+	struct amdgpu_ctx *ctx;
+	u32 i;
+
+	ctx = container_of(ref, struct amdgpu_ctx, refcount);
+
+	for (i = 0; i < ctx->adev->num_rings; i++) {
+
+		if (ctx->adev->rings[i] == &ctx->adev->gfx.kiq.ring)
+			continue;
+
+		drm_sched_entity_destroy(&ctx->rings[i].entity);
+	}
+
+	amdgpu_ctx_fini(ref);
+}
+
+static int amdgpu_ctx_free(struct amdgpu_fpriv *fpriv, uint32_t id)
+{
+	struct amdgpu_ctx_mgr *mgr = &fpriv->ctx_mgr;
+	struct amdgpu_ctx *ctx;
+
+	mutex_lock(&mgr->lock);
+	ctx = idr_remove(&mgr->ctx_handles, id);
+	if (ctx)
+		kref_put(&ctx->refcount, amdgpu_ctx_do_release);
+	mutex_unlock(&mgr->lock);
+	return ctx ? 0 : -EINVAL;
+}
+
+static int amdgpu_ctx_query(struct amdgpu_device *adev,
+			    struct amdgpu_fpriv *fpriv, uint32_t id,
+			    union drm_amdgpu_ctx_out *out)
+{
+	struct amdgpu_ctx *ctx;
+	struct amdgpu_ctx_mgr *mgr;
+	unsigned reset_counter;
+
+	if (!fpriv)
+		return -EINVAL;
+
+	mgr = &fpriv->ctx_mgr;
+	mutex_lock(&mgr->lock);
+	ctx = idr_find(&mgr->ctx_handles, id);
+	if (!ctx) {
+		mutex_unlock(&mgr->lock);
+		return -EINVAL;
+	}
+
+	/* TODO: these two are always zero */
+	out->state.flags = 0x0;
+	out->state.hangs = 0x0;
+
+	/* determine if a GPU reset has occured since the last call */
+	reset_counter = atomic_read(&adev->gpu_reset_counter);
+	/* TODO: this should ideally return NO, GUILTY, or INNOCENT. */
+	if (ctx->reset_counter_query == reset_counter)
+		out->state.reset_status = AMDGPU_CTX_NO_RESET;
+	else
+		out->state.reset_status = AMDGPU_CTX_UNKNOWN_RESET;
+	ctx->reset_counter_query = reset_counter;
+
+	mutex_unlock(&mgr->lock);
+	return 0;
+}
+
+static int amdgpu_ctx_query2(struct amdgpu_device *adev,
+	struct amdgpu_fpriv *fpriv, uint32_t id,
+	union drm_amdgpu_ctx_out *out)
+{
+	struct amdgpu_ctx *ctx;
+	struct amdgpu_ctx_mgr *mgr;
+
+	if (!fpriv)
+		return -EINVAL;
+
+	mgr = &fpriv->ctx_mgr;
+	mutex_lock(&mgr->lock);
+	ctx = idr_find(&mgr->ctx_handles, id);
+	if (!ctx) {
+		mutex_unlock(&mgr->lock);
+		return -EINVAL;
+	}
+
+	out->state.flags = 0x0;
+	out->state.hangs = 0x0;
+
+	if (ctx->reset_counter != atomic_read(&adev->gpu_reset_counter))
+		out->state.flags |= AMDGPU_CTX_QUERY2_FLAGS_RESET;
+
+	if (ctx->vram_lost_counter != atomic_read(&adev->vram_lost_counter))
+		out->state.flags |= AMDGPU_CTX_QUERY2_FLAGS_VRAMLOST;
+
+	if (atomic_read(&ctx->guilty))
+		out->state.flags |= AMDGPU_CTX_QUERY2_FLAGS_GUILTY;
+
+	mutex_unlock(&mgr->lock);
+	return 0;
+}
+
+int amdgpu_ctx_ioctl(struct drm_device *dev, void *data,
+		     struct drm_file *filp)
+{
+	int r;
+	uint32_t id;
+	enum drm_sched_priority priority;
+
+	union drm_amdgpu_ctx *args = data;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+
+	r = 0;
+	id = args->in.ctx_id;
+	priority = amdgpu_to_sched_priority(args->in.priority);
+
+	/* For backwards compatibility reasons, we need to accept
+	 * ioctls with garbage in the priority field */
+	if (priority == DRM_SCHED_PRIORITY_INVALID)
+		priority = DRM_SCHED_PRIORITY_NORMAL;
+
+	switch (args->in.op) {
+	case AMDGPU_CTX_OP_ALLOC_CTX:
+		r = amdgpu_ctx_alloc(adev, fpriv, filp, priority, &id);
+		args->out.alloc.ctx_id = id;
+		break;
+	case AMDGPU_CTX_OP_FREE_CTX:
+		r = amdgpu_ctx_free(fpriv, id);
+		break;
+	case AMDGPU_CTX_OP_QUERY_STATE:
+		r = amdgpu_ctx_query(adev, fpriv, id, &args->out);
+		break;
+	case AMDGPU_CTX_OP_QUERY_STATE2:
+		r = amdgpu_ctx_query2(adev, fpriv, id, &args->out);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return r;
+}
+
+struct amdgpu_ctx *amdgpu_ctx_get(struct amdgpu_fpriv *fpriv, uint32_t id)
+{
+	struct amdgpu_ctx *ctx;
+	struct amdgpu_ctx_mgr *mgr;
+
+	if (!fpriv)
+		return NULL;
+
+	mgr = &fpriv->ctx_mgr;
+
+	mutex_lock(&mgr->lock);
+	ctx = idr_find(&mgr->ctx_handles, id);
+	if (ctx)
+		kref_get(&ctx->refcount);
+	mutex_unlock(&mgr->lock);
+	return ctx;
+}
+
+int amdgpu_ctx_put(struct amdgpu_ctx *ctx)
+{
+	if (ctx == NULL)
+		return -EINVAL;
+
+	kref_put(&ctx->refcount, amdgpu_ctx_do_release);
+	return 0;
+}
+
+int amdgpu_ctx_add_fence(struct amdgpu_ctx *ctx, struct amdgpu_ring *ring,
+			      struct dma_fence *fence, uint64_t* handler)
+{
+	struct amdgpu_ctx_ring *cring = & ctx->rings[ring->idx];
+	uint64_t seq = cring->sequence;
+	unsigned idx = 0;
+	struct dma_fence *other = NULL;
+
+	idx = seq & (amdgpu_sched_jobs - 1);
+	other = cring->fences[idx];
+	if (other)
+		BUG_ON(!dma_fence_is_signaled(other));
+
+	dma_fence_get(fence);
+
+	spin_lock(&ctx->ring_lock);
+	cring->fences[idx] = fence;
+	cring->sequence++;
+	spin_unlock(&ctx->ring_lock);
+
+	dma_fence_put(other);
+	if (handler)
+		*handler = seq;
+
+	return 0;
+}
+
+struct dma_fence *amdgpu_ctx_get_fence(struct amdgpu_ctx *ctx,
+				       struct amdgpu_ring *ring, uint64_t seq)
+{
+	struct amdgpu_ctx_ring *cring = & ctx->rings[ring->idx];
+	struct dma_fence *fence;
+
+	spin_lock(&ctx->ring_lock);
+
+	if (seq == ~0ull)
+		seq = ctx->rings[ring->idx].sequence - 1;
+
+	if (seq >= cring->sequence) {
+		spin_unlock(&ctx->ring_lock);
+		return ERR_PTR(-EINVAL);
+	}
+
+
+	if (seq + amdgpu_sched_jobs < cring->sequence) {
+		spin_unlock(&ctx->ring_lock);
+		return NULL;
+	}
+
+	fence = dma_fence_get(cring->fences[seq & (amdgpu_sched_jobs - 1)]);
+	spin_unlock(&ctx->ring_lock);
+
+	return fence;
+}
+
+void amdgpu_ctx_priority_override(struct amdgpu_ctx *ctx,
+				  enum drm_sched_priority priority)
+{
+	int i;
+	struct amdgpu_device *adev = ctx->adev;
+	struct drm_sched_rq *rq;
+	struct drm_sched_entity *entity;
+	struct amdgpu_ring *ring;
+	enum drm_sched_priority ctx_prio;
+
+	ctx->override_priority = priority;
+
+	ctx_prio = (ctx->override_priority == DRM_SCHED_PRIORITY_UNSET) ?
+			ctx->init_priority : ctx->override_priority;
+
+	for (i = 0; i < adev->num_rings; i++) {
+		ring = adev->rings[i];
+		entity = &ctx->rings[i].entity;
+		rq = &ring->sched.sched_rq[ctx_prio];
+
+		if (ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
+			continue;
+
+		drm_sched_entity_set_rq(entity, rq);
+	}
+}
+
+int amdgpu_ctx_wait_prev_fence(struct amdgpu_ctx *ctx, unsigned ring_id)
+{
+	struct amdgpu_ctx_ring *cring = &ctx->rings[ring_id];
+	unsigned idx = cring->sequence & (amdgpu_sched_jobs - 1);
+	struct dma_fence *other = cring->fences[idx];
+
+	if (other) {
+		signed long r;
+		r = dma_fence_wait(other, true);
+		if (r < 0) {
+			if (r != -ERESTARTSYS)
+				DRM_ERROR("Error (%ld) waiting for fence!\n", r);
+
+			return r;
+		}
+	}
+
+	return 0;
+}
+
+void amdgpu_ctx_mgr_init(struct amdgpu_ctx_mgr *mgr)
+{
+	rw_init(&mgr->lock, "mgrlk");
+	idr_init(&mgr->ctx_handles);
+}
+
+void amdgpu_ctx_mgr_entity_flush(struct amdgpu_ctx_mgr *mgr)
+{
+	struct amdgpu_ctx *ctx;
+	struct idr *idp;
+	uint32_t id, i;
+	long max_wait = MAX_WAIT_SCHED_ENTITY_Q_EMPTY;
+
+	idp = &mgr->ctx_handles;
+
+	mutex_lock(&mgr->lock);
+	idr_for_each_entry(idp, ctx, id) {
+
+		if (!ctx->adev) {
+			mutex_unlock(&mgr->lock);
+			return;
+		}
+
+		for (i = 0; i < ctx->adev->num_rings; i++) {
+
+			if (ctx->adev->rings[i] == &ctx->adev->gfx.kiq.ring)
+				continue;
+
+			max_wait = drm_sched_entity_flush(&ctx->rings[i].entity,
+							  max_wait);
+		}
+	}
+	mutex_unlock(&mgr->lock);
+}
+
+void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr)
+{
+	struct amdgpu_ctx *ctx;
+	struct idr *idp;
+	uint32_t id, i;
+
+	idp = &mgr->ctx_handles;
+
+	idr_for_each_entry(idp, ctx, id) {
+
+		if (!ctx->adev)
+			return;
+
+		for (i = 0; i < ctx->adev->num_rings; i++) {
+
+			if (ctx->adev->rings[i] == &ctx->adev->gfx.kiq.ring)
+				continue;
+
+			if (kref_read(&ctx->refcount) == 1)
+				drm_sched_entity_fini(&ctx->rings[i].entity);
+			else
+				DRM_ERROR("ctx %p is still alive\n", ctx);
+		}
+	}
+}
+
+void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr)
+{
+	struct amdgpu_ctx *ctx;
+	struct idr *idp;
+	uint32_t id;
+
+	amdgpu_ctx_mgr_entity_fini(mgr);
+
+	idp = &mgr->ctx_handles;
+
+	idr_for_each_entry(idp, ctx, id) {
+		if (kref_put(&ctx->refcount, amdgpu_ctx_fini) != 1)
+			DRM_ERROR("ctx %p is still alive\n", ctx);
+	}
+
+	idr_destroy(&mgr->ctx_handles);
+	mutex_destroy(&mgr->lock);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_debugfs.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_debugfs.c
new file mode 100644
index 00000000000..f5fb93795a6
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_debugfs.c
@@ -0,0 +1,951 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/kthread.h>
+#include <drm/drmP.h>
+#include <linux/debugfs.h>
+#include "amdgpu.h"
+
+/**
+ * amdgpu_debugfs_add_files - Add simple debugfs entries
+ *
+ * @adev:  Device to attach debugfs entries to
+ * @files:  Array of function callbacks that respond to reads
+ * @nfiles: Number of callbacks to register
+ *
+ */
+int amdgpu_debugfs_add_files(struct amdgpu_device *adev,
+			     const struct drm_info_list *files,
+			     unsigned nfiles)
+{
+	unsigned i;
+
+	for (i = 0; i < adev->debugfs_count; i++) {
+		if (adev->debugfs[i].files == files) {
+			/* Already registered */
+			return 0;
+		}
+	}
+
+	i = adev->debugfs_count + 1;
+	if (i > AMDGPU_DEBUGFS_MAX_COMPONENTS) {
+		DRM_ERROR("Reached maximum number of debugfs components.\n");
+		DRM_ERROR("Report so we increase "
+			  "AMDGPU_DEBUGFS_MAX_COMPONENTS.\n");
+		return -EINVAL;
+	}
+	adev->debugfs[adev->debugfs_count].files = files;
+	adev->debugfs[adev->debugfs_count].num_files = nfiles;
+	adev->debugfs_count = i;
+#if defined(CONFIG_DEBUG_FS)
+	drm_debugfs_create_files(files, nfiles,
+				 adev->ddev->primary->debugfs_root,
+				 adev->ddev->primary);
+#endif
+	return 0;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+/**
+ * amdgpu_debugfs_process_reg_op - Handle MMIO register reads/writes
+ *
+ * @read: True if reading
+ * @f: open file handle
+ * @buf: User buffer to write/read to
+ * @size: Number of bytes to write/read
+ * @pos:  Offset to seek to
+ *
+ * This debugfs entry has special meaning on the offset being sought.
+ * Various bits have different meanings:
+ *
+ * Bit 62:  Indicates a GRBM bank switch is needed
+ * Bit 61:  Indicates a SRBM bank switch is needed (implies bit 62 is
+ * 			zero)
+ * Bits 24..33: The SE or ME selector if needed
+ * Bits 34..43: The SH (or SA) or PIPE selector if needed
+ * Bits 44..53: The INSTANCE (or CU/WGP) or QUEUE selector if needed
+ *
+ * Bit 23:  Indicates that the PM power gating lock should be held
+ * 			This is necessary to read registers that might be
+ * 			unreliable during a power gating transistion.
+ *
+ * The lower bits are the BYTE offset of the register to read.  This
+ * allows reading multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static int  amdgpu_debugfs_process_reg_op(bool read, struct file *f,
+		char __user *buf, size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+	bool pm_pg_lock, use_bank, use_ring;
+	unsigned instance_bank, sh_bank, se_bank, me, pipe, queue;
+
+	pm_pg_lock = use_bank = use_ring = false;
+	instance_bank = sh_bank = se_bank = me = pipe = queue = 0;
+
+	if (size & 0x3 || *pos & 0x3 ||
+			((*pos & (1ULL << 62)) && (*pos & (1ULL << 61))))
+		return -EINVAL;
+
+	/* are we reading registers for which a PG lock is necessary? */
+	pm_pg_lock = (*pos >> 23) & 1;
+
+	if (*pos & (1ULL << 62)) {
+		se_bank = (*pos & GENMASK_ULL(33, 24)) >> 24;
+		sh_bank = (*pos & GENMASK_ULL(43, 34)) >> 34;
+		instance_bank = (*pos & GENMASK_ULL(53, 44)) >> 44;
+
+		if (se_bank == 0x3FF)
+			se_bank = 0xFFFFFFFF;
+		if (sh_bank == 0x3FF)
+			sh_bank = 0xFFFFFFFF;
+		if (instance_bank == 0x3FF)
+			instance_bank = 0xFFFFFFFF;
+		use_bank = 1;
+	} else if (*pos & (1ULL << 61)) {
+
+		me = (*pos & GENMASK_ULL(33, 24)) >> 24;
+		pipe = (*pos & GENMASK_ULL(43, 34)) >> 34;
+		queue = (*pos & GENMASK_ULL(53, 44)) >> 44;
+
+		use_ring = 1;
+	} else {
+		use_bank = use_ring = 0;
+	}
+
+	*pos &= (1UL << 22) - 1;
+
+	if (use_bank) {
+		if ((sh_bank != 0xFFFFFFFF && sh_bank >= adev->gfx.config.max_sh_per_se) ||
+		    (se_bank != 0xFFFFFFFF && se_bank >= adev->gfx.config.max_shader_engines))
+			return -EINVAL;
+		mutex_lock(&adev->grbm_idx_mutex);
+		amdgpu_gfx_select_se_sh(adev, se_bank,
+					sh_bank, instance_bank);
+	} else if (use_ring) {
+		mutex_lock(&adev->srbm_mutex);
+		amdgpu_gfx_select_me_pipe_q(adev, me, pipe, queue);
+	}
+
+	if (pm_pg_lock)
+		mutex_lock(&adev->pm.mutex);
+
+	while (size) {
+		uint32_t value;
+
+		if (*pos > adev->rmmio_size)
+			goto end;
+
+		if (read) {
+			value = RREG32(*pos >> 2);
+			r = put_user(value, (uint32_t *)buf);
+		} else {
+			r = get_user(value, (uint32_t *)buf);
+			if (!r)
+				WREG32(*pos >> 2, value);
+		}
+		if (r) {
+			result = r;
+			goto end;
+		}
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+end:
+	if (use_bank) {
+		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		mutex_unlock(&adev->grbm_idx_mutex);
+	} else if (use_ring) {
+		amdgpu_gfx_select_me_pipe_q(adev, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	}
+
+	if (pm_pg_lock)
+		mutex_unlock(&adev->pm.mutex);
+
+	return result;
+}
+
+/**
+ * amdgpu_debugfs_regs_read - Callback for reading MMIO registers
+ */
+static ssize_t amdgpu_debugfs_regs_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	return amdgpu_debugfs_process_reg_op(true, f, buf, size, pos);
+}
+
+/**
+ * amdgpu_debugfs_regs_write - Callback for writing MMIO registers
+ */
+static ssize_t amdgpu_debugfs_regs_write(struct file *f, const char __user *buf,
+					 size_t size, loff_t *pos)
+{
+	return amdgpu_debugfs_process_reg_op(false, f, (char __user *)buf, size, pos);
+}
+
+
+/**
+ * amdgpu_debugfs_regs_pcie_read - Read from a PCIE register
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * The lower bits are the BYTE offset of the register to read.  This
+ * allows reading multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static ssize_t amdgpu_debugfs_regs_pcie_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	while (size) {
+		uint32_t value;
+
+		value = RREG32_PCIE(*pos >> 2);
+		r = put_user(value, (uint32_t *)buf);
+		if (r)
+			return r;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	return result;
+}
+
+/**
+ * amdgpu_debugfs_regs_pcie_write - Write to a PCIE register
+ *
+ * @f: open file handle
+ * @buf: User buffer to write data from
+ * @size: Number of bytes to write
+ * @pos:  Offset to seek to
+ *
+ * The lower bits are the BYTE offset of the register to write.  This
+ * allows writing multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static ssize_t amdgpu_debugfs_regs_pcie_write(struct file *f, const char __user *buf,
+					 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	while (size) {
+		uint32_t value;
+
+		r = get_user(value, (uint32_t *)buf);
+		if (r)
+			return r;
+
+		WREG32_PCIE(*pos >> 2, value);
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	return result;
+}
+
+/**
+ * amdgpu_debugfs_regs_didt_read - Read from a DIDT register
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * The lower bits are the BYTE offset of the register to read.  This
+ * allows reading multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static ssize_t amdgpu_debugfs_regs_didt_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	while (size) {
+		uint32_t value;
+
+		value = RREG32_DIDT(*pos >> 2);
+		r = put_user(value, (uint32_t *)buf);
+		if (r)
+			return r;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	return result;
+}
+
+/**
+ * amdgpu_debugfs_regs_didt_write - Write to a DIDT register
+ *
+ * @f: open file handle
+ * @buf: User buffer to write data from
+ * @size: Number of bytes to write
+ * @pos:  Offset to seek to
+ *
+ * The lower bits are the BYTE offset of the register to write.  This
+ * allows writing multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static ssize_t amdgpu_debugfs_regs_didt_write(struct file *f, const char __user *buf,
+					 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	while (size) {
+		uint32_t value;
+
+		r = get_user(value, (uint32_t *)buf);
+		if (r)
+			return r;
+
+		WREG32_DIDT(*pos >> 2, value);
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	return result;
+}
+
+/**
+ * amdgpu_debugfs_regs_smc_read - Read from a SMC register
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * The lower bits are the BYTE offset of the register to read.  This
+ * allows reading multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static ssize_t amdgpu_debugfs_regs_smc_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	while (size) {
+		uint32_t value;
+
+		value = RREG32_SMC(*pos);
+		r = put_user(value, (uint32_t *)buf);
+		if (r)
+			return r;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	return result;
+}
+
+/**
+ * amdgpu_debugfs_regs_smc_write - Write to a SMC register
+ *
+ * @f: open file handle
+ * @buf: User buffer to write data from
+ * @size: Number of bytes to write
+ * @pos:  Offset to seek to
+ *
+ * The lower bits are the BYTE offset of the register to write.  This
+ * allows writing multiple registers in a single call and having
+ * the returned size reflect that.
+ */
+static ssize_t amdgpu_debugfs_regs_smc_write(struct file *f, const char __user *buf,
+					 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	while (size) {
+		uint32_t value;
+
+		r = get_user(value, (uint32_t *)buf);
+		if (r)
+			return r;
+
+		WREG32_SMC(*pos, value);
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	return result;
+}
+
+/**
+ * amdgpu_debugfs_gca_config_read - Read from gfx config data
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * This file is used to access configuration data in a somewhat
+ * stable fashion.  The format is a series of DWORDs with the first
+ * indicating which revision it is.  New content is appended to the
+ * end so that older software can still read the data.
+ */
+
+static ssize_t amdgpu_debugfs_gca_config_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+	uint32_t *config, no_regs = 0;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	config = kmalloc_array(256, sizeof(*config), GFP_KERNEL);
+	if (!config)
+		return -ENOMEM;
+
+	/* version, increment each time something is added */
+	config[no_regs++] = 3;
+	config[no_regs++] = adev->gfx.config.max_shader_engines;
+	config[no_regs++] = adev->gfx.config.max_tile_pipes;
+	config[no_regs++] = adev->gfx.config.max_cu_per_sh;
+	config[no_regs++] = adev->gfx.config.max_sh_per_se;
+	config[no_regs++] = adev->gfx.config.max_backends_per_se;
+	config[no_regs++] = adev->gfx.config.max_texture_channel_caches;
+	config[no_regs++] = adev->gfx.config.max_gprs;
+	config[no_regs++] = adev->gfx.config.max_gs_threads;
+	config[no_regs++] = adev->gfx.config.max_hw_contexts;
+	config[no_regs++] = adev->gfx.config.sc_prim_fifo_size_frontend;
+	config[no_regs++] = adev->gfx.config.sc_prim_fifo_size_backend;
+	config[no_regs++] = adev->gfx.config.sc_hiz_tile_fifo_size;
+	config[no_regs++] = adev->gfx.config.sc_earlyz_tile_fifo_size;
+	config[no_regs++] = adev->gfx.config.num_tile_pipes;
+	config[no_regs++] = adev->gfx.config.backend_enable_mask;
+	config[no_regs++] = adev->gfx.config.mem_max_burst_length_bytes;
+	config[no_regs++] = adev->gfx.config.mem_row_size_in_kb;
+	config[no_regs++] = adev->gfx.config.shader_engine_tile_size;
+	config[no_regs++] = adev->gfx.config.num_gpus;
+	config[no_regs++] = adev->gfx.config.multi_gpu_tile_size;
+	config[no_regs++] = adev->gfx.config.mc_arb_ramcfg;
+	config[no_regs++] = adev->gfx.config.gb_addr_config;
+	config[no_regs++] = adev->gfx.config.num_rbs;
+
+	/* rev==1 */
+	config[no_regs++] = adev->rev_id;
+	config[no_regs++] = adev->pg_flags;
+	config[no_regs++] = adev->cg_flags;
+
+	/* rev==2 */
+	config[no_regs++] = adev->family;
+	config[no_regs++] = adev->external_rev_id;
+
+	/* rev==3 */
+	config[no_regs++] = adev->pdev->device;
+	config[no_regs++] = adev->pdev->revision;
+	config[no_regs++] = adev->pdev->subsystem_device;
+	config[no_regs++] = adev->pdev->subsystem_vendor;
+
+	while (size && (*pos < no_regs * 4)) {
+		uint32_t value;
+
+		value = config[*pos >> 2];
+		r = put_user(value, (uint32_t *)buf);
+		if (r) {
+			kfree(config);
+			return r;
+		}
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	kfree(config);
+	return result;
+}
+
+/**
+ * amdgpu_debugfs_sensor_read - Read from the powerplay sensors
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * The offset is treated as the BYTE address of one of the sensors
+ * enumerated in amd/include/kgd_pp_interface.h under the
+ * 'amd_pp_sensors' enumeration.  For instance to read the UVD VCLK
+ * you would use the offset 3 * 4 = 12.
+ */
+static ssize_t amdgpu_debugfs_sensor_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	int idx, x, outsize, r, valuesize;
+	uint32_t values[16];
+
+	if (size & 3 || *pos & 0x3)
+		return -EINVAL;
+
+	if (!adev->pm.dpm_enabled)
+		return -EINVAL;
+
+	/* convert offset to sensor number */
+	idx = *pos >> 2;
+
+	valuesize = sizeof(values);
+	if (adev->powerplay.pp_funcs && adev->powerplay.pp_funcs->read_sensor)
+		r = amdgpu_dpm_read_sensor(adev, idx, &values[0], &valuesize);
+	else
+		return -EINVAL;
+
+	if (size > valuesize)
+		return -EINVAL;
+
+	outsize = 0;
+	x = 0;
+	if (!r) {
+		while (size) {
+			r = put_user(values[x++], (int32_t *)buf);
+			buf += 4;
+			size -= 4;
+			outsize += 4;
+		}
+	}
+
+	return !r ? outsize : r;
+}
+
+/** amdgpu_debugfs_wave_read - Read WAVE STATUS data
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * The offset being sought changes which wave that the status data
+ * will be returned for.  The bits are used as follows:
+ *
+ * Bits 0..6: 	Byte offset into data
+ * Bits 7..14:	SE selector
+ * Bits 15..22:	SH/SA selector
+ * Bits 23..30: CU/{WGP+SIMD} selector
+ * Bits 31..36: WAVE ID selector
+ * Bits 37..44: SIMD ID selector
+ *
+ * The returned data begins with one DWORD of version information
+ * Followed by WAVE STATUS registers relevant to the GFX IP version
+ * being used.  See gfx_v8_0_read_wave_data() for an example output.
+ */
+static ssize_t amdgpu_debugfs_wave_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = f->f_inode->i_private;
+	int r, x;
+	ssize_t result=0;
+	uint32_t offset, se, sh, cu, wave, simd, data[32];
+
+	if (size & 3 || *pos & 3)
+		return -EINVAL;
+
+	/* decode offset */
+	offset = (*pos & GENMASK_ULL(6, 0));
+	se = (*pos & GENMASK_ULL(14, 7)) >> 7;
+	sh = (*pos & GENMASK_ULL(22, 15)) >> 15;
+	cu = (*pos & GENMASK_ULL(30, 23)) >> 23;
+	wave = (*pos & GENMASK_ULL(36, 31)) >> 31;
+	simd = (*pos & GENMASK_ULL(44, 37)) >> 37;
+
+	/* switch to the specific se/sh/cu */
+	mutex_lock(&adev->grbm_idx_mutex);
+	amdgpu_gfx_select_se_sh(adev, se, sh, cu);
+
+	x = 0;
+	if (adev->gfx.funcs->read_wave_data)
+		adev->gfx.funcs->read_wave_data(adev, simd, wave, data, &x);
+
+	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	if (!x)
+		return -EINVAL;
+
+	while (size && (offset < x * 4)) {
+		uint32_t value;
+
+		value = data[offset >> 2];
+		r = put_user(value, (uint32_t *)buf);
+		if (r)
+			return r;
+
+		result += 4;
+		buf += 4;
+		offset += 4;
+		size -= 4;
+	}
+
+	return result;
+}
+
+/** amdgpu_debugfs_gpr_read - Read wave gprs
+ *
+ * @f: open file handle
+ * @buf: User buffer to store read data in
+ * @size: Number of bytes to read
+ * @pos:  Offset to seek to
+ *
+ * The offset being sought changes which wave that the status data
+ * will be returned for.  The bits are used as follows:
+ *
+ * Bits 0..11:	Byte offset into data
+ * Bits 12..19:	SE selector
+ * Bits 20..27:	SH/SA selector
+ * Bits 28..35: CU/{WGP+SIMD} selector
+ * Bits 36..43: WAVE ID selector
+ * Bits 37..44: SIMD ID selector
+ * Bits 52..59: Thread selector
+ * Bits 60..61: Bank selector (VGPR=0,SGPR=1)
+ *
+ * The return data comes from the SGPR or VGPR register bank for
+ * the selected operational unit.
+ */
+static ssize_t amdgpu_debugfs_gpr_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = f->f_inode->i_private;
+	int r;
+	ssize_t result = 0;
+	uint32_t offset, se, sh, cu, wave, simd, thread, bank, *data;
+
+	if (size & 3 || *pos & 3)
+		return -EINVAL;
+
+	/* decode offset */
+	offset = *pos & GENMASK_ULL(11, 0);
+	se = (*pos & GENMASK_ULL(19, 12)) >> 12;
+	sh = (*pos & GENMASK_ULL(27, 20)) >> 20;
+	cu = (*pos & GENMASK_ULL(35, 28)) >> 28;
+	wave = (*pos & GENMASK_ULL(43, 36)) >> 36;
+	simd = (*pos & GENMASK_ULL(51, 44)) >> 44;
+	thread = (*pos & GENMASK_ULL(59, 52)) >> 52;
+	bank = (*pos & GENMASK_ULL(61, 60)) >> 60;
+
+	data = kmalloc_array(1024, sizeof(*data), GFP_KERNEL);
+	if (!data)
+		return -ENOMEM;
+
+	/* switch to the specific se/sh/cu */
+	mutex_lock(&adev->grbm_idx_mutex);
+	amdgpu_gfx_select_se_sh(adev, se, sh, cu);
+
+	if (bank == 0) {
+		if (adev->gfx.funcs->read_wave_vgprs)
+			adev->gfx.funcs->read_wave_vgprs(adev, simd, wave, thread, offset, size>>2, data);
+	} else {
+		if (adev->gfx.funcs->read_wave_sgprs)
+			adev->gfx.funcs->read_wave_sgprs(adev, simd, wave, offset, size>>2, data);
+	}
+
+	amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	while (size) {
+		uint32_t value;
+
+		value = data[offset++];
+		r = put_user(value, (uint32_t *)buf);
+		if (r) {
+			result = r;
+			goto err;
+		}
+
+		result += 4;
+		buf += 4;
+		size -= 4;
+	}
+
+err:
+	kfree(data);
+	return result;
+}
+
+static const struct file_operations amdgpu_debugfs_regs_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_regs_read,
+	.write = amdgpu_debugfs_regs_write,
+	.llseek = default_llseek
+};
+static const struct file_operations amdgpu_debugfs_regs_didt_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_regs_didt_read,
+	.write = amdgpu_debugfs_regs_didt_write,
+	.llseek = default_llseek
+};
+static const struct file_operations amdgpu_debugfs_regs_pcie_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_regs_pcie_read,
+	.write = amdgpu_debugfs_regs_pcie_write,
+	.llseek = default_llseek
+};
+static const struct file_operations amdgpu_debugfs_regs_smc_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_regs_smc_read,
+	.write = amdgpu_debugfs_regs_smc_write,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_debugfs_gca_config_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_gca_config_read,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_debugfs_sensors_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_sensor_read,
+	.llseek = default_llseek
+};
+
+static const struct file_operations amdgpu_debugfs_wave_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_wave_read,
+	.llseek = default_llseek
+};
+static const struct file_operations amdgpu_debugfs_gpr_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_gpr_read,
+	.llseek = default_llseek
+};
+
+static const struct file_operations *debugfs_regs[] = {
+	&amdgpu_debugfs_regs_fops,
+	&amdgpu_debugfs_regs_didt_fops,
+	&amdgpu_debugfs_regs_pcie_fops,
+	&amdgpu_debugfs_regs_smc_fops,
+	&amdgpu_debugfs_gca_config_fops,
+	&amdgpu_debugfs_sensors_fops,
+	&amdgpu_debugfs_wave_fops,
+	&amdgpu_debugfs_gpr_fops,
+};
+
+static const char *debugfs_regs_names[] = {
+	"amdgpu_regs",
+	"amdgpu_regs_didt",
+	"amdgpu_regs_pcie",
+	"amdgpu_regs_smc",
+	"amdgpu_gca_config",
+	"amdgpu_sensors",
+	"amdgpu_wave",
+	"amdgpu_gpr",
+};
+
+/**
+ * amdgpu_debugfs_regs_init -	Initialize debugfs entries that provide
+ * 								register access.
+ *
+ * @adev: The device to attach the debugfs entries to
+ */
+int amdgpu_debugfs_regs_init(struct amdgpu_device *adev)
+{
+	struct drm_minor *minor = adev->ddev->primary;
+	struct dentry *ent, *root = minor->debugfs_root;
+	unsigned i, j;
+
+	for (i = 0; i < ARRAY_SIZE(debugfs_regs); i++) {
+		ent = debugfs_create_file(debugfs_regs_names[i],
+					  S_IFREG | S_IRUGO, root,
+					  adev, debugfs_regs[i]);
+		if (IS_ERR(ent)) {
+			for (j = 0; j < i; j++) {
+				debugfs_remove(adev->debugfs_regs[i]);
+				adev->debugfs_regs[i] = NULL;
+			}
+			return PTR_ERR(ent);
+		}
+
+		if (!i)
+			i_size_write(ent->d_inode, adev->rmmio_size);
+		adev->debugfs_regs[i] = ent;
+	}
+
+	return 0;
+}
+
+void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	for (i = 0; i < ARRAY_SIZE(debugfs_regs); i++) {
+		if (adev->debugfs_regs[i]) {
+			debugfs_remove(adev->debugfs_regs[i]);
+			adev->debugfs_regs[i] = NULL;
+		}
+	}
+}
+
+static int amdgpu_debugfs_test_ib(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int r = 0, i;
+
+	/* hold on the scheduler */
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!ring || !ring->sched.thread)
+			continue;
+		kthread_park(ring->sched.thread);
+	}
+
+	seq_printf(m, "run ib test:\n");
+	r = amdgpu_ib_ring_tests(adev);
+	if (r)
+		seq_printf(m, "ib ring tests failed (%d).\n", r);
+	else
+		seq_printf(m, "ib ring tests passed.\n");
+
+	/* go on the scheduler */
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!ring || !ring->sched.thread)
+			continue;
+		kthread_unpark(ring->sched.thread);
+	}
+
+	return 0;
+}
+
+static int amdgpu_debugfs_get_vbios_dump(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	seq_write(m, adev->bios, adev->bios_size);
+	return 0;
+}
+
+static int amdgpu_debugfs_evict_vram(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *)m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	seq_printf(m, "(%d)\n", amdgpu_bo_evict_vram(adev));
+	return 0;
+}
+
+static int amdgpu_debugfs_evict_gtt(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *)m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	seq_printf(m, "(%d)\n", ttm_bo_evict_mm(&adev->mman.bdev, TTM_PL_TT));
+	return 0;
+}
+
+static const struct drm_info_list amdgpu_debugfs_list[] = {
+	{"amdgpu_vbios", amdgpu_debugfs_get_vbios_dump},
+	{"amdgpu_test_ib", &amdgpu_debugfs_test_ib},
+	{"amdgpu_evict_vram", &amdgpu_debugfs_evict_vram},
+	{"amdgpu_evict_gtt", &amdgpu_debugfs_evict_gtt},
+};
+
+int amdgpu_debugfs_init(struct amdgpu_device *adev)
+{
+	return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_list,
+					ARRAY_SIZE(amdgpu_debugfs_list));
+}
+
+#else
+int amdgpu_debugfs_init(struct amdgpu_device *adev)
+{
+	return 0;
+}
+int amdgpu_debugfs_regs_init(struct amdgpu_device *adev)
+{
+	return 0;
+}
+void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev) { }
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_debugfs.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_debugfs.h
new file mode 100644
index 00000000000..8260d8073c2
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_debugfs.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+/*
+ * Debugfs
+ */
+struct amdgpu_debugfs {
+	const struct drm_info_list	*files;
+	unsigned		num_files;
+};
+
+int amdgpu_debugfs_regs_init(struct amdgpu_device *adev);
+void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev);
+int amdgpu_debugfs_init(struct amdgpu_device *adev);
+int amdgpu_debugfs_add_files(struct amdgpu_device *adev,
+			     const struct drm_info_list *files,
+			     unsigned nfiles);
+int amdgpu_debugfs_fence_init(struct amdgpu_device *adev);
+int amdgpu_debugfs_firmware_init(struct amdgpu_device *adev);
+int amdgpu_debugfs_gem_init(struct amdgpu_device *adev);
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_device.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_device.c
new file mode 100644
index 00000000000..7996d714f51
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_device.c
@@ -0,0 +1,3605 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <linux/power_supply.h>
+#include <linux/kthread.h>
+#include <linux/console.h>
+#include <linux/slab.h>
+#include <drm/drmP.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/drm_atomic_helper.h>
+#include <drm/amdgpu_drm.h>
+#include <linux/vgaarb.h>
+#include <linux/vga_switcheroo.h>
+#include <linux/efi.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+#include "amdgpu_i2c.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_atomfirmware.h"
+#include "amd_pcie.h"
+#ifdef CONFIG_DRM_AMDGPU_SI
+#include "si.h"
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+#include "cik.h"
+#endif
+#include "vi.h"
+#include "soc15.h"
+#include "bif/bif_4_1_d.h"
+#include <linux/pci.h>
+#include <linux/firmware.h>
+#include "amdgpu_vf_error.h"
+
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_pm.h"
+
+MODULE_FIRMWARE("amdgpu/vega10_gpu_info.bin");
+MODULE_FIRMWARE("amdgpu/vega12_gpu_info.bin");
+MODULE_FIRMWARE("amdgpu/raven_gpu_info.bin");
+
+#define AMDGPU_RESUME_MS		2000
+
+static const char *amdgpu_asic_name[] = {
+	"TAHITI",
+	"PITCAIRN",
+	"VERDE",
+	"OLAND",
+	"HAINAN",
+	"BONAIRE",
+	"KAVERI",
+	"KABINI",
+	"HAWAII",
+	"MULLINS",
+	"TOPAZ",
+	"TONGA",
+	"FIJI",
+	"CARRIZO",
+	"STONEY",
+	"POLARIS10",
+	"POLARIS11",
+	"POLARIS12",
+	"VEGAM",
+	"VEGA10",
+	"VEGA12",
+	"VEGA20",
+	"RAVEN",
+	"LAST",
+};
+
+static void amdgpu_device_get_pcie_info(struct amdgpu_device *adev);
+
+/**
+ * amdgpu_device_is_px - Is the device is a dGPU with HG/PX power control
+ *
+ * @dev: drm_device pointer
+ *
+ * Returns true if the device is a dGPU with HG/PX power control,
+ * otherwise return false.
+ */
+bool amdgpu_device_is_px(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (adev->flags & AMD_IS_PX)
+		return true;
+	return false;
+}
+
+/*
+ * MMIO register access helper functions.
+ */
+/**
+ * amdgpu_mm_rreg - read a memory mapped IO register
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: dword aligned register offset
+ * @acc_flags: access flags which require special behavior
+ *
+ * Returns the 32 bit value from the offset specified.
+ */
+uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg,
+			uint32_t acc_flags)
+{
+	uint32_t ret;
+
+	if (!(acc_flags & AMDGPU_REGS_NO_KIQ) && amdgpu_sriov_runtime(adev))
+		return amdgpu_virt_kiq_rreg(adev, reg);
+
+	if ((reg * 4) < adev->rmmio_size && !(acc_flags & AMDGPU_REGS_IDX))
+		ret = bus_space_read_4(adev->rmmio_bst, adev->rmmio_bsh,
+		    reg * 4);
+	else {
+		unsigned long flags;
+
+		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
+		bus_space_write_4(adev->rmmio_bst, adev->rmmio_bsh,
+		    mmMM_INDEX * 4, reg * 4);
+		ret = bus_space_read_4(adev->rmmio_bst, adev->rmmio_bsh,
+		    mmMM_DATA * 4);
+		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
+	}
+	trace_amdgpu_mm_rreg(adev->pdev->device, reg, ret);
+	return ret;
+}
+
+/*
+ * MMIO register read with bytes helper functions
+ * @offset:bytes offset from MMIO start
+ *
+*/
+
+/**
+ * amdgpu_mm_rreg8 - read a memory mapped IO register
+ *
+ * @adev: amdgpu_device pointer
+ * @offset: byte aligned register offset
+ *
+ * Returns the 8 bit value from the offset specified.
+ */
+uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset) {
+	if (offset < adev->rmmio_size)
+		return bus_space_read_1(adev->rmmio_bst, adev->rmmio_bsh, 
+		    offset);
+	BUG();
+}
+
+/*
+ * MMIO register write with bytes helper functions
+ * @offset:bytes offset from MMIO start
+ * @value: the value want to be written to the register
+ *
+*/
+/**
+ * amdgpu_mm_wreg8 - read a memory mapped IO register
+ *
+ * @adev: amdgpu_device pointer
+ * @offset: byte aligned register offset
+ * @value: 8 bit value to write
+ *
+ * Writes the value specified to the offset specified.
+ */
+void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value) {
+	if (offset < adev->rmmio_size)
+		bus_space_write_1(adev->rmmio_bst, adev->rmmio_bsh,
+		    offset, value);
+	else
+		BUG();
+}
+
+/**
+ * amdgpu_mm_wreg - write to a memory mapped IO register
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: dword aligned register offset
+ * @v: 32 bit value to write to the register
+ * @acc_flags: access flags which require special behavior
+ *
+ * Writes the value specified to the offset specified.
+ */
+void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
+		    uint32_t acc_flags)
+{
+	trace_amdgpu_mm_wreg(adev->pdev->device, reg, v);
+
+	if (adev->asic_type >= CHIP_VEGA10 && reg == 0) {
+		adev->last_mm_index = v;
+	}
+
+	if (!(acc_flags & AMDGPU_REGS_NO_KIQ) && amdgpu_sriov_runtime(adev))
+		return amdgpu_virt_kiq_wreg(adev, reg, v);
+
+	if ((reg * 4) < adev->rmmio_size && !(acc_flags & AMDGPU_REGS_IDX))
+		bus_space_write_4(adev->rmmio_bst, adev->rmmio_bsh,
+		    reg * 4, v);
+	else {
+		unsigned long flags;
+
+		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
+		bus_space_write_4(adev->rmmio_bst, adev->rmmio_bsh,
+		    mmMM_INDEX * 4, reg * 4);
+		bus_space_write_4(adev->rmmio_bst, adev->rmmio_bsh,
+		    mmMM_DATA * 4, v);
+		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
+	}
+
+	if (adev->asic_type >= CHIP_VEGA10 && reg == 1 && adev->last_mm_index == 0x5702C) {
+		udelay(500);
+	}
+}
+
+/**
+ * amdgpu_io_rreg - read an IO register
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: dword aligned register offset
+ *
+ * Returns the 32 bit value from the offset specified.
+ */
+u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	if ((reg * 4) < adev->rio_mem_size)
+		return bus_space_read_4(adev->rio_mem_bst, adev->rio_mem_bsh, reg);
+	else {
+		bus_space_write_4(adev->rio_mem_bst, adev->rio_mem_bsh,
+		    mmMM_INDEX * 4, reg * 4);
+		return bus_space_read_4(adev->rio_mem_bst, adev->rio_mem_bsh,
+		    mmMM_INDEX * 4);
+	}
+}
+
+/**
+ * amdgpu_io_wreg - write to an IO register
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: dword aligned register offset
+ * @v: 32 bit value to write to the register
+ *
+ * Writes the value specified to the offset specified.
+ */
+void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	if (adev->asic_type >= CHIP_VEGA10 && reg == 0) {
+		adev->last_mm_index = v;
+	}
+
+	if ((reg * 4) < adev->rio_mem_size)
+		bus_space_write_4(adev->rio_mem_bst, adev->rio_mem_bsh,
+		    reg * 4, v);
+	else {
+		bus_space_write_4(adev->rio_mem_bst, adev->rio_mem_bsh,
+		    mmMM_INDEX * 4, reg * 4);
+		bus_space_write_4(adev->rio_mem_bst, adev->rio_mem_bsh,
+		    mmMM_DATA * 4, v);
+		
+	}
+
+	if (adev->asic_type >= CHIP_VEGA10 && reg == 1 && adev->last_mm_index == 0x5702C) {
+		udelay(500);
+	}
+}
+
+/**
+ * amdgpu_mm_rdoorbell - read a doorbell dword
+ *
+ * @adev: amdgpu_device pointer
+ * @index: doorbell index
+ *
+ * Returns the value in the doorbell aperture at the
+ * requested doorbell index (CIK).
+ */
+u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index)
+{
+	if (index < adev->doorbell.num_doorbells) {
+		return bus_space_read_4(adev->doorbell.bst, adev->doorbell.bsh,
+		    index * 4);
+	} else {
+		DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", index);
+		return 0;
+	}
+}
+
+/**
+ * amdgpu_mm_wdoorbell - write a doorbell dword
+ *
+ * @adev: amdgpu_device pointer
+ * @index: doorbell index
+ * @v: value to write
+ *
+ * Writes @v to the doorbell aperture at the
+ * requested doorbell index (CIK).
+ */
+void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v)
+{
+	if (index < adev->doorbell.num_doorbells) {
+		bus_space_write_4(adev->doorbell.bst, adev->doorbell.bsh,
+		    index * 4, v);
+	} else {
+		DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index);
+	}
+}
+
+/**
+ * amdgpu_mm_rdoorbell64 - read a doorbell Qword
+ *
+ * @adev: amdgpu_device pointer
+ * @index: doorbell index
+ *
+ * Returns the value in the doorbell aperture at the
+ * requested doorbell index (VEGA10+).
+ */
+u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index)
+{
+	if (index < adev->doorbell.num_doorbells) {
+		return bus_space_read_8(adev->doorbell.bst, adev->doorbell.bsh,
+		    index * 4);
+	} else {
+		DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", index);
+		return 0;
+	}
+}
+
+/**
+ * amdgpu_mm_wdoorbell64 - write a doorbell Qword
+ *
+ * @adev: amdgpu_device pointer
+ * @index: doorbell index
+ * @v: value to write
+ *
+ * Writes @v to the doorbell aperture at the
+ * requested doorbell index (VEGA10+).
+ */
+void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v)
+{
+	if (index < adev->doorbell.num_doorbells) {
+		bus_space_write_8(adev->doorbell.bst, adev->doorbell.bsh,
+		    index * 4, v);
+	} else {
+		DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index);
+	}
+}
+
+/**
+ * amdgpu_invalid_rreg - dummy reg read function
+ *
+ * @adev: amdgpu device pointer
+ * @reg: offset of register
+ *
+ * Dummy register read function.  Used for register blocks
+ * that certain asics don't have (all asics).
+ * Returns the value in the register.
+ */
+static uint32_t amdgpu_invalid_rreg(struct amdgpu_device *adev, uint32_t reg)
+{
+	DRM_ERROR("Invalid callback to read register 0x%04X\n", reg);
+	BUG();
+	return 0;
+}
+
+/**
+ * amdgpu_invalid_wreg - dummy reg write function
+ *
+ * @adev: amdgpu device pointer
+ * @reg: offset of register
+ * @v: value to write to the register
+ *
+ * Dummy register read function.  Used for register blocks
+ * that certain asics don't have (all asics).
+ */
+static void amdgpu_invalid_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v)
+{
+	DRM_ERROR("Invalid callback to write register 0x%04X with 0x%08X\n",
+		  reg, v);
+	BUG();
+}
+
+/**
+ * amdgpu_block_invalid_rreg - dummy reg read function
+ *
+ * @adev: amdgpu device pointer
+ * @block: offset of instance
+ * @reg: offset of register
+ *
+ * Dummy register read function.  Used for register blocks
+ * that certain asics don't have (all asics).
+ * Returns the value in the register.
+ */
+static uint32_t amdgpu_block_invalid_rreg(struct amdgpu_device *adev,
+					  uint32_t block, uint32_t reg)
+{
+	DRM_ERROR("Invalid callback to read register 0x%04X in block 0x%04X\n",
+		  reg, block);
+	BUG();
+	return 0;
+}
+
+/**
+ * amdgpu_block_invalid_wreg - dummy reg write function
+ *
+ * @adev: amdgpu device pointer
+ * @block: offset of instance
+ * @reg: offset of register
+ * @v: value to write to the register
+ *
+ * Dummy register read function.  Used for register blocks
+ * that certain asics don't have (all asics).
+ */
+static void amdgpu_block_invalid_wreg(struct amdgpu_device *adev,
+				      uint32_t block,
+				      uint32_t reg, uint32_t v)
+{
+	DRM_ERROR("Invalid block callback to write register 0x%04X in block 0x%04X with 0x%08X\n",
+		  reg, block, v);
+	BUG();
+}
+
+/**
+ * amdgpu_device_vram_scratch_init - allocate the VRAM scratch page
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Allocates a scratch page of VRAM for use by various things in the
+ * driver.
+ */
+static int amdgpu_device_vram_scratch_init(struct amdgpu_device *adev)
+{
+	return amdgpu_bo_create_kernel(adev, AMDGPU_GPU_PAGE_SIZE,
+				       PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
+				       &adev->vram_scratch.robj,
+				       &adev->vram_scratch.gpu_addr,
+				       (void **)&adev->vram_scratch.ptr);
+}
+
+/**
+ * amdgpu_device_vram_scratch_fini - Free the VRAM scratch page
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Frees the VRAM scratch page.
+ */
+static void amdgpu_device_vram_scratch_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->vram_scratch.robj, NULL, NULL);
+}
+
+/**
+ * amdgpu_device_program_register_sequence - program an array of registers.
+ *
+ * @adev: amdgpu_device pointer
+ * @registers: pointer to the register array
+ * @array_size: size of the register array
+ *
+ * Programs an array or registers with and and or masks.
+ * This is a helper for setting golden registers.
+ */
+void amdgpu_device_program_register_sequence(struct amdgpu_device *adev,
+					     const u32 *registers,
+					     const u32 array_size)
+{
+	u32 tmp, reg, and_mask, or_mask;
+	int i;
+
+	if (array_size % 3)
+		return;
+
+	for (i = 0; i < array_size; i +=3) {
+		reg = registers[i + 0];
+		and_mask = registers[i + 1];
+		or_mask = registers[i + 2];
+
+		if (and_mask == 0xffffffff) {
+			tmp = or_mask;
+		} else {
+			tmp = RREG32(reg);
+			tmp &= ~and_mask;
+			tmp |= or_mask;
+		}
+		WREG32(reg, tmp);
+	}
+}
+
+/**
+ * amdgpu_device_pci_config_reset - reset the GPU
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Resets the GPU using the pci config reset sequence.
+ * Only applicable to asics prior to vega10.
+ */
+void amdgpu_device_pci_config_reset(struct amdgpu_device *adev)
+{
+	pci_write_config_dword(adev->pdev, 0x7c, AMDGPU_ASIC_RESET_DATA);
+}
+
+/*
+ * GPU doorbell aperture helpers function.
+ */
+/**
+ * amdgpu_device_doorbell_init - Init doorbell driver information.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Init doorbell driver information (CIK)
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_device_doorbell_init(struct amdgpu_device *adev)
+{
+	/* No doorbell on SI hardware generation */
+	if (adev->asic_type < CHIP_BONAIRE) {
+		adev->doorbell.base = 0;
+		adev->doorbell.size = 0;
+		adev->doorbell.num_doorbells = 0;
+#ifdef __linux__
+		adev->doorbell.ptr = NULL;
+#endif
+		return 0;
+	}
+
+#ifdef __linux__
+	if (pci_resource_flags(adev->pdev, 2) & IORESOURCE_UNSET)
+		return -EINVAL;
+
+	/* doorbell bar mapping */
+	adev->doorbell.base = pci_resource_start(adev->pdev, 2);
+	adev->doorbell.size = pci_resource_len(adev->pdev, 2);
+#endif
+
+	adev->doorbell.num_doorbells = min_t(u32, adev->doorbell.size / sizeof(u32),
+					     AMDGPU_DOORBELL_MAX_ASSIGNMENT+1);
+	if (adev->doorbell.num_doorbells == 0)
+		return -EINVAL;
+
+#ifdef __linux__
+	adev->doorbell.ptr = ioremap(adev->doorbell.base,
+				     adev->doorbell.num_doorbells *
+				     sizeof(u32));
+	if (adev->doorbell.ptr == NULL)
+		return -ENOMEM;
+#endif
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_doorbell_fini - Tear down doorbell driver information.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down doorbell driver information (CIK)
+ */
+static void amdgpu_device_doorbell_fini(struct amdgpu_device *adev)
+{
+#ifdef __linux__
+	iounmap(adev->doorbell.ptr);
+	adev->doorbell.ptr = NULL;
+#else
+	if (adev->doorbell.size > 0)
+		bus_space_unmap(adev->doorbell.bst, adev->doorbell.bsh,
+		    adev->doorbell.size);
+#endif
+}
+
+
+
+/*
+ * amdgpu_device_wb_*()
+ * Writeback is the method by which the GPU updates special pages in memory
+ * with the status of certain GPU events (fences, ring pointers,etc.).
+ */
+
+/**
+ * amdgpu_device_wb_fini - Disable Writeback and free memory
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disables Writeback and frees the Writeback memory (all asics).
+ * Used at driver shutdown.
+ */
+static void amdgpu_device_wb_fini(struct amdgpu_device *adev)
+{
+	if (adev->wb.wb_obj) {
+		amdgpu_bo_free_kernel(&adev->wb.wb_obj,
+				      &adev->wb.gpu_addr,
+				      (void **)&adev->wb.wb);
+		adev->wb.wb_obj = NULL;
+	}
+}
+
+/**
+ * amdgpu_device_wb_init- Init Writeback driver info and allocate memory
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initializes writeback and allocates writeback memory (all asics).
+ * Used at driver startup.
+ * Returns 0 on success or an -error on failure.
+ */
+static int amdgpu_device_wb_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->wb.wb_obj == NULL) {
+		/* AMDGPU_MAX_WB * sizeof(uint32_t) * 8 = AMDGPU_MAX_WB 256bit slots */
+		r = amdgpu_bo_create_kernel(adev, AMDGPU_MAX_WB * sizeof(uint32_t) * 8,
+					    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+					    &adev->wb.wb_obj, &adev->wb.gpu_addr,
+					    (void **)&adev->wb.wb);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create WB bo failed\n", r);
+			return r;
+		}
+
+		adev->wb.num_wb = AMDGPU_MAX_WB;
+		memset(&adev->wb.used, 0, sizeof(adev->wb.used));
+
+		/* clear wb memory */
+		memset((char *)adev->wb.wb, 0, AMDGPU_MAX_WB * sizeof(uint32_t) * 8);
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_wb_get - Allocate a wb entry
+ *
+ * @adev: amdgpu_device pointer
+ * @wb: wb index
+ *
+ * Allocate a wb slot for use by the driver (all asics).
+ * Returns 0 on success or -EINVAL on failure.
+ */
+int amdgpu_device_wb_get(struct amdgpu_device *adev, u32 *wb)
+{
+	unsigned long offset = find_first_zero_bit(adev->wb.used, adev->wb.num_wb);
+
+	if (offset < adev->wb.num_wb) {
+		__set_bit(offset, adev->wb.used);
+		*wb = offset << 3; /* convert to dw offset */
+		return 0;
+	} else {
+		return -EINVAL;
+	}
+}
+
+/**
+ * amdgpu_device_wb_free - Free a wb entry
+ *
+ * @adev: amdgpu_device pointer
+ * @wb: wb index
+ *
+ * Free a wb slot allocated for use by the driver (all asics)
+ */
+void amdgpu_device_wb_free(struct amdgpu_device *adev, u32 wb)
+{
+	wb >>= 3;
+	if (wb < adev->wb.num_wb)
+		__clear_bit(wb, adev->wb.used);
+}
+
+/**
+ * amdgpu_device_vram_location - try to find VRAM location
+ *
+ * @adev: amdgpu device structure holding all necessary informations
+ * @mc: memory controller structure holding memory informations
+ * @base: base address at which to put VRAM
+ *
+ * Function will try to place VRAM at base address provided
+ * as parameter.
+ */
+void amdgpu_device_vram_location(struct amdgpu_device *adev,
+				 struct amdgpu_gmc *mc, u64 base)
+{
+	uint64_t limit = (uint64_t)amdgpu_vram_limit << 20;
+
+	mc->vram_start = base;
+	mc->vram_end = mc->vram_start + mc->mc_vram_size - 1;
+	if (limit && limit < mc->real_vram_size)
+		mc->real_vram_size = limit;
+	dev_info(adev->dev, "VRAM: %lluM 0x%016llX - 0x%016llX (%lluM used)\n",
+			mc->mc_vram_size >> 20, mc->vram_start,
+			mc->vram_end, mc->real_vram_size >> 20);
+}
+
+/**
+ * amdgpu_device_gart_location - try to find GART location
+ *
+ * @adev: amdgpu device structure holding all necessary informations
+ * @mc: memory controller structure holding memory informations
+ *
+ * Function will place try to place GART before or after VRAM.
+ *
+ * If GART size is bigger than space left then we ajust GART size.
+ * Thus function will never fails.
+ */
+void amdgpu_device_gart_location(struct amdgpu_device *adev,
+				 struct amdgpu_gmc *mc)
+{
+	u64 size_af, size_bf;
+
+	mc->gart_size += adev->pm.smu_prv_buffer_size;
+
+	size_af = adev->gmc.mc_mask - mc->vram_end;
+	size_bf = mc->vram_start;
+	if (size_bf > size_af) {
+		if (mc->gart_size > size_bf) {
+			dev_warn(adev->dev, "limiting GART\n");
+			mc->gart_size = size_bf;
+		}
+		mc->gart_start = 0;
+	} else {
+		if (mc->gart_size > size_af) {
+			dev_warn(adev->dev, "limiting GART\n");
+			mc->gart_size = size_af;
+		}
+		/* VCE doesn't like it when BOs cross a 4GB segment, so align
+		 * the GART base on a 4GB boundary as well.
+		 */
+		mc->gart_start = roundup2(mc->vram_end + 1, 0x100000000ULL);
+	}
+	mc->gart_end = mc->gart_start + mc->gart_size - 1;
+	dev_info(adev->dev, "GART: %lluM 0x%016llX - 0x%016llX\n",
+			mc->gart_size >> 20, mc->gart_start, mc->gart_end);
+}
+
+/**
+ * amdgpu_device_resize_fb_bar - try to resize FB BAR
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Try to resize FB BAR to make all VRAM CPU accessible. We try very hard not
+ * to fail, but if any of the BARs is not accessible after the size we abort
+ * driver loading by returning -ENODEV.
+ */
+int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev)
+{
+	u64 space_needed = roundup_pow_of_two(adev->gmc.real_vram_size);
+	u32 rbar_size = order_base_2(((space_needed >> 20) | 1)) - 1;
+	struct pci_bus *root;
+	struct resource *res;
+	unsigned i;
+	u16 cmd;
+	int r;
+	pcireg_t type;
+
+	/* XXX not right yet */
+	STUB();
+	return 0;
+
+	/* Bypass for VF */
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+#ifdef notyet
+
+	/* Check if the root BUS has 64bit memory resources */
+	root = adev->pdev->bus;
+	while (root->parent)
+		root = root->parent;
+
+	pci_bus_for_each_resource(root, res, i) {
+		if (res && res->flags & (IORESOURCE_MEM | IORESOURCE_MEM_64) &&
+		    res->start > 0x100000000ull)
+			break;
+	}
+
+	/* Trying to resize is pointless without a root hub window above 4GB */
+	if (!res)
+		return 0;
+#endif
+
+	/* Disable memory decoding while we change the BAR addresses and size */
+	pci_read_config_word(adev->pdev, PCI_COMMAND, &cmd);
+	pci_write_config_word(adev->pdev, PCI_COMMAND,
+			      cmd & ~PCI_COMMAND_MEMORY);
+
+	/* Free the VRAM and doorbell BAR, we most likely need to move both. */
+	amdgpu_device_doorbell_fini(adev);
+#ifdef __linux__
+	if (adev->asic_type >= CHIP_BONAIRE)
+		pci_release_resource(adev->pdev, 2);
+
+	pci_release_resource(adev->pdev, 0);
+#endif
+
+	r = pci_resize_resource(adev->pdev, 0, rbar_size);
+	if (r == -ENOSPC)
+		DRM_INFO("Not enough PCI address space for a large BAR.");
+	else if (r && r != -ENOTSUPP)
+		DRM_ERROR("Problem resizing BAR0 (%d).", r);
+
+#ifdef __linux__
+	pci_assign_unassigned_bus_resources(adev->pdev->bus);
+#else
+#define AMDGPU_PCI_MEM		0x10
+
+	type = pci_mapreg_type(adev->pc, adev->pa_tag, AMDGPU_PCI_MEM);
+	if (PCI_MAPREG_TYPE(type) != PCI_MAPREG_TYPE_MEM ||
+	    pci_mapreg_info(adev->pc, adev->pa_tag, AMDGPU_PCI_MEM,
+	    type, NULL, &adev->fb_aper_size, NULL)) {
+		printf(": can't get frambuffer info\n");
+		return -ENODEV;
+	}
+#endif
+	/* When the doorbell or fb BAR isn't available we have no chance of
+	 * using the device.
+	 */
+	r = amdgpu_device_doorbell_init(adev);
+#ifdef notyet
+	if (r || (pci_resource_flags(adev->pdev, 0) & IORESOURCE_UNSET))
+#else
+	if (r)
+#endif
+		return -ENODEV;
+
+	pci_write_config_word(adev->pdev, PCI_COMMAND, cmd);
+
+	return 0;
+}
+
+/*
+ * GPU helpers function.
+ */
+/**
+ * amdgpu_device_need_post - check if the hw need post or not
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Check if the asic has been initialized (all asics) at driver startup
+ * or post is needed if  hw reset is performed.
+ * Returns true if need or false if not.
+ */
+bool amdgpu_device_need_post(struct amdgpu_device *adev)
+{
+	uint32_t reg;
+
+	if (amdgpu_sriov_vf(adev))
+		return false;
+
+	if (amdgpu_passthrough(adev)) {
+		/* for FIJI: In whole GPU pass-through virtualization case, after VM reboot
+		 * some old smc fw still need driver do vPost otherwise gpu hang, while
+		 * those smc fw version above 22.15 doesn't have this flaw, so we force
+		 * vpost executed for smc version below 22.15
+		 */
+		if (adev->asic_type == CHIP_FIJI) {
+			int err;
+			uint32_t fw_ver;
+			err = request_firmware(&adev->pm.fw, "amdgpu/fiji_smc.bin", adev->dev);
+			/* force vPost if error occured */
+			if (err)
+				return true;
+
+			fw_ver = *((uint32_t *)adev->pm.fw->data + 69);
+			if (fw_ver < 0x00160e00)
+				return true;
+		}
+	}
+
+	if (adev->has_hw_reset) {
+		adev->has_hw_reset = false;
+		return true;
+	}
+
+	/* bios scratch used on CIK+ */
+	if (adev->asic_type >= CHIP_BONAIRE)
+		return amdgpu_atombios_scratch_need_asic_init(adev);
+
+	/* check MEM_SIZE for older asics */
+	reg = amdgpu_asic_get_config_memsize(adev);
+
+	if ((reg != 0) && (reg != 0xffffffff))
+		return false;
+
+	return true;
+}
+
+/* if we get transitioned to only one device, take VGA back */
+/**
+ * amdgpu_device_vga_set_decode - enable/disable vga decode
+ *
+ * @cookie: amdgpu_device pointer
+ * @state: enable/disable vga decode
+ *
+ * Enable/disable vga decode (all asics).
+ * Returns VGA resource flags.
+ */
+#ifdef notyet
+static unsigned int amdgpu_device_vga_set_decode(void *cookie, bool state)
+{
+	struct amdgpu_device *adev = cookie;
+	amdgpu_asic_set_vga_state(adev, state);
+	if (state)
+		return VGA_RSRC_LEGACY_IO | VGA_RSRC_LEGACY_MEM |
+		       VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM;
+	else
+		return VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM;
+}
+#endif
+
+/**
+ * amdgpu_device_check_block_size - validate the vm block size
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Validates the vm block size specified via module parameter.
+ * The vm block size defines number of bits in page table versus page directory,
+ * a page is 4KB so we have 12 bits offset, minimum 9 bits in the
+ * page table and the remaining bits are in the page directory.
+ */
+static void amdgpu_device_check_block_size(struct amdgpu_device *adev)
+{
+	/* defines number of bits in page table versus page directory,
+	 * a page is 4KB so we have 12 bits offset, minimum 9 bits in the
+	 * page table and the remaining bits are in the page directory */
+	if (amdgpu_vm_block_size == -1)
+		return;
+
+	if (amdgpu_vm_block_size < 9) {
+		dev_warn(adev->dev, "VM page table size (%d) too small\n",
+			 amdgpu_vm_block_size);
+		amdgpu_vm_block_size = -1;
+	}
+}
+
+/**
+ * amdgpu_device_check_vm_size - validate the vm size
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Validates the vm size in GB specified via module parameter.
+ * The VM size is the size of the GPU virtual memory space in GB.
+ */
+static void amdgpu_device_check_vm_size(struct amdgpu_device *adev)
+{
+	/* no need to check the default value */
+	if (amdgpu_vm_size == -1)
+		return;
+
+	if (amdgpu_vm_size < 1) {
+		dev_warn(adev->dev, "VM size (%d) too small, min is 1GB\n",
+			 amdgpu_vm_size);
+		amdgpu_vm_size = -1;
+	}
+}
+
+static void amdgpu_device_check_smu_prv_buffer_size(struct amdgpu_device *adev)
+{
+#ifdef __linux__
+	struct sysinfo si;
+#endif
+	bool is_os_64 = (sizeof(void *) == 8) ? true : false;
+	uint64_t total_memory;
+	uint64_t dram_size_seven_GB = 0x1B8000000;
+	uint64_t dram_size_three_GB = 0xB8000000;
+
+	if (amdgpu_smu_memory_pool_size == 0)
+		return;
+
+	if (!is_os_64) {
+		DRM_WARN("Not 64-bit OS, feature not supported\n");
+		goto def_value;
+	}
+#ifdef __linux__
+	si_meminfo(&si);
+	total_memory = (uint64_t)si.totalram * si.mem_unit;
+#else
+	total_memory = ptoa(physmem);
+#endif
+
+	if ((amdgpu_smu_memory_pool_size == 1) ||
+		(amdgpu_smu_memory_pool_size == 2)) {
+		if (total_memory < dram_size_three_GB)
+			goto def_value1;
+	} else if ((amdgpu_smu_memory_pool_size == 4) ||
+		(amdgpu_smu_memory_pool_size == 8)) {
+		if (total_memory < dram_size_seven_GB)
+			goto def_value1;
+	} else {
+		DRM_WARN("Smu memory pool size not supported\n");
+		goto def_value;
+	}
+	adev->pm.smu_prv_buffer_size = amdgpu_smu_memory_pool_size << 28;
+
+	return;
+
+def_value1:
+	DRM_WARN("No enough system memory\n");
+def_value:
+	adev->pm.smu_prv_buffer_size = 0;
+}
+
+/**
+ * amdgpu_device_check_arguments - validate module params
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Validates certain module parameters and updates
+ * the associated values used by the driver (all asics).
+ */
+static void amdgpu_device_check_arguments(struct amdgpu_device *adev)
+{
+	if (amdgpu_sched_jobs < 4) {
+		dev_warn(adev->dev, "sched jobs (%d) must be at least 4\n",
+			 amdgpu_sched_jobs);
+		amdgpu_sched_jobs = 4;
+	} else if (!is_power_of_2(amdgpu_sched_jobs)){
+		dev_warn(adev->dev, "sched jobs (%d) must be a power of 2\n",
+			 amdgpu_sched_jobs);
+		amdgpu_sched_jobs = roundup_pow_of_two(amdgpu_sched_jobs);
+	}
+
+	if (amdgpu_gart_size != -1 && amdgpu_gart_size < 32) {
+		/* gart size must be greater or equal to 32M */
+		dev_warn(adev->dev, "gart size (%d) too small\n",
+			 amdgpu_gart_size);
+		amdgpu_gart_size = -1;
+	}
+
+	if (amdgpu_gtt_size != -1 && amdgpu_gtt_size < 32) {
+		/* gtt size must be greater or equal to 32M */
+		dev_warn(adev->dev, "gtt size (%d) too small\n",
+				 amdgpu_gtt_size);
+		amdgpu_gtt_size = -1;
+	}
+
+	/* valid range is between 4 and 9 inclusive */
+	if (amdgpu_vm_fragment_size != -1 &&
+	    (amdgpu_vm_fragment_size > 9 || amdgpu_vm_fragment_size < 4)) {
+		dev_warn(adev->dev, "valid range is between 4 and 9\n");
+		amdgpu_vm_fragment_size = -1;
+	}
+
+	amdgpu_device_check_smu_prv_buffer_size(adev);
+
+	amdgpu_device_check_vm_size(adev);
+
+	amdgpu_device_check_block_size(adev);
+
+	if (amdgpu_vram_page_split != -1 && (amdgpu_vram_page_split < 16 ||
+	    !is_power_of_2(amdgpu_vram_page_split))) {
+		dev_warn(adev->dev, "invalid VRAM page split (%d)\n",
+			 amdgpu_vram_page_split);
+		amdgpu_vram_page_split = 1024;
+	}
+
+	if (amdgpu_lockup_timeout == 0) {
+		dev_warn(adev->dev, "lockup_timeout msut be > 0, adjusting to 10000\n");
+		amdgpu_lockup_timeout = 10000;
+	}
+
+	adev->firmware.load_type = amdgpu_ucode_get_load_type(adev, amdgpu_fw_load_type);
+}
+
+#ifdef __linux__
+/**
+ * amdgpu_switcheroo_set_state - set switcheroo state
+ *
+ * @pdev: pci dev pointer
+ * @state: vga_switcheroo state
+ *
+ * Callback for the switcheroo driver.  Suspends or resumes the
+ * the asics before or after it is powered up using ACPI methods.
+ */
+static void amdgpu_switcheroo_set_state(struct pci_dev *pdev, enum vga_switcheroo_state state)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+
+	if (amdgpu_device_is_px(dev) && state == VGA_SWITCHEROO_OFF)
+		return;
+
+	if (state == VGA_SWITCHEROO_ON) {
+		pr_info("amdgpu: switched on\n");
+		/* don't suspend or resume card normally */
+		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
+
+		amdgpu_device_resume(dev, true, true);
+
+		dev->switch_power_state = DRM_SWITCH_POWER_ON;
+		drm_kms_helper_poll_enable(dev);
+	} else {
+		pr_info("amdgpu: switched off\n");
+		drm_kms_helper_poll_disable(dev);
+		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
+		amdgpu_device_suspend(dev, true, true);
+		dev->switch_power_state = DRM_SWITCH_POWER_OFF;
+	}
+}
+
+/**
+ * amdgpu_switcheroo_can_switch - see if switcheroo state can change
+ *
+ * @pdev: pci dev pointer
+ *
+ * Callback for the switcheroo driver.  Check of the switcheroo
+ * state can be changed.
+ * Returns true if the state can be changed, false if not.
+ */
+static bool amdgpu_switcheroo_can_switch(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+
+	/*
+	* FIXME: open_count is protected by drm_global_mutex but that would lead to
+	* locking inversion with the driver load path. And the access here is
+	* completely racy anyway. So don't bother with locking for now.
+	*/
+	return dev->open_count == 0;
+}
+
+static const struct vga_switcheroo_client_ops amdgpu_switcheroo_ops = {
+	.set_gpu_state = amdgpu_switcheroo_set_state,
+	.reprobe = NULL,
+	.can_switch = amdgpu_switcheroo_can_switch,
+};
+#endif /* __linux__ */
+
+/**
+ * amdgpu_device_ip_set_clockgating_state - set the CG state
+ *
+ * @dev: amdgpu_device pointer
+ * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.)
+ * @state: clockgating state (gate or ungate)
+ *
+ * Sets the requested clockgating state for all instances of
+ * the hardware IP specified.
+ * Returns the error code from the last instance.
+ */
+int amdgpu_device_ip_set_clockgating_state(void *dev,
+					   enum amd_ip_block_type block_type,
+					   enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = dev;
+	int i, r = 0;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].version->type != block_type)
+			continue;
+		if (!adev->ip_blocks[i].version->funcs->set_clockgating_state)
+			continue;
+		r = adev->ip_blocks[i].version->funcs->set_clockgating_state(
+			(void *)adev, state);
+		if (r)
+			DRM_ERROR("set_clockgating_state of IP block <%s> failed %d\n",
+				  adev->ip_blocks[i].version->funcs->name, r);
+	}
+	return r;
+}
+
+/**
+ * amdgpu_device_ip_set_powergating_state - set the PG state
+ *
+ * @dev: amdgpu_device pointer
+ * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.)
+ * @state: powergating state (gate or ungate)
+ *
+ * Sets the requested powergating state for all instances of
+ * the hardware IP specified.
+ * Returns the error code from the last instance.
+ */
+int amdgpu_device_ip_set_powergating_state(void *dev,
+					   enum amd_ip_block_type block_type,
+					   enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = dev;
+	int i, r = 0;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].version->type != block_type)
+			continue;
+		if (!adev->ip_blocks[i].version->funcs->set_powergating_state)
+			continue;
+		r = adev->ip_blocks[i].version->funcs->set_powergating_state(
+			(void *)adev, state);
+		if (r)
+			DRM_ERROR("set_powergating_state of IP block <%s> failed %d\n",
+				  adev->ip_blocks[i].version->funcs->name, r);
+	}
+	return r;
+}
+
+/**
+ * amdgpu_device_ip_get_clockgating_state - get the CG state
+ *
+ * @adev: amdgpu_device pointer
+ * @flags: clockgating feature flags
+ *
+ * Walks the list of IPs on the device and updates the clockgating
+ * flags for each IP.
+ * Updates @flags with the feature flags for each hardware IP where
+ * clockgating is enabled.
+ */
+void amdgpu_device_ip_get_clockgating_state(struct amdgpu_device *adev,
+					    u32 *flags)
+{
+	int i;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].version->funcs->get_clockgating_state)
+			adev->ip_blocks[i].version->funcs->get_clockgating_state((void *)adev, flags);
+	}
+}
+
+/**
+ * amdgpu_device_ip_wait_for_idle - wait for idle
+ *
+ * @adev: amdgpu_device pointer
+ * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.)
+ *
+ * Waits for the request hardware IP to be idle.
+ * Returns 0 for success or a negative error code on failure.
+ */
+int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev,
+				   enum amd_ip_block_type block_type)
+{
+	int i, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].version->type == block_type) {
+			r = adev->ip_blocks[i].version->funcs->wait_for_idle((void *)adev);
+			if (r)
+				return r;
+			break;
+		}
+	}
+	return 0;
+
+}
+
+/**
+ * amdgpu_device_ip_is_idle - is the hardware IP idle
+ *
+ * @adev: amdgpu_device pointer
+ * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.)
+ *
+ * Check if the hardware IP is idle or not.
+ * Returns true if it the IP is idle, false if not.
+ */
+bool amdgpu_device_ip_is_idle(struct amdgpu_device *adev,
+			      enum amd_ip_block_type block_type)
+{
+	int i;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].version->type == block_type)
+			return adev->ip_blocks[i].version->funcs->is_idle((void *)adev);
+	}
+	return true;
+
+}
+
+/**
+ * amdgpu_device_ip_get_ip_block - get a hw IP pointer
+ *
+ * @adev: amdgpu_device pointer
+ * @type: Type of hardware IP (SMU, GFX, UVD, etc.)
+ *
+ * Returns a pointer to the hardware IP block structure
+ * if it exists for the asic, otherwise NULL.
+ */
+struct amdgpu_ip_block *
+amdgpu_device_ip_get_ip_block(struct amdgpu_device *adev,
+			      enum amd_ip_block_type type)
+{
+	int i;
+
+	for (i = 0; i < adev->num_ip_blocks; i++)
+		if (adev->ip_blocks[i].version->type == type)
+			return &adev->ip_blocks[i];
+
+	return NULL;
+}
+
+/**
+ * amdgpu_device_ip_block_version_cmp
+ *
+ * @adev: amdgpu_device pointer
+ * @type: enum amd_ip_block_type
+ * @major: major version
+ * @minor: minor version
+ *
+ * return 0 if equal or greater
+ * return 1 if smaller or the ip_block doesn't exist
+ */
+int amdgpu_device_ip_block_version_cmp(struct amdgpu_device *adev,
+				       enum amd_ip_block_type type,
+				       u32 major, u32 minor)
+{
+	struct amdgpu_ip_block *ip_block = amdgpu_device_ip_get_ip_block(adev, type);
+
+	if (ip_block && ((ip_block->version->major > major) ||
+			((ip_block->version->major == major) &&
+			(ip_block->version->minor >= minor))))
+		return 0;
+
+	return 1;
+}
+
+/**
+ * amdgpu_device_ip_block_add
+ *
+ * @adev: amdgpu_device pointer
+ * @ip_block_version: pointer to the IP to add
+ *
+ * Adds the IP block driver information to the collection of IPs
+ * on the asic.
+ */
+int amdgpu_device_ip_block_add(struct amdgpu_device *adev,
+			       const struct amdgpu_ip_block_version *ip_block_version)
+{
+	if (!ip_block_version)
+		return -EINVAL;
+
+	DRM_INFO("add ip block number %d <%s>\n", adev->num_ip_blocks,
+		  ip_block_version->funcs->name);
+
+	adev->ip_blocks[adev->num_ip_blocks++].version = ip_block_version;
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_enable_virtual_display - enable virtual display feature
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enabled the virtual display feature if the user has enabled it via
+ * the module parameter virtual_display.  This feature provides a virtual
+ * display hardware on headless boards or in virtualized environments.
+ * This function parses and validates the configuration string specified by
+ * the user and configues the virtual display configuration (number of
+ * virtual connectors, crtcs, etc.) specified.
+ */
+static void amdgpu_device_enable_virtual_display(struct amdgpu_device *adev)
+{
+	adev->enable_virtual_display = false;
+
+#ifdef notyet
+	if (amdgpu_virtual_display) {
+		struct drm_device *ddev = adev->ddev;
+		const char *pci_address_name = pci_name(ddev->pdev);
+		char *pciaddstr, *pciaddstr_tmp, *pciaddname_tmp, *pciaddname;
+
+		pciaddstr = kstrdup(amdgpu_virtual_display, GFP_KERNEL);
+		pciaddstr_tmp = pciaddstr;
+		while ((pciaddname_tmp = strsep(&pciaddstr_tmp, ";"))) {
+			pciaddname = strsep(&pciaddname_tmp, ",");
+			if (!strcmp("all", pciaddname)
+			    || !strcmp(pci_address_name, pciaddname)) {
+				long num_crtc;
+				int res = -1;
+
+				adev->enable_virtual_display = true;
+
+				if (pciaddname_tmp)
+					res = kstrtol(pciaddname_tmp, 10,
+						      &num_crtc);
+
+				if (!res) {
+					if (num_crtc < 1)
+						num_crtc = 1;
+					if (num_crtc > 6)
+						num_crtc = 6;
+					adev->mode_info.num_crtc = num_crtc;
+				} else {
+					adev->mode_info.num_crtc = 1;
+				}
+				break;
+			}
+		}
+
+		DRM_INFO("virtual display string:%s, %s:virtual_display:%d, num_crtc:%d\n",
+			 amdgpu_virtual_display, pci_address_name,
+			 adev->enable_virtual_display, adev->mode_info.num_crtc);
+
+		kfree(pciaddstr);
+	}
+#endif
+}
+
+/**
+ * amdgpu_device_parse_gpu_info_fw - parse gpu info firmware
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Parses the asic configuration parameters specified in the gpu info
+ * firmware and makes them availale to the driver for use in configuring
+ * the asic.
+ * Returns 0 on success, -EINVAL on failure.
+ */
+static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+	const struct gpu_info_firmware_header_v1_0 *hdr;
+
+	adev->firmware.gpu_info_fw = NULL;
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_VERDE:
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_OLAND:
+	case CHIP_HAINAN:
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+#endif
+	case CHIP_VEGA20:
+	default:
+		return 0;
+	case CHIP_VEGA10:
+		chip_name = "vega10";
+		break;
+	case CHIP_VEGA12:
+		chip_name = "vega12";
+		break;
+	case CHIP_RAVEN:
+		chip_name = "raven";
+		break;
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_gpu_info.bin", chip_name);
+	err = request_firmware(&adev->firmware.gpu_info_fw, fw_name, adev->dev);
+	if (err) {
+		dev_err(adev->dev,
+			"Failed to load gpu_info firmware \"%s\"\n",
+			fw_name);
+		goto out;
+	}
+	err = amdgpu_ucode_validate(adev->firmware.gpu_info_fw);
+	if (err) {
+		dev_err(adev->dev,
+			"Failed to validate gpu_info firmware \"%s\"\n",
+			fw_name);
+		goto out;
+	}
+
+	hdr = (const struct gpu_info_firmware_header_v1_0 *)adev->firmware.gpu_info_fw->data;
+	amdgpu_ucode_print_gpu_info_hdr(&hdr->header);
+
+	switch (hdr->version_major) {
+	case 1:
+	{
+		const struct gpu_info_firmware_v1_0 *gpu_info_fw =
+			(const struct gpu_info_firmware_v1_0 *)(adev->firmware.gpu_info_fw->data +
+								le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+		adev->gfx.config.max_shader_engines = le32_to_cpu(gpu_info_fw->gc_num_se);
+		adev->gfx.config.max_cu_per_sh = le32_to_cpu(gpu_info_fw->gc_num_cu_per_sh);
+		adev->gfx.config.max_sh_per_se = le32_to_cpu(gpu_info_fw->gc_num_sh_per_se);
+		adev->gfx.config.max_backends_per_se = le32_to_cpu(gpu_info_fw->gc_num_rb_per_se);
+		adev->gfx.config.max_texture_channel_caches =
+			le32_to_cpu(gpu_info_fw->gc_num_tccs);
+		adev->gfx.config.max_gprs = le32_to_cpu(gpu_info_fw->gc_num_gprs);
+		adev->gfx.config.max_gs_threads = le32_to_cpu(gpu_info_fw->gc_num_max_gs_thds);
+		adev->gfx.config.gs_vgt_table_depth = le32_to_cpu(gpu_info_fw->gc_gs_table_depth);
+		adev->gfx.config.gs_prim_buffer_depth = le32_to_cpu(gpu_info_fw->gc_gsprim_buff_depth);
+		adev->gfx.config.double_offchip_lds_buf =
+			le32_to_cpu(gpu_info_fw->gc_double_offchip_lds_buffer);
+		adev->gfx.cu_info.wave_front_size = le32_to_cpu(gpu_info_fw->gc_wave_size);
+		adev->gfx.cu_info.max_waves_per_simd =
+			le32_to_cpu(gpu_info_fw->gc_max_waves_per_simd);
+		adev->gfx.cu_info.max_scratch_slots_per_cu =
+			le32_to_cpu(gpu_info_fw->gc_max_scratch_slots_per_cu);
+		adev->gfx.cu_info.lds_size = le32_to_cpu(gpu_info_fw->gc_lds_size);
+		break;
+	}
+	default:
+		dev_err(adev->dev,
+			"Unsupported gpu_info table %d\n", hdr->header.ucode_version);
+		err = -EINVAL;
+		goto out;
+	}
+out:
+	return err;
+}
+
+/**
+ * amdgpu_device_ip_early_init - run early init for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Early initialization pass for hardware IPs.  The hardware IPs that make
+ * up each asic are discovered each IP's early_init callback is run.  This
+ * is the first stage in initializing the asic.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	amdgpu_device_enable_virtual_display(adev);
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		if (adev->asic_type == CHIP_CARRIZO || adev->asic_type == CHIP_STONEY)
+			adev->family = AMDGPU_FAMILY_CZ;
+		else
+			adev->family = AMDGPU_FAMILY_VI;
+
+		r = vi_set_ip_blocks(adev);
+		if (r)
+			return r;
+		break;
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_VERDE:
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_OLAND:
+	case CHIP_HAINAN:
+		adev->family = AMDGPU_FAMILY_SI;
+		r = si_set_ip_blocks(adev);
+		if (r)
+			return r;
+		break;
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		if ((adev->asic_type == CHIP_BONAIRE) || (adev->asic_type == CHIP_HAWAII))
+			adev->family = AMDGPU_FAMILY_CI;
+		else
+			adev->family = AMDGPU_FAMILY_KV;
+
+		r = cik_set_ip_blocks(adev);
+		if (r)
+			return r;
+		break;
+#endif
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+	case CHIP_RAVEN:
+		if (adev->asic_type == CHIP_RAVEN)
+			adev->family = AMDGPU_FAMILY_RV;
+		else
+			adev->family = AMDGPU_FAMILY_AI;
+
+		r = soc15_set_ip_blocks(adev);
+		if (r)
+			return r;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	r = amdgpu_device_parse_gpu_info_fw(adev);
+	if (r)
+		return r;
+
+	amdgpu_amdkfd_device_probe(adev);
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = amdgpu_virt_request_full_gpu(adev, true);
+		if (r)
+			return -EAGAIN;
+	}
+
+	adev->powerplay.pp_feature = amdgpu_pp_feature_mask;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if ((amdgpu_ip_block_mask & (1 << i)) == 0) {
+			DRM_ERROR("disabled ip block: %d <%s>\n",
+				  i, adev->ip_blocks[i].version->funcs->name);
+			adev->ip_blocks[i].status.valid = false;
+		} else {
+			if (adev->ip_blocks[i].version->funcs->early_init) {
+				r = adev->ip_blocks[i].version->funcs->early_init((void *)adev);
+				if (r == -ENOENT) {
+					adev->ip_blocks[i].status.valid = false;
+				} else if (r) {
+					DRM_ERROR("early_init of IP block <%s> failed %d\n",
+						  adev->ip_blocks[i].version->funcs->name, r);
+					return r;
+				} else {
+					adev->ip_blocks[i].status.valid = true;
+				}
+			} else {
+				adev->ip_blocks[i].status.valid = true;
+			}
+		}
+	}
+
+	adev->cg_flags &= amdgpu_cg_mask;
+	adev->pg_flags &= amdgpu_pg_mask;
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_init - run init for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Main initialization pass for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked and the sw_init and hw_init callbacks
+ * are run.  sw_init initializes the software state associated with each IP
+ * and hw_init initializes the hardware associated with each IP.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_init(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		r = adev->ip_blocks[i].version->funcs->sw_init((void *)adev);
+		if (r) {
+			DRM_ERROR("sw_init of IP block <%s> failed %d\n",
+				  adev->ip_blocks[i].version->funcs->name, r);
+			return r;
+		}
+		adev->ip_blocks[i].status.sw = true;
+
+		/* need to do gmc hw init early so we can allocate gpu mem */
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC) {
+			r = amdgpu_device_vram_scratch_init(adev);
+			if (r) {
+				DRM_ERROR("amdgpu_vram_scratch_init failed %d\n", r);
+				return r;
+			}
+			r = adev->ip_blocks[i].version->funcs->hw_init((void *)adev);
+			if (r) {
+				DRM_ERROR("hw_init %d failed %d\n", i, r);
+				return r;
+			}
+			r = amdgpu_device_wb_init(adev);
+			if (r) {
+				DRM_ERROR("amdgpu_device_wb_init failed %d\n", r);
+				return r;
+			}
+			adev->ip_blocks[i].status.hw = true;
+
+			/* right after GMC hw init, we create CSA */
+			if (amdgpu_sriov_vf(adev)) {
+				r = amdgpu_allocate_static_csa(adev);
+				if (r) {
+					DRM_ERROR("allocate CSA failed %d\n", r);
+					return r;
+				}
+			}
+		}
+	}
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.sw)
+			continue;
+		if (adev->ip_blocks[i].status.hw)
+			continue;
+		r = adev->ip_blocks[i].version->funcs->hw_init((void *)adev);
+		if (r) {
+			DRM_ERROR("hw_init of IP block <%s> failed %d\n",
+				  adev->ip_blocks[i].version->funcs->name, r);
+			return r;
+		}
+		adev->ip_blocks[i].status.hw = true;
+	}
+
+	amdgpu_amdkfd_device_init(adev);
+
+	if (amdgpu_sriov_vf(adev)) {
+		amdgpu_virt_init_data_exchange(adev);
+		amdgpu_virt_release_full_gpu(adev, true);
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_fill_reset_magic - writes reset magic to gart pointer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Writes a reset magic value to the gart pointer in VRAM.  The driver calls
+ * this function before a GPU reset.  If the value is retained after a
+ * GPU reset, VRAM has not been lost.  Some GPU resets may destry VRAM contents.
+ */
+static void amdgpu_device_fill_reset_magic(struct amdgpu_device *adev)
+{
+	memcpy(adev->reset_magic, adev->gart.ptr, AMDGPU_RESET_MAGIC_NUM);
+}
+
+/**
+ * amdgpu_device_check_vram_lost - check if vram is valid
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Checks the reset magic value written to the gart pointer in VRAM.
+ * The driver calls this after a GPU reset to see if the contents of
+ * VRAM is lost or now.
+ * returns true if vram is lost, false if not.
+ */
+static bool amdgpu_device_check_vram_lost(struct amdgpu_device *adev)
+{
+	return !!memcmp(adev->gart.ptr, adev->reset_magic,
+			AMDGPU_RESET_MAGIC_NUM);
+}
+
+/**
+ * amdgpu_device_ip_late_set_cg_state - late init for clockgating
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Late initialization pass enabling clockgating for hardware IPs.
+ * The list of all the hardware IPs that make up the asic is walked and the
+ * set_clockgating_state callbacks are run.  This stage is run late
+ * in the init process.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_late_set_cg_state(struct amdgpu_device *adev)
+{
+	int i = 0, r;
+
+	if (amdgpu_emu_mode == 1)
+		return 0;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		/* skip CG for VCE/UVD, it's handled specially */
+		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD &&
+		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCE &&
+		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCN &&
+		    adev->ip_blocks[i].version->funcs->set_clockgating_state) {
+			/* enable clockgating to save power */
+			r = adev->ip_blocks[i].version->funcs->set_clockgating_state((void *)adev,
+										     AMD_CG_STATE_GATE);
+			if (r) {
+				DRM_ERROR("set_clockgating_state(gate) of IP block <%s> failed %d\n",
+					  adev->ip_blocks[i].version->funcs->name, r);
+				return r;
+			}
+		}
+	}
+
+	return 0;
+}
+
+static int amdgpu_device_ip_late_set_pg_state(struct amdgpu_device *adev)
+{
+	int i = 0, r;
+
+	if (amdgpu_emu_mode == 1)
+		return 0;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		/* skip CG for VCE/UVD, it's handled specially */
+		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD &&
+		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCE &&
+		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCN &&
+		    adev->ip_blocks[i].version->funcs->set_powergating_state) {
+			/* enable powergating to save power */
+			r = adev->ip_blocks[i].version->funcs->set_powergating_state((void *)adev,
+										     AMD_PG_STATE_GATE);
+			if (r) {
+				DRM_ERROR("set_powergating_state(gate) of IP block <%s> failed %d\n",
+					  adev->ip_blocks[i].version->funcs->name, r);
+				return r;
+			}
+		}
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_late_init - run late init for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Late initialization pass for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked and the late_init callbacks are run.
+ * late_init covers any special initialization that an IP requires
+ * after all of the have been initialized or something that needs to happen
+ * late in the init process.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_late_init(struct amdgpu_device *adev)
+{
+	int i = 0, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].version->funcs->late_init) {
+			r = adev->ip_blocks[i].version->funcs->late_init((void *)adev);
+			if (r) {
+				DRM_ERROR("late_init of IP block <%s> failed %d\n",
+					  adev->ip_blocks[i].version->funcs->name, r);
+				return r;
+			}
+			adev->ip_blocks[i].status.late_initialized = true;
+		}
+	}
+
+	amdgpu_device_ip_late_set_cg_state(adev);
+	amdgpu_device_ip_late_set_pg_state(adev);
+
+	queue_delayed_work(system_wq, &adev->late_init_work,
+			   msecs_to_jiffies(AMDGPU_RESUME_MS));
+
+	amdgpu_device_fill_reset_magic(adev);
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_fini - run fini for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Main teardown pass for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked and the hw_fini and sw_fini callbacks
+ * are run.  hw_fini tears down the hardware associated with each IP
+ * and sw_fini tears down any software state associated with each IP.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_fini(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	amdgpu_amdkfd_device_fini(adev);
+	/* need to disable SMC first */
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.hw)
+			continue;
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC &&
+			adev->ip_blocks[i].version->funcs->set_clockgating_state) {
+			/* ungate blocks before hw fini so that we can shutdown the blocks safely */
+			r = adev->ip_blocks[i].version->funcs->set_clockgating_state((void *)adev,
+										     AMD_CG_STATE_UNGATE);
+			if (r) {
+				DRM_ERROR("set_clockgating_state(ungate) of IP block <%s> failed %d\n",
+					  adev->ip_blocks[i].version->funcs->name, r);
+				return r;
+			}
+			if (adev->powerplay.pp_funcs->set_powergating_by_smu)
+				amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_GFX, false);
+			r = adev->ip_blocks[i].version->funcs->hw_fini((void *)adev);
+			/* XXX handle errors */
+			if (r) {
+				DRM_DEBUG("hw_fini of IP block <%s> failed %d\n",
+					  adev->ip_blocks[i].version->funcs->name, r);
+			}
+			adev->ip_blocks[i].status.hw = false;
+			break;
+		}
+	}
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!adev->ip_blocks[i].status.hw)
+			continue;
+
+		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD &&
+			adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCE &&
+			adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCN &&
+			adev->ip_blocks[i].version->funcs->set_clockgating_state) {
+			/* ungate blocks before hw fini so that we can shutdown the blocks safely */
+			r = adev->ip_blocks[i].version->funcs->set_clockgating_state((void *)adev,
+										     AMD_CG_STATE_UNGATE);
+			if (r) {
+				DRM_ERROR("set_clockgating_state(ungate) of IP block <%s> failed %d\n",
+					  adev->ip_blocks[i].version->funcs->name, r);
+				return r;
+			}
+		}
+
+		r = adev->ip_blocks[i].version->funcs->hw_fini((void *)adev);
+		/* XXX handle errors */
+		if (r) {
+			DRM_DEBUG("hw_fini of IP block <%s> failed %d\n",
+				  adev->ip_blocks[i].version->funcs->name, r);
+		}
+
+		adev->ip_blocks[i].status.hw = false;
+	}
+
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!adev->ip_blocks[i].status.sw)
+			continue;
+
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC) {
+			amdgpu_free_static_csa(adev);
+			amdgpu_device_wb_fini(adev);
+			amdgpu_device_vram_scratch_fini(adev);
+		}
+
+		r = adev->ip_blocks[i].version->funcs->sw_fini((void *)adev);
+		/* XXX handle errors */
+		if (r) {
+			DRM_DEBUG("sw_fini of IP block <%s> failed %d\n",
+				  adev->ip_blocks[i].version->funcs->name, r);
+		}
+		adev->ip_blocks[i].status.sw = false;
+		adev->ip_blocks[i].status.valid = false;
+	}
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!adev->ip_blocks[i].status.late_initialized)
+			continue;
+		if (adev->ip_blocks[i].version->funcs->late_fini)
+			adev->ip_blocks[i].version->funcs->late_fini((void *)adev);
+		adev->ip_blocks[i].status.late_initialized = false;
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		if (amdgpu_virt_release_full_gpu(adev, false))
+			DRM_ERROR("failed to release exclusive mode on fini\n");
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_late_init_func_handler - work handler for clockgating
+ *
+ * @work: work_struct
+ *
+ * Work handler for amdgpu_device_ip_late_set_cg_state.  We put the
+ * clockgating setup into a worker thread to speed up driver init and
+ * resume from suspend.
+ */
+static void amdgpu_device_ip_late_init_func_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, late_init_work.work);
+	int r;
+
+	r = amdgpu_ib_ring_tests(adev);
+	if (r)
+		DRM_ERROR("ib ring test failed (%d).\n", r);
+}
+
+/**
+ * amdgpu_device_ip_suspend_phase1 - run suspend for hardware IPs (phase 1)
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Main suspend function for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked, clockgating is disabled and the
+ * suspend callbacks are run.  suspend puts the hardware and software state
+ * in each IP into a state suitable for suspend.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_request_full_gpu(adev, false);
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		/* displays are handled separately */
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_DCE) {
+			/* ungate blocks so that suspend can properly shut them down */
+			if (adev->ip_blocks[i].version->funcs->set_clockgating_state) {
+				r = adev->ip_blocks[i].version->funcs->set_clockgating_state((void *)adev,
+											     AMD_CG_STATE_UNGATE);
+				if (r) {
+					DRM_ERROR("set_clockgating_state(ungate) of IP block <%s> failed %d\n",
+						  adev->ip_blocks[i].version->funcs->name, r);
+				}
+			}
+			/* XXX handle errors */
+			r = adev->ip_blocks[i].version->funcs->suspend(adev);
+			/* XXX handle errors */
+			if (r) {
+				DRM_ERROR("suspend of IP block <%s> failed %d\n",
+					  adev->ip_blocks[i].version->funcs->name, r);
+			}
+		}
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_release_full_gpu(adev, false);
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_suspend_phase2 - run suspend for hardware IPs (phase 2)
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Main suspend function for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked, clockgating is disabled and the
+ * suspend callbacks are run.  suspend puts the hardware and software state
+ * in each IP into a state suitable for suspend.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_request_full_gpu(adev, false);
+
+	/* ungate SMC block first */
+	r = amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_SMC,
+						   AMD_CG_STATE_UNGATE);
+	if (r) {
+		DRM_ERROR("set_clockgating_state(ungate) SMC failed %d\n", r);
+	}
+
+	/* call smu to disable gfx off feature first when suspend */
+	if (adev->powerplay.pp_funcs->set_powergating_by_smu)
+		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_GFX, false);
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		/* displays are handled in phase1 */
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_DCE)
+			continue;
+		/* ungate blocks so that suspend can properly shut them down */
+		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_SMC &&
+			adev->ip_blocks[i].version->funcs->set_clockgating_state) {
+			r = adev->ip_blocks[i].version->funcs->set_clockgating_state((void *)adev,
+										     AMD_CG_STATE_UNGATE);
+			if (r) {
+				DRM_ERROR("set_clockgating_state(ungate) of IP block <%s> failed %d\n",
+					  adev->ip_blocks[i].version->funcs->name, r);
+			}
+		}
+		/* XXX handle errors */
+		r = adev->ip_blocks[i].version->funcs->suspend(adev);
+		/* XXX handle errors */
+		if (r) {
+			DRM_ERROR("suspend of IP block <%s> failed %d\n",
+				  adev->ip_blocks[i].version->funcs->name, r);
+		}
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_release_full_gpu(adev, false);
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_suspend - run suspend for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Main suspend function for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked, clockgating is disabled and the
+ * suspend callbacks are run.  suspend puts the hardware and software state
+ * in each IP into a state suitable for suspend.
+ * Returns 0 on success, negative error code on failure.
+ */
+int amdgpu_device_ip_suspend(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_device_ip_suspend_phase1(adev);
+	if (r)
+		return r;
+	r = amdgpu_device_ip_suspend_phase2(adev);
+
+	return r;
+}
+
+static int amdgpu_device_ip_reinit_early_sriov(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	static enum amd_ip_block_type ip_order[] = {
+		AMD_IP_BLOCK_TYPE_GMC,
+		AMD_IP_BLOCK_TYPE_COMMON,
+		AMD_IP_BLOCK_TYPE_PSP,
+		AMD_IP_BLOCK_TYPE_IH,
+	};
+
+	for (i = 0; i < ARRAY_SIZE(ip_order); i++) {
+		int j;
+		struct amdgpu_ip_block *block;
+
+		for (j = 0; j < adev->num_ip_blocks; j++) {
+			block = &adev->ip_blocks[j];
+
+			if (block->version->type != ip_order[i] ||
+				!block->status.valid)
+				continue;
+
+			r = block->version->funcs->hw_init(adev);
+			DRM_INFO("RE-INIT: %s %s\n", block->version->funcs->name, r?"failed":"succeeded");
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+static int amdgpu_device_ip_reinit_late_sriov(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	static enum amd_ip_block_type ip_order[] = {
+		AMD_IP_BLOCK_TYPE_SMC,
+		AMD_IP_BLOCK_TYPE_DCE,
+		AMD_IP_BLOCK_TYPE_GFX,
+		AMD_IP_BLOCK_TYPE_SDMA,
+		AMD_IP_BLOCK_TYPE_UVD,
+		AMD_IP_BLOCK_TYPE_VCE
+	};
+
+	for (i = 0; i < ARRAY_SIZE(ip_order); i++) {
+		int j;
+		struct amdgpu_ip_block *block;
+
+		for (j = 0; j < adev->num_ip_blocks; j++) {
+			block = &adev->ip_blocks[j];
+
+			if (block->version->type != ip_order[i] ||
+				!block->status.valid)
+				continue;
+
+			r = block->version->funcs->hw_init(adev);
+			DRM_INFO("RE-INIT: %s %s\n", block->version->funcs->name, r?"failed":"succeeded");
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_resume_phase1 - run resume for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * First resume function for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked and the resume callbacks are run for
+ * COMMON, GMC, and IH.  resume puts the hardware into a functional state
+ * after a suspend and updates the software state as necessary.  This
+ * function is also used for restoring the GPU after a GPU reset.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_resume_phase1(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON ||
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC ||
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH) {
+			r = adev->ip_blocks[i].version->funcs->resume(adev);
+			if (r) {
+				DRM_ERROR("resume of IP block <%s> failed %d\n",
+					  adev->ip_blocks[i].version->funcs->name, r);
+				return r;
+			}
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_resume_phase2 - run resume for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * First resume function for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked and the resume callbacks are run for
+ * all blocks except COMMON, GMC, and IH.  resume puts the hardware into a
+ * functional state after a suspend and updates the software state as
+ * necessary.  This function is also used for restoring the GPU after a GPU
+ * reset.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_resume_phase2(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON ||
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC ||
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH)
+			continue;
+		r = adev->ip_blocks[i].version->funcs->resume(adev);
+		if (r) {
+			DRM_ERROR("resume of IP block <%s> failed %d\n",
+				  adev->ip_blocks[i].version->funcs->name, r);
+			return r;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_resume - run resume for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Main resume function for hardware IPs.  The hardware IPs
+ * are split into two resume functions because they are
+ * are also used in in recovering from a GPU reset and some additional
+ * steps need to be take between them.  In this case (S3/S4) they are
+ * run sequentially.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_resume(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_device_ip_resume_phase1(adev);
+	if (r)
+		return r;
+	r = amdgpu_device_ip_resume_phase2(adev);
+
+	return r;
+}
+
+/**
+ * amdgpu_device_detect_sriov_bios - determine if the board supports SR-IOV
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Query the VBIOS data tables to determine if the board supports SR-IOV.
+ */
+static void amdgpu_device_detect_sriov_bios(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev)) {
+		if (adev->is_atom_fw) {
+			if (amdgpu_atomfirmware_gpu_supports_virtualization(adev))
+				adev->virt.caps |= AMDGPU_SRIOV_CAPS_SRIOV_VBIOS;
+		} else {
+			if (amdgpu_atombios_has_gpu_virtualization_table(adev))
+				adev->virt.caps |= AMDGPU_SRIOV_CAPS_SRIOV_VBIOS;
+		}
+
+		if (!(adev->virt.caps & AMDGPU_SRIOV_CAPS_SRIOV_VBIOS))
+			amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_NO_VBIOS, 0, 0);
+	}
+}
+
+/**
+ * amdgpu_device_asic_has_dc_support - determine if DC supports the asic
+ *
+ * @asic_type: AMD asic type
+ *
+ * Check if there is DC (new modesetting infrastructre) support for an asic.
+ * returns true if DC has support, false if not.
+ */
+bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type)
+{
+	switch (asic_type) {
+#if defined(CONFIG_DRM_AMD_DC)
+	case CHIP_BONAIRE:
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		/*
+		 * We have systems in the wild with these ASICs that require
+		 * LVDS and VGA support which is not supported with DC.
+		 *
+		 * Fallback to the non-DC driver here by default so as not to
+		 * cause regressions.
+		 */
+		return amdgpu_dc > 0;
+	case CHIP_HAWAII:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+#if defined(CONFIG_DRM_AMD_DC_DCN1_0)
+	case CHIP_RAVEN:
+#endif
+		return amdgpu_dc != 0;
+#endif
+	default:
+		return false;
+	}
+}
+
+/**
+ * amdgpu_device_has_dc_support - check if dc is supported
+ *
+ * @adev: amdgpu_device_pointer
+ *
+ * Returns true for supported, false for not supported
+ */
+bool amdgpu_device_has_dc_support(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev))
+		return false;
+
+	return amdgpu_device_asic_has_dc_support(adev->asic_type);
+}
+
+/**
+ * amdgpu_device_init - initialize the driver
+ *
+ * @adev: amdgpu_device pointer
+ * @ddev: drm dev pointer
+ * @pdev: pci dev pointer
+ * @flags: driver flags
+ *
+ * Initializes the driver info and hw (all asics).
+ * Returns 0 for success or an error on failure.
+ * Called at driver startup.
+ */
+int amdgpu_device_init(struct amdgpu_device *adev,
+		       struct drm_device *ddev,
+		       struct pci_dev *pdev,
+		       uint32_t flags)
+{
+	int r, i;
+	bool runtime = false;
+	u32 max_MBps;
+
+	adev->shutdown = false;
+#ifdef __linux__
+	adev->dev = &pdev->dev;
+#endif
+	adev->ddev = ddev;
+	adev->pdev = pdev;
+	adev->flags = flags;
+	adev->asic_type = flags & AMD_ASIC_MASK;
+	adev->usec_timeout = AMDGPU_MAX_USEC_TIMEOUT;
+	if (amdgpu_emu_mode == 1)
+		adev->usec_timeout *= 2;
+	adev->gmc.gart_size = 512 * 1024 * 1024;
+	adev->accel_working = false;
+	adev->num_rings = 0;
+	adev->mman.buffer_funcs = NULL;
+	adev->mman.buffer_funcs_ring = NULL;
+	adev->vm_manager.vm_pte_funcs = NULL;
+	adev->vm_manager.vm_pte_num_rings = 0;
+	adev->gmc.gmc_funcs = NULL;
+	adev->fence_context = dma_fence_context_alloc(AMDGPU_MAX_RINGS);
+	bitmap_zero(adev->gfx.pipe_reserve_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+
+	adev->smc_rreg = &amdgpu_invalid_rreg;
+	adev->smc_wreg = &amdgpu_invalid_wreg;
+	adev->pcie_rreg = &amdgpu_invalid_rreg;
+	adev->pcie_wreg = &amdgpu_invalid_wreg;
+	adev->pciep_rreg = &amdgpu_invalid_rreg;
+	adev->pciep_wreg = &amdgpu_invalid_wreg;
+	adev->uvd_ctx_rreg = &amdgpu_invalid_rreg;
+	adev->uvd_ctx_wreg = &amdgpu_invalid_wreg;
+	adev->didt_rreg = &amdgpu_invalid_rreg;
+	adev->didt_wreg = &amdgpu_invalid_wreg;
+	adev->gc_cac_rreg = &amdgpu_invalid_rreg;
+	adev->gc_cac_wreg = &amdgpu_invalid_wreg;
+	adev->audio_endpt_rreg = &amdgpu_block_invalid_rreg;
+	adev->audio_endpt_wreg = &amdgpu_block_invalid_wreg;
+
+	DRM_INFO("initializing kernel modesetting (%s 0x%04X:0x%04X 0x%04X:0x%04X 0x%02X).\n",
+		 amdgpu_asic_name[adev->asic_type], pdev->vendor, pdev->device,
+		 pdev->subsystem_vendor, pdev->subsystem_device, pdev->revision);
+
+	/* mutex initialization are all done here so we
+	 * can recall function without having locking issues */
+	atomic_set(&adev->irq.ih.lock, 0);
+	rw_init(&adev->firmware.mutex, "agfw");
+	rw_init(&adev->pm.mutex, "agpm");
+	rw_init(&adev->gfx.gpu_clock_mutex, "gfxclk");
+	rw_init(&adev->srbm_mutex, "srbm");
+	rw_init(&adev->gfx.pipe_reserve_mutex, "pipers");
+	rw_init(&adev->grbm_idx_mutex, "grbmidx");
+	rw_init(&adev->mn_lock, "agpumn");
+	rw_init(&adev->virt.vf_errors.lock, "vferr");
+	hash_init(adev->mn_hash);
+	rw_init(&adev->lock_reset, "aglkrst");
+
+	amdgpu_device_check_arguments(adev);
+
+	mtx_init(&adev->mmio_idx_lock, IPL_TTY);
+	mtx_init(&adev->smc_idx_lock, IPL_TTY);
+	mtx_init(&adev->pcie_idx_lock, IPL_TTY);
+	mtx_init(&adev->uvd_ctx_idx_lock, IPL_TTY);
+	mtx_init(&adev->didt_idx_lock, IPL_TTY);
+	mtx_init(&adev->gc_cac_idx_lock, IPL_TTY);
+	mtx_init(&adev->se_cac_idx_lock, IPL_TTY);
+	mtx_init(&adev->audio_endpt_idx_lock, IPL_TTY);
+	mtx_init(&adev->mm_stats.lock, IPL_TTY);
+
+	INIT_LIST_HEAD(&adev->shadow_list);
+	rw_init(&adev->shadow_list_lock, "sdwlst");
+
+	INIT_LIST_HEAD(&adev->ring_lru_list);
+	mtx_init(&adev->ring_lru_list_lock, IPL_TTY);
+
+	INIT_DELAYED_WORK(&adev->late_init_work,
+			  amdgpu_device_ip_late_init_func_handler);
+
+	adev->pm.ac_power = power_supply_is_system_supplied() > 0 ? true : false;
+
+#ifdef __linux__
+	/* Registers mapping */
+	/* TODO: block userspace mapping of io register */
+	if (adev->asic_type >= CHIP_BONAIRE) {
+		adev->rmmio_base = pci_resource_start(adev->pdev, 5);
+		adev->rmmio_size = pci_resource_len(adev->pdev, 5);
+	} else {
+		adev->rmmio_base = pci_resource_start(adev->pdev, 2);
+		adev->rmmio_size = pci_resource_len(adev->pdev, 2);
+	}
+
+	adev->rmmio = ioremap(adev->rmmio_base, adev->rmmio_size);
+	if (adev->rmmio == NULL) {
+		return -ENOMEM;
+	}
+#endif
+	DRM_INFO("register mmio base: 0x%08X\n", (uint32_t)adev->rmmio_base);
+	DRM_INFO("register mmio size: %u\n", (unsigned)adev->rmmio_size);
+
+	/* doorbell bar mapping */
+	amdgpu_device_doorbell_init(adev);
+
+	/* io port mapping */
+#ifdef __linux__
+	for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) {
+		if (pci_resource_flags(adev->pdev, i) & IORESOURCE_IO) {
+			adev->rio_mem_size = pci_resource_len(adev->pdev, i);
+			adev->rio_mem = pci_iomap(adev->pdev, i, adev->rio_mem_size);
+			break;
+		}
+	}
+	if (adev->rio_mem == NULL)
+		DRM_INFO("PCI I/O BAR is not found.\n");
+#endif
+
+	amdgpu_device_get_pcie_info(adev);
+
+	/* early init functions */
+	r = amdgpu_device_ip_early_init(adev);
+	if (r)
+		return r;
+
+	/* if we have > 1 VGA cards, then disable the amdgpu VGA resources */
+	/* this will fail for cards that aren't VGA class devices, just
+	 * ignore it */
+#ifdef notyet
+	vga_client_register(adev->pdev, adev, NULL, amdgpu_device_vga_set_decode);
+#endif
+
+	if (amdgpu_device_is_px(ddev))
+		runtime = true;
+#ifdef notyet
+	if (!pci_is_thunderbolt_attached(adev->pdev))
+		vga_switcheroo_register_client(adev->pdev,
+					       &amdgpu_switcheroo_ops, runtime);
+	if (runtime)
+		vga_switcheroo_init_domain_pm_ops(adev->dev, &adev->vga_pm_domain);
+#endif
+
+	if (amdgpu_emu_mode == 1) {
+		/* post the asic on emulation mode */
+		emu_soc_asic_init(adev);
+		goto fence_driver_init;
+	}
+
+	/* Read BIOS */
+	if (!amdgpu_get_bios(adev)) {
+		r = -EINVAL;
+		goto failed;
+	}
+
+	r = amdgpu_atombios_init(adev);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_atombios_init failed\n");
+		amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_ATOMBIOS_INIT_FAIL, 0, 0);
+		goto failed;
+	}
+
+	/* detect if we are with an SRIOV vbios */
+	amdgpu_device_detect_sriov_bios(adev);
+
+	/* Post card if necessary */
+	if (amdgpu_device_need_post(adev)) {
+		if (!adev->bios) {
+			dev_err(adev->dev, "no vBIOS found\n");
+			r = -EINVAL;
+			goto failed;
+		}
+		DRM_INFO("GPU posting now...\n");
+		r = amdgpu_atom_asic_init(adev->mode_info.atom_context);
+		if (r) {
+			dev_err(adev->dev, "gpu post error!\n");
+			goto failed;
+		}
+	}
+
+	if (adev->is_atom_fw) {
+		/* Initialize clocks */
+		r = amdgpu_atomfirmware_get_clock_info(adev);
+		if (r) {
+			dev_err(adev->dev, "amdgpu_atomfirmware_get_clock_info failed\n");
+			amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_ATOMBIOS_GET_CLOCK_FAIL, 0, 0);
+			goto failed;
+		}
+	} else {
+		/* Initialize clocks */
+		r = amdgpu_atombios_get_clock_info(adev);
+		if (r) {
+			dev_err(adev->dev, "amdgpu_atombios_get_clock_info failed\n");
+			amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_ATOMBIOS_GET_CLOCK_FAIL, 0, 0);
+			goto failed;
+		}
+		/* init i2c buses */
+		if (!amdgpu_device_has_dc_support(adev))
+			amdgpu_atombios_i2c_init(adev);
+	}
+
+fence_driver_init:
+	/* Fence driver */
+	r = amdgpu_fence_driver_init(adev);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_fence_driver_init failed\n");
+		amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_FENCE_INIT_FAIL, 0, 0);
+		goto failed;
+	}
+
+	/* init the mode config */
+	drm_mode_config_init(adev->ddev);
+
+	r = amdgpu_device_ip_init(adev);
+	if (r) {
+		/* failed in exclusive mode due to timeout */
+		if (amdgpu_sriov_vf(adev) &&
+		    !amdgpu_sriov_runtime(adev) &&
+		    amdgpu_virt_mmio_blocked(adev) &&
+		    !amdgpu_virt_wait_reset(adev)) {
+			dev_err(adev->dev, "VF exclusive mode timeout\n");
+			/* Don't send request since VF is inactive. */
+			adev->virt.caps &= ~AMDGPU_SRIOV_CAPS_RUNTIME;
+			adev->virt.ops = NULL;
+			r = -EAGAIN;
+			goto failed;
+		}
+		dev_err(adev->dev, "amdgpu_device_ip_init failed\n");
+		amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_AMDGPU_INIT_FAIL, 0, 0);
+		goto failed;
+	}
+
+	adev->accel_working = true;
+
+	amdgpu_vm_check_compute_bug(adev);
+
+	/* Initialize the buffer migration limit. */
+	if (amdgpu_moverate >= 0)
+		max_MBps = amdgpu_moverate;
+	else
+		max_MBps = 8; /* Allow 8 MB/s. */
+	/* Get a log2 for easy divisions. */
+	adev->mm_stats.log2_max_MBps = ilog2(max(1u, max_MBps));
+
+	r = amdgpu_ib_pool_init(adev);
+	if (r) {
+		dev_err(adev->dev, "IB initialization failed (%d).\n", r);
+		amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_IB_INIT_FAIL, 0, r);
+		goto failed;
+	}
+
+	amdgpu_fbdev_init(adev);
+
+	r = amdgpu_pm_sysfs_init(adev);
+	if (r)
+		DRM_ERROR("registering pm debugfs failed (%d).\n", r);
+
+	r = amdgpu_debugfs_gem_init(adev);
+	if (r)
+		DRM_ERROR("registering gem debugfs failed (%d).\n", r);
+
+	r = amdgpu_debugfs_regs_init(adev);
+	if (r)
+		DRM_ERROR("registering register debugfs failed (%d).\n", r);
+
+	r = amdgpu_debugfs_firmware_init(adev);
+	if (r)
+		DRM_ERROR("registering firmware debugfs failed (%d).\n", r);
+
+	r = amdgpu_debugfs_init(adev);
+	if (r)
+		DRM_ERROR("Creating debugfs files failed (%d).\n", r);
+
+	if ((amdgpu_testing & 1)) {
+		if (adev->accel_working)
+			amdgpu_test_moves(adev);
+		else
+			DRM_INFO("amdgpu: acceleration disabled, skipping move tests\n");
+	}
+	if (amdgpu_benchmarking) {
+		if (adev->accel_working)
+			amdgpu_benchmark(adev, amdgpu_benchmarking);
+		else
+			DRM_INFO("amdgpu: acceleration disabled, skipping benchmarks\n");
+	}
+
+	/* enable clockgating, etc. after ib tests, etc. since some blocks require
+	 * explicit gating rather than handling it automatically.
+	 */
+	r = amdgpu_device_ip_late_init(adev);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_device_ip_late_init failed\n");
+		amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_AMDGPU_LATE_INIT_FAIL, 0, r);
+		goto failed;
+	}
+
+	return 0;
+
+failed:
+	amdgpu_vf_error_trans_all(adev);
+	if (runtime)
+		vga_switcheroo_fini_domain_pm_ops(adev->dev);
+
+	return r;
+}
+
+/**
+ * amdgpu_device_fini - tear down the driver
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the driver info (all asics).
+ * Called at driver shutdown.
+ */
+void amdgpu_device_fini(struct amdgpu_device *adev)
+{
+	int r;
+
+	DRM_INFO("amdgpu: finishing device.\n");
+	adev->shutdown = true;
+	/* disable all interrupts */
+	amdgpu_irq_disable_all(adev);
+	if (adev->mode_info.mode_config_initialized){
+		if (!amdgpu_device_has_dc_support(adev))
+			drm_crtc_force_disable_all(adev->ddev);
+		else
+			drm_atomic_helper_shutdown(adev->ddev);
+	}
+	amdgpu_ib_pool_fini(adev);
+	amdgpu_fence_driver_fini(adev);
+	amdgpu_pm_sysfs_fini(adev);
+	amdgpu_fbdev_fini(adev);
+	r = amdgpu_device_ip_fini(adev);
+	if (adev->firmware.gpu_info_fw) {
+		release_firmware(adev->firmware.gpu_info_fw);
+		adev->firmware.gpu_info_fw = NULL;
+	}
+	adev->accel_working = false;
+	cancel_delayed_work_sync(&adev->late_init_work);
+	/* free i2c buses */
+	if (!amdgpu_device_has_dc_support(adev))
+		amdgpu_i2c_fini(adev);
+
+	if (amdgpu_emu_mode != 1)
+		amdgpu_atombios_fini(adev);
+
+	kfree(adev->bios);
+	adev->bios = NULL;
+	if (!pci_is_thunderbolt_attached(adev->pdev))
+		vga_switcheroo_unregister_client(adev->pdev);
+	if (adev->flags & AMD_IS_PX)
+		vga_switcheroo_fini_domain_pm_ops(adev->dev);
+	vga_client_register(adev->pdev, NULL, NULL, NULL);
+#ifdef __linux__
+	if (adev->rio_mem)
+		pci_iounmap(adev->pdev, adev->rio_mem);
+	adev->rio_mem = NULL;
+	iounmap(adev->rmmio);
+	adev->rmmio = NULL;
+#else
+	if (adev->rio_mem_size > 0)
+		bus_space_unmap(adev->rio_mem_bst, adev->rio_mem_bsh,
+		    adev->rio_mem_size);
+	adev->rio_mem_size = 0;
+
+	if (adev->rmmio_size > 0)
+		bus_space_unmap(adev->rmmio_bst, adev->rmmio_bsh,
+		    adev->rmmio_size);
+	adev->rmmio_size = 0;
+#endif
+	amdgpu_device_doorbell_fini(adev);
+	amdgpu_debugfs_regs_cleanup(adev);
+}
+
+
+/*
+ * Suspend & resume.
+ */
+/**
+ * amdgpu_device_suspend - initiate device suspend
+ *
+ * @dev: drm dev pointer
+ * @suspend: suspend state
+ * @fbcon : notify the fbdev of suspend
+ *
+ * Puts the hw in the suspend state (all asics).
+ * Returns 0 for success or an error on failure.
+ * Called at driver suspend.
+ */
+int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon)
+{
+	struct amdgpu_device *adev;
+	struct drm_crtc *crtc;
+	struct drm_connector *connector;
+	int r;
+
+	if (dev == NULL || dev->dev_private == NULL) {
+		return -ENODEV;
+	}
+
+	adev = dev->dev_private;
+
+#ifdef notyet
+	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
+		return 0;
+#endif
+
+	drm_kms_helper_poll_disable(dev);
+
+	if (fbcon)
+		amdgpu_fbdev_set_suspend(adev, 1);
+
+	if (!amdgpu_device_has_dc_support(adev)) {
+		/* turn off display hw */
+		drm_modeset_lock_all(dev);
+		list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+			drm_helper_connector_dpms(connector, DRM_MODE_DPMS_OFF);
+		}
+		drm_modeset_unlock_all(dev);
+			/* unpin the front buffers and cursors */
+		list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+			struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+			struct drm_framebuffer *fb = crtc->primary->fb;
+			struct amdgpu_bo *robj;
+
+			if (amdgpu_crtc->cursor_bo) {
+				struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+				r = amdgpu_bo_reserve(aobj, true);
+				if (r == 0) {
+					amdgpu_bo_unpin(aobj);
+					amdgpu_bo_unreserve(aobj);
+				}
+			}
+
+			if (fb == NULL || fb->obj[0] == NULL) {
+				continue;
+			}
+			robj = gem_to_amdgpu_bo(fb->obj[0]);
+			/* don't unpin kernel fb objects */
+			if (!amdgpu_fbdev_robj_is_fb(adev, robj)) {
+				r = amdgpu_bo_reserve(robj, true);
+				if (r == 0) {
+					amdgpu_bo_unpin(robj);
+					amdgpu_bo_unreserve(robj);
+				}
+			}
+		}
+	}
+
+	amdgpu_amdkfd_suspend(adev);
+
+	r = amdgpu_device_ip_suspend_phase1(adev);
+
+	/* evict vram memory */
+	amdgpu_bo_evict_vram(adev);
+
+	amdgpu_fence_driver_suspend(adev);
+
+	r = amdgpu_device_ip_suspend_phase2(adev);
+
+	/* evict remaining vram memory
+	 * This second call to evict vram is to evict the gart page table
+	 * using the CPU.
+	 */
+	amdgpu_bo_evict_vram(adev);
+
+	pci_save_state(dev->pdev);
+	if (suspend) {
+		/* Shut down the device */
+		pci_disable_device(dev->pdev);
+		pci_set_power_state(dev->pdev, PCI_D3hot);
+	} else {
+		r = amdgpu_asic_reset(adev);
+		if (r)
+			DRM_ERROR("amdgpu asic reset failed\n");
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_resume - initiate device resume
+ *
+ * @dev: drm dev pointer
+ * @resume: resume state
+ * @fbcon : notify the fbdev of resume
+ *
+ * Bring the hw back to operating state (all asics).
+ * Returns 0 for success or an error on failure.
+ * Called at driver resume.
+ */
+int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)
+{
+	struct drm_connector *connector;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_crtc *crtc;
+	int r = 0;
+
+#ifdef notyet
+	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
+		return 0;
+#endif
+
+	if (resume) {
+		pci_set_power_state(dev->pdev, PCI_D0);
+		pci_restore_state(dev->pdev);
+		r = pci_enable_device(dev->pdev);
+		if (r)
+			return r;
+	}
+
+	/* post card */
+	if (amdgpu_device_need_post(adev)) {
+		r = amdgpu_atom_asic_init(adev->mode_info.atom_context);
+		if (r)
+			DRM_ERROR("amdgpu asic init failed\n");
+	}
+
+	r = amdgpu_device_ip_resume(adev);
+	if (r) {
+		DRM_ERROR("amdgpu_device_ip_resume failed (%d).\n", r);
+		return r;
+	}
+	amdgpu_fence_driver_resume(adev);
+
+
+	r = amdgpu_device_ip_late_init(adev);
+	if (r)
+		return r;
+
+	if (!amdgpu_device_has_dc_support(adev)) {
+		/* pin cursors */
+		list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+			struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+			if (amdgpu_crtc->cursor_bo) {
+				struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+				r = amdgpu_bo_reserve(aobj, true);
+				if (r == 0) {
+					r = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM);
+					if (r != 0)
+						DRM_ERROR("Failed to pin cursor BO (%d)\n", r);
+					amdgpu_crtc->cursor_addr = amdgpu_bo_gpu_offset(aobj);
+					amdgpu_bo_unreserve(aobj);
+				}
+			}
+		}
+	}
+	r = amdgpu_amdkfd_resume(adev);
+	if (r)
+		return r;
+
+	/* Make sure IB tests flushed */
+	flush_delayed_work(&adev->late_init_work);
+
+	/* blat the mode back in */
+	if (fbcon) {
+		if (!amdgpu_device_has_dc_support(adev)) {
+			/* pre DCE11 */
+			drm_helper_resume_force_mode(dev);
+
+			/* turn on display hw */
+			drm_modeset_lock_all(dev);
+			list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+				drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON);
+			}
+			drm_modeset_unlock_all(dev);
+		}
+		amdgpu_fbdev_set_suspend(adev, 0);
+	}
+
+	drm_kms_helper_poll_enable(dev);
+
+	/*
+	 * Most of the connector probing functions try to acquire runtime pm
+	 * refs to ensure that the GPU is powered on when connector polling is
+	 * performed. Since we're calling this from a runtime PM callback,
+	 * trying to acquire rpm refs will cause us to deadlock.
+	 *
+	 * Since we're guaranteed to be holding the rpm lock, it's safe to
+	 * temporarily disable the rpm helpers so this doesn't deadlock us.
+	 */
+#if defined(CONFIG_PM) && defined(__linux__)
+	dev->dev->power.disable_depth++;
+#endif
+	if (!amdgpu_device_has_dc_support(adev))
+		drm_helper_hpd_irq_event(dev);
+	else
+		drm_kms_helper_hotplug_event(dev);
+#if defined(CONFIG_PM) && defined(__linux__)
+	dev->dev->power.disable_depth--;
+#endif
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_check_soft_reset - did soft reset succeed
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * The list of all the hardware IPs that make up the asic is walked and
+ * the check_soft_reset callbacks are run.  check_soft_reset determines
+ * if the asic is still hung or not.
+ * Returns true if any of the IPs are still in a hung state, false if not.
+ */
+static bool amdgpu_device_ip_check_soft_reset(struct amdgpu_device *adev)
+{
+	int i;
+	bool asic_hang = false;
+
+	if (amdgpu_sriov_vf(adev))
+		return true;
+
+	if (amdgpu_asic_need_full_reset(adev))
+		return true;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].version->funcs->check_soft_reset)
+			adev->ip_blocks[i].status.hang =
+				adev->ip_blocks[i].version->funcs->check_soft_reset(adev);
+		if (adev->ip_blocks[i].status.hang) {
+			DRM_INFO("IP block:%s is hung!\n", adev->ip_blocks[i].version->funcs->name);
+			asic_hang = true;
+		}
+	}
+	return asic_hang;
+}
+
+/**
+ * amdgpu_device_ip_pre_soft_reset - prepare for soft reset
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * The list of all the hardware IPs that make up the asic is walked and the
+ * pre_soft_reset callbacks are run if the block is hung.  pre_soft_reset
+ * handles any IP specific hardware or software state changes that are
+ * necessary for a soft reset to succeed.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_pre_soft_reset(struct amdgpu_device *adev)
+{
+	int i, r = 0;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].status.hang &&
+		    adev->ip_blocks[i].version->funcs->pre_soft_reset) {
+			r = adev->ip_blocks[i].version->funcs->pre_soft_reset(adev);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_need_full_reset - check if a full asic reset is needed
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Some hardware IPs cannot be soft reset.  If they are hung, a full gpu
+ * reset is necessary to recover.
+ * Returns true if a full asic reset is required, false if not.
+ */
+static bool amdgpu_device_ip_need_full_reset(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (amdgpu_asic_need_full_reset(adev))
+		return true;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if ((adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC) ||
+		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) ||
+		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_ACP) ||
+		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_DCE) ||
+		     adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP) {
+			if (adev->ip_blocks[i].status.hang) {
+				DRM_INFO("Some block need full reset!\n");
+				return true;
+			}
+		}
+	}
+	return false;
+}
+
+/**
+ * amdgpu_device_ip_soft_reset - do a soft reset
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * The list of all the hardware IPs that make up the asic is walked and the
+ * soft_reset callbacks are run if the block is hung.  soft_reset handles any
+ * IP specific hardware or software state changes that are necessary to soft
+ * reset the IP.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_soft_reset(struct amdgpu_device *adev)
+{
+	int i, r = 0;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].status.hang &&
+		    adev->ip_blocks[i].version->funcs->soft_reset) {
+			r = adev->ip_blocks[i].version->funcs->soft_reset(adev);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_ip_post_soft_reset - clean up from soft reset
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * The list of all the hardware IPs that make up the asic is walked and the
+ * post_soft_reset callbacks are run if the asic was hung.  post_soft_reset
+ * handles any IP specific hardware or software state changes that are
+ * necessary after the IP has been soft reset.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_post_soft_reset(struct amdgpu_device *adev)
+{
+	int i, r = 0;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].status.hang &&
+		    adev->ip_blocks[i].version->funcs->post_soft_reset)
+			r = adev->ip_blocks[i].version->funcs->post_soft_reset(adev);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_recover_vram_from_shadow - restore shadowed VRAM buffers
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring for the engine handling the buffer operations
+ * @bo: amdgpu_bo buffer whose shadow is being restored
+ * @fence: dma_fence associated with the operation
+ *
+ * Restores the VRAM buffer contents from the shadow in GTT.  Used to
+ * restore things like GPUVM page tables after a GPU reset where
+ * the contents of VRAM might be lost.
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_recover_vram_from_shadow(struct amdgpu_device *adev,
+						  struct amdgpu_ring *ring,
+						  struct amdgpu_bo *bo,
+						  struct dma_fence **fence)
+{
+	uint32_t domain;
+	int r;
+
+	if (!bo->shadow)
+		return 0;
+
+	r = amdgpu_bo_reserve(bo, true);
+	if (r)
+		return r;
+	domain = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type);
+	/* if bo has been evicted, then no need to recover */
+	if (domain == AMDGPU_GEM_DOMAIN_VRAM) {
+		r = amdgpu_bo_validate(bo->shadow);
+		if (r) {
+			DRM_ERROR("bo validate failed!\n");
+			goto err;
+		}
+
+		r = amdgpu_bo_restore_from_shadow(adev, ring, bo,
+						 NULL, fence, true);
+		if (r) {
+			DRM_ERROR("recover page table failed!\n");
+			goto err;
+		}
+	}
+err:
+	amdgpu_bo_unreserve(bo);
+	return r;
+}
+
+/**
+ * amdgpu_device_handle_vram_lost - Handle the loss of VRAM contents
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Restores the contents of VRAM buffers from the shadows in GTT.  Used to
+ * restore things like GPUVM page tables after a GPU reset where
+ * the contents of VRAM might be lost.
+ * Returns 0 on success, 1 on failure.
+ */
+static int amdgpu_device_handle_vram_lost(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+	struct amdgpu_bo *bo, *tmp;
+	struct dma_fence *fence = NULL, *next = NULL;
+	long r = 1;
+	int i = 0;
+	long tmo;
+
+	if (amdgpu_sriov_runtime(adev))
+		tmo = msecs_to_jiffies(8000);
+	else
+		tmo = msecs_to_jiffies(100);
+
+	DRM_INFO("recover vram bo from shadow start\n");
+	mutex_lock(&adev->shadow_list_lock);
+	list_for_each_entry_safe(bo, tmp, &adev->shadow_list, shadow_list) {
+		next = NULL;
+		amdgpu_device_recover_vram_from_shadow(adev, ring, bo, &next);
+		if (fence) {
+			r = dma_fence_wait_timeout(fence, false, tmo);
+			if (r == 0)
+				pr_err("wait fence %p[%d] timeout\n", fence, i);
+			else if (r < 0)
+				pr_err("wait fence %p[%d] interrupted\n", fence, i);
+			if (r < 1) {
+				dma_fence_put(fence);
+				fence = next;
+				break;
+			}
+			i++;
+		}
+
+		dma_fence_put(fence);
+		fence = next;
+	}
+	mutex_unlock(&adev->shadow_list_lock);
+
+	if (fence) {
+		r = dma_fence_wait_timeout(fence, false, tmo);
+		if (r == 0)
+			pr_err("wait fence %p[%d] timeout\n", fence, i);
+		else if (r < 0)
+			pr_err("wait fence %p[%d] interrupted\n", fence, i);
+
+	}
+	dma_fence_put(fence);
+
+	if (r > 0)
+		DRM_INFO("recover vram bo from shadow done\n");
+	else
+		DRM_ERROR("recover vram bo from shadow failed\n");
+
+	return (r > 0) ? 0 : 1;
+}
+
+/**
+ * amdgpu_device_reset - reset ASIC/GPU for bare-metal or passthrough
+ *
+ * @adev: amdgpu device pointer
+ *
+ * attempt to do soft-reset or full-reset and reinitialize Asic
+ * return 0 means succeeded otherwise failed
+ */
+static int amdgpu_device_reset(struct amdgpu_device *adev)
+{
+	bool need_full_reset, vram_lost = 0;
+	int r;
+
+	need_full_reset = amdgpu_device_ip_need_full_reset(adev);
+
+	if (!need_full_reset) {
+		amdgpu_device_ip_pre_soft_reset(adev);
+		r = amdgpu_device_ip_soft_reset(adev);
+		amdgpu_device_ip_post_soft_reset(adev);
+		if (r || amdgpu_device_ip_check_soft_reset(adev)) {
+			DRM_INFO("soft reset failed, will fallback to full reset!\n");
+			need_full_reset = true;
+		}
+	}
+
+	if (need_full_reset) {
+		r = amdgpu_device_ip_suspend(adev);
+
+retry:
+		r = amdgpu_asic_reset(adev);
+		/* post card */
+		amdgpu_atom_asic_init(adev->mode_info.atom_context);
+
+		if (!r) {
+			dev_info(adev->dev, "GPU reset succeeded, trying to resume\n");
+			r = amdgpu_device_ip_resume_phase1(adev);
+			if (r)
+				goto out;
+
+			vram_lost = amdgpu_device_check_vram_lost(adev);
+			if (vram_lost) {
+				DRM_ERROR("VRAM is lost!\n");
+				atomic_inc(&adev->vram_lost_counter);
+			}
+
+			r = amdgpu_gtt_mgr_recover(
+				&adev->mman.bdev.man[TTM_PL_TT]);
+			if (r)
+				goto out;
+
+			r = amdgpu_device_ip_resume_phase2(adev);
+			if (r)
+				goto out;
+
+			if (vram_lost)
+				amdgpu_device_fill_reset_magic(adev);
+		}
+	}
+
+out:
+	if (!r) {
+		amdgpu_irq_gpu_reset_resume_helper(adev);
+		r = amdgpu_ib_ring_tests(adev);
+		if (r) {
+			dev_err(adev->dev, "ib ring test failed (%d).\n", r);
+			r = amdgpu_device_ip_suspend(adev);
+			need_full_reset = true;
+			goto retry;
+		}
+	}
+
+	if (!r && ((need_full_reset && !(adev->flags & AMD_IS_APU)) || vram_lost))
+		r = amdgpu_device_handle_vram_lost(adev);
+
+	return r;
+}
+
+/**
+ * amdgpu_device_reset_sriov - reset ASIC for SR-IOV vf
+ *
+ * @adev: amdgpu device pointer
+ * @from_hypervisor: request from hypervisor
+ *
+ * do VF FLR and reinitialize Asic
+ * return 0 means succeeded otherwise failed
+ */
+static int amdgpu_device_reset_sriov(struct amdgpu_device *adev,
+				     bool from_hypervisor)
+{
+	int r;
+
+	if (from_hypervisor)
+		r = amdgpu_virt_request_full_gpu(adev, true);
+	else
+		r = amdgpu_virt_reset_gpu(adev);
+	if (r)
+		return r;
+
+	/* Resume IP prior to SMC */
+	r = amdgpu_device_ip_reinit_early_sriov(adev);
+	if (r)
+		goto error;
+
+	/* we need recover gart prior to run SMC/CP/SDMA resume */
+	amdgpu_gtt_mgr_recover(&adev->mman.bdev.man[TTM_PL_TT]);
+
+	/* now we are okay to resume SMC/CP/SDMA */
+	r = amdgpu_device_ip_reinit_late_sriov(adev);
+	if (r)
+		goto error;
+
+	amdgpu_irq_gpu_reset_resume_helper(adev);
+	r = amdgpu_ib_ring_tests(adev);
+
+error:
+	amdgpu_virt_init_data_exchange(adev);
+	amdgpu_virt_release_full_gpu(adev, true);
+	if (!r && adev->virt.gim_feature & AMDGIM_FEATURE_GIM_FLR_VRAMLOST) {
+		atomic_inc(&adev->vram_lost_counter);
+		r = amdgpu_device_handle_vram_lost(adev);
+	}
+
+	return r;
+}
+
+/**
+ * amdgpu_device_gpu_recover - reset the asic and recover scheduler
+ *
+ * @adev: amdgpu device pointer
+ * @job: which job trigger hang
+ * @force: forces reset regardless of amdgpu_gpu_recovery
+ *
+ * Attempt to reset the GPU if it has hung (all asics).
+ * Returns 0 for success or an error on failure.
+ */
+int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
+			      struct amdgpu_job *job, bool force)
+{
+	int i, r, resched;
+
+	if (!force && !amdgpu_device_ip_check_soft_reset(adev)) {
+		DRM_INFO("No hardware hang detected. Did some blocks stall?\n");
+		return 0;
+	}
+
+	if (!force && (amdgpu_gpu_recovery == 0 ||
+			(amdgpu_gpu_recovery == -1  && !amdgpu_sriov_vf(adev)))) {
+		DRM_INFO("GPU recovery disabled.\n");
+		return 0;
+	}
+
+	dev_info(adev->dev, "GPU reset begin!\n");
+
+	mutex_lock(&adev->lock_reset);
+	atomic_inc(&adev->gpu_reset_counter);
+	adev->in_gpu_reset = 1;
+
+	/* Block kfd */
+	amdgpu_amdkfd_pre_reset(adev);
+
+	/* block TTM */
+	resched = ttm_bo_lock_delayed_workqueue(&adev->mman.bdev);
+
+	/* block all schedulers and reset given job's ring */
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!ring || !ring->sched.thread)
+			continue;
+
+		kthread_park(ring->sched.thread);
+
+		if (job && job->base.sched == &ring->sched)
+			continue;
+
+		drm_sched_hw_job_reset(&ring->sched, job ? &job->base : NULL);
+
+		/* after all hw jobs are reset, hw fence is meaningless, so force_completion */
+		amdgpu_fence_driver_force_completion(ring);
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		r = amdgpu_device_reset_sriov(adev, job ? false : true);
+	else
+		r = amdgpu_device_reset(adev);
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!ring || !ring->sched.thread)
+			continue;
+
+		/* only need recovery sched of the given job's ring
+		 * or all rings (in the case @job is NULL)
+		 * after above amdgpu_reset accomplished
+		 */
+		if ((!job || job->base.sched == &ring->sched) && !r)
+			drm_sched_job_recovery(&ring->sched);
+
+		kthread_unpark(ring->sched.thread);
+	}
+
+	if (!amdgpu_device_has_dc_support(adev)) {
+		drm_helper_resume_force_mode(adev->ddev);
+	}
+
+	ttm_bo_unlock_delayed_workqueue(&adev->mman.bdev, resched);
+
+	if (r) {
+		/* bad news, how to tell it to userspace ? */
+		dev_info(adev->dev, "GPU reset(%d) failed\n", atomic_read(&adev->gpu_reset_counter));
+		amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_GPU_RESET_FAIL, 0, r);
+	} else {
+		dev_info(adev->dev, "GPU reset(%d) succeeded!\n",atomic_read(&adev->gpu_reset_counter));
+	}
+
+	/*unlock kfd */
+	amdgpu_amdkfd_post_reset(adev);
+	amdgpu_vf_error_trans_all(adev);
+	adev->in_gpu_reset = 0;
+	mutex_unlock(&adev->lock_reset);
+	return r;
+}
+
+/**
+ * amdgpu_device_get_pcie_info - fence pcie info about the PCIE slot
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Fetchs and stores in the driver the PCIE capabilities (gen speed
+ * and lanes) of the slot the device is in. Handles APUs and
+ * virtualized environments where PCIE config space may not be available.
+ */
+static void amdgpu_device_get_pcie_info(struct amdgpu_device *adev)
+{
+	struct pci_dev *pdev;
+	enum pci_bus_speed speed_cap;
+	enum pcie_link_width link_width;
+
+	if (amdgpu_pcie_gen_cap)
+		adev->pm.pcie_gen_mask = amdgpu_pcie_gen_cap;
+
+	if (amdgpu_pcie_lane_cap)
+		adev->pm.pcie_mlw_mask = amdgpu_pcie_lane_cap;
+
+	/* covers APUs as well */
+	if (pci_is_root_bus(adev->pdev->bus)) {
+		if (adev->pm.pcie_gen_mask == 0)
+			adev->pm.pcie_gen_mask = AMDGPU_DEFAULT_PCIE_GEN_MASK;
+		if (adev->pm.pcie_mlw_mask == 0)
+			adev->pm.pcie_mlw_mask = AMDGPU_DEFAULT_PCIE_MLW_MASK;
+		return;
+	}
+
+	if (adev->pm.pcie_gen_mask == 0) {
+		/* asic caps */
+		pdev = adev->pdev;
+		speed_cap = pcie_get_speed_cap(pdev);
+		if (speed_cap == PCI_SPEED_UNKNOWN) {
+			adev->pm.pcie_gen_mask |= (CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+						  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+						  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN3);
+		} else {
+			if (speed_cap == PCIE_SPEED_16_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN3 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN4);
+			else if (speed_cap == PCIE_SPEED_8_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN3);
+			else if (speed_cap == PCIE_SPEED_5_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							  CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN2);
+			else
+				adev->pm.pcie_gen_mask |= CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1;
+		}
+		/* platform caps */
+		pdev = adev->ddev->pdev->bus->self;
+		speed_cap = pcie_get_speed_cap(pdev);
+		if (speed_cap == PCI_SPEED_UNKNOWN) {
+			adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+						   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2);
+		} else {
+			if (speed_cap == PCIE_SPEED_16_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN4);
+			else if (speed_cap == PCIE_SPEED_8_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3);
+			else if (speed_cap == PCIE_SPEED_5_0GT)
+				adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
+							   CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2);
+			else
+				adev->pm.pcie_gen_mask |= CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1;
+
+		}
+	}
+	if (adev->pm.pcie_mlw_mask == 0) {
+		pdev = adev->ddev->pdev->bus->self;
+		link_width = pcie_get_width_cap(pdev);
+		if (link_width == PCIE_LNK_WIDTH_UNKNOWN) {
+			adev->pm.pcie_mlw_mask |= AMDGPU_DEFAULT_PCIE_MLW_MASK;
+		} else {
+			switch (link_width) {
+			case PCIE_LNK_X32:
+				adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X32 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X16 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+				break;
+			case PCIE_LNK_X16:
+				adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X16 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+				break;
+			case PCIE_LNK_X12:
+				adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+				break;
+			case PCIE_LNK_X8:
+				adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+				break;
+			case PCIE_LNK_X4:
+				adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+				break;
+			case PCIE_LNK_X2:
+				adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 |
+							  CAIL_PCIE_LINK_WIDTH_SUPPORT_X1);
+				break;
+			case PCIE_LNK_X1:
+				adev->pm.pcie_mlw_mask = CAIL_PCIE_LINK_WIDTH_SUPPORT_X1;
+				break;
+			default:
+				break;
+			}
+		}
+	}
+}
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_display.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_display.c
new file mode 100644
index 00000000000..686a26de50f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_display.c
@@ -0,0 +1,890 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_i2c.h"
+#include "atom.h"
+#include "amdgpu_connectors.h"
+#include "amdgpu_display.h"
+#include <asm/div64.h>
+
+#include <linux/pm_runtime.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/drm_edid.h>
+#include <drm/drm_gem_framebuffer_helper.h>
+#include <drm/drm_fb_helper.h>
+
+static void amdgpu_display_flip_callback(struct dma_fence *f,
+					 struct dma_fence_cb *cb)
+{
+	struct amdgpu_flip_work *work =
+		container_of(cb, struct amdgpu_flip_work, cb);
+
+	dma_fence_put(f);
+	schedule_work(&work->flip_work.work);
+}
+
+static bool amdgpu_display_flip_handle_fence(struct amdgpu_flip_work *work,
+					     struct dma_fence **f)
+{
+	struct dma_fence *fence= *f;
+
+	if (fence == NULL)
+		return false;
+
+	*f = NULL;
+
+	if (!dma_fence_add_callback(fence, &work->cb,
+				    amdgpu_display_flip_callback))
+		return true;
+
+	dma_fence_put(fence);
+	return false;
+}
+
+static void amdgpu_display_flip_work_func(struct work_struct *__work)
+{
+	struct delayed_work *delayed_work =
+		container_of(__work, struct delayed_work, work);
+	struct amdgpu_flip_work *work =
+		container_of(delayed_work, struct amdgpu_flip_work, flip_work);
+	struct amdgpu_device *adev = work->adev;
+	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[work->crtc_id];
+
+	struct drm_crtc *crtc = &amdgpu_crtc->base;
+	unsigned long flags;
+	unsigned i;
+	int vpos, hpos;
+
+	if (amdgpu_display_flip_handle_fence(work, &work->excl))
+		return;
+
+	for (i = 0; i < work->shared_count; ++i)
+		if (amdgpu_display_flip_handle_fence(work, &work->shared[i]))
+			return;
+
+	/* Wait until we're out of the vertical blank period before the one
+	 * targeted by the flip
+	 */
+	if (amdgpu_crtc->enabled &&
+	    (amdgpu_display_get_crtc_scanoutpos(adev->ddev, work->crtc_id, 0,
+						&vpos, &hpos, NULL, NULL,
+						&crtc->hwmode)
+	     & (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_IN_VBLANK)) ==
+	    (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_IN_VBLANK) &&
+	    (int)(work->target_vblank -
+		  amdgpu_get_vblank_counter_kms(adev->ddev, amdgpu_crtc->crtc_id)) > 0) {
+		schedule_delayed_work(&work->flip_work, usecs_to_jiffies(1000));
+		return;
+	}
+
+	/* We borrow the event spin lock for protecting flip_status */
+	spin_lock_irqsave(&crtc->dev->event_lock, flags);
+
+	/* Do the flip (mmio) */
+	adev->mode_info.funcs->page_flip(adev, work->crtc_id, work->base, work->async);
+
+	/* Set the flip status */
+	amdgpu_crtc->pflip_status = AMDGPU_FLIP_SUBMITTED;
+	spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+
+
+	DRM_DEBUG_DRIVER("crtc:%d[%p], pflip_stat:AMDGPU_FLIP_SUBMITTED, work: %p,\n",
+					 amdgpu_crtc->crtc_id, amdgpu_crtc, work);
+
+}
+
+/*
+ * Handle unpin events outside the interrupt handler proper.
+ */
+static void amdgpu_display_unpin_work_func(struct work_struct *__work)
+{
+	struct amdgpu_flip_work *work =
+		container_of(__work, struct amdgpu_flip_work, unpin_work);
+	int r;
+
+	/* unpin of the old buffer */
+	r = amdgpu_bo_reserve(work->old_abo, true);
+	if (likely(r == 0)) {
+		r = amdgpu_bo_unpin(work->old_abo);
+		if (unlikely(r != 0)) {
+			DRM_ERROR("failed to unpin buffer after flip\n");
+		}
+		amdgpu_bo_unreserve(work->old_abo);
+	} else
+		DRM_ERROR("failed to reserve buffer after flip\n");
+
+	amdgpu_bo_unref(&work->old_abo);
+	kfree(work->shared);
+	kfree(work);
+}
+
+int amdgpu_display_crtc_page_flip_target(struct drm_crtc *crtc,
+				struct drm_framebuffer *fb,
+				struct drm_pending_vblank_event *event,
+				uint32_t page_flip_flags, uint32_t target,
+				struct drm_modeset_acquire_ctx *ctx)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_gem_object *obj;
+	struct amdgpu_flip_work *work;
+	struct amdgpu_bo *new_abo;
+	unsigned long flags;
+	u64 tiling_flags;
+	int i, r;
+
+	work = kzalloc(sizeof *work, GFP_KERNEL);
+	if (work == NULL)
+		return -ENOMEM;
+
+	INIT_DELAYED_WORK(&work->flip_work, amdgpu_display_flip_work_func);
+	INIT_WORK(&work->unpin_work, amdgpu_display_unpin_work_func);
+
+	work->event = event;
+	work->adev = adev;
+	work->crtc_id = amdgpu_crtc->crtc_id;
+	work->async = (page_flip_flags & DRM_MODE_PAGE_FLIP_ASYNC) != 0;
+
+	/* schedule unpin of the old buffer */
+	obj = crtc->primary->fb->obj[0];
+
+	/* take a reference to the old object */
+	work->old_abo = gem_to_amdgpu_bo(obj);
+	amdgpu_bo_ref(work->old_abo);
+
+	obj = fb->obj[0];
+	new_abo = gem_to_amdgpu_bo(obj);
+
+	/* pin the new buffer */
+	r = amdgpu_bo_reserve(new_abo, false);
+	if (unlikely(r != 0)) {
+		DRM_ERROR("failed to reserve new abo buffer before flip\n");
+		goto cleanup;
+	}
+
+	r = amdgpu_bo_pin(new_abo, amdgpu_display_supported_domains(adev));
+	if (unlikely(r != 0)) {
+		DRM_ERROR("failed to pin new abo buffer before flip\n");
+		goto unreserve;
+	}
+
+	r = amdgpu_ttm_alloc_gart(&new_abo->tbo);
+	if (unlikely(r != 0)) {
+		DRM_ERROR("%p bind failed\n", new_abo);
+		goto unpin;
+	}
+
+	r = reservation_object_get_fences_rcu(new_abo->tbo.resv, &work->excl,
+					      &work->shared_count,
+					      &work->shared);
+	if (unlikely(r != 0)) {
+		DRM_ERROR("failed to get fences for buffer\n");
+		goto unpin;
+	}
+
+	amdgpu_bo_get_tiling_flags(new_abo, &tiling_flags);
+	amdgpu_bo_unreserve(new_abo);
+
+	work->base = amdgpu_bo_gpu_offset(new_abo);
+	work->target_vblank = target - (uint32_t)drm_crtc_vblank_count(crtc) +
+		amdgpu_get_vblank_counter_kms(dev, work->crtc_id);
+
+	/* we borrow the event spin lock for protecting flip_wrok */
+	spin_lock_irqsave(&crtc->dev->event_lock, flags);
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_NONE) {
+		DRM_DEBUG_DRIVER("flip queue: crtc already busy\n");
+		spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+		r = -EBUSY;
+		goto pflip_cleanup;
+	}
+
+	amdgpu_crtc->pflip_status = AMDGPU_FLIP_PENDING;
+	amdgpu_crtc->pflip_works = work;
+
+
+	DRM_DEBUG_DRIVER("crtc:%d[%p], pflip_stat:AMDGPU_FLIP_PENDING, work: %p,\n",
+					 amdgpu_crtc->crtc_id, amdgpu_crtc, work);
+	/* update crtc fb */
+	crtc->primary->fb = fb;
+	spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+	amdgpu_display_flip_work_func(&work->flip_work.work);
+	return 0;
+
+pflip_cleanup:
+	if (unlikely(amdgpu_bo_reserve(new_abo, false) != 0)) {
+		DRM_ERROR("failed to reserve new abo in error path\n");
+		goto cleanup;
+	}
+unpin:
+	if (unlikely(amdgpu_bo_unpin(new_abo) != 0)) {
+		DRM_ERROR("failed to unpin new abo in error path\n");
+	}
+unreserve:
+	amdgpu_bo_unreserve(new_abo);
+
+cleanup:
+	amdgpu_bo_unref(&work->old_abo);
+	dma_fence_put(work->excl);
+	for (i = 0; i < work->shared_count; ++i)
+		dma_fence_put(work->shared[i]);
+	kfree(work->shared);
+	kfree(work);
+
+	return r;
+}
+
+int amdgpu_display_crtc_set_config(struct drm_mode_set *set,
+				   struct drm_modeset_acquire_ctx *ctx)
+{
+	struct drm_device *dev;
+	struct amdgpu_device *adev;
+	struct drm_crtc *crtc;
+	bool active = false;
+	int ret;
+
+	if (!set || !set->crtc)
+		return -EINVAL;
+
+	dev = set->crtc->dev;
+
+	ret = pm_runtime_get_sync(dev->dev);
+	if (ret < 0)
+		return ret;
+
+	ret = drm_crtc_helper_set_config(set, ctx);
+
+	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head)
+		if (crtc->enabled)
+			active = true;
+
+	pm_runtime_mark_last_busy(dev->dev);
+
+	adev = dev->dev_private;
+	/* if we have active crtcs and we don't have a power ref,
+	   take the current one */
+	if (active && !adev->have_disp_power_ref) {
+		adev->have_disp_power_ref = true;
+		return ret;
+	}
+	/* if we have no active crtcs, then drop the power ref
+	   we got before */
+	if (!active && adev->have_disp_power_ref) {
+		pm_runtime_put_autosuspend(dev->dev);
+		adev->have_disp_power_ref = false;
+	}
+
+	/* drop the power reference we got coming in here */
+	pm_runtime_put_autosuspend(dev->dev);
+	return ret;
+}
+
+static const char *encoder_names[41] = {
+	"NONE",
+	"INTERNAL_LVDS",
+	"INTERNAL_TMDS1",
+	"INTERNAL_TMDS2",
+	"INTERNAL_DAC1",
+	"INTERNAL_DAC2",
+	"INTERNAL_SDVOA",
+	"INTERNAL_SDVOB",
+	"SI170B",
+	"CH7303",
+	"CH7301",
+	"INTERNAL_DVO1",
+	"EXTERNAL_SDVOA",
+	"EXTERNAL_SDVOB",
+	"TITFP513",
+	"INTERNAL_LVTM1",
+	"VT1623",
+	"HDMI_SI1930",
+	"HDMI_INTERNAL",
+	"INTERNAL_KLDSCP_TMDS1",
+	"INTERNAL_KLDSCP_DVO1",
+	"INTERNAL_KLDSCP_DAC1",
+	"INTERNAL_KLDSCP_DAC2",
+	"SI178",
+	"MVPU_FPGA",
+	"INTERNAL_DDI",
+	"VT1625",
+	"HDMI_SI1932",
+	"DP_AN9801",
+	"DP_DP501",
+	"INTERNAL_UNIPHY",
+	"INTERNAL_KLDSCP_LVTMA",
+	"INTERNAL_UNIPHY1",
+	"INTERNAL_UNIPHY2",
+	"NUTMEG",
+	"TRAVIS",
+	"INTERNAL_VCE",
+	"INTERNAL_UNIPHY3",
+	"HDMI_ANX9805",
+	"INTERNAL_AMCLK",
+	"VIRTUAL",
+};
+
+static const char *hpd_names[6] = {
+	"HPD1",
+	"HPD2",
+	"HPD3",
+	"HPD4",
+	"HPD5",
+	"HPD6",
+};
+
+void amdgpu_display_print_display_setup(struct drm_device *dev)
+{
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+	uint32_t devices;
+	int i = 0;
+
+	DRM_INFO("AMDGPU Display Connectors\n");
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		amdgpu_connector = to_amdgpu_connector(connector);
+		DRM_INFO("Connector %d:\n", i);
+		DRM_INFO("  %s\n", connector->name);
+		if (amdgpu_connector->hpd.hpd != AMDGPU_HPD_NONE)
+			DRM_INFO("  %s\n", hpd_names[amdgpu_connector->hpd.hpd]);
+		if (amdgpu_connector->ddc_bus) {
+			DRM_INFO("  DDC: 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x\n",
+				 amdgpu_connector->ddc_bus->rec.mask_clk_reg,
+				 amdgpu_connector->ddc_bus->rec.mask_data_reg,
+				 amdgpu_connector->ddc_bus->rec.a_clk_reg,
+				 amdgpu_connector->ddc_bus->rec.a_data_reg,
+				 amdgpu_connector->ddc_bus->rec.en_clk_reg,
+				 amdgpu_connector->ddc_bus->rec.en_data_reg,
+				 amdgpu_connector->ddc_bus->rec.y_clk_reg,
+				 amdgpu_connector->ddc_bus->rec.y_data_reg);
+			if (amdgpu_connector->router.ddc_valid)
+				DRM_INFO("  DDC Router 0x%x/0x%x\n",
+					 amdgpu_connector->router.ddc_mux_control_pin,
+					 amdgpu_connector->router.ddc_mux_state);
+			if (amdgpu_connector->router.cd_valid)
+				DRM_INFO("  Clock/Data Router 0x%x/0x%x\n",
+					 amdgpu_connector->router.cd_mux_control_pin,
+					 amdgpu_connector->router.cd_mux_state);
+		} else {
+			if (connector->connector_type == DRM_MODE_CONNECTOR_VGA ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_DVII ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_DVID ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_DVIA ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_HDMIA ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_HDMIB)
+				DRM_INFO("  DDC: no ddc bus - possible BIOS bug - please report to xorg-driver-ati@lists.x.org\n");
+		}
+		DRM_INFO("  Encoders:\n");
+		list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+			amdgpu_encoder = to_amdgpu_encoder(encoder);
+			devices = amdgpu_encoder->devices & amdgpu_connector->devices;
+			if (devices) {
+				if (devices & ATOM_DEVICE_CRT1_SUPPORT)
+					DRM_INFO("    CRT1: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_CRT2_SUPPORT)
+					DRM_INFO("    CRT2: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_LCD1_SUPPORT)
+					DRM_INFO("    LCD1: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_DFP1_SUPPORT)
+					DRM_INFO("    DFP1: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_DFP2_SUPPORT)
+					DRM_INFO("    DFP2: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_DFP3_SUPPORT)
+					DRM_INFO("    DFP3: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_DFP4_SUPPORT)
+					DRM_INFO("    DFP4: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_DFP5_SUPPORT)
+					DRM_INFO("    DFP5: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_DFP6_SUPPORT)
+					DRM_INFO("    DFP6: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_TV1_SUPPORT)
+					DRM_INFO("    TV1: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_CV_SUPPORT)
+					DRM_INFO("    CV: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+			}
+		}
+		i++;
+	}
+}
+
+/**
+ * amdgpu_display_ddc_probe
+ *
+ */
+bool amdgpu_display_ddc_probe(struct amdgpu_connector *amdgpu_connector,
+			      bool use_aux)
+{
+	u8 out = 0x0;
+	u8 buf[8];
+	int ret;
+	struct i2c_msg msgs[] = {
+		{
+			.addr = DDC_ADDR,
+			.flags = 0,
+			.len = 1,
+			.buf = &out,
+		},
+		{
+			.addr = DDC_ADDR,
+			.flags = I2C_M_RD,
+			.len = 8,
+			.buf = buf,
+		}
+	};
+
+	/* on hw with routers, select right port */
+	if (amdgpu_connector->router.ddc_valid)
+		amdgpu_i2c_router_select_ddc_port(amdgpu_connector);
+
+	if (use_aux) {
+		ret = i2c_transfer(&amdgpu_connector->ddc_bus->aux.ddc, msgs, 2);
+	} else {
+		ret = i2c_transfer(&amdgpu_connector->ddc_bus->adapter, msgs, 2);
+	}
+
+	if (ret != 2)
+		/* Couldn't find an accessible DDC on this connector */
+		return false;
+	/* Probe also for valid EDID header
+	 * EDID header starts with:
+	 * 0x00,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0x00.
+	 * Only the first 6 bytes must be valid as
+	 * drm_edid_block_valid() can fix the last 2 bytes */
+	if (drm_edid_header_is_valid(buf) < 6) {
+		/* Couldn't find an accessible EDID on this
+		 * connector */
+		return false;
+	}
+	return true;
+}
+
+static const struct drm_framebuffer_funcs amdgpu_fb_funcs = {
+	.destroy = drm_gem_fb_destroy,
+	.create_handle = drm_gem_fb_create_handle,
+};
+
+uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev)
+{
+	uint32_t domain = AMDGPU_GEM_DOMAIN_VRAM;
+
+#if defined(CONFIG_DRM_AMD_DC)
+	if (adev->asic_type >= CHIP_CARRIZO && adev->asic_type < CHIP_RAVEN &&
+	    adev->flags & AMD_IS_APU &&
+	    amdgpu_device_asic_has_dc_support(adev->asic_type))
+		domain |= AMDGPU_GEM_DOMAIN_GTT;
+#endif
+
+	return domain;
+}
+
+int amdgpu_display_framebuffer_init(struct drm_device *dev,
+				    struct amdgpu_framebuffer *rfb,
+				    const struct drm_mode_fb_cmd2 *mode_cmd,
+				    struct drm_gem_object *obj)
+{
+	int ret;
+	rfb->base.obj[0] = obj;
+	drm_helper_mode_fill_fb_struct(dev, &rfb->base, mode_cmd);
+	ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs);
+	if (ret) {
+		rfb->base.obj[0] = NULL;
+		return ret;
+	}
+	return 0;
+}
+
+struct drm_framebuffer *
+amdgpu_display_user_framebuffer_create(struct drm_device *dev,
+				       struct drm_file *file_priv,
+				       const struct drm_mode_fb_cmd2 *mode_cmd)
+{
+	struct drm_gem_object *obj;
+	struct amdgpu_framebuffer *amdgpu_fb;
+	int ret;
+
+	obj = drm_gem_object_lookup(file_priv, mode_cmd->handles[0]);
+	if (obj ==  NULL) {
+		dev_err(&dev->pdev->dev, "No GEM object associated to handle 0x%08X, "
+			"can't create framebuffer\n", mode_cmd->handles[0]);
+		return ERR_PTR(-ENOENT);
+	}
+
+	/* Handle is imported dma-buf, so cannot be migrated to VRAM for scanout */
+	if (obj->import_attach) {
+		DRM_DEBUG_KMS("Cannot create framebuffer from imported dma_buf\n");
+		return ERR_PTR(-EINVAL);
+	}
+
+	amdgpu_fb = kzalloc(sizeof(*amdgpu_fb), GFP_KERNEL);
+	if (amdgpu_fb == NULL) {
+		drm_gem_object_put_unlocked(obj);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	ret = amdgpu_display_framebuffer_init(dev, amdgpu_fb, mode_cmd, obj);
+	if (ret) {
+		kfree(amdgpu_fb);
+		drm_gem_object_put_unlocked(obj);
+		return ERR_PTR(ret);
+	}
+
+	return &amdgpu_fb->base;
+}
+
+const struct drm_mode_config_funcs amdgpu_mode_funcs = {
+	.fb_create = amdgpu_display_user_framebuffer_create,
+	.output_poll_changed = drm_fb_helper_output_poll_changed,
+};
+
+static const struct drm_prop_enum_list amdgpu_underscan_enum_list[] =
+{	{ UNDERSCAN_OFF, "off" },
+	{ UNDERSCAN_ON, "on" },
+	{ UNDERSCAN_AUTO, "auto" },
+};
+
+static const struct drm_prop_enum_list amdgpu_audio_enum_list[] =
+{	{ AMDGPU_AUDIO_DISABLE, "off" },
+	{ AMDGPU_AUDIO_ENABLE, "on" },
+	{ AMDGPU_AUDIO_AUTO, "auto" },
+};
+
+/* XXX support different dither options? spatial, temporal, both, etc. */
+static const struct drm_prop_enum_list amdgpu_dither_enum_list[] =
+{	{ AMDGPU_FMT_DITHER_DISABLE, "off" },
+	{ AMDGPU_FMT_DITHER_ENABLE, "on" },
+};
+
+int amdgpu_display_modeset_create_props(struct amdgpu_device *adev)
+{
+	int sz;
+
+	adev->mode_info.coherent_mode_property =
+		drm_property_create_range(adev->ddev, 0 , "coherent", 0, 1);
+	if (!adev->mode_info.coherent_mode_property)
+		return -ENOMEM;
+
+	adev->mode_info.load_detect_property =
+		drm_property_create_range(adev->ddev, 0, "load detection", 0, 1);
+	if (!adev->mode_info.load_detect_property)
+		return -ENOMEM;
+
+	drm_mode_create_scaling_mode_property(adev->ddev);
+
+	sz = ARRAY_SIZE(amdgpu_underscan_enum_list);
+	adev->mode_info.underscan_property =
+		drm_property_create_enum(adev->ddev, 0,
+				    "underscan",
+				    amdgpu_underscan_enum_list, sz);
+
+	adev->mode_info.underscan_hborder_property =
+		drm_property_create_range(adev->ddev, 0,
+					"underscan hborder", 0, 128);
+	if (!adev->mode_info.underscan_hborder_property)
+		return -ENOMEM;
+
+	adev->mode_info.underscan_vborder_property =
+		drm_property_create_range(adev->ddev, 0,
+					"underscan vborder", 0, 128);
+	if (!adev->mode_info.underscan_vborder_property)
+		return -ENOMEM;
+
+	sz = ARRAY_SIZE(amdgpu_audio_enum_list);
+	adev->mode_info.audio_property =
+		drm_property_create_enum(adev->ddev, 0,
+					 "audio",
+					 amdgpu_audio_enum_list, sz);
+
+	sz = ARRAY_SIZE(amdgpu_dither_enum_list);
+	adev->mode_info.dither_property =
+		drm_property_create_enum(adev->ddev, 0,
+					 "dither",
+					 amdgpu_dither_enum_list, sz);
+
+	if (amdgpu_device_has_dc_support(adev)) {
+		adev->mode_info.max_bpc_property =
+			drm_property_create_range(adev->ddev, 0, "max bpc", 8, 16);
+		if (!adev->mode_info.max_bpc_property)
+			return -ENOMEM;
+	}
+
+	return 0;
+}
+
+void amdgpu_display_update_priority(struct amdgpu_device *adev)
+{
+	/* adjustment options for the display watermarks */
+	if ((amdgpu_disp_priority == 0) || (amdgpu_disp_priority > 2))
+		adev->mode_info.disp_priority = 0;
+	else
+		adev->mode_info.disp_priority = amdgpu_disp_priority;
+
+}
+
+static bool amdgpu_display_is_hdtv_mode(const struct drm_display_mode *mode)
+{
+	/* try and guess if this is a tv or a monitor */
+	if ((mode->vdisplay == 480 && mode->hdisplay == 720) || /* 480p */
+	    (mode->vdisplay == 576) || /* 576p */
+	    (mode->vdisplay == 720) || /* 720p */
+	    (mode->vdisplay == 1080)) /* 1080p */
+		return true;
+	else
+		return false;
+}
+
+bool amdgpu_display_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
+					const struct drm_display_mode *mode,
+					struct drm_display_mode *adjusted_mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct drm_encoder *encoder;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_encoder *amdgpu_encoder;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+	u32 src_v = 1, dst_v = 1;
+	u32 src_h = 1, dst_h = 1;
+
+	amdgpu_crtc->h_border = 0;
+	amdgpu_crtc->v_border = 0;
+
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		if (encoder->crtc != crtc)
+			continue;
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		connector = amdgpu_get_connector_for_encoder(encoder);
+		amdgpu_connector = to_amdgpu_connector(connector);
+
+		/* set scaling */
+		if (amdgpu_encoder->rmx_type == RMX_OFF)
+			amdgpu_crtc->rmx_type = RMX_OFF;
+		else if (mode->hdisplay < amdgpu_encoder->native_mode.hdisplay ||
+			 mode->vdisplay < amdgpu_encoder->native_mode.vdisplay)
+			amdgpu_crtc->rmx_type = amdgpu_encoder->rmx_type;
+		else
+			amdgpu_crtc->rmx_type = RMX_OFF;
+		/* copy native mode */
+		memcpy(&amdgpu_crtc->native_mode,
+		       &amdgpu_encoder->native_mode,
+		       sizeof(struct drm_display_mode));
+		src_v = crtc->mode.vdisplay;
+		dst_v = amdgpu_crtc->native_mode.vdisplay;
+		src_h = crtc->mode.hdisplay;
+		dst_h = amdgpu_crtc->native_mode.hdisplay;
+
+		/* fix up for overscan on hdmi */
+		if ((!(mode->flags & DRM_MODE_FLAG_INTERLACE)) &&
+		    ((amdgpu_encoder->underscan_type == UNDERSCAN_ON) ||
+		     ((amdgpu_encoder->underscan_type == UNDERSCAN_AUTO) &&
+		      drm_detect_hdmi_monitor(amdgpu_connector_edid(connector)) &&
+		      amdgpu_display_is_hdtv_mode(mode)))) {
+			if (amdgpu_encoder->underscan_hborder != 0)
+				amdgpu_crtc->h_border = amdgpu_encoder->underscan_hborder;
+			else
+				amdgpu_crtc->h_border = (mode->hdisplay >> 5) + 16;
+			if (amdgpu_encoder->underscan_vborder != 0)
+				amdgpu_crtc->v_border = amdgpu_encoder->underscan_vborder;
+			else
+				amdgpu_crtc->v_border = (mode->vdisplay >> 5) + 16;
+			amdgpu_crtc->rmx_type = RMX_FULL;
+			src_v = crtc->mode.vdisplay;
+			dst_v = crtc->mode.vdisplay - (amdgpu_crtc->v_border * 2);
+			src_h = crtc->mode.hdisplay;
+			dst_h = crtc->mode.hdisplay - (amdgpu_crtc->h_border * 2);
+		}
+	}
+	if (amdgpu_crtc->rmx_type != RMX_OFF) {
+		fixed20_12 a, b;
+		a.full = dfixed_const(src_v);
+		b.full = dfixed_const(dst_v);
+		amdgpu_crtc->vsc.full = dfixed_div(a, b);
+		a.full = dfixed_const(src_h);
+		b.full = dfixed_const(dst_h);
+		amdgpu_crtc->hsc.full = dfixed_div(a, b);
+	} else {
+		amdgpu_crtc->vsc.full = dfixed_const(1);
+		amdgpu_crtc->hsc.full = dfixed_const(1);
+	}
+	return true;
+}
+
+/*
+ * Retrieve current video scanout position of crtc on a given gpu, and
+ * an optional accurate timestamp of when query happened.
+ *
+ * \param dev Device to query.
+ * \param pipe Crtc to query.
+ * \param flags Flags from caller (DRM_CALLED_FROM_VBLIRQ or 0).
+ *              For driver internal use only also supports these flags:
+ *
+ *              USE_REAL_VBLANKSTART to use the real start of vblank instead
+ *              of a fudged earlier start of vblank.
+ *
+ *              GET_DISTANCE_TO_VBLANKSTART to return distance to the
+ *              fudged earlier start of vblank in *vpos and the distance
+ *              to true start of vblank in *hpos.
+ *
+ * \param *vpos Location where vertical scanout position should be stored.
+ * \param *hpos Location where horizontal scanout position should go.
+ * \param *stime Target location for timestamp taken immediately before
+ *               scanout position query. Can be NULL to skip timestamp.
+ * \param *etime Target location for timestamp taken immediately after
+ *               scanout position query. Can be NULL to skip timestamp.
+ *
+ * Returns vpos as a positive number while in active scanout area.
+ * Returns vpos as a negative number inside vblank, counting the number
+ * of scanlines to go until end of vblank, e.g., -1 means "one scanline
+ * until start of active scanout / end of vblank."
+ *
+ * \return Flags, or'ed together as follows:
+ *
+ * DRM_SCANOUTPOS_VALID = Query successful.
+ * DRM_SCANOUTPOS_INVBL = Inside vblank.
+ * DRM_SCANOUTPOS_ACCURATE = Returned position is accurate. A lack of
+ * this flag means that returned position may be offset by a constant but
+ * unknown small number of scanlines wrt. real scanout position.
+ *
+ */
+int amdgpu_display_get_crtc_scanoutpos(struct drm_device *dev,
+			unsigned int pipe, unsigned int flags, int *vpos,
+			int *hpos, ktime_t *stime, ktime_t *etime,
+			const struct drm_display_mode *mode)
+{
+	u32 vbl = 0, position = 0;
+	int vbl_start, vbl_end, vtotal, ret = 0;
+	bool in_vbl = true;
+
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* preempt_disable_rt() should go right here in PREEMPT_RT patchset. */
+
+	/* Get optional system timestamp before query. */
+	if (stime)
+		*stime = ktime_get();
+
+	if (amdgpu_display_page_flip_get_scanoutpos(adev, pipe, &vbl, &position) == 0)
+		ret |= DRM_SCANOUTPOS_VALID;
+
+	/* Get optional system timestamp after query. */
+	if (etime)
+		*etime = ktime_get();
+
+	/* preempt_enable_rt() should go right here in PREEMPT_RT patchset. */
+
+	/* Decode into vertical and horizontal scanout position. */
+	*vpos = position & 0x1fff;
+	*hpos = (position >> 16) & 0x1fff;
+
+	/* Valid vblank area boundaries from gpu retrieved? */
+	if (vbl > 0) {
+		/* Yes: Decode. */
+		ret |= DRM_SCANOUTPOS_ACCURATE;
+		vbl_start = vbl & 0x1fff;
+		vbl_end = (vbl >> 16) & 0x1fff;
+	}
+	else {
+		/* No: Fake something reasonable which gives at least ok results. */
+		vbl_start = mode->crtc_vdisplay;
+		vbl_end = 0;
+	}
+
+	/* Called from driver internal vblank counter query code? */
+	if (flags & GET_DISTANCE_TO_VBLANKSTART) {
+	    /* Caller wants distance from real vbl_start in *hpos */
+	    *hpos = *vpos - vbl_start;
+	}
+
+	/* Fudge vblank to start a few scanlines earlier to handle the
+	 * problem that vblank irqs fire a few scanlines before start
+	 * of vblank. Some driver internal callers need the true vblank
+	 * start to be used and signal this via the USE_REAL_VBLANKSTART flag.
+	 *
+	 * The cause of the "early" vblank irq is that the irq is triggered
+	 * by the line buffer logic when the line buffer read position enters
+	 * the vblank, whereas our crtc scanout position naturally lags the
+	 * line buffer read position.
+	 */
+	if (!(flags & USE_REAL_VBLANKSTART))
+		vbl_start -= adev->mode_info.crtcs[pipe]->lb_vblank_lead_lines;
+
+	/* Test scanout position against vblank region. */
+	if ((*vpos < vbl_start) && (*vpos >= vbl_end))
+		in_vbl = false;
+
+	/* In vblank? */
+	if (in_vbl)
+	    ret |= DRM_SCANOUTPOS_IN_VBLANK;
+
+	/* Called from driver internal vblank counter query code? */
+	if (flags & GET_DISTANCE_TO_VBLANKSTART) {
+		/* Caller wants distance from fudged earlier vbl_start */
+		*vpos -= vbl_start;
+		return ret;
+	}
+
+	/* Check if inside vblank area and apply corrective offsets:
+	 * vpos will then be >=0 in video scanout area, but negative
+	 * within vblank area, counting down the number of lines until
+	 * start of scanout.
+	 */
+
+	/* Inside "upper part" of vblank area? Apply corrective offset if so: */
+	if (in_vbl && (*vpos >= vbl_start)) {
+		vtotal = mode->crtc_vtotal;
+		*vpos = *vpos - vtotal;
+	}
+
+	/* Correct for shifted end of vbl at vbl_end. */
+	*vpos = *vpos - vbl_end;
+
+	return ret;
+}
+
+int amdgpu_display_crtc_idx_to_irq_type(struct amdgpu_device *adev, int crtc)
+{
+	if (crtc < 0 || crtc >= adev->mode_info.num_crtc)
+		return AMDGPU_CRTC_IRQ_NONE;
+
+	switch (crtc) {
+	case 0:
+		return AMDGPU_CRTC_IRQ_VBLANK1;
+	case 1:
+		return AMDGPU_CRTC_IRQ_VBLANK2;
+	case 2:
+		return AMDGPU_CRTC_IRQ_VBLANK3;
+	case 3:
+		return AMDGPU_CRTC_IRQ_VBLANK4;
+	case 4:
+		return AMDGPU_CRTC_IRQ_VBLANK5;
+	case 5:
+		return AMDGPU_CRTC_IRQ_VBLANK6;
+	default:
+		return AMDGPU_CRTC_IRQ_NONE;
+	}
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_display.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_display.h
new file mode 100644
index 00000000000..f66e3e3fef0
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_display.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_DISPLAY_H__
+#define __AMDGPU_DISPLAY_H__
+
+uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev);
+struct drm_framebuffer *
+amdgpu_display_user_framebuffer_create(struct drm_device *dev,
+				       struct drm_file *file_priv,
+				       const struct drm_mode_fb_cmd2 *mode_cmd);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_dpm.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_dpm.c
new file mode 100644
index 00000000000..4bf6f212447
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_dpm.c
@@ -0,0 +1,996 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_i2c.h"
+#include "amdgpu_dpm.h"
+#include "atom.h"
+#include "amd_pcie.h"
+
+void amdgpu_dpm_print_class_info(u32 class, u32 class2)
+{
+	const char *s;
+
+	switch (class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) {
+	case ATOM_PPLIB_CLASSIFICATION_UI_NONE:
+	default:
+		s = "none";
+		break;
+	case ATOM_PPLIB_CLASSIFICATION_UI_BATTERY:
+		s = "battery";
+		break;
+	case ATOM_PPLIB_CLASSIFICATION_UI_BALANCED:
+		s = "balanced";
+		break;
+	case ATOM_PPLIB_CLASSIFICATION_UI_PERFORMANCE:
+		s = "performance";
+		break;
+	}
+	printk("\tui class: %s\n", s);
+	printk("\tinternal class:");
+	if (((class & ~ATOM_PPLIB_CLASSIFICATION_UI_MASK) == 0) &&
+	    (class2 == 0))
+		pr_cont(" none");
+	else {
+		if (class & ATOM_PPLIB_CLASSIFICATION_BOOT)
+			pr_cont(" boot");
+		if (class & ATOM_PPLIB_CLASSIFICATION_THERMAL)
+			pr_cont(" thermal");
+		if (class & ATOM_PPLIB_CLASSIFICATION_LIMITEDPOWERSOURCE)
+			pr_cont(" limited_pwr");
+		if (class & ATOM_PPLIB_CLASSIFICATION_REST)
+			pr_cont(" rest");
+		if (class & ATOM_PPLIB_CLASSIFICATION_FORCED)
+			pr_cont(" forced");
+		if (class & ATOM_PPLIB_CLASSIFICATION_3DPERFORMANCE)
+			pr_cont(" 3d_perf");
+		if (class & ATOM_PPLIB_CLASSIFICATION_OVERDRIVETEMPLATE)
+			pr_cont(" ovrdrv");
+		if (class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
+			pr_cont(" uvd");
+		if (class & ATOM_PPLIB_CLASSIFICATION_3DLOW)
+			pr_cont(" 3d_low");
+		if (class & ATOM_PPLIB_CLASSIFICATION_ACPI)
+			pr_cont(" acpi");
+		if (class & ATOM_PPLIB_CLASSIFICATION_HD2STATE)
+			pr_cont(" uvd_hd2");
+		if (class & ATOM_PPLIB_CLASSIFICATION_HDSTATE)
+			pr_cont(" uvd_hd");
+		if (class & ATOM_PPLIB_CLASSIFICATION_SDSTATE)
+			pr_cont(" uvd_sd");
+		if (class2 & ATOM_PPLIB_CLASSIFICATION2_LIMITEDPOWERSOURCE_2)
+			pr_cont(" limited_pwr2");
+		if (class2 & ATOM_PPLIB_CLASSIFICATION2_ULV)
+			pr_cont(" ulv");
+		if (class2 & ATOM_PPLIB_CLASSIFICATION2_MVC)
+			pr_cont(" uvd_mvc");
+	}
+	pr_cont("\n");
+}
+
+void amdgpu_dpm_print_cap_info(u32 caps)
+{
+	printk("\tcaps:");
+	if (caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY)
+		pr_cont(" single_disp");
+	if (caps & ATOM_PPLIB_SUPPORTS_VIDEO_PLAYBACK)
+		pr_cont(" video");
+	if (caps & ATOM_PPLIB_DISALLOW_ON_DC)
+		pr_cont(" no_dc");
+	pr_cont("\n");
+}
+
+void amdgpu_dpm_print_ps_status(struct amdgpu_device *adev,
+				struct amdgpu_ps *rps)
+{
+	printk("\tstatus:");
+	if (rps == adev->pm.dpm.current_ps)
+		pr_cont(" c");
+	if (rps == adev->pm.dpm.requested_ps)
+		pr_cont(" r");
+	if (rps == adev->pm.dpm.boot_ps)
+		pr_cont(" b");
+	pr_cont("\n");
+}
+
+void amdgpu_dpm_get_active_displays(struct amdgpu_device *adev)
+{
+	struct drm_device *ddev = adev->ddev;
+	struct drm_crtc *crtc;
+	struct amdgpu_crtc *amdgpu_crtc;
+
+	adev->pm.dpm.new_active_crtcs = 0;
+	adev->pm.dpm.new_active_crtc_count = 0;
+	if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) {
+		list_for_each_entry(crtc,
+				    &ddev->mode_config.crtc_list, head) {
+			amdgpu_crtc = to_amdgpu_crtc(crtc);
+			if (amdgpu_crtc->enabled) {
+				adev->pm.dpm.new_active_crtcs |= (1 << amdgpu_crtc->crtc_id);
+				adev->pm.dpm.new_active_crtc_count++;
+			}
+		}
+	}
+}
+
+
+u32 amdgpu_dpm_get_vblank_time(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_crtc *crtc;
+	struct amdgpu_crtc *amdgpu_crtc;
+	u32 vblank_in_pixels;
+	u32 vblank_time_us = 0xffffffff; /* if the displays are off, vblank time is max */
+
+	if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) {
+		list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+			amdgpu_crtc = to_amdgpu_crtc(crtc);
+			if (crtc->enabled && amdgpu_crtc->enabled && amdgpu_crtc->hw_mode.clock) {
+				vblank_in_pixels =
+					amdgpu_crtc->hw_mode.crtc_htotal *
+					(amdgpu_crtc->hw_mode.crtc_vblank_end -
+					amdgpu_crtc->hw_mode.crtc_vdisplay +
+					(amdgpu_crtc->v_border * 2));
+
+				vblank_time_us = vblank_in_pixels * 1000 / amdgpu_crtc->hw_mode.clock;
+				break;
+			}
+		}
+	}
+
+	return vblank_time_us;
+}
+
+u32 amdgpu_dpm_get_vrefresh(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_crtc *crtc;
+	struct amdgpu_crtc *amdgpu_crtc;
+	u32 vrefresh = 0;
+
+	if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) {
+		list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+			amdgpu_crtc = to_amdgpu_crtc(crtc);
+			if (crtc->enabled && amdgpu_crtc->enabled && amdgpu_crtc->hw_mode.clock) {
+				vrefresh = drm_mode_vrefresh(&amdgpu_crtc->hw_mode);
+				break;
+			}
+		}
+	}
+
+	return vrefresh;
+}
+
+void amdgpu_calculate_u_and_p(u32 i, u32 r_c, u32 p_b,
+			      u32 *p, u32 *u)
+{
+	u32 b_c = 0;
+	u32 i_c;
+	u32 tmp;
+
+	i_c = (i * r_c) / 100;
+	tmp = i_c >> p_b;
+
+	while (tmp) {
+		b_c++;
+		tmp >>= 1;
+	}
+
+	*u = (b_c + 1) / 2;
+	*p = i_c / (1 << (2 * (*u)));
+}
+
+int amdgpu_calculate_at(u32 t, u32 h, u32 fh, u32 fl, u32 *tl, u32 *th)
+{
+	u32 k, a, ah, al;
+	u32 t1;
+
+	if ((fl == 0) || (fh == 0) || (fl > fh))
+		return -EINVAL;
+
+	k = (100 * fh) / fl;
+	t1 = (t * (k - 100));
+	a = (1000 * (100 * h + t1)) / (10000 + (t1 / 100));
+	a = (a + 5) / 10;
+	ah = ((a * t) + 5000) / 10000;
+	al = a - ah;
+
+	*th = t - ah;
+	*tl = t + al;
+
+	return 0;
+}
+
+bool amdgpu_is_uvd_state(u32 class, u32 class2)
+{
+	if (class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
+		return true;
+	if (class & ATOM_PPLIB_CLASSIFICATION_HD2STATE)
+		return true;
+	if (class & ATOM_PPLIB_CLASSIFICATION_HDSTATE)
+		return true;
+	if (class & ATOM_PPLIB_CLASSIFICATION_SDSTATE)
+		return true;
+	if (class2 & ATOM_PPLIB_CLASSIFICATION2_MVC)
+		return true;
+	return false;
+}
+
+bool amdgpu_is_internal_thermal_sensor(enum amdgpu_int_thermal_type sensor)
+{
+	switch (sensor) {
+	case THERMAL_TYPE_RV6XX:
+	case THERMAL_TYPE_RV770:
+	case THERMAL_TYPE_EVERGREEN:
+	case THERMAL_TYPE_SUMO:
+	case THERMAL_TYPE_NI:
+	case THERMAL_TYPE_SI:
+	case THERMAL_TYPE_CI:
+	case THERMAL_TYPE_KV:
+		return true;
+	case THERMAL_TYPE_ADT7473_WITH_INTERNAL:
+	case THERMAL_TYPE_EMC2103_WITH_INTERNAL:
+		return false; /* need special handling */
+	case THERMAL_TYPE_NONE:
+	case THERMAL_TYPE_EXTERNAL:
+	case THERMAL_TYPE_EXTERNAL_GPIO:
+	default:
+		return false;
+	}
+}
+
+union power_info {
+	struct _ATOM_POWERPLAY_INFO info;
+	struct _ATOM_POWERPLAY_INFO_V2 info_2;
+	struct _ATOM_POWERPLAY_INFO_V3 info_3;
+	struct _ATOM_PPLIB_POWERPLAYTABLE pplib;
+	struct _ATOM_PPLIB_POWERPLAYTABLE2 pplib2;
+	struct _ATOM_PPLIB_POWERPLAYTABLE3 pplib3;
+	struct _ATOM_PPLIB_POWERPLAYTABLE4 pplib4;
+	struct _ATOM_PPLIB_POWERPLAYTABLE5 pplib5;
+};
+
+union fan_info {
+	struct _ATOM_PPLIB_FANTABLE fan;
+	struct _ATOM_PPLIB_FANTABLE2 fan2;
+	struct _ATOM_PPLIB_FANTABLE3 fan3;
+};
+
+static int amdgpu_parse_clk_voltage_dep_table(struct amdgpu_clock_voltage_dependency_table *amdgpu_table,
+					      ATOM_PPLIB_Clock_Voltage_Dependency_Table *atom_table)
+{
+	u32 size = atom_table->ucNumEntries *
+		sizeof(struct amdgpu_clock_voltage_dependency_entry);
+	int i;
+	ATOM_PPLIB_Clock_Voltage_Dependency_Record *entry;
+
+	amdgpu_table->entries = kzalloc(size, GFP_KERNEL);
+	if (!amdgpu_table->entries)
+		return -ENOMEM;
+
+	entry = &atom_table->entries[0];
+	for (i = 0; i < atom_table->ucNumEntries; i++) {
+		amdgpu_table->entries[i].clk = le16_to_cpu(entry->usClockLow) |
+			(entry->ucClockHigh << 16);
+		amdgpu_table->entries[i].v = le16_to_cpu(entry->usVoltage);
+		entry = (ATOM_PPLIB_Clock_Voltage_Dependency_Record *)
+			((u8 *)entry + sizeof(ATOM_PPLIB_Clock_Voltage_Dependency_Record));
+	}
+	amdgpu_table->count = atom_table->ucNumEntries;
+
+	return 0;
+}
+
+int amdgpu_get_platform_caps(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	union power_info *power_info;
+	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
+	u16 data_offset;
+	u8 frev, crev;
+
+	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset))
+		return -EINVAL;
+	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
+
+	adev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps);
+	adev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime);
+	adev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime);
+
+	return 0;
+}
+
+/* sizeof(ATOM_PPLIB_EXTENDEDHEADER) */
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V2 12
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V3 14
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V4 16
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V5 18
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V6 20
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V7 22
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V8 24
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V9 26
+
+int amdgpu_parse_extended_power_table(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	union power_info *power_info;
+	union fan_info *fan_info;
+	ATOM_PPLIB_Clock_Voltage_Dependency_Table *dep_table;
+	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
+	u16 data_offset;
+	u8 frev, crev;
+	int ret, i;
+
+	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset))
+		return -EINVAL;
+	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
+
+	/* fan table */
+	if (le16_to_cpu(power_info->pplib.usTableSize) >=
+	    sizeof(struct _ATOM_PPLIB_POWERPLAYTABLE3)) {
+		if (power_info->pplib3.usFanTableOffset) {
+			fan_info = (union fan_info *)(mode_info->atom_context->bios + data_offset +
+						      le16_to_cpu(power_info->pplib3.usFanTableOffset));
+			adev->pm.dpm.fan.t_hyst = fan_info->fan.ucTHyst;
+			adev->pm.dpm.fan.t_min = le16_to_cpu(fan_info->fan.usTMin);
+			adev->pm.dpm.fan.t_med = le16_to_cpu(fan_info->fan.usTMed);
+			adev->pm.dpm.fan.t_high = le16_to_cpu(fan_info->fan.usTHigh);
+			adev->pm.dpm.fan.pwm_min = le16_to_cpu(fan_info->fan.usPWMMin);
+			adev->pm.dpm.fan.pwm_med = le16_to_cpu(fan_info->fan.usPWMMed);
+			adev->pm.dpm.fan.pwm_high = le16_to_cpu(fan_info->fan.usPWMHigh);
+			if (fan_info->fan.ucFanTableFormat >= 2)
+				adev->pm.dpm.fan.t_max = le16_to_cpu(fan_info->fan2.usTMax);
+			else
+				adev->pm.dpm.fan.t_max = 10900;
+			adev->pm.dpm.fan.cycle_delay = 100000;
+			if (fan_info->fan.ucFanTableFormat >= 3) {
+				adev->pm.dpm.fan.control_mode = fan_info->fan3.ucFanControlMode;
+				adev->pm.dpm.fan.default_max_fan_pwm =
+					le16_to_cpu(fan_info->fan3.usFanPWMMax);
+				adev->pm.dpm.fan.default_fan_output_sensitivity = 4836;
+				adev->pm.dpm.fan.fan_output_sensitivity =
+					le16_to_cpu(fan_info->fan3.usFanOutputSensitivity);
+			}
+			adev->pm.dpm.fan.ucode_fan_control = true;
+		}
+	}
+
+	/* clock dependancy tables, shedding tables */
+	if (le16_to_cpu(power_info->pplib.usTableSize) >=
+	    sizeof(struct _ATOM_PPLIB_POWERPLAYTABLE4)) {
+		if (power_info->pplib4.usVddcDependencyOnSCLKOffset) {
+			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib4.usVddcDependencyOnSCLKOffset));
+			ret = amdgpu_parse_clk_voltage_dep_table(&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk,
+								 dep_table);
+			if (ret) {
+				amdgpu_free_extended_power_table(adev);
+				return ret;
+			}
+		}
+		if (power_info->pplib4.usVddciDependencyOnMCLKOffset) {
+			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib4.usVddciDependencyOnMCLKOffset));
+			ret = amdgpu_parse_clk_voltage_dep_table(&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk,
+								 dep_table);
+			if (ret) {
+				amdgpu_free_extended_power_table(adev);
+				return ret;
+			}
+		}
+		if (power_info->pplib4.usVddcDependencyOnMCLKOffset) {
+			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib4.usVddcDependencyOnMCLKOffset));
+			ret = amdgpu_parse_clk_voltage_dep_table(&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk,
+								 dep_table);
+			if (ret) {
+				amdgpu_free_extended_power_table(adev);
+				return ret;
+			}
+		}
+		if (power_info->pplib4.usMvddDependencyOnMCLKOffset) {
+			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib4.usMvddDependencyOnMCLKOffset));
+			ret = amdgpu_parse_clk_voltage_dep_table(&adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk,
+								 dep_table);
+			if (ret) {
+				amdgpu_free_extended_power_table(adev);
+				return ret;
+			}
+		}
+		if (power_info->pplib4.usMaxClockVoltageOnDCOffset) {
+			ATOM_PPLIB_Clock_Voltage_Limit_Table *clk_v =
+				(ATOM_PPLIB_Clock_Voltage_Limit_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib4.usMaxClockVoltageOnDCOffset));
+			if (clk_v->ucNumEntries) {
+				adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.sclk =
+					le16_to_cpu(clk_v->entries[0].usSclkLow) |
+					(clk_v->entries[0].ucSclkHigh << 16);
+				adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.mclk =
+					le16_to_cpu(clk_v->entries[0].usMclkLow) |
+					(clk_v->entries[0].ucMclkHigh << 16);
+				adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.vddc =
+					le16_to_cpu(clk_v->entries[0].usVddc);
+				adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.vddci =
+					le16_to_cpu(clk_v->entries[0].usVddci);
+			}
+		}
+		if (power_info->pplib4.usVddcPhaseShedLimitsTableOffset) {
+			ATOM_PPLIB_PhaseSheddingLimits_Table *psl =
+				(ATOM_PPLIB_PhaseSheddingLimits_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib4.usVddcPhaseShedLimitsTableOffset));
+			ATOM_PPLIB_PhaseSheddingLimits_Record *entry;
+
+			adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries =
+				kcalloc(psl->ucNumEntries,
+					sizeof(struct amdgpu_phase_shedding_limits_entry),
+					GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+
+			entry = &psl->entries[0];
+			for (i = 0; i < psl->ucNumEntries; i++) {
+				adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries[i].sclk =
+					le16_to_cpu(entry->usSclkLow) | (entry->ucSclkHigh << 16);
+				adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries[i].mclk =
+					le16_to_cpu(entry->usMclkLow) | (entry->ucMclkHigh << 16);
+				adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries[i].voltage =
+					le16_to_cpu(entry->usVoltage);
+				entry = (ATOM_PPLIB_PhaseSheddingLimits_Record *)
+					((u8 *)entry + sizeof(ATOM_PPLIB_PhaseSheddingLimits_Record));
+			}
+			adev->pm.dpm.dyn_state.phase_shedding_limits_table.count =
+				psl->ucNumEntries;
+		}
+	}
+
+	/* cac data */
+	if (le16_to_cpu(power_info->pplib.usTableSize) >=
+	    sizeof(struct _ATOM_PPLIB_POWERPLAYTABLE5)) {
+		adev->pm.dpm.tdp_limit = le32_to_cpu(power_info->pplib5.ulTDPLimit);
+		adev->pm.dpm.near_tdp_limit = le32_to_cpu(power_info->pplib5.ulNearTDPLimit);
+		adev->pm.dpm.near_tdp_limit_adjusted = adev->pm.dpm.near_tdp_limit;
+		adev->pm.dpm.tdp_od_limit = le16_to_cpu(power_info->pplib5.usTDPODLimit);
+		if (adev->pm.dpm.tdp_od_limit)
+			adev->pm.dpm.power_control = true;
+		else
+			adev->pm.dpm.power_control = false;
+		adev->pm.dpm.tdp_adjustment = 0;
+		adev->pm.dpm.sq_ramping_threshold = le32_to_cpu(power_info->pplib5.ulSQRampingThreshold);
+		adev->pm.dpm.cac_leakage = le32_to_cpu(power_info->pplib5.ulCACLeakage);
+		adev->pm.dpm.load_line_slope = le16_to_cpu(power_info->pplib5.usLoadLineSlope);
+		if (power_info->pplib5.usCACLeakageTableOffset) {
+			ATOM_PPLIB_CAC_Leakage_Table *cac_table =
+				(ATOM_PPLIB_CAC_Leakage_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib5.usCACLeakageTableOffset));
+			ATOM_PPLIB_CAC_Leakage_Record *entry;
+			u32 size = cac_table->ucNumEntries * sizeof(struct amdgpu_cac_leakage_table);
+			adev->pm.dpm.dyn_state.cac_leakage_table.entries = kzalloc(size, GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.cac_leakage_table.entries) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			entry = &cac_table->entries[0];
+			for (i = 0; i < cac_table->ucNumEntries; i++) {
+				if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_EVV) {
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc1 =
+						le16_to_cpu(entry->usVddc1);
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc2 =
+						le16_to_cpu(entry->usVddc2);
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc3 =
+						le16_to_cpu(entry->usVddc3);
+				} else {
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc =
+						le16_to_cpu(entry->usVddc);
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].leakage =
+						le32_to_cpu(entry->ulLeakageValue);
+				}
+				entry = (ATOM_PPLIB_CAC_Leakage_Record *)
+					((u8 *)entry + sizeof(ATOM_PPLIB_CAC_Leakage_Record));
+			}
+			adev->pm.dpm.dyn_state.cac_leakage_table.count = cac_table->ucNumEntries;
+		}
+	}
+
+	/* ext tables */
+	if (le16_to_cpu(power_info->pplib.usTableSize) >=
+	    sizeof(struct _ATOM_PPLIB_POWERPLAYTABLE3)) {
+		ATOM_PPLIB_EXTENDEDHEADER *ext_hdr = (ATOM_PPLIB_EXTENDEDHEADER *)
+			(mode_info->atom_context->bios + data_offset +
+			 le16_to_cpu(power_info->pplib3.usExtendendedHeaderOffset));
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V2) &&
+			ext_hdr->usVCETableOffset) {
+			VCEClockInfoArray *array = (VCEClockInfoArray *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usVCETableOffset) + 1);
+			ATOM_PPLIB_VCE_Clock_Voltage_Limit_Table *limits =
+				(ATOM_PPLIB_VCE_Clock_Voltage_Limit_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usVCETableOffset) + 1 +
+				 1 + array->ucNumEntries * sizeof(VCEClockInfo));
+			ATOM_PPLIB_VCE_State_Table *states =
+				(ATOM_PPLIB_VCE_State_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usVCETableOffset) + 1 +
+				 1 + (array->ucNumEntries * sizeof (VCEClockInfo)) +
+				 1 + (limits->numEntries * sizeof(ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record)));
+			ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record *entry;
+			ATOM_PPLIB_VCE_State_Record *state_entry;
+			VCEClockInfo *vce_clk;
+			u32 size = limits->numEntries *
+				sizeof(struct amdgpu_vce_clock_voltage_dependency_entry);
+			adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries =
+				kzalloc(size, GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.count =
+				limits->numEntries;
+			entry = &limits->entries[0];
+			state_entry = &states->entries[0];
+			for (i = 0; i < limits->numEntries; i++) {
+				vce_clk = (VCEClockInfo *)
+					((u8 *)&array->entries[0] +
+					 (entry->ucVCEClockInfoIndex * sizeof(VCEClockInfo)));
+				adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].evclk =
+					le16_to_cpu(vce_clk->usEVClkLow) | (vce_clk->ucEVClkHigh << 16);
+				adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].ecclk =
+					le16_to_cpu(vce_clk->usECClkLow) | (vce_clk->ucECClkHigh << 16);
+				adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].v =
+					le16_to_cpu(entry->usVoltage);
+				entry = (ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record *)
+					((u8 *)entry + sizeof(ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record));
+			}
+			adev->pm.dpm.num_of_vce_states =
+					states->numEntries > AMD_MAX_VCE_LEVELS ?
+					AMD_MAX_VCE_LEVELS : states->numEntries;
+			for (i = 0; i < adev->pm.dpm.num_of_vce_states; i++) {
+				vce_clk = (VCEClockInfo *)
+					((u8 *)&array->entries[0] +
+					 (state_entry->ucVCEClockInfoIndex * sizeof(VCEClockInfo)));
+				adev->pm.dpm.vce_states[i].evclk =
+					le16_to_cpu(vce_clk->usEVClkLow) | (vce_clk->ucEVClkHigh << 16);
+				adev->pm.dpm.vce_states[i].ecclk =
+					le16_to_cpu(vce_clk->usECClkLow) | (vce_clk->ucECClkHigh << 16);
+				adev->pm.dpm.vce_states[i].clk_idx =
+					state_entry->ucClockInfoIndex & 0x3f;
+				adev->pm.dpm.vce_states[i].pstate =
+					(state_entry->ucClockInfoIndex & 0xc0) >> 6;
+				state_entry = (ATOM_PPLIB_VCE_State_Record *)
+					((u8 *)state_entry + sizeof(ATOM_PPLIB_VCE_State_Record));
+			}
+		}
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V3) &&
+			ext_hdr->usUVDTableOffset) {
+			UVDClockInfoArray *array = (UVDClockInfoArray *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usUVDTableOffset) + 1);
+			ATOM_PPLIB_UVD_Clock_Voltage_Limit_Table *limits =
+				(ATOM_PPLIB_UVD_Clock_Voltage_Limit_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usUVDTableOffset) + 1 +
+				 1 + (array->ucNumEntries * sizeof (UVDClockInfo)));
+			ATOM_PPLIB_UVD_Clock_Voltage_Limit_Record *entry;
+			u32 size = limits->numEntries *
+				sizeof(struct amdgpu_uvd_clock_voltage_dependency_entry);
+			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries =
+				kzalloc(size, GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count =
+				limits->numEntries;
+			entry = &limits->entries[0];
+			for (i = 0; i < limits->numEntries; i++) {
+				UVDClockInfo *uvd_clk = (UVDClockInfo *)
+					((u8 *)&array->entries[0] +
+					 (entry->ucUVDClockInfoIndex * sizeof(UVDClockInfo)));
+				adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[i].vclk =
+					le16_to_cpu(uvd_clk->usVClkLow) | (uvd_clk->ucVClkHigh << 16);
+				adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[i].dclk =
+					le16_to_cpu(uvd_clk->usDClkLow) | (uvd_clk->ucDClkHigh << 16);
+				adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[i].v =
+					le16_to_cpu(entry->usVoltage);
+				entry = (ATOM_PPLIB_UVD_Clock_Voltage_Limit_Record *)
+					((u8 *)entry + sizeof(ATOM_PPLIB_UVD_Clock_Voltage_Limit_Record));
+			}
+		}
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V4) &&
+			ext_hdr->usSAMUTableOffset) {
+			ATOM_PPLIB_SAMClk_Voltage_Limit_Table *limits =
+				(ATOM_PPLIB_SAMClk_Voltage_Limit_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usSAMUTableOffset) + 1);
+			ATOM_PPLIB_SAMClk_Voltage_Limit_Record *entry;
+			u32 size = limits->numEntries *
+				sizeof(struct amdgpu_clock_voltage_dependency_entry);
+			adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries =
+				kzalloc(size, GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.count =
+				limits->numEntries;
+			entry = &limits->entries[0];
+			for (i = 0; i < limits->numEntries; i++) {
+				adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[i].clk =
+					le16_to_cpu(entry->usSAMClockLow) | (entry->ucSAMClockHigh << 16);
+				adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[i].v =
+					le16_to_cpu(entry->usVoltage);
+				entry = (ATOM_PPLIB_SAMClk_Voltage_Limit_Record *)
+					((u8 *)entry + sizeof(ATOM_PPLIB_SAMClk_Voltage_Limit_Record));
+			}
+		}
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V5) &&
+		    ext_hdr->usPPMTableOffset) {
+			ATOM_PPLIB_PPM_Table *ppm = (ATOM_PPLIB_PPM_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usPPMTableOffset));
+			adev->pm.dpm.dyn_state.ppm_table =
+				kzalloc(sizeof(struct amdgpu_ppm_table), GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.ppm_table) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			adev->pm.dpm.dyn_state.ppm_table->ppm_design = ppm->ucPpmDesign;
+			adev->pm.dpm.dyn_state.ppm_table->cpu_core_number =
+				le16_to_cpu(ppm->usCpuCoreNumber);
+			adev->pm.dpm.dyn_state.ppm_table->platform_tdp =
+				le32_to_cpu(ppm->ulPlatformTDP);
+			adev->pm.dpm.dyn_state.ppm_table->small_ac_platform_tdp =
+				le32_to_cpu(ppm->ulSmallACPlatformTDP);
+			adev->pm.dpm.dyn_state.ppm_table->platform_tdc =
+				le32_to_cpu(ppm->ulPlatformTDC);
+			adev->pm.dpm.dyn_state.ppm_table->small_ac_platform_tdc =
+				le32_to_cpu(ppm->ulSmallACPlatformTDC);
+			adev->pm.dpm.dyn_state.ppm_table->apu_tdp =
+				le32_to_cpu(ppm->ulApuTDP);
+			adev->pm.dpm.dyn_state.ppm_table->dgpu_tdp =
+				le32_to_cpu(ppm->ulDGpuTDP);
+			adev->pm.dpm.dyn_state.ppm_table->dgpu_ulv_power =
+				le32_to_cpu(ppm->ulDGpuUlvPower);
+			adev->pm.dpm.dyn_state.ppm_table->tj_max =
+				le32_to_cpu(ppm->ulTjmax);
+		}
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V6) &&
+			ext_hdr->usACPTableOffset) {
+			ATOM_PPLIB_ACPClk_Voltage_Limit_Table *limits =
+				(ATOM_PPLIB_ACPClk_Voltage_Limit_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usACPTableOffset) + 1);
+			ATOM_PPLIB_ACPClk_Voltage_Limit_Record *entry;
+			u32 size = limits->numEntries *
+				sizeof(struct amdgpu_clock_voltage_dependency_entry);
+			adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries =
+				kzalloc(size, GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.count =
+				limits->numEntries;
+			entry = &limits->entries[0];
+			for (i = 0; i < limits->numEntries; i++) {
+				adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[i].clk =
+					le16_to_cpu(entry->usACPClockLow) | (entry->ucACPClockHigh << 16);
+				adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[i].v =
+					le16_to_cpu(entry->usVoltage);
+				entry = (ATOM_PPLIB_ACPClk_Voltage_Limit_Record *)
+					((u8 *)entry + sizeof(ATOM_PPLIB_ACPClk_Voltage_Limit_Record));
+			}
+		}
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V7) &&
+			ext_hdr->usPowerTuneTableOffset) {
+			u8 rev = *(u8 *)(mode_info->atom_context->bios + data_offset +
+					 le16_to_cpu(ext_hdr->usPowerTuneTableOffset));
+			ATOM_PowerTune_Table *pt;
+			adev->pm.dpm.dyn_state.cac_tdp_table =
+				kzalloc(sizeof(struct amdgpu_cac_tdp_table), GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.cac_tdp_table) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			if (rev > 0) {
+				ATOM_PPLIB_POWERTUNE_Table_V1 *ppt = (ATOM_PPLIB_POWERTUNE_Table_V1 *)
+					(mode_info->atom_context->bios + data_offset +
+					 le16_to_cpu(ext_hdr->usPowerTuneTableOffset));
+				adev->pm.dpm.dyn_state.cac_tdp_table->maximum_power_delivery_limit =
+					ppt->usMaximumPowerDeliveryLimit;
+				pt = &ppt->power_tune_table;
+			} else {
+				ATOM_PPLIB_POWERTUNE_Table *ppt = (ATOM_PPLIB_POWERTUNE_Table *)
+					(mode_info->atom_context->bios + data_offset +
+					 le16_to_cpu(ext_hdr->usPowerTuneTableOffset));
+				adev->pm.dpm.dyn_state.cac_tdp_table->maximum_power_delivery_limit = 255;
+				pt = &ppt->power_tune_table;
+			}
+			adev->pm.dpm.dyn_state.cac_tdp_table->tdp = le16_to_cpu(pt->usTDP);
+			adev->pm.dpm.dyn_state.cac_tdp_table->configurable_tdp =
+				le16_to_cpu(pt->usConfigurableTDP);
+			adev->pm.dpm.dyn_state.cac_tdp_table->tdc = le16_to_cpu(pt->usTDC);
+			adev->pm.dpm.dyn_state.cac_tdp_table->battery_power_limit =
+				le16_to_cpu(pt->usBatteryPowerLimit);
+			adev->pm.dpm.dyn_state.cac_tdp_table->small_power_limit =
+				le16_to_cpu(pt->usSmallPowerLimit);
+			adev->pm.dpm.dyn_state.cac_tdp_table->low_cac_leakage =
+				le16_to_cpu(pt->usLowCACLeakage);
+			adev->pm.dpm.dyn_state.cac_tdp_table->high_cac_leakage =
+				le16_to_cpu(pt->usHighCACLeakage);
+		}
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V8) &&
+				ext_hdr->usSclkVddgfxTableOffset) {
+			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usSclkVddgfxTableOffset));
+			ret = amdgpu_parse_clk_voltage_dep_table(
+					&adev->pm.dpm.dyn_state.vddgfx_dependency_on_sclk,
+					dep_table);
+			if (ret) {
+				kfree(adev->pm.dpm.dyn_state.vddgfx_dependency_on_sclk.entries);
+				return ret;
+			}
+		}
+	}
+
+	return 0;
+}
+
+void amdgpu_free_extended_power_table(struct amdgpu_device *adev)
+{
+	struct amdgpu_dpm_dynamic_state *dyn_state = &adev->pm.dpm.dyn_state;
+
+	kfree(dyn_state->vddc_dependency_on_sclk.entries);
+	kfree(dyn_state->vddci_dependency_on_mclk.entries);
+	kfree(dyn_state->vddc_dependency_on_mclk.entries);
+	kfree(dyn_state->mvdd_dependency_on_mclk.entries);
+	kfree(dyn_state->cac_leakage_table.entries);
+	kfree(dyn_state->phase_shedding_limits_table.entries);
+	kfree(dyn_state->ppm_table);
+	kfree(dyn_state->cac_tdp_table);
+	kfree(dyn_state->vce_clock_voltage_dependency_table.entries);
+	kfree(dyn_state->uvd_clock_voltage_dependency_table.entries);
+	kfree(dyn_state->samu_clock_voltage_dependency_table.entries);
+	kfree(dyn_state->acp_clock_voltage_dependency_table.entries);
+	kfree(dyn_state->vddgfx_dependency_on_sclk.entries);
+}
+
+static const char *pp_lib_thermal_controller_names[] = {
+	"NONE",
+	"lm63",
+	"adm1032",
+	"adm1030",
+	"max6649",
+	"lm64",
+	"f75375",
+	"RV6xx",
+	"RV770",
+	"adt7473",
+	"NONE",
+	"External GPIO",
+	"Evergreen",
+	"emc2103",
+	"Sumo",
+	"Northern Islands",
+	"Southern Islands",
+	"lm96163",
+	"Sea Islands",
+	"Kaveri/Kabini",
+};
+
+void amdgpu_add_thermal_controller(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	ATOM_PPLIB_POWERPLAYTABLE *power_table;
+	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
+	ATOM_PPLIB_THERMALCONTROLLER *controller;
+	struct amdgpu_i2c_bus_rec i2c_bus;
+	u16 data_offset;
+	u8 frev, crev;
+
+	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset))
+		return;
+	power_table = (ATOM_PPLIB_POWERPLAYTABLE *)
+		(mode_info->atom_context->bios + data_offset);
+	controller = &power_table->sThermalController;
+
+	/* add the i2c bus for thermal/fan chip */
+	if (controller->ucType > 0) {
+		if (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN)
+			adev->pm.no_fan = true;
+		adev->pm.fan_pulses_per_revolution =
+			controller->ucFanParameters & ATOM_PP_FANPARAMETERS_TACHOMETER_PULSES_PER_REVOLUTION_MASK;
+		if (adev->pm.fan_pulses_per_revolution) {
+			adev->pm.fan_min_rpm = controller->ucFanMinRPM;
+			adev->pm.fan_max_rpm = controller->ucFanMaxRPM;
+		}
+		if (controller->ucType == ATOM_PP_THERMALCONTROLLER_RV6xx) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_RV6XX;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_RV770) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_RV770;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_EVERGREEN) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_EVERGREEN;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_SUMO) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_SUMO;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_NISLANDS) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_NI;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_SISLANDS) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_SI;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_CISLANDS) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_CI;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_KAVERI) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_KV;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_EXTERNAL_GPIO) {
+			DRM_INFO("External GPIO thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_EXTERNAL_GPIO;
+		} else if (controller->ucType ==
+			   ATOM_PP_THERMALCONTROLLER_ADT7473_WITH_INTERNAL) {
+			DRM_INFO("ADT7473 with internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_ADT7473_WITH_INTERNAL;
+		} else if (controller->ucType ==
+			   ATOM_PP_THERMALCONTROLLER_EMC2103_WITH_INTERNAL) {
+			DRM_INFO("EMC2103 with internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_EMC2103_WITH_INTERNAL;
+		} else if (controller->ucType < ARRAY_SIZE(pp_lib_thermal_controller_names)) {
+			DRM_INFO("Possible %s thermal controller at 0x%02x %s fan control\n",
+				 pp_lib_thermal_controller_names[controller->ucType],
+				 controller->ucI2cAddress >> 1,
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_EXTERNAL;
+			i2c_bus = amdgpu_atombios_lookup_i2c_gpio(adev, controller->ucI2cLine);
+			adev->pm.i2c_bus = amdgpu_i2c_lookup(adev, &i2c_bus);
+#ifdef notyet
+			if (adev->pm.i2c_bus) {
+				struct i2c_board_info info = { };
+				const char *name = pp_lib_thermal_controller_names[controller->ucType];
+				info.addr = controller->ucI2cAddress >> 1;
+				strlcpy(info.type, name, sizeof(info.type));
+				i2c_new_device(&adev->pm.i2c_bus->adapter, &info);
+			}
+#endif
+		} else {
+			DRM_INFO("Unknown thermal controller type %d at 0x%02x %s fan control\n",
+				 controller->ucType,
+				 controller->ucI2cAddress >> 1,
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+		}
+	}
+}
+
+enum amdgpu_pcie_gen amdgpu_get_pcie_gen_support(struct amdgpu_device *adev,
+						 u32 sys_mask,
+						 enum amdgpu_pcie_gen asic_gen,
+						 enum amdgpu_pcie_gen default_gen)
+{
+	switch (asic_gen) {
+	case AMDGPU_PCIE_GEN1:
+		return AMDGPU_PCIE_GEN1;
+	case AMDGPU_PCIE_GEN2:
+		return AMDGPU_PCIE_GEN2;
+	case AMDGPU_PCIE_GEN3:
+		return AMDGPU_PCIE_GEN3;
+	default:
+		if ((sys_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3) &&
+		    (default_gen == AMDGPU_PCIE_GEN3))
+			return AMDGPU_PCIE_GEN3;
+		else if ((sys_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2) &&
+			 (default_gen == AMDGPU_PCIE_GEN2))
+			return AMDGPU_PCIE_GEN2;
+		else
+			return AMDGPU_PCIE_GEN1;
+	}
+	return AMDGPU_PCIE_GEN1;
+}
+
+u16 amdgpu_get_pcie_lane_support(struct amdgpu_device *adev,
+				 u16 asic_lanes,
+				 u16 default_lanes)
+{
+	switch (asic_lanes) {
+	case 0:
+	default:
+		return default_lanes;
+	case 1:
+		return 1;
+	case 2:
+		return 2;
+	case 4:
+		return 4;
+	case 8:
+		return 8;
+	case 12:
+		return 12;
+	case 16:
+		return 16;
+	}
+}
+
+u8 amdgpu_encode_pci_lane_width(u32 lanes)
+{
+	u8 encoded_lanes[] = { 0, 1, 2, 0, 3, 0, 0, 0, 4, 0, 0, 0, 5, 0, 0, 0, 6 };
+
+	if (lanes > 16)
+		return 0;
+
+	return encoded_lanes[lanes];
+}
+
+struct amd_vce_state*
+amdgpu_get_vce_clock_state(void *handle, u32 idx)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (idx < adev->pm.dpm.num_of_vce_states)
+		return &adev->pm.dpm.vce_states[idx];
+
+	return NULL;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_dpm.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_dpm.h
new file mode 100644
index 00000000000..cdfd4550989
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_dpm.h
@@ -0,0 +1,501 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_DPM_H__
+#define __AMDGPU_DPM_H__
+
+enum amdgpu_int_thermal_type {
+	THERMAL_TYPE_NONE,
+	THERMAL_TYPE_EXTERNAL,
+	THERMAL_TYPE_EXTERNAL_GPIO,
+	THERMAL_TYPE_RV6XX,
+	THERMAL_TYPE_RV770,
+	THERMAL_TYPE_ADT7473_WITH_INTERNAL,
+	THERMAL_TYPE_EVERGREEN,
+	THERMAL_TYPE_SUMO,
+	THERMAL_TYPE_NI,
+	THERMAL_TYPE_SI,
+	THERMAL_TYPE_EMC2103_WITH_INTERNAL,
+	THERMAL_TYPE_CI,
+	THERMAL_TYPE_KV,
+};
+
+enum amdgpu_dpm_auto_throttle_src {
+	AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL,
+	AMDGPU_DPM_AUTO_THROTTLE_SRC_EXTERNAL
+};
+
+enum amdgpu_dpm_event_src {
+	AMDGPU_DPM_EVENT_SRC_ANALOG = 0,
+	AMDGPU_DPM_EVENT_SRC_EXTERNAL = 1,
+	AMDGPU_DPM_EVENT_SRC_DIGITAL = 2,
+	AMDGPU_DPM_EVENT_SRC_ANALOG_OR_EXTERNAL = 3,
+	AMDGPU_DPM_EVENT_SRC_DIGIAL_OR_EXTERNAL = 4
+};
+
+struct amdgpu_ps {
+	u32 caps; /* vbios flags */
+	u32 class; /* vbios flags */
+	u32 class2; /* vbios flags */
+	/* UVD clocks */
+	u32 vclk;
+	u32 dclk;
+	/* VCE clocks */
+	u32 evclk;
+	u32 ecclk;
+	bool vce_active;
+	enum amd_vce_level vce_level;
+	/* asic priv */
+	void *ps_priv;
+};
+
+struct amdgpu_dpm_thermal {
+	/* thermal interrupt work */
+	struct work_struct work;
+	/* low temperature threshold */
+	int                min_temp;
+	/* high temperature threshold */
+	int                max_temp;
+	/* was last interrupt low to high or high to low */
+	bool               high_to_low;
+	/* interrupt source */
+	struct amdgpu_irq_src	irq;
+};
+
+enum amdgpu_clk_action
+{
+	AMDGPU_SCLK_UP = 1,
+	AMDGPU_SCLK_DOWN
+};
+
+struct amdgpu_blacklist_clocks
+{
+	u32 sclk;
+	u32 mclk;
+	enum amdgpu_clk_action action;
+};
+
+struct amdgpu_clock_and_voltage_limits {
+	u32 sclk;
+	u32 mclk;
+	u16 vddc;
+	u16 vddci;
+};
+
+struct amdgpu_clock_array {
+	u32 count;
+	u32 *values;
+};
+
+struct amdgpu_clock_voltage_dependency_entry {
+	u32 clk;
+	u16 v;
+};
+
+struct amdgpu_clock_voltage_dependency_table {
+	u32 count;
+	struct amdgpu_clock_voltage_dependency_entry *entries;
+};
+
+union amdgpu_cac_leakage_entry {
+	struct {
+		u16 vddc;
+		u32 leakage;
+	};
+	struct {
+		u16 vddc1;
+		u16 vddc2;
+		u16 vddc3;
+	};
+};
+
+struct amdgpu_cac_leakage_table {
+	u32 count;
+	union amdgpu_cac_leakage_entry *entries;
+};
+
+struct amdgpu_phase_shedding_limits_entry {
+	u16 voltage;
+	u32 sclk;
+	u32 mclk;
+};
+
+struct amdgpu_phase_shedding_limits_table {
+	u32 count;
+	struct amdgpu_phase_shedding_limits_entry *entries;
+};
+
+struct amdgpu_uvd_clock_voltage_dependency_entry {
+	u32 vclk;
+	u32 dclk;
+	u16 v;
+};
+
+struct amdgpu_uvd_clock_voltage_dependency_table {
+	u8 count;
+	struct amdgpu_uvd_clock_voltage_dependency_entry *entries;
+};
+
+struct amdgpu_vce_clock_voltage_dependency_entry {
+	u32 ecclk;
+	u32 evclk;
+	u16 v;
+};
+
+struct amdgpu_vce_clock_voltage_dependency_table {
+	u8 count;
+	struct amdgpu_vce_clock_voltage_dependency_entry *entries;
+};
+
+struct amdgpu_ppm_table {
+	u8 ppm_design;
+	u16 cpu_core_number;
+	u32 platform_tdp;
+	u32 small_ac_platform_tdp;
+	u32 platform_tdc;
+	u32 small_ac_platform_tdc;
+	u32 apu_tdp;
+	u32 dgpu_tdp;
+	u32 dgpu_ulv_power;
+	u32 tj_max;
+};
+
+struct amdgpu_cac_tdp_table {
+	u16 tdp;
+	u16 configurable_tdp;
+	u16 tdc;
+	u16 battery_power_limit;
+	u16 small_power_limit;
+	u16 low_cac_leakage;
+	u16 high_cac_leakage;
+	u16 maximum_power_delivery_limit;
+};
+
+struct amdgpu_dpm_dynamic_state {
+	struct amdgpu_clock_voltage_dependency_table vddc_dependency_on_sclk;
+	struct amdgpu_clock_voltage_dependency_table vddci_dependency_on_mclk;
+	struct amdgpu_clock_voltage_dependency_table vddc_dependency_on_mclk;
+	struct amdgpu_clock_voltage_dependency_table mvdd_dependency_on_mclk;
+	struct amdgpu_clock_voltage_dependency_table vddc_dependency_on_dispclk;
+	struct amdgpu_uvd_clock_voltage_dependency_table uvd_clock_voltage_dependency_table;
+	struct amdgpu_vce_clock_voltage_dependency_table vce_clock_voltage_dependency_table;
+	struct amdgpu_clock_voltage_dependency_table samu_clock_voltage_dependency_table;
+	struct amdgpu_clock_voltage_dependency_table acp_clock_voltage_dependency_table;
+	struct amdgpu_clock_voltage_dependency_table vddgfx_dependency_on_sclk;
+	struct amdgpu_clock_array valid_sclk_values;
+	struct amdgpu_clock_array valid_mclk_values;
+	struct amdgpu_clock_and_voltage_limits max_clock_voltage_on_dc;
+	struct amdgpu_clock_and_voltage_limits max_clock_voltage_on_ac;
+	u32 mclk_sclk_ratio;
+	u32 sclk_mclk_delta;
+	u16 vddc_vddci_delta;
+	u16 min_vddc_for_pcie_gen2;
+	struct amdgpu_cac_leakage_table cac_leakage_table;
+	struct amdgpu_phase_shedding_limits_table phase_shedding_limits_table;
+	struct amdgpu_ppm_table *ppm_table;
+	struct amdgpu_cac_tdp_table *cac_tdp_table;
+};
+
+struct amdgpu_dpm_fan {
+	u16 t_min;
+	u16 t_med;
+	u16 t_high;
+	u16 pwm_min;
+	u16 pwm_med;
+	u16 pwm_high;
+	u8 t_hyst;
+	u32 cycle_delay;
+	u16 t_max;
+	u8 control_mode;
+	u16 default_max_fan_pwm;
+	u16 default_fan_output_sensitivity;
+	u16 fan_output_sensitivity;
+	bool ucode_fan_control;
+};
+
+enum amdgpu_pcie_gen {
+	AMDGPU_PCIE_GEN1 = 0,
+	AMDGPU_PCIE_GEN2 = 1,
+	AMDGPU_PCIE_GEN3 = 2,
+	AMDGPU_PCIE_GEN_INVALID = 0xffff
+};
+
+#define amdgpu_dpm_pre_set_power_state(adev) \
+		((adev)->powerplay.pp_funcs->pre_set_power_state((adev)->powerplay.pp_handle))
+
+#define amdgpu_dpm_set_power_state(adev) \
+		((adev)->powerplay.pp_funcs->set_power_state((adev)->powerplay.pp_handle))
+
+#define amdgpu_dpm_post_set_power_state(adev) \
+		((adev)->powerplay.pp_funcs->post_set_power_state((adev)->powerplay.pp_handle))
+
+#define amdgpu_dpm_display_configuration_changed(adev) \
+		((adev)->powerplay.pp_funcs->display_configuration_changed((adev)->powerplay.pp_handle))
+
+#define amdgpu_dpm_print_power_state(adev, ps) \
+		((adev)->powerplay.pp_funcs->print_power_state((adev)->powerplay.pp_handle, (ps)))
+
+#define amdgpu_dpm_vblank_too_short(adev) \
+		((adev)->powerplay.pp_funcs->vblank_too_short((adev)->powerplay.pp_handle))
+
+#define amdgpu_dpm_enable_bapm(adev, e) \
+		((adev)->powerplay.pp_funcs->enable_bapm((adev)->powerplay.pp_handle, (e)))
+
+#define amdgpu_dpm_read_sensor(adev, idx, value, size) \
+		((adev)->powerplay.pp_funcs->read_sensor((adev)->powerplay.pp_handle, (idx), (value), (size)))
+
+#define amdgpu_dpm_set_fan_control_mode(adev, m) \
+		((adev)->powerplay.pp_funcs->set_fan_control_mode((adev)->powerplay.pp_handle, (m)))
+
+#define amdgpu_dpm_get_fan_control_mode(adev) \
+		((adev)->powerplay.pp_funcs->get_fan_control_mode((adev)->powerplay.pp_handle))
+
+#define amdgpu_dpm_set_fan_speed_percent(adev, s) \
+		((adev)->powerplay.pp_funcs->set_fan_speed_percent((adev)->powerplay.pp_handle, (s)))
+
+#define amdgpu_dpm_get_fan_speed_percent(adev, s) \
+		((adev)->powerplay.pp_funcs->get_fan_speed_percent((adev)->powerplay.pp_handle, (s)))
+
+#define amdgpu_dpm_get_fan_speed_rpm(adev, s) \
+		((adev)->powerplay.pp_funcs->get_fan_speed_rpm)((adev)->powerplay.pp_handle, (s))
+
+#define amdgpu_dpm_get_sclk(adev, l) \
+		((adev)->powerplay.pp_funcs->get_sclk((adev)->powerplay.pp_handle, (l)))
+
+#define amdgpu_dpm_get_mclk(adev, l)  \
+		((adev)->powerplay.pp_funcs->get_mclk((adev)->powerplay.pp_handle, (l)))
+
+#define amdgpu_dpm_force_performance_level(adev, l) \
+		((adev)->powerplay.pp_funcs->force_performance_level((adev)->powerplay.pp_handle, (l)))
+
+#define amdgpu_dpm_get_current_power_state(adev) \
+		((adev)->powerplay.pp_funcs->get_current_power_state((adev)->powerplay.pp_handle))
+
+#define amdgpu_dpm_get_pp_num_states(adev, data) \
+		((adev)->powerplay.pp_funcs->get_pp_num_states((adev)->powerplay.pp_handle, data))
+
+#define amdgpu_dpm_get_pp_table(adev, table) \
+		((adev)->powerplay.pp_funcs->get_pp_table((adev)->powerplay.pp_handle, table))
+
+#define amdgpu_dpm_set_pp_table(adev, buf, size) \
+		((adev)->powerplay.pp_funcs->set_pp_table((adev)->powerplay.pp_handle, buf, size))
+
+#define amdgpu_dpm_print_clock_levels(adev, type, buf) \
+		((adev)->powerplay.pp_funcs->print_clock_levels((adev)->powerplay.pp_handle, type, buf))
+
+#define amdgpu_dpm_force_clock_level(adev, type, level) \
+		((adev)->powerplay.pp_funcs->force_clock_level((adev)->powerplay.pp_handle, type, level))
+
+#define amdgpu_dpm_get_sclk_od(adev) \
+		((adev)->powerplay.pp_funcs->get_sclk_od((adev)->powerplay.pp_handle))
+
+#define amdgpu_dpm_set_sclk_od(adev, value) \
+		((adev)->powerplay.pp_funcs->set_sclk_od((adev)->powerplay.pp_handle, value))
+
+#define amdgpu_dpm_get_mclk_od(adev) \
+		((adev)->powerplay.pp_funcs->get_mclk_od((adev)->powerplay.pp_handle))
+
+#define amdgpu_dpm_set_mclk_od(adev, value) \
+		((adev)->powerplay.pp_funcs->set_mclk_od((adev)->powerplay.pp_handle, value))
+
+#define amdgpu_dpm_dispatch_task(adev, task_id, user_state)		\
+		((adev)->powerplay.pp_funcs->dispatch_tasks)((adev)->powerplay.pp_handle, (task_id), (user_state))
+
+#define amdgpu_dpm_check_state_equal(adev, cps, rps, equal) \
+		((adev)->powerplay.pp_funcs->check_state_equal((adev)->powerplay.pp_handle, (cps), (rps), (equal)))
+
+#define amdgpu_dpm_get_vce_clock_state(adev, i)				\
+		((adev)->powerplay.pp_funcs->get_vce_clock_state((adev)->powerplay.pp_handle, (i)))
+
+#define amdgpu_dpm_get_performance_level(adev)				\
+		((adev)->powerplay.pp_funcs->get_performance_level((adev)->powerplay.pp_handle))
+
+#define amdgpu_dpm_reset_power_profile_state(adev, request) \
+		((adev)->powerplay.pp_funcs->reset_power_profile_state(\
+			(adev)->powerplay.pp_handle, request))
+
+#define amdgpu_dpm_switch_power_profile(adev, type, en) \
+		((adev)->powerplay.pp_funcs->switch_power_profile(\
+			(adev)->powerplay.pp_handle, type, en))
+
+#define amdgpu_dpm_set_clockgating_by_smu(adev, msg_id) \
+		((adev)->powerplay.pp_funcs->set_clockgating_by_smu(\
+			(adev)->powerplay.pp_handle, msg_id))
+
+#define amdgpu_dpm_set_powergating_by_smu(adev, block_type, gate) \
+		((adev)->powerplay.pp_funcs->set_powergating_by_smu(\
+			(adev)->powerplay.pp_handle, block_type, gate))
+
+#define amdgpu_dpm_get_power_profile_mode(adev, buf) \
+		((adev)->powerplay.pp_funcs->get_power_profile_mode(\
+			(adev)->powerplay.pp_handle, buf))
+
+#define amdgpu_dpm_set_power_profile_mode(adev, parameter, size) \
+		((adev)->powerplay.pp_funcs->set_power_profile_mode(\
+			(adev)->powerplay.pp_handle, parameter, size))
+
+#define amdgpu_dpm_odn_edit_dpm_table(adev, type, parameter, size) \
+		((adev)->powerplay.pp_funcs->odn_edit_dpm_table(\
+			(adev)->powerplay.pp_handle, type, parameter, size))
+
+struct amdgpu_dpm {
+	struct amdgpu_ps        *ps;
+	/* number of valid power states */
+	int                     num_ps;
+	/* current power state that is active */
+	struct amdgpu_ps        *current_ps;
+	/* requested power state */
+	struct amdgpu_ps        *requested_ps;
+	/* boot up power state */
+	struct amdgpu_ps        *boot_ps;
+	/* default uvd power state */
+	struct amdgpu_ps        *uvd_ps;
+	/* vce requirements */
+	u32                  num_of_vce_states;
+	struct amd_vce_state vce_states[AMD_MAX_VCE_LEVELS];
+	enum amd_vce_level vce_level;
+	enum amd_pm_state_type state;
+	enum amd_pm_state_type user_state;
+	enum amd_pm_state_type last_state;
+	enum amd_pm_state_type last_user_state;
+	u32                     platform_caps;
+	u32                     voltage_response_time;
+	u32                     backbias_response_time;
+	void                    *priv;
+	u32			new_active_crtcs;
+	int			new_active_crtc_count;
+	u32			current_active_crtcs;
+	int			current_active_crtc_count;
+	struct amdgpu_dpm_dynamic_state dyn_state;
+	struct amdgpu_dpm_fan fan;
+	u32 tdp_limit;
+	u32 near_tdp_limit;
+	u32 near_tdp_limit_adjusted;
+	u32 sq_ramping_threshold;
+	u32 cac_leakage;
+	u16 tdp_od_limit;
+	u32 tdp_adjustment;
+	u16 load_line_slope;
+	bool power_control;
+	/* special states active */
+	bool                    thermal_active;
+	bool                    uvd_active;
+	bool                    vce_active;
+	/* thermal handling */
+	struct amdgpu_dpm_thermal thermal;
+	/* forced levels */
+	enum amd_dpm_forced_level forced_level;
+};
+
+struct amdgpu_pm {
+	struct rwlock		mutex;
+	u32                     current_sclk;
+	u32                     current_mclk;
+	u32                     default_sclk;
+	u32                     default_mclk;
+	struct amdgpu_i2c_chan *i2c_bus;
+	/* internal thermal controller on rv6xx+ */
+	enum amdgpu_int_thermal_type int_thermal_type;
+	struct device	        *int_hwmon_dev;
+	/* fan control parameters */
+	bool                    no_fan;
+	u8                      fan_pulses_per_revolution;
+	u8                      fan_min_rpm;
+	u8                      fan_max_rpm;
+	/* dpm */
+	bool                    dpm_enabled;
+	bool                    sysfs_initialized;
+	struct amdgpu_dpm       dpm;
+	const struct firmware	*fw;	/* SMC firmware */
+	uint32_t                fw_version;
+	uint32_t                pcie_gen_mask;
+	uint32_t                pcie_mlw_mask;
+	struct amd_pp_display_configuration pm_display_cfg;/* set by dc */
+	uint32_t                smu_prv_buffer_size;
+	struct amdgpu_bo        *smu_prv_buffer;
+	bool ac_power;
+};
+
+#define R600_SSTU_DFLT                               0
+#define R600_SST_DFLT                                0x00C8
+
+/* XXX are these ok? */
+#define R600_TEMP_RANGE_MIN (90 * 1000)
+#define R600_TEMP_RANGE_MAX (120 * 1000)
+
+#define FDO_PWM_MODE_STATIC  1
+#define FDO_PWM_MODE_STATIC_RPM 5
+
+enum amdgpu_td {
+	AMDGPU_TD_AUTO,
+	AMDGPU_TD_UP,
+	AMDGPU_TD_DOWN,
+};
+
+enum amdgpu_display_watermark {
+	AMDGPU_DISPLAY_WATERMARK_LOW = 0,
+	AMDGPU_DISPLAY_WATERMARK_HIGH = 1,
+};
+
+enum amdgpu_display_gap
+{
+    AMDGPU_PM_DISPLAY_GAP_VBLANK_OR_WM = 0,
+    AMDGPU_PM_DISPLAY_GAP_VBLANK       = 1,
+    AMDGPU_PM_DISPLAY_GAP_WATERMARK    = 2,
+    AMDGPU_PM_DISPLAY_GAP_IGNORE       = 3,
+};
+
+void amdgpu_dpm_print_class_info(u32 class, u32 class2);
+void amdgpu_dpm_print_cap_info(u32 caps);
+void amdgpu_dpm_print_ps_status(struct amdgpu_device *adev,
+				struct amdgpu_ps *rps);
+u32 amdgpu_dpm_get_vblank_time(struct amdgpu_device *adev);
+u32 amdgpu_dpm_get_vrefresh(struct amdgpu_device *adev);
+void amdgpu_dpm_get_active_displays(struct amdgpu_device *adev);
+bool amdgpu_is_uvd_state(u32 class, u32 class2);
+void amdgpu_calculate_u_and_p(u32 i, u32 r_c, u32 p_b,
+			      u32 *p, u32 *u);
+int amdgpu_calculate_at(u32 t, u32 h, u32 fh, u32 fl, u32 *tl, u32 *th);
+
+bool amdgpu_is_internal_thermal_sensor(enum amdgpu_int_thermal_type sensor);
+
+int amdgpu_get_platform_caps(struct amdgpu_device *adev);
+
+int amdgpu_parse_extended_power_table(struct amdgpu_device *adev);
+void amdgpu_free_extended_power_table(struct amdgpu_device *adev);
+
+void amdgpu_add_thermal_controller(struct amdgpu_device *adev);
+
+enum amdgpu_pcie_gen amdgpu_get_pcie_gen_support(struct amdgpu_device *adev,
+						 u32 sys_mask,
+						 enum amdgpu_pcie_gen asic_gen,
+						 enum amdgpu_pcie_gen default_gen);
+
+u16 amdgpu_get_pcie_lane_support(struct amdgpu_device *adev,
+				 u16 asic_lanes,
+				 u16 default_lanes);
+u8 amdgpu_encode_pci_lane_width(u32 lanes);
+
+struct amd_vce_state*
+amdgpu_get_vce_clock_state(void *handle, u32 idx);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_drv.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_drv.c
new file mode 100644
index 00000000000..b97a3d30ebe
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_drv.c
@@ -0,0 +1,1229 @@
+/*
+ * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include <drm/drm_gem.h>
+#include "amdgpu_drv.h"
+
+#include <drm/drm_pciids.h>
+#include <linux/console.h>
+#include <linux/module.h>
+#include <linux/pm_runtime.h>
+#include <linux/vga_switcheroo.h>
+#include <drm/drm_crtc_helper.h>
+
+#include "amdgpu.h"
+#include "amdgpu_irq.h"
+
+#include "amdgpu_amdkfd.h"
+
+/*
+ * KMS wrapper.
+ * - 3.0.0 - initial driver
+ * - 3.1.0 - allow reading more status registers (GRBM, SRBM, SDMA, CP)
+ * - 3.2.0 - GFX8: Uses EOP_TC_WB_ACTION_EN, so UMDs don't have to do the same
+ *           at the end of IBs.
+ * - 3.3.0 - Add VM support for UVD on supported hardware.
+ * - 3.4.0 - Add AMDGPU_INFO_NUM_EVICTIONS.
+ * - 3.5.0 - Add support for new UVD_NO_OP register.
+ * - 3.6.0 - kmd involves use CONTEXT_CONTROL in ring buffer.
+ * - 3.7.0 - Add support for VCE clock list packet
+ * - 3.8.0 - Add support raster config init in the kernel
+ * - 3.9.0 - Add support for memory query info about VRAM and GTT.
+ * - 3.10.0 - Add support for new fences ioctl, new gem ioctl flags
+ * - 3.11.0 - Add support for sensor query info (clocks, temp, etc).
+ * - 3.12.0 - Add query for double offchip LDS buffers
+ * - 3.13.0 - Add PRT support
+ * - 3.14.0 - Fix race in amdgpu_ctx_get_fence() and note new functionality
+ * - 3.15.0 - Export more gpu info for gfx9
+ * - 3.16.0 - Add reserved vmid support
+ * - 3.17.0 - Add AMDGPU_NUM_VRAM_CPU_PAGE_FAULTS.
+ * - 3.18.0 - Export gpu always on cu bitmap
+ * - 3.19.0 - Add support for UVD MJPEG decode
+ * - 3.20.0 - Add support for local BOs
+ * - 3.21.0 - Add DRM_AMDGPU_FENCE_TO_HANDLE ioctl
+ * - 3.22.0 - Add DRM_AMDGPU_SCHED ioctl
+ * - 3.23.0 - Add query for VRAM lost counter
+ * - 3.24.0 - Add high priority compute support for gfx9
+ * - 3.25.0 - Add support for sensor query info (stable pstate sclk/mclk).
+ * - 3.26.0 - GFX9: Process AMDGPU_IB_FLAG_TC_WB_NOT_INVALIDATE.
+ * - 3.27.0 - Add new chunk to to AMDGPU_CS to enable BO_LIST creation.
+ */
+#define KMS_DRIVER_MAJOR	3
+#define KMS_DRIVER_MINOR	27
+#define KMS_DRIVER_PATCHLEVEL	0
+
+int amdgpu_vram_limit = 0;
+int amdgpu_vis_vram_limit = 0;
+int amdgpu_gart_size = -1; /* auto */
+int amdgpu_gtt_size = -1; /* auto */
+int amdgpu_moverate = -1; /* auto */
+int amdgpu_benchmarking = 0;
+int amdgpu_testing = 0;
+int amdgpu_audio = -1;
+int amdgpu_disp_priority = 0;
+int amdgpu_hw_i2c = 0;
+int amdgpu_pcie_gen2 = -1;
+int amdgpu_msi = -1;
+int amdgpu_lockup_timeout = 10000;
+int amdgpu_dpm = -1;
+int amdgpu_fw_load_type = -1;
+int amdgpu_aspm = -1;
+int amdgpu_runtime_pm = -1;
+uint amdgpu_ip_block_mask = 0xffffffff;
+int amdgpu_bapm = -1;
+int amdgpu_deep_color = 0;
+int amdgpu_vm_size = -1;
+int amdgpu_vm_fragment_size = -1;
+int amdgpu_vm_block_size = -1;
+int amdgpu_vm_fault_stop = 0;
+int amdgpu_vm_debug = 0;
+int amdgpu_vram_page_split = 512;
+int amdgpu_vm_update_mode = -1;
+int amdgpu_exp_hw_support = 0;
+int amdgpu_dc = -1;
+int amdgpu_sched_jobs = 32;
+int amdgpu_sched_hw_submission = 2;
+uint amdgpu_pcie_gen_cap = 0;
+uint amdgpu_pcie_lane_cap = 0;
+uint amdgpu_cg_mask = 0xffffffff;
+uint amdgpu_pg_mask = 0xffffffff;
+uint amdgpu_sdma_phase_quantum = 32;
+char *amdgpu_disable_cu = NULL;
+char *amdgpu_virtual_display = NULL;
+/* OverDrive(bit 14),gfxoff(bit 15),stutter mode(bit 17) disabled by default*/
+uint amdgpu_pp_feature_mask = 0xfffd3fff;
+int amdgpu_ngg = 0;
+int amdgpu_prim_buf_per_se = 0;
+int amdgpu_pos_buf_per_se = 0;
+int amdgpu_cntl_sb_buf_per_se = 0;
+int amdgpu_param_buf_per_se = 0;
+int amdgpu_job_hang_limit = 0;
+int amdgpu_lbpw = -1;
+int amdgpu_compute_multipipe = -1;
+int amdgpu_gpu_recovery = -1; /* auto */
+int amdgpu_emu_mode = 0;
+uint amdgpu_smu_memory_pool_size = 0;
+
+/**
+ * DOC: vramlimit (int)
+ * Restrict the total amount of VRAM in MiB for testing.  The default is 0 (Use full VRAM).
+ */
+MODULE_PARM_DESC(vramlimit, "Restrict VRAM for testing, in megabytes");
+module_param_named(vramlimit, amdgpu_vram_limit, int, 0600);
+
+/**
+ * DOC: vis_vramlimit (int)
+ * Restrict the amount of CPU visible VRAM in MiB for testing.  The default is 0 (Use full CPU visible VRAM).
+ */
+MODULE_PARM_DESC(vis_vramlimit, "Restrict visible VRAM for testing, in megabytes");
+module_param_named(vis_vramlimit, amdgpu_vis_vram_limit, int, 0444);
+
+/**
+ * DOC: gartsize (uint)
+ * Restrict the size of GART in Mib (32, 64, etc.) for testing. The default is -1 (The size depends on asic).
+ */
+MODULE_PARM_DESC(gartsize, "Size of GART to setup in megabytes (32, 64, etc., -1=auto)");
+module_param_named(gartsize, amdgpu_gart_size, uint, 0600);
+
+/**
+ * DOC: gttsize (int)
+ * Restrict the size of GTT domain in MiB for testing. The default is -1 (It's VRAM size if 3GB < VRAM < 3/4 RAM,
+ * otherwise 3/4 RAM size).
+ */
+MODULE_PARM_DESC(gttsize, "Size of the GTT domain in megabytes (-1 = auto)");
+module_param_named(gttsize, amdgpu_gtt_size, int, 0600);
+
+/**
+ * DOC: moverate (int)
+ * Set maximum buffer migration rate in MB/s. The default is -1 (8 MB/s).
+ */
+MODULE_PARM_DESC(moverate, "Maximum buffer migration rate in MB/s. (32, 64, etc., -1=auto, 0=1=disabled)");
+module_param_named(moverate, amdgpu_moverate, int, 0600);
+
+/**
+ * DOC: benchmark (int)
+ * Run benchmarks. The default is 0 (Skip benchmarks).
+ */
+MODULE_PARM_DESC(benchmark, "Run benchmark");
+module_param_named(benchmark, amdgpu_benchmarking, int, 0444);
+
+/**
+ * DOC: test (int)
+ * Test BO GTT->VRAM and VRAM->GTT GPU copies. The default is 0 (Skip test, only set 1 to run test).
+ */
+MODULE_PARM_DESC(test, "Run tests");
+module_param_named(test, amdgpu_testing, int, 0444);
+
+/**
+ * DOC: audio (int)
+ * Set HDMI/DPAudio. Only affects non-DC display handling. The default is -1 (Enabled), set 0 to disabled it.
+ */
+MODULE_PARM_DESC(audio, "Audio enable (-1 = auto, 0 = disable, 1 = enable)");
+module_param_named(audio, amdgpu_audio, int, 0444);
+
+/**
+ * DOC: disp_priority (int)
+ * Set display Priority (1 = normal, 2 = high). Only affects non-DC display handling. The default is 0 (auto).
+ */
+MODULE_PARM_DESC(disp_priority, "Display Priority (0 = auto, 1 = normal, 2 = high)");
+module_param_named(disp_priority, amdgpu_disp_priority, int, 0444);
+
+/**
+ * DOC: hw_i2c (int)
+ * To enable hw i2c engine. Only affects non-DC display handling. The default is 0 (Disabled).
+ */
+MODULE_PARM_DESC(hw_i2c, "hw i2c engine enable (0 = disable)");
+module_param_named(hw_i2c, amdgpu_hw_i2c, int, 0444);
+
+/**
+ * DOC: pcie_gen2 (int)
+ * To disable PCIE Gen2/3 mode (0 = disable, 1 = enable). The default is -1 (auto, enabled).
+ */
+MODULE_PARM_DESC(pcie_gen2, "PCIE Gen2 mode (-1 = auto, 0 = disable, 1 = enable)");
+module_param_named(pcie_gen2, amdgpu_pcie_gen2, int, 0444);
+
+/**
+ * DOC: msi (int)
+ * To disable Message Signaled Interrupts (MSI) functionality (1 = enable, 0 = disable). The default is -1 (auto, enabled).
+ */
+MODULE_PARM_DESC(msi, "MSI support (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(msi, amdgpu_msi, int, 0444);
+
+/**
+ * DOC: lockup_timeout (int)
+ * Set GPU scheduler timeout value in ms. Value 0 is invalidated, will be adjusted to 10000.
+ * Negative values mean 'infinite timeout' (MAX_JIFFY_OFFSET). The default is 10000.
+ */
+MODULE_PARM_DESC(lockup_timeout, "GPU lockup timeout in ms > 0 (default 10000)");
+module_param_named(lockup_timeout, amdgpu_lockup_timeout, int, 0444);
+
+/**
+ * DOC: dpm (int)
+ * Override for dynamic power management setting (1 = enable, 0 = disable). The default is -1 (auto).
+ */
+MODULE_PARM_DESC(dpm, "DPM support (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(dpm, amdgpu_dpm, int, 0444);
+
+/**
+ * DOC: fw_load_type (int)
+ * Set different firmware loading type for debugging (0 = direct, 1 = SMU, 2 = PSP). The default is -1 (auto).
+ */
+MODULE_PARM_DESC(fw_load_type, "firmware loading type (0 = direct, 1 = SMU, 2 = PSP, -1 = auto)");
+module_param_named(fw_load_type, amdgpu_fw_load_type, int, 0444);
+
+/**
+ * DOC: aspm (int)
+ * To disable ASPM (1 = enable, 0 = disable). The default is -1 (auto, enabled).
+ */
+MODULE_PARM_DESC(aspm, "ASPM support (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(aspm, amdgpu_aspm, int, 0444);
+
+/**
+ * DOC: runpm (int)
+ * Override for runtime power management control for dGPUs in PX/HG laptops. The amdgpu driver can dynamically power down
+ * the dGPU on PX/HG laptops when it is idle. The default is -1 (auto enable). Setting the value to 0 disables this functionality.
+ */
+MODULE_PARM_DESC(runpm, "PX runtime pm (1 = force enable, 0 = disable, -1 = PX only default)");
+module_param_named(runpm, amdgpu_runtime_pm, int, 0444);
+
+/**
+ * DOC: ip_block_mask (uint)
+ * Override what IP blocks are enabled on the GPU. Each GPU is a collection of IP blocks (gfx, display, video, etc.).
+ * Use this parameter to disable specific blocks. Note that the IP blocks do not have a fixed index. Some asics may not have
+ * some IPs or may include multiple instances of an IP so the ordering various from asic to asic. See the driver output in
+ * the kernel log for the list of IPs on the asic. The default is 0xffffffff (enable all blocks on a device).
+ */
+MODULE_PARM_DESC(ip_block_mask, "IP Block Mask (all blocks enabled (default))");
+module_param_named(ip_block_mask, amdgpu_ip_block_mask, uint, 0444);
+
+/**
+ * DOC: bapm (int)
+ * Bidirectional Application Power Management (BAPM) used to dynamically share TDP between CPU and GPU. Set value 0 to disable it.
+ * The default -1 (auto, enabled)
+ */
+MODULE_PARM_DESC(bapm, "BAPM support (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(bapm, amdgpu_bapm, int, 0444);
+
+/**
+ * DOC: deep_color (int)
+ * Set 1 to enable Deep Color support. Only affects non-DC display handling. The default is 0 (disabled).
+ */
+MODULE_PARM_DESC(deep_color, "Deep Color support (1 = enable, 0 = disable (default))");
+module_param_named(deep_color, amdgpu_deep_color, int, 0444);
+
+/**
+ * DOC: vm_size (int)
+ * Override the size of the GPU's per client virtual address space in GiB.  The default is -1 (automatic for each asic).
+ */
+MODULE_PARM_DESC(vm_size, "VM address space size in gigabytes (default 64GB)");
+module_param_named(vm_size, amdgpu_vm_size, int, 0444);
+
+/**
+ * DOC: vm_fragment_size (int)
+ * Override VM fragment size in bits (4, 5, etc. 4 = 64K, 9 = 2M). The default is -1 (automatic for each asic).
+ */
+MODULE_PARM_DESC(vm_fragment_size, "VM fragment size in bits (4, 5, etc. 4 = 64K (default), Max 9 = 2M)");
+module_param_named(vm_fragment_size, amdgpu_vm_fragment_size, int, 0444);
+
+/**
+ * DOC: vm_block_size (int)
+ * Override VM page table size in bits (default depending on vm_size and hw setup). The default is -1 (automatic for each asic).
+ */
+MODULE_PARM_DESC(vm_block_size, "VM page table size in bits (default depending on vm_size)");
+module_param_named(vm_block_size, amdgpu_vm_block_size, int, 0444);
+
+/**
+ * DOC: vm_fault_stop (int)
+ * Stop on VM fault for debugging (0 = never, 1 = print first, 2 = always). The default is 0 (No stop).
+ */
+MODULE_PARM_DESC(vm_fault_stop, "Stop on VM fault (0 = never (default), 1 = print first, 2 = always)");
+module_param_named(vm_fault_stop, amdgpu_vm_fault_stop, int, 0444);
+
+/**
+ * DOC: vm_debug (int)
+ * Debug VM handling (0 = disabled, 1 = enabled). The default is 0 (Disabled).
+ */
+MODULE_PARM_DESC(vm_debug, "Debug VM handling (0 = disabled (default), 1 = enabled)");
+module_param_named(vm_debug, amdgpu_vm_debug, int, 0644);
+
+/**
+ * DOC: vm_update_mode (int)
+ * Override VM update mode. VM updated by using CPU (0 = never, 1 = Graphics only, 2 = Compute only, 3 = Both). The default
+ * is -1 (Only in large BAR(LB) systems Compute VM tables will be updated by CPU, otherwise 0, never).
+ */
+MODULE_PARM_DESC(vm_update_mode, "VM update using CPU (0 = never (default except for large BAR(LB)), 1 = Graphics only, 2 = Compute only (default for LB), 3 = Both");
+module_param_named(vm_update_mode, amdgpu_vm_update_mode, int, 0444);
+
+/**
+ * DOC: vram_page_split (int)
+ * Override the number of pages after we split VRAM allocations (default 512, -1 = disable). The default is 512.
+ */
+MODULE_PARM_DESC(vram_page_split, "Number of pages after we split VRAM allocations (default 512, -1 = disable)");
+module_param_named(vram_page_split, amdgpu_vram_page_split, int, 0444);
+
+/**
+ * DOC: exp_hw_support (int)
+ * Enable experimental hw support (1 = enable). The default is 0 (disabled).
+ */
+MODULE_PARM_DESC(exp_hw_support, "experimental hw support (1 = enable, 0 = disable (default))");
+module_param_named(exp_hw_support, amdgpu_exp_hw_support, int, 0444);
+
+/**
+ * DOC: dc (int)
+ * Disable/Enable Display Core driver for debugging (1 = enable, 0 = disable). The default is -1 (automatic for each asic).
+ */
+MODULE_PARM_DESC(dc, "Display Core driver (1 = enable, 0 = disable, -1 = auto (default))");
+module_param_named(dc, amdgpu_dc, int, 0444);
+
+/**
+ * DOC: sched_jobs (int)
+ * Override the max number of jobs supported in the sw queue. The default is 32.
+ */
+MODULE_PARM_DESC(sched_jobs, "the max number of jobs supported in the sw queue (default 32)");
+module_param_named(sched_jobs, amdgpu_sched_jobs, int, 0444);
+
+/**
+ * DOC: sched_hw_submission (int)
+ * Override the max number of HW submissions. The default is 2.
+ */
+MODULE_PARM_DESC(sched_hw_submission, "the max number of HW submissions (default 2)");
+module_param_named(sched_hw_submission, amdgpu_sched_hw_submission, int, 0444);
+
+/**
+ * DOC: ppfeaturemask (uint)
+ * Override power features enabled. See enum PP_FEATURE_MASK in drivers/gpu/drm/amd/include/amd_shared.h.
+ * The default is the current set of stable power features.
+ */
+MODULE_PARM_DESC(ppfeaturemask, "all power features enabled (default))");
+module_param_named(ppfeaturemask, amdgpu_pp_feature_mask, uint, 0444);
+
+/**
+ * DOC: pcie_gen_cap (uint)
+ * Override PCIE gen speed capabilities. See the CAIL flags in drivers/gpu/drm/amd/include/amd_pcie.h.
+ * The default is 0 (automatic for each asic).
+ */
+MODULE_PARM_DESC(pcie_gen_cap, "PCIE Gen Caps (0: autodetect (default))");
+module_param_named(pcie_gen_cap, amdgpu_pcie_gen_cap, uint, 0444);
+
+/**
+ * DOC: pcie_lane_cap (uint)
+ * Override PCIE lanes capabilities. See the CAIL flags in drivers/gpu/drm/amd/include/amd_pcie.h.
+ * The default is 0 (automatic for each asic).
+ */
+MODULE_PARM_DESC(pcie_lane_cap, "PCIE Lane Caps (0: autodetect (default))");
+module_param_named(pcie_lane_cap, amdgpu_pcie_lane_cap, uint, 0444);
+
+/**
+ * DOC: cg_mask (uint)
+ * Override Clockgating features enabled on GPU (0 = disable clock gating). See the AMD_CG_SUPPORT flags in
+ * drivers/gpu/drm/amd/include/amd_shared.h. The default is 0xffffffff (all enabled).
+ */
+MODULE_PARM_DESC(cg_mask, "Clockgating flags mask (0 = disable clock gating)");
+module_param_named(cg_mask, amdgpu_cg_mask, uint, 0444);
+
+/**
+ * DOC: pg_mask (uint)
+ * Override Powergating features enabled on GPU (0 = disable power gating). See the AMD_PG_SUPPORT flags in
+ * drivers/gpu/drm/amd/include/amd_shared.h. The default is 0xffffffff (all enabled).
+ */
+MODULE_PARM_DESC(pg_mask, "Powergating flags mask (0 = disable power gating)");
+module_param_named(pg_mask, amdgpu_pg_mask, uint, 0444);
+
+/**
+ * DOC: sdma_phase_quantum (uint)
+ * Override SDMA context switch phase quantum (x 1K GPU clock cycles, 0 = no change). The default is 32.
+ */
+MODULE_PARM_DESC(sdma_phase_quantum, "SDMA context switch phase quantum (x 1K GPU clock cycles, 0 = no change (default 32))");
+module_param_named(sdma_phase_quantum, amdgpu_sdma_phase_quantum, uint, 0444);
+
+/**
+ * DOC: disable_cu (charp)
+ * Set to disable CUs (It's set like se.sh.cu,...). The default is NULL.
+ */
+MODULE_PARM_DESC(disable_cu, "Disable CUs (se.sh.cu,...)");
+module_param_named(disable_cu, amdgpu_disable_cu, charp, 0444);
+
+/**
+ * DOC: virtual_display (charp)
+ * Set to enable virtual display feature. This feature provides a virtual display hardware on headless boards
+ * or in virtualized environments. It will be set like xxxx:xx:xx.x,x;xxxx:xx:xx.x,x. It's the pci address of
+ * the device, plus the number of crtcs to expose. E.g., 0000:26:00.0,4 would enable 4 virtual crtcs on the pci
+ * device at 26:00.0. The default is NULL.
+ */
+MODULE_PARM_DESC(virtual_display,
+		 "Enable virtual display feature (the virtual_display will be set like xxxx:xx:xx.x,x;xxxx:xx:xx.x,x)");
+module_param_named(virtual_display, amdgpu_virtual_display, charp, 0444);
+
+/**
+ * DOC: ngg (int)
+ * Set to enable Next Generation Graphics (1 = enable). The default is 0 (disabled).
+ */
+MODULE_PARM_DESC(ngg, "Next Generation Graphics (1 = enable, 0 = disable(default depending on gfx))");
+module_param_named(ngg, amdgpu_ngg, int, 0444);
+
+/**
+ * DOC: prim_buf_per_se (int)
+ * Override the size of Primitive Buffer per Shader Engine in Byte. The default is 0 (depending on gfx).
+ */
+MODULE_PARM_DESC(prim_buf_per_se, "the size of Primitive Buffer per Shader Engine (default depending on gfx)");
+module_param_named(prim_buf_per_se, amdgpu_prim_buf_per_se, int, 0444);
+
+/**
+ * DOC: pos_buf_per_se (int)
+ * Override the size of Position Buffer per Shader Engine in Byte. The default is 0 (depending on gfx).
+ */
+MODULE_PARM_DESC(pos_buf_per_se, "the size of Position Buffer per Shader Engine (default depending on gfx)");
+module_param_named(pos_buf_per_se, amdgpu_pos_buf_per_se, int, 0444);
+
+/**
+ * DOC: cntl_sb_buf_per_se (int)
+ * Override the size of Control Sideband per Shader Engine in Byte. The default is 0 (depending on gfx).
+ */
+MODULE_PARM_DESC(cntl_sb_buf_per_se, "the size of Control Sideband per Shader Engine (default depending on gfx)");
+module_param_named(cntl_sb_buf_per_se, amdgpu_cntl_sb_buf_per_se, int, 0444);
+
+/**
+ * DOC: param_buf_per_se (int)
+ * Override the size of Off-Chip Pramater Cache per Shader Engine in Byte. The default is 0 (depending on gfx).
+ */
+MODULE_PARM_DESC(param_buf_per_se, "the size of Off-Chip Pramater Cache per Shader Engine (default depending on gfx)");
+module_param_named(param_buf_per_se, amdgpu_param_buf_per_se, int, 0444);
+
+/**
+ * DOC: job_hang_limit (int)
+ * Set how much time allow a job hang and not drop it. The default is 0.
+ */
+MODULE_PARM_DESC(job_hang_limit, "how much time allow a job hang and not drop it (default 0)");
+module_param_named(job_hang_limit, amdgpu_job_hang_limit, int ,0444);
+
+/**
+ * DOC: lbpw (int)
+ * Override Load Balancing Per Watt (LBPW) support (1 = enable, 0 = disable). The default is -1 (auto, enabled).
+ */
+MODULE_PARM_DESC(lbpw, "Load Balancing Per Watt (LBPW) support (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(lbpw, amdgpu_lbpw, int, 0444);
+
+MODULE_PARM_DESC(compute_multipipe, "Force compute queues to be spread across pipes (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(compute_multipipe, amdgpu_compute_multipipe, int, 0444);
+
+/**
+ * DOC: gpu_recovery (int)
+ * Set to enable GPU recovery mechanism (1 = enable, 0 = disable). The default is -1 (auto, disabled except SRIOV).
+ */
+MODULE_PARM_DESC(gpu_recovery, "Enable GPU recovery mechanism, (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(gpu_recovery, amdgpu_gpu_recovery, int, 0444);
+
+/**
+ * DOC: emu_mode (int)
+ * Set value 1 to enable emulation mode. This is only needed when running on an emulator. The default is 0 (disabled).
+ */
+MODULE_PARM_DESC(emu_mode, "Emulation mode, (1 = enable, 0 = disable)");
+module_param_named(emu_mode, amdgpu_emu_mode, int, 0444);
+
+/**
+ * DOC: si_support (int)
+ * Set SI support driver. This parameter works after set config CONFIG_DRM_AMDGPU_SI. For SI asic, when radeon driver is enabled,
+ * set value 0 to use radeon driver, while set value 1 to use amdgpu driver. The default is using radeon driver when it available,
+ * otherwise using amdgpu driver.
+ */
+#ifdef CONFIG_DRM_AMDGPU_SI
+
+#if defined(CONFIG_DRM_RADEON) || defined(CONFIG_DRM_RADEON_MODULE)
+int amdgpu_si_support = 0;
+MODULE_PARM_DESC(si_support, "SI support (1 = enabled, 0 = disabled (default))");
+#else
+int amdgpu_si_support = 1;
+MODULE_PARM_DESC(si_support, "SI support (1 = enabled (default), 0 = disabled)");
+#endif
+
+module_param_named(si_support, amdgpu_si_support, int, 0444);
+#endif
+
+/**
+ * DOC: cik_support (int)
+ * Set CIK support driver. This parameter works after set config CONFIG_DRM_AMDGPU_CIK. For CIK asic, when radeon driver is enabled,
+ * set value 0 to use radeon driver, while set value 1 to use amdgpu driver. The default is using radeon driver when it available,
+ * otherwise using amdgpu driver.
+ */
+#ifdef CONFIG_DRM_AMDGPU_CIK
+
+#if defined(CONFIG_DRM_RADEON) || defined(CONFIG_DRM_RADEON_MODULE)
+int amdgpu_cik_support = 0;
+MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled, 0 = disabled (default))");
+#else
+int amdgpu_cik_support = 1;
+MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled (default), 0 = disabled)");
+#endif
+
+module_param_named(cik_support, amdgpu_cik_support, int, 0444);
+#endif
+
+/**
+ * DOC: smu_memory_pool_size (uint)
+ * It is used to reserve gtt for smu debug usage, setting value 0 to disable it. The actual size is value * 256MiB.
+ * E.g. 0x1 = 256Mbyte, 0x2 = 512Mbyte, 0x4 = 1 Gbyte, 0x8 = 2GByte. The default is 0 (disabled).
+ */
+MODULE_PARM_DESC(smu_memory_pool_size,
+	"reserve gtt for smu debug usage, 0 = disable,"
+		"0x1 = 256Mbyte, 0x2 = 512Mbyte, 0x4 = 1 Gbyte, 0x8 = 2GByte");
+module_param_named(smu_memory_pool_size, amdgpu_smu_memory_pool_size, uint, 0444);
+
+const struct drm_pcidev amdgpu_pciidlist[] = {
+#ifdef  CONFIG_DRM_AMDGPU_SI
+	{0x1002, 0x6780, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6784, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6788, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x678A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6790, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6791, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6792, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6798, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6799, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x679A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x679B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x679E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x679F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6800, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN|AMD_IS_MOBILITY},
+	{0x1002, 0x6801, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN|AMD_IS_MOBILITY},
+	{0x1002, 0x6802, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN|AMD_IS_MOBILITY},
+	{0x1002, 0x6806, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6808, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6809, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6810, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6811, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6816, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6817, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6818, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6819, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6600, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6601, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6602, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6603, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6604, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6605, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6606, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6607, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6608, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND},
+	{0x1002, 0x6610, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND},
+	{0x1002, 0x6611, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND},
+	{0x1002, 0x6613, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND},
+	{0x1002, 0x6617, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6620, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6621, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6623, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6631, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND},
+	{0x1002, 0x6820, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6821, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6822, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6823, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6824, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6825, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6826, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6827, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6828, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x6829, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x682A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x682B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x682C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x682D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x682F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6830, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6831, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6835, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x6837, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x6838, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x6839, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x683B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x683D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x683F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x6660, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
+	{0x1002, 0x6663, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
+	{0x1002, 0x6664, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
+	{0x1002, 0x6665, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
+	{0x1002, 0x6667, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
+	{0x1002, 0x666F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	/* Kaveri */
+	{0x1002, 0x1304, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x1305, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1306, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x1307, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1309, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x130A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x130B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x130C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x130D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x130E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x130F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1310, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1311, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1312, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1313, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1315, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1316, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1317, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x1318, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x131B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x131C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x131D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	/* Bonaire */
+	{0x1002, 0x6640, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE|AMD_IS_MOBILITY},
+	{0x1002, 0x6641, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE|AMD_IS_MOBILITY},
+	{0x1002, 0x6646, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE|AMD_IS_MOBILITY},
+	{0x1002, 0x6647, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE|AMD_IS_MOBILITY},
+	{0x1002, 0x6649, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	{0x1002, 0x6650, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	{0x1002, 0x6651, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	{0x1002, 0x6658, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	{0x1002, 0x665c, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	{0x1002, 0x665d, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	{0x1002, 0x665f, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	/* Hawaii */
+	{0x1002, 0x67A0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67A1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67A2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67A8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67A9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67AA, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67B0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67B1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67B8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67B9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67BA, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67BE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	/* Kabini */
+	{0x1002, 0x9830, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9831, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x9832, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9833, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x9834, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9835, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x9836, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9837, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x9838, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9839, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x983a, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x983b, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x983c, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x983d, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x983e, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x983f, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	/* mullins */
+	{0x1002, 0x9850, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9851, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9852, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9853, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9854, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9855, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9856, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9857, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9858, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9859, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x985A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x985B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x985C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x985D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x985E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x985F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+#endif
+	/* topaz */
+	{0x1002, 0x6900, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
+	{0x1002, 0x6901, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
+	{0x1002, 0x6902, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
+	{0x1002, 0x6903, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
+	{0x1002, 0x6907, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
+	/* tonga */
+	{0x1002, 0x6920, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x6921, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x6928, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x6929, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x692B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x692F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x6930, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x6938, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x6939, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	/* fiji */
+	{0x1002, 0x7300, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_FIJI},
+	{0x1002, 0x730F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_FIJI},
+	/* carrizo */
+	{0x1002, 0x9870, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CARRIZO|AMD_IS_APU},
+	{0x1002, 0x9874, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CARRIZO|AMD_IS_APU},
+	{0x1002, 0x9875, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CARRIZO|AMD_IS_APU},
+	{0x1002, 0x9876, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CARRIZO|AMD_IS_APU},
+	{0x1002, 0x9877, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CARRIZO|AMD_IS_APU},
+	/* stoney */
+	{0x1002, 0x98E4, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_STONEY|AMD_IS_APU},
+	/* Polaris11 */
+	{0x1002, 0x67E0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},
+	{0x1002, 0x67E3, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},
+	{0x1002, 0x67E8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},
+	{0x1002, 0x67EB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},
+	{0x1002, 0x67EF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},
+	{0x1002, 0x67FF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},
+	{0x1002, 0x67E1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},
+	{0x1002, 0x67E7, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},
+	{0x1002, 0x67E9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS11},
+	/* Polaris10 */
+	{0x1002, 0x67C0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x67C1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x67C2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x67C4, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x67C7, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x67D0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x67DF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x67C8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x67C9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x67CA, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x67CC, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x67CF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	{0x1002, 0x6FDF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS10},
+	/* Polaris12 */
+	{0x1002, 0x6980, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
+	{0x1002, 0x6981, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
+	{0x1002, 0x6985, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
+	{0x1002, 0x6986, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
+	{0x1002, 0x6987, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
+	{0x1002, 0x6995, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
+	{0x1002, 0x6997, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
+	{0x1002, 0x699F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_POLARIS12},
+	/* VEGAM */
+	{0x1002, 0x694C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGAM},
+	{0x1002, 0x694E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGAM},
+	{0x1002, 0x694F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGAM},
+	/* Vega 10 */
+	{0x1002, 0x6860, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6861, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6862, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6863, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6864, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6867, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6868, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x6869, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x686a, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x686b, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x686c, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x686d, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x686e, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x686f, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	{0x1002, 0x687f, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA10},
+	/* Vega 12 */
+	{0x1002, 0x69A0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12},
+	{0x1002, 0x69A1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12},
+	{0x1002, 0x69A2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12},
+	{0x1002, 0x69A3, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12},
+	{0x1002, 0x69AF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA12},
+	/* Vega 20 */
+	{0x1002, 0x66A0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20|AMD_EXP_HW_SUPPORT},
+	{0x1002, 0x66A1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20|AMD_EXP_HW_SUPPORT},
+	{0x1002, 0x66A2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20|AMD_EXP_HW_SUPPORT},
+	{0x1002, 0x66A3, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20|AMD_EXP_HW_SUPPORT},
+	{0x1002, 0x66A7, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20|AMD_EXP_HW_SUPPORT},
+	{0x1002, 0x66AF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VEGA20|AMD_EXP_HW_SUPPORT},
+	/* Raven */
+	{0x1002, 0x15dd, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RAVEN|AMD_IS_APU},
+
+	{0, 0, 0}
+};
+
+MODULE_DEVICE_TABLE(pci, pciidlist);
+
+struct drm_driver amdgpu_kms_driver;
+
+#ifdef notyet
+static int amdgpu_kick_out_firmware_fb(struct pci_dev *pdev)
+{
+	struct apertures_struct *ap;
+	bool primary = false;
+
+	ap = alloc_apertures(1);
+	if (!ap)
+		return -ENOMEM;
+
+	ap->ranges[0].base = pci_resource_start(pdev, 0);
+	ap->ranges[0].size = pci_resource_len(pdev, 0);
+
+#ifdef CONFIG_X86
+	primary = pdev->resource[PCI_ROM_RESOURCE].flags & IORESOURCE_ROM_SHADOW;
+#endif
+	drm_fb_helper_remove_conflicting_framebuffers(ap, "amdgpudrmfb", primary);
+	kfree(ap);
+
+	return 0;
+}
+#endif
+
+
+#ifdef notyet
+static int amdgpu_pci_probe(struct pci_dev *pdev,
+			    const struct pci_device_id *ent)
+{
+	struct drm_device *dev;
+	unsigned long flags = ent->driver_data;
+	int ret, retry = 0;
+	bool supports_atomic = false;
+
+	if (!amdgpu_virtual_display &&
+	    amdgpu_device_asic_has_dc_support(flags & AMD_ASIC_MASK))
+		supports_atomic = true;
+
+	if ((flags & AMD_EXP_HW_SUPPORT) && !amdgpu_exp_hw_support) {
+		DRM_INFO("This hardware requires experimental hardware support.\n"
+			 "See modparam exp_hw_support\n");
+		return -ENODEV;
+	}
+
+	/*
+	 * Initialize amdkfd before starting radeon. If it was not loaded yet,
+	 * defer radeon probing
+	 */
+	ret = amdgpu_amdkfd_init();
+	if (ret == -EPROBE_DEFER)
+		return ret;
+
+	/* Get rid of things like offb */
+	ret = amdgpu_kick_out_firmware_fb(pdev);
+	if (ret)
+		return ret;
+
+	/* warn the user if they mix atomic and non-atomic capable GPUs */
+	if ((kms_driver.driver_features & DRIVER_ATOMIC) && !supports_atomic)
+		DRM_ERROR("Mixing atomic and non-atomic capable GPUs!\n");
+	/* support atomic early so the atomic debugfs stuff gets created */
+	if (supports_atomic)
+		kms_driver.driver_features |= DRIVER_ATOMIC;
+
+	dev = drm_dev_alloc(&kms_driver, &pdev->dev);
+	if (IS_ERR(dev))
+		return PTR_ERR(dev);
+
+	ret = pci_enable_device(pdev);
+	if (ret)
+		goto err_free;
+
+	dev->pdev = pdev;
+
+	pci_set_drvdata(pdev, dev);
+
+retry_init:
+	ret = drm_dev_register(dev, ent->driver_data);
+	if (ret == -EAGAIN && ++retry <= 3) {
+		DRM_INFO("retry init %d\n", retry);
+		/* Don't request EX mode too frequently which is attacking */
+		drm_msleep(5000);
+		goto retry_init;
+	} else if (ret)
+		goto err_pci;
+
+	return 0;
+
+err_pci:
+	pci_disable_device(pdev);
+err_free:
+	drm_dev_put(dev);
+	return ret;
+}
+
+static void
+amdgpu_pci_remove(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+
+	drm_dev_unregister(dev);
+	drm_dev_put(dev);
+	pci_disable_device(pdev);
+	pci_set_drvdata(pdev, NULL);
+}
+
+static void
+amdgpu_pci_shutdown(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* if we are running in a VM, make sure the device
+	 * torn down properly on reboot/shutdown.
+	 * unfortunately we can't detect certain
+	 * hypervisors so just do this all the time.
+	 */
+	amdgpu_device_ip_suspend(adev);
+}
+
+static int amdgpu_pmops_suspend(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	return amdgpu_device_suspend(drm_dev, true, true);
+}
+
+static int amdgpu_pmops_resume(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+
+	/* GPU comes up enabled by the bios on resume */
+	if (amdgpu_device_is_px(drm_dev)) {
+		pm_runtime_disable(dev);
+		pm_runtime_set_active(dev);
+		pm_runtime_enable(dev);
+	}
+
+	return amdgpu_device_resume(drm_dev, true, true);
+}
+
+static int amdgpu_pmops_freeze(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	return amdgpu_device_suspend(drm_dev, false, true);
+}
+
+static int amdgpu_pmops_thaw(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	return amdgpu_device_resume(drm_dev, false, true);
+}
+
+static int amdgpu_pmops_poweroff(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	return amdgpu_device_suspend(drm_dev, true, true);
+}
+
+static int amdgpu_pmops_restore(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	return amdgpu_device_resume(drm_dev, false, true);
+}
+
+static int amdgpu_pmops_runtime_suspend(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	int ret;
+
+	if (!amdgpu_device_is_px(drm_dev)) {
+		pm_runtime_forbid(dev);
+		return -EBUSY;
+	}
+
+	drm_dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
+	drm_kms_helper_poll_disable(drm_dev);
+
+	ret = amdgpu_device_suspend(drm_dev, false, false);
+	pci_save_state(pdev);
+	pci_disable_device(pdev);
+	pci_ignore_hotplug(pdev);
+	if (amdgpu_is_atpx_hybrid())
+		pci_set_power_state(pdev, PCI_D3cold);
+	else if (!amdgpu_has_atpx_dgpu_power_cntl())
+		pci_set_power_state(pdev, PCI_D3hot);
+	drm_dev->switch_power_state = DRM_SWITCH_POWER_DYNAMIC_OFF;
+
+	return 0;
+}
+
+static int amdgpu_pmops_runtime_resume(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	int ret;
+
+	if (!amdgpu_device_is_px(drm_dev))
+		return -EINVAL;
+
+	drm_dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
+
+	if (amdgpu_is_atpx_hybrid() ||
+	    !amdgpu_has_atpx_dgpu_power_cntl())
+		pci_set_power_state(pdev, PCI_D0);
+	pci_restore_state(pdev);
+	ret = pci_enable_device(pdev);
+	if (ret)
+		return ret;
+	pci_set_master(pdev);
+
+	ret = amdgpu_device_resume(drm_dev, false, false);
+	drm_kms_helper_poll_enable(drm_dev);
+	drm_dev->switch_power_state = DRM_SWITCH_POWER_ON;
+	return 0;
+}
+
+static int amdgpu_pmops_runtime_idle(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	struct drm_crtc *crtc;
+
+	if (!amdgpu_device_is_px(drm_dev)) {
+		pm_runtime_forbid(dev);
+		return -EBUSY;
+	}
+
+	list_for_each_entry(crtc, &drm_dev->mode_config.crtc_list, head) {
+		if (crtc->enabled) {
+			DRM_DEBUG_DRIVER("failing to power off - crtc active\n");
+			return -EBUSY;
+		}
+	}
+
+	pm_runtime_mark_last_busy(dev);
+	pm_runtime_autosuspend(dev);
+	/* we don't want the main rpm_idle to call suspend - we want to autosuspend */
+	return 1;
+}
+#endif /* notyet */
+
+#ifdef __linux__
+long amdgpu_drm_ioctl(struct file *filp,
+		      unsigned int cmd, unsigned long arg)
+{
+	struct drm_file *file_priv = filp->private_data;
+	struct drm_device *dev;
+	long ret;
+	dev = file_priv->minor->dev;
+	ret = pm_runtime_get_sync(dev->dev);
+	if (ret < 0)
+		return ret;
+
+	ret = drm_ioctl(filp, cmd, arg);
+
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
+	return ret;
+}
+
+static const struct dev_pm_ops amdgpu_pm_ops = {
+	.suspend = amdgpu_pmops_suspend,
+	.resume = amdgpu_pmops_resume,
+	.freeze = amdgpu_pmops_freeze,
+	.thaw = amdgpu_pmops_thaw,
+	.poweroff = amdgpu_pmops_poweroff,
+	.restore = amdgpu_pmops_restore,
+	.runtime_suspend = amdgpu_pmops_runtime_suspend,
+	.runtime_resume = amdgpu_pmops_runtime_resume,
+	.runtime_idle = amdgpu_pmops_runtime_idle,
+};
+
+static int amdgpu_flush(struct file *f, fl_owner_t id)
+{
+	struct drm_file *file_priv = f->private_data;
+	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
+
+	amdgpu_ctx_mgr_entity_flush(&fpriv->ctx_mgr);
+
+	return 0;
+}
+
+static const struct file_operations amdgpu_driver_kms_fops = {
+	.owner = THIS_MODULE,
+	.open = drm_open,
+	.flush = amdgpu_flush,
+	.release = drm_release,
+	.unlocked_ioctl = amdgpu_drm_ioctl,
+	.mmap = amdgpu_mmap,
+	.poll = drm_poll,
+	.read = drm_read,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl = amdgpu_kms_compat_ioctl,
+#endif
+};
+#endif /* __linux__ */
+
+static bool
+amdgpu_get_crtc_scanout_position(struct drm_device *dev, unsigned int pipe,
+				 bool in_vblank_irq, int *vpos, int *hpos,
+				 ktime_t *stime, ktime_t *etime,
+				 const struct drm_display_mode *mode)
+{
+	return amdgpu_display_get_crtc_scanoutpos(dev, pipe, 0, vpos, hpos,
+						  stime, etime, mode);
+}
+
+struct drm_driver amdgpu_kms_driver = {
+	.driver_features =
+	    DRIVER_USE_AGP |
+	    DRIVER_HAVE_IRQ | DRIVER_IRQ_SHARED | DRIVER_GEM |
+	    DRIVER_PRIME | DRIVER_RENDER | DRIVER_MODESET | DRIVER_SYNCOBJ,
+#ifdef notyet
+	.load = amdgpu_driver_load_kms,
+#endif
+	.open = amdgpu_driver_open_kms,
+	.mmap = amdgpu_mmap,
+	.postclose = amdgpu_driver_postclose_kms,
+	.lastclose = amdgpu_driver_lastclose_kms,
+#ifdef notyet
+	.unload = amdgpu_driver_unload_kms,
+#endif
+	.get_vblank_counter = amdgpu_get_vblank_counter_kms,
+	.enable_vblank = amdgpu_enable_vblank_kms,
+	.disable_vblank = amdgpu_disable_vblank_kms,
+	.get_vblank_timestamp = drm_calc_vbltimestamp_from_scanoutpos,
+	.get_scanout_position = amdgpu_get_crtc_scanout_position,
+#ifdef notyet
+	.irq_handler = amdgpu_irq_handler,
+#endif
+	.ioctls = amdgpu_ioctls_kms,
+	.gem_free_object_unlocked = amdgpu_gem_object_free,
+	.gem_open_object = amdgpu_gem_object_open,
+	.gem_close_object = amdgpu_gem_object_close,
+	.gem_size = sizeof(struct amdgpu_bo),
+	.dumb_create = amdgpu_mode_dumb_create,
+	.dumb_map_offset = amdgpu_mode_dumb_mmap,
+#ifdef __linux__
+	.fops = &amdgpu_driver_kms_fops,
+#endif
+
+	.prime_handle_to_fd = drm_gem_prime_handle_to_fd,
+	.prime_fd_to_handle = drm_gem_prime_fd_to_handle,
+	.gem_prime_export = amdgpu_gem_prime_export,
+	.gem_prime_import = amdgpu_gem_prime_import,
+	.gem_prime_res_obj = amdgpu_gem_prime_res_obj,
+#ifdef notyet
+	.gem_prime_get_sg_table = amdgpu_gem_prime_get_sg_table,
+	.gem_prime_import_sg_table = amdgpu_gem_prime_import_sg_table,
+	.gem_prime_vmap = amdgpu_gem_prime_vmap,
+	.gem_prime_vunmap = amdgpu_gem_prime_vunmap,
+	.gem_prime_mmap = amdgpu_gem_prime_mmap,
+#endif
+
+	.name = DRIVER_NAME,
+	.desc = DRIVER_DESC,
+	.date = DRIVER_DATE,
+	.major = KMS_DRIVER_MAJOR,
+	.minor = KMS_DRIVER_MINOR,
+	.patchlevel = KMS_DRIVER_PATCHLEVEL,
+};
+
+static struct drm_driver *driver;
+#ifdef __linux__
+static struct pci_driver *pdriver;
+
+static struct pci_driver amdgpu_kms_pci_driver = {
+	.name = DRIVER_NAME,
+	.id_table = pciidlist,
+	.probe = amdgpu_pci_probe,
+	.remove = amdgpu_pci_remove,
+	.shutdown = amdgpu_pci_shutdown,
+	.driver.pm = &amdgpu_pm_ops,
+};
+
+static int __init amdgpu_init(void)
+{
+	int r;
+
+	if (vgacon_text_force()) {
+		DRM_ERROR("VGACON disables amdgpu kernel modesetting.\n");
+		return -EINVAL;
+	}
+
+	r = amdgpu_sync_init();
+	if (r)
+		goto error_sync;
+
+	r = amdgpu_fence_slab_init();
+	if (r)
+		goto error_fence;
+
+	DRM_INFO("amdgpu kernel modesetting enabled.\n");
+	driver = &kms_driver;
+	pdriver = &amdgpu_kms_pci_driver;
+	driver->num_ioctls = amdgpu_max_kms_ioctl;
+	amdgpu_register_atpx_handler();
+	/* let modprobe override vga console setting */
+	return pci_register_driver(pdriver);
+
+error_fence:
+	amdgpu_sync_fini();
+
+error_sync:
+	return r;
+}
+
+static void __exit amdgpu_exit(void)
+{
+	amdgpu_amdkfd_fini();
+	pci_unregister_driver(pdriver);
+	amdgpu_unregister_atpx_handler();
+	amdgpu_sync_fini();
+	amdgpu_fence_slab_fini();
+}
+
+module_init(amdgpu_init);
+module_exit(amdgpu_exit);
+
+MODULE_AUTHOR(DRIVER_AUTHOR);
+MODULE_DESCRIPTION(DRIVER_DESC);
+MODULE_LICENSE("GPL and additional rights");
+#endif /* __linux__ */
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_drv.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_drv.h
new file mode 100644
index 00000000000..e3a4f704804
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_drv.h
@@ -0,0 +1,48 @@
+/* amdgpu_drv.h -- Private header for amdgpu driver -*- linux-c -*-
+ *
+ * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas.
+ * Copyright 2000 VA Linux Systems, Inc., Fremont, California.
+ * All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_DRV_H__
+#define __AMDGPU_DRV_H__
+
+#include <linux/firmware.h>
+#include <linux/platform_device.h>
+
+#include "amd_shared.h"
+
+/* General customization:
+ */
+
+#define DRIVER_AUTHOR		"AMD linux driver team"
+
+#define DRIVER_NAME		"amdgpu"
+#define DRIVER_DESC		"AMD GPU"
+#define DRIVER_DATE		"20150101"
+
+long amdgpu_drm_ioctl(struct file *filp,
+		      unsigned int cmd, unsigned long arg);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_encoders.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_encoders.c
new file mode 100644
index 00000000000..ae8fac34f7a
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_encoders.c
@@ -0,0 +1,245 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <drm/drmP.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_connectors.h"
+#include "atom.h"
+#include "atombios_encoders.h"
+
+void
+amdgpu_link_encoder_connector(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	/* walk the list and link encoders to connectors */
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		amdgpu_connector = to_amdgpu_connector(connector);
+		list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+			amdgpu_encoder = to_amdgpu_encoder(encoder);
+			if (amdgpu_encoder->devices & amdgpu_connector->devices) {
+				drm_connector_attach_encoder(connector, encoder);
+				if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+					amdgpu_atombios_encoder_init_backlight(amdgpu_encoder, connector);
+					adev->mode_info.bl_encoder = amdgpu_encoder;
+				}
+			}
+		}
+	}
+}
+
+void amdgpu_encoder_set_active_device(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+			amdgpu_encoder->active_device = amdgpu_encoder->devices & amdgpu_connector->devices;
+			DRM_DEBUG_KMS("setting active device to %08x from %08x %08x for encoder %d\n",
+				  amdgpu_encoder->active_device, amdgpu_encoder->devices,
+				  amdgpu_connector->devices, encoder->encoder_type);
+		}
+	}
+}
+
+struct drm_connector *
+amdgpu_get_connector_for_encoder(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		amdgpu_connector = to_amdgpu_connector(connector);
+		if (amdgpu_encoder->active_device & amdgpu_connector->devices)
+			return connector;
+	}
+	return NULL;
+}
+
+struct drm_connector *
+amdgpu_get_connector_for_encoder_init(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		amdgpu_connector = to_amdgpu_connector(connector);
+		if (amdgpu_encoder->devices & amdgpu_connector->devices)
+			return connector;
+	}
+	return NULL;
+}
+
+struct drm_encoder *amdgpu_get_external_encoder(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_encoder *other_encoder;
+	struct amdgpu_encoder *other_amdgpu_encoder;
+
+	if (amdgpu_encoder->is_ext_encoder)
+		return NULL;
+
+	list_for_each_entry(other_encoder, &dev->mode_config.encoder_list, head) {
+		if (other_encoder == encoder)
+			continue;
+		other_amdgpu_encoder = to_amdgpu_encoder(other_encoder);
+		if (other_amdgpu_encoder->is_ext_encoder &&
+		    (amdgpu_encoder->devices & other_amdgpu_encoder->devices))
+			return other_encoder;
+	}
+	return NULL;
+}
+
+u16 amdgpu_encoder_get_dp_bridge_encoder_id(struct drm_encoder *encoder)
+{
+	struct drm_encoder *other_encoder = amdgpu_get_external_encoder(encoder);
+
+	if (other_encoder) {
+		struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(other_encoder);
+
+		switch (amdgpu_encoder->encoder_id) {
+		case ENCODER_OBJECT_ID_TRAVIS:
+		case ENCODER_OBJECT_ID_NUTMEG:
+			return amdgpu_encoder->encoder_id;
+		default:
+			return ENCODER_OBJECT_ID_NONE;
+		}
+	}
+	return ENCODER_OBJECT_ID_NONE;
+}
+
+void amdgpu_panel_mode_fixup(struct drm_encoder *encoder,
+			     struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+	unsigned hblank = native_mode->htotal - native_mode->hdisplay;
+	unsigned vblank = native_mode->vtotal - native_mode->vdisplay;
+	unsigned hover = native_mode->hsync_start - native_mode->hdisplay;
+	unsigned vover = native_mode->vsync_start - native_mode->vdisplay;
+	unsigned hsync_width = native_mode->hsync_end - native_mode->hsync_start;
+	unsigned vsync_width = native_mode->vsync_end - native_mode->vsync_start;
+
+	adjusted_mode->clock = native_mode->clock;
+	adjusted_mode->flags = native_mode->flags;
+
+	adjusted_mode->hdisplay = native_mode->hdisplay;
+	adjusted_mode->vdisplay = native_mode->vdisplay;
+
+	adjusted_mode->htotal = native_mode->hdisplay + hblank;
+	adjusted_mode->hsync_start = native_mode->hdisplay + hover;
+	adjusted_mode->hsync_end = adjusted_mode->hsync_start + hsync_width;
+
+	adjusted_mode->vtotal = native_mode->vdisplay + vblank;
+	adjusted_mode->vsync_start = native_mode->vdisplay + vover;
+	adjusted_mode->vsync_end = adjusted_mode->vsync_start + vsync_width;
+
+	drm_mode_set_crtcinfo(adjusted_mode, CRTC_INTERLACE_HALVE_V);
+
+	adjusted_mode->crtc_hdisplay = native_mode->hdisplay;
+	adjusted_mode->crtc_vdisplay = native_mode->vdisplay;
+
+	adjusted_mode->crtc_htotal = adjusted_mode->crtc_hdisplay + hblank;
+	adjusted_mode->crtc_hsync_start = adjusted_mode->crtc_hdisplay + hover;
+	adjusted_mode->crtc_hsync_end = adjusted_mode->crtc_hsync_start + hsync_width;
+
+	adjusted_mode->crtc_vtotal = adjusted_mode->crtc_vdisplay + vblank;
+	adjusted_mode->crtc_vsync_start = adjusted_mode->crtc_vdisplay + vover;
+	adjusted_mode->crtc_vsync_end = adjusted_mode->crtc_vsync_start + vsync_width;
+
+}
+
+bool amdgpu_dig_monitor_is_duallink(struct drm_encoder *encoder,
+				    u32 pixel_clock)
+{
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+	struct amdgpu_connector_atom_dig *dig_connector;
+
+	connector = amdgpu_get_connector_for_encoder(encoder);
+	/* if we don't have an active device yet, just use one of
+	 * the connectors tied to the encoder.
+	 */
+	if (!connector)
+		connector = amdgpu_get_connector_for_encoder_init(encoder);
+	amdgpu_connector = to_amdgpu_connector(connector);
+
+	switch (connector->connector_type) {
+	case DRM_MODE_CONNECTOR_DVII:
+	case DRM_MODE_CONNECTOR_HDMIB:
+		if (amdgpu_connector->use_digital) {
+			/* HDMI 1.3 supports up to 340 Mhz over single link */
+			if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+				if (pixel_clock > 340000)
+					return true;
+				else
+					return false;
+			} else {
+				if (pixel_clock > 165000)
+					return true;
+				else
+					return false;
+			}
+		} else
+			return false;
+	case DRM_MODE_CONNECTOR_DVID:
+	case DRM_MODE_CONNECTOR_HDMIA:
+	case DRM_MODE_CONNECTOR_DisplayPort:
+		dig_connector = amdgpu_connector->con_priv;
+		if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) ||
+		    (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP))
+			return false;
+		else {
+			/* HDMI 1.3 supports up to 340 Mhz over single link */
+			if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+				if (pixel_clock > 340000)
+					return true;
+				else
+					return false;
+			} else {
+				if (pixel_clock > 165000)
+					return true;
+				else
+					return false;
+			}
+		}
+	default:
+		return false;
+	}
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_fb.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_fb.c
new file mode 100644
index 00000000000..9d1c158f17e
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_fb.c
@@ -0,0 +1,472 @@
+/*
+ * Copyright © 2007 David Airlie
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ *     David Airlie
+ */
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/pm_runtime.h>
+
+#include <drm/drmP.h>
+#include <drm/drm_crtc.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "cikd.h"
+
+#include <drm/drm_fb_helper.h>
+
+#include <linux/vga_switcheroo.h>
+
+#include "amdgpu_display.h"
+
+/* object hierarchy -
+   this contains a helper + a amdgpu fb
+   the helper contains a pointer to amdgpu framebuffer baseclass.
+*/
+
+static int
+amdgpufb_open(struct fb_info *info, int user)
+{
+	struct amdgpu_fbdev *rfbdev = info->par;
+	struct amdgpu_device *adev = rfbdev->adev;
+	int ret = pm_runtime_get_sync(adev->ddev->dev);
+	if (ret < 0 && ret != -EACCES) {
+		pm_runtime_mark_last_busy(adev->ddev->dev);
+		pm_runtime_put_autosuspend(adev->ddev->dev);
+		return ret;
+	}
+	return 0;
+}
+
+static int
+amdgpufb_release(struct fb_info *info, int user)
+{
+	struct amdgpu_fbdev *rfbdev = info->par;
+	struct amdgpu_device *adev = rfbdev->adev;
+
+	pm_runtime_mark_last_busy(adev->ddev->dev);
+	pm_runtime_put_autosuspend(adev->ddev->dev);
+	return 0;
+}
+
+#ifdef notyet
+static struct fb_ops amdgpufb_ops = {
+	.owner = THIS_MODULE,
+	DRM_FB_HELPER_DEFAULT_OPS,
+	.fb_open = amdgpufb_open,
+	.fb_release = amdgpufb_release,
+	.fb_fillrect = drm_fb_helper_cfb_fillrect,
+	.fb_copyarea = drm_fb_helper_cfb_copyarea,
+	.fb_imageblit = drm_fb_helper_cfb_imageblit,
+};
+#endif
+
+void amdgpu_burner_cb(void *);
+
+int amdgpu_align_pitch(struct amdgpu_device *adev, int width, int cpp, bool tiled)
+{
+	int aligned = width;
+	int pitch_mask = 0;
+
+	switch (cpp) {
+	case 1:
+		pitch_mask = 255;
+		break;
+	case 2:
+		pitch_mask = 127;
+		break;
+	case 3:
+	case 4:
+		pitch_mask = 63;
+		break;
+	}
+
+	aligned += pitch_mask;
+	aligned &= ~pitch_mask;
+	return aligned * cpp;
+}
+
+static void amdgpufb_destroy_pinned_object(struct drm_gem_object *gobj)
+{
+	struct amdgpu_bo *abo = gem_to_amdgpu_bo(gobj);
+	int ret;
+
+	ret = amdgpu_bo_reserve(abo, true);
+	if (likely(ret == 0)) {
+		amdgpu_bo_kunmap(abo);
+		amdgpu_bo_unpin(abo);
+		amdgpu_bo_unreserve(abo);
+	}
+	drm_gem_object_put_unlocked(gobj);
+}
+
+static int amdgpufb_create_pinned_object(struct amdgpu_fbdev *rfbdev,
+					 struct drm_mode_fb_cmd2 *mode_cmd,
+					 struct drm_gem_object **gobj_p)
+{
+	struct amdgpu_device *adev = rfbdev->adev;
+	struct drm_gem_object *gobj = NULL;
+	struct amdgpu_bo *abo = NULL;
+	bool fb_tiled = false; /* useful for testing */
+	u32 tiling_flags = 0, domain;
+	int ret;
+	int aligned_size, size;
+	int height = mode_cmd->height;
+	u32 cpp;
+
+	cpp = drm_format_plane_cpp(mode_cmd->pixel_format, 0);
+
+	/* need to align pitch with crtc limits */
+	mode_cmd->pitches[0] = amdgpu_align_pitch(adev, mode_cmd->width, cpp,
+						  fb_tiled);
+	domain = amdgpu_display_supported_domains(adev);
+
+	height = roundup2(mode_cmd->height, 8);
+	size = mode_cmd->pitches[0] * height;
+	aligned_size = roundup2(size, PAGE_SIZE);
+	ret = amdgpu_gem_object_create(adev, aligned_size, 0, domain,
+				       AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
+				       AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
+				       AMDGPU_GEM_CREATE_VRAM_CLEARED,
+				       ttm_bo_type_kernel, NULL, &gobj);
+	if (ret) {
+		pr_err("failed to allocate framebuffer (%d)\n", aligned_size);
+		return -ENOMEM;
+	}
+	abo = gem_to_amdgpu_bo(gobj);
+
+	if (fb_tiled)
+		tiling_flags = AMDGPU_TILING_SET(ARRAY_MODE, GRPH_ARRAY_2D_TILED_THIN1);
+
+	ret = amdgpu_bo_reserve(abo, false);
+	if (unlikely(ret != 0))
+		goto out_unref;
+
+	if (tiling_flags) {
+		ret = amdgpu_bo_set_tiling_flags(abo,
+						 tiling_flags);
+		if (ret)
+			dev_err(adev->dev, "FB failed to set tiling flags\n");
+	}
+
+
+	ret = amdgpu_bo_pin(abo, domain);
+	if (ret) {
+		amdgpu_bo_unreserve(abo);
+		goto out_unref;
+	}
+
+	ret = amdgpu_ttm_alloc_gart(&abo->tbo);
+	if (ret) {
+		amdgpu_bo_unreserve(abo);
+		dev_err(adev->dev, "%p bind failed\n", abo);
+		goto out_unref;
+	}
+
+	ret = amdgpu_bo_kmap(abo, NULL);
+	amdgpu_bo_unreserve(abo);
+	if (ret) {
+		goto out_unref;
+	}
+
+	*gobj_p = gobj;
+	return 0;
+out_unref:
+	amdgpufb_destroy_pinned_object(gobj);
+	*gobj_p = NULL;
+	return ret;
+}
+
+static int amdgpufb_create(struct drm_fb_helper *helper,
+			   struct drm_fb_helper_surface_size *sizes)
+{
+	struct amdgpu_fbdev *rfbdev = (struct amdgpu_fbdev *)helper;
+	struct amdgpu_device *adev = rfbdev->adev;
+	struct fb_info *info;
+	struct rasops_info *ri = &adev->ro;
+	struct drm_framebuffer *fb = NULL;
+	struct drm_mode_fb_cmd2 mode_cmd;
+	struct drm_gem_object *gobj = NULL;
+	struct amdgpu_bo *abo = NULL;
+	int ret;
+	unsigned long tmp;
+
+	mode_cmd.width = sizes->surface_width;
+	mode_cmd.height = sizes->surface_height;
+
+	if (sizes->surface_bpp == 24)
+		sizes->surface_bpp = 32;
+
+	mode_cmd.pixel_format = drm_mode_legacy_fb_format(sizes->surface_bpp,
+							  sizes->surface_depth);
+
+	ret = amdgpufb_create_pinned_object(rfbdev, &mode_cmd, &gobj);
+	if (ret) {
+		DRM_ERROR("failed to create fbcon object %d\n", ret);
+		return ret;
+	}
+
+	abo = gem_to_amdgpu_bo(gobj);
+
+	/* okay we have an object now allocate the framebuffer */
+	info = drm_fb_helper_alloc_fbi(helper);
+	if (IS_ERR(info)) {
+		ret = PTR_ERR(info);
+		goto out;
+	}
+
+	info->par = rfbdev;
+	info->skip_vt_switch = true;
+
+	ret = amdgpu_display_framebuffer_init(adev->ddev, &rfbdev->rfb,
+					      &mode_cmd, gobj);
+	if (ret) {
+		DRM_ERROR("failed to initialize framebuffer %d\n", ret);
+		goto out;
+	}
+
+	fb = &rfbdev->rfb.base;
+
+	/* setup helper */
+	rfbdev->helper.fb = fb;
+
+#ifdef __linux__
+	strcpy(info->fix.id, "amdgpudrmfb");
+
+	drm_fb_helper_fill_fix(info, fb->pitches[0], fb->format->depth);
+
+	info->fbops = &amdgpufb_ops;
+#endif
+
+	tmp = amdgpu_bo_gpu_offset(abo) - adev->gmc.vram_start;
+#ifdef __linux__
+	info->fix.smem_start = adev->gmc.aper_base + tmp;
+	info->fix.smem_len = amdgpu_bo_size(abo);
+	info->screen_base = amdgpu_bo_kptr(abo);
+	info->screen_size = amdgpu_bo_size(abo);
+
+	drm_fb_helper_fill_var(info, &rfbdev->helper, sizes->fb_width, sizes->fb_height);
+
+	/* setup aperture base/size for vesafb takeover */
+	info->apertures->ranges[0].base = adev->ddev->mode_config.fb_base;
+	info->apertures->ranges[0].size = adev->gmc.aper_size;
+#endif
+
+	/* Use default scratch pixmap (info->pixmap.flags = FB_PIXMAP_SYSTEM) */
+
+#ifdef __linux__
+	if (info->screen_base == NULL) {
+		ret = -ENOSPC;
+		goto out;
+	}
+
+	DRM_INFO("fb mappable at 0x%lX\n",  info->fix.smem_start);
+#endif
+	DRM_INFO("vram apper at 0x%lX\n",  (unsigned long)adev->gmc.aper_base);
+	DRM_INFO("size %lu\n", (unsigned long)amdgpu_bo_size(abo));
+	DRM_INFO("fb depth is %d\n", fb->format->depth);
+	DRM_INFO("   pitch is %d\n", fb->pitches[0]);
+
+	ri->ri_bits = amdgpu_bo_kptr(abo);
+	ri->ri_depth = fb->format->cpp[0] * 8;
+	ri->ri_stride = fb->pitches[0];
+	ri->ri_width = sizes->fb_width;
+	ri->ri_height = sizes->fb_height;
+
+	switch (fb->format->format) {
+	case DRM_FORMAT_XRGB8888:
+		ri->ri_rnum = 8;
+		ri->ri_rpos = 16;
+		ri->ri_gnum = 8;
+		ri->ri_gpos = 8;
+		ri->ri_bnum = 8;
+		ri->ri_bpos = 0;
+		break;
+	case DRM_FORMAT_RGB565:
+		ri->ri_rnum = 5;
+		ri->ri_rpos = 11;
+		ri->ri_gnum = 6;
+		ri->ri_gpos = 5;
+		ri->ri_bnum = 5;
+		ri->ri_bpos = 0;
+		break;
+	}
+
+	vga_switcheroo_client_fb_set(adev->ddev->pdev, info);
+	return 0;
+
+out:
+	if (abo) {
+
+	}
+	if (fb && ret) {
+		drm_gem_object_put_unlocked(gobj);
+		drm_framebuffer_unregister_private(fb);
+		drm_framebuffer_cleanup(fb);
+		kfree(fb);
+	}
+	return ret;
+}
+
+static int amdgpu_fbdev_destroy(struct drm_device *dev, struct amdgpu_fbdev *rfbdev)
+{
+	struct amdgpu_framebuffer *rfb = &rfbdev->rfb;
+
+	drm_fb_helper_unregister_fbi(&rfbdev->helper);
+
+	if (rfb->base.obj[0]) {
+		amdgpufb_destroy_pinned_object(rfb->base.obj[0]);
+		rfb->base.obj[0] = NULL;
+		drm_framebuffer_unregister_private(&rfb->base);
+		drm_framebuffer_cleanup(&rfb->base);
+	}
+	drm_fb_helper_fini(&rfbdev->helper);
+
+	return 0;
+}
+
+static const struct drm_fb_helper_funcs amdgpu_fb_helper_funcs = {
+	.fb_probe = amdgpufb_create,
+};
+
+int amdgpu_fbdev_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_fbdev *rfbdev;
+	int bpp_sel = 32;
+	int ret;
+
+	/* don't init fbdev on hw without DCE */
+	if (!adev->mode_info.mode_config_initialized)
+		return 0;
+
+	/* don't init fbdev if there are no connectors */
+	if (list_empty(&adev->ddev->mode_config.connector_list))
+		return 0;
+
+	/* select 8 bpp console on low vram cards */
+	if (adev->gmc.real_vram_size <= (32*1024*1024))
+		bpp_sel = 8;
+
+	rfbdev = kzalloc(sizeof(struct amdgpu_fbdev), GFP_KERNEL);
+	if (!rfbdev)
+		return -ENOMEM;
+
+	rfbdev->adev = adev;
+	adev->mode_info.rfbdev = rfbdev;
+
+	drm_fb_helper_prepare(adev->ddev, &rfbdev->helper,
+			&amdgpu_fb_helper_funcs);
+
+	ret = drm_fb_helper_init(adev->ddev, &rfbdev->helper,
+				 AMDGPUFB_CONN_LIMIT);
+	if (ret) {
+		kfree(rfbdev);
+		return ret;
+	}
+
+	task_set(&adev->burner_task, amdgpu_burner_cb, adev);
+
+	drm_fb_helper_single_add_all_connectors(&rfbdev->helper);
+
+	/* disable all the possible outputs/crtcs before entering KMS mode */
+	if (!amdgpu_device_has_dc_support(adev))
+		drm_helper_disable_unused_functions(adev->ddev);
+
+	drm_fb_helper_initial_config(&rfbdev->helper, bpp_sel);
+	return 0;
+}
+
+void amdgpu_fbdev_fini(struct amdgpu_device *adev)
+{
+	if (!adev->mode_info.rfbdev)
+		return;
+
+	task_del(systq, &adev->burner_task);
+
+	amdgpu_fbdev_destroy(adev->ddev, adev->mode_info.rfbdev);
+	kfree(adev->mode_info.rfbdev);
+	adev->mode_info.rfbdev = NULL;
+}
+
+void amdgpu_fbdev_set_suspend(struct amdgpu_device *adev, int state)
+{
+	if (adev->mode_info.rfbdev)
+		drm_fb_helper_set_suspend_unlocked(&adev->mode_info.rfbdev->helper,
+						   state);
+}
+
+int amdgpu_fbdev_total_size(struct amdgpu_device *adev)
+{
+	struct amdgpu_bo *robj;
+	int size = 0;
+
+	if (!adev->mode_info.rfbdev)
+		return 0;
+
+	robj = gem_to_amdgpu_bo(adev->mode_info.rfbdev->rfb.base.obj[0]);
+	size += amdgpu_bo_size(robj);
+	return size;
+}
+
+bool amdgpu_fbdev_robj_is_fb(struct amdgpu_device *adev, struct amdgpu_bo *robj)
+{
+	if (!adev->mode_info.rfbdev)
+		return false;
+	if (robj == gem_to_amdgpu_bo(adev->mode_info.rfbdev->rfb.base.obj[0]))
+		return true;
+	return false;
+}
+
+void
+amdgpu_burner(void *v, u_int on, u_int flags)
+{
+	struct rasops_info *ri = v;
+	struct amdgpu_device *adev = ri->ri_hw;
+
+	task_del(systq, &adev->burner_task);
+
+	if (on)
+		adev->burner_fblank = FB_BLANK_UNBLANK;
+	else {
+		if (flags & WSDISPLAY_BURN_VBLANK)
+			adev->burner_fblank = FB_BLANK_VSYNC_SUSPEND;
+		else
+			adev->burner_fblank = FB_BLANK_NORMAL;
+	}
+
+	/*
+	 * Setting the DPMS mode may sleep while waiting for vblank so
+	 * hand things off to a taskq.
+	 */
+	task_add(systq, &adev->burner_task);
+}
+
+void
+amdgpu_burner_cb(void *arg1)
+{
+	struct amdgpu_device *adev = arg1;
+	struct drm_fb_helper *helper = &adev->mode_info.rfbdev->helper;
+
+	drm_fb_helper_blank(adev->burner_fblank, helper->fbdev);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_fence.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_fence.c
new file mode 100644
index 00000000000..fcb4dfacd26
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_fence.c
@@ -0,0 +1,750 @@
+/*
+ * Copyright 2009 Jerome Glisse.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Jerome Glisse <glisse@freedesktop.org>
+ *    Dave Airlie
+ */
+#include <linux/seq_file.h>
+#include <linux/atomic.h>
+#include <linux/wait.h>
+#include <linux/kref.h>
+#include <linux/slab.h>
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+/*
+ * Fences
+ * Fences mark an event in the GPUs pipeline and are used
+ * for GPU/CPU synchronization.  When the fence is written,
+ * it is expected that all buffers associated with that fence
+ * are no longer in use by the associated ring on the GPU and
+ * that the the relevant GPU caches have been flushed.
+ */
+
+struct amdgpu_fence {
+	struct dma_fence base;
+
+	/* RB, DMA, etc. */
+	struct amdgpu_ring		*ring;
+};
+
+static struct pool amdgpu_fence_slab;
+
+int amdgpu_fence_slab_init(void)
+{
+#ifdef __linux__
+	amdgpu_fence_slab = kmem_cache_create(
+		"amdgpu_fence", sizeof(struct amdgpu_fence), 0,
+		SLAB_HWCACHE_ALIGN, NULL);
+	if (!amdgpu_fence_slab)
+		return -ENOMEM;
+#else
+	pool_init(&amdgpu_fence_slab, sizeof(struct amdgpu_fence),
+	    0, IPL_TTY, 0, "amdgpu_fence", NULL);
+#endif
+	return 0;
+}
+
+void amdgpu_fence_slab_fini(void)
+{
+	rcu_barrier();
+#ifdef __linux__
+	kmem_cache_destroy(amdgpu_fence_slab);
+#else
+	pool_destroy(&amdgpu_fence_slab);
+#endif
+}
+/*
+ * Cast helper
+ */
+static const struct dma_fence_ops amdgpu_fence_ops;
+static inline struct amdgpu_fence *to_amdgpu_fence(struct dma_fence *f)
+{
+	struct amdgpu_fence *__f = container_of(f, struct amdgpu_fence, base);
+
+	if (__f->base.ops == &amdgpu_fence_ops)
+		return __f;
+
+	return NULL;
+}
+
+/**
+ * amdgpu_fence_write - write a fence value
+ *
+ * @ring: ring the fence is associated with
+ * @seq: sequence number to write
+ *
+ * Writes a fence value to memory (all asics).
+ */
+static void amdgpu_fence_write(struct amdgpu_ring *ring, u32 seq)
+{
+	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+
+	if (drv->cpu_addr)
+		*drv->cpu_addr = cpu_to_le32(seq);
+}
+
+/**
+ * amdgpu_fence_read - read a fence value
+ *
+ * @ring: ring the fence is associated with
+ *
+ * Reads a fence value from memory (all asics).
+ * Returns the value of the fence read from memory.
+ */
+static u32 amdgpu_fence_read(struct amdgpu_ring *ring)
+{
+	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+	u32 seq = 0;
+
+	if (drv->cpu_addr)
+		seq = le32_to_cpu(*drv->cpu_addr);
+	else
+		seq = atomic_read(&drv->last_seq);
+
+	return seq;
+}
+
+/**
+ * amdgpu_fence_emit - emit a fence on the requested ring
+ *
+ * @ring: ring the fence is associated with
+ * @f: resulting fence object
+ *
+ * Emits a fence command on the requested ring (all asics).
+ * Returns 0 on success, -ENOMEM on failure.
+ */
+int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f,
+		      unsigned flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_fence *fence;
+	struct dma_fence *old, **ptr;
+	uint32_t seq;
+
+#ifdef __linux__
+	fence = kmem_cache_alloc(amdgpu_fence_slab, GFP_KERNEL);
+#else
+	fence = pool_get(&amdgpu_fence_slab, PR_WAITOK);
+#endif
+	if (fence == NULL)
+		return -ENOMEM;
+
+	seq = ++ring->fence_drv.sync_seq;
+	fence->ring = ring;
+	dma_fence_init(&fence->base, &amdgpu_fence_ops,
+		       &ring->fence_drv.lock,
+		       adev->fence_context + ring->idx,
+		       seq);
+	amdgpu_ring_emit_fence(ring, ring->fence_drv.gpu_addr,
+			       seq, flags | AMDGPU_FENCE_FLAG_INT);
+
+	ptr = &ring->fence_drv.fences[seq & ring->fence_drv.num_fences_mask];
+	/* This function can't be called concurrently anyway, otherwise
+	 * emitting the fence would mess up the hardware ring buffer.
+	 */
+	old = rcu_dereference_protected(*ptr, 1);
+	if (old && !dma_fence_is_signaled(old)) {
+		DRM_INFO("rcu slot is busy\n");
+		dma_fence_wait(old, false);
+	}
+
+	rcu_assign_pointer(*ptr, dma_fence_get(&fence->base));
+
+	*f = &fence->base;
+
+	return 0;
+}
+
+/**
+ * amdgpu_fence_emit_polling - emit a fence on the requeste ring
+ *
+ * @ring: ring the fence is associated with
+ * @s: resulting sequence number
+ *
+ * Emits a fence command on the requested ring (all asics).
+ * Used For polling fence.
+ * Returns 0 on success, -ENOMEM on failure.
+ */
+int amdgpu_fence_emit_polling(struct amdgpu_ring *ring, uint32_t *s)
+{
+	uint32_t seq;
+
+	if (!s)
+		return -EINVAL;
+
+	seq = ++ring->fence_drv.sync_seq;
+	amdgpu_ring_emit_fence(ring, ring->fence_drv.gpu_addr,
+			       seq, 0);
+
+	*s = seq;
+
+	return 0;
+}
+
+/**
+ * amdgpu_fence_schedule_fallback - schedule fallback check
+ *
+ * @ring: pointer to struct amdgpu_ring
+ *
+ * Start a timer as fallback to our interrupts.
+ */
+static void amdgpu_fence_schedule_fallback(struct amdgpu_ring *ring)
+{
+	mod_timer(&ring->fence_drv.fallback_timer,
+		  jiffies + AMDGPU_FENCE_JIFFIES_TIMEOUT);
+}
+
+/**
+ * amdgpu_fence_process - check for fence activity
+ *
+ * @ring: pointer to struct amdgpu_ring
+ *
+ * Checks the current fence value and calculates the last
+ * signalled fence value. Wakes the fence queue if the
+ * sequence number has increased.
+ */
+void amdgpu_fence_process(struct amdgpu_ring *ring)
+{
+	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+	uint32_t seq, last_seq;
+	int r;
+
+	do {
+		last_seq = atomic_read(&ring->fence_drv.last_seq);
+		seq = amdgpu_fence_read(ring);
+
+	} while (atomic_cmpxchg(&drv->last_seq, last_seq, seq) != last_seq);
+
+	if (seq != ring->fence_drv.sync_seq)
+		amdgpu_fence_schedule_fallback(ring);
+
+	if (unlikely(seq == last_seq))
+		return;
+
+	last_seq &= drv->num_fences_mask;
+	seq &= drv->num_fences_mask;
+
+	do {
+		struct dma_fence *fence, **ptr;
+
+		++last_seq;
+		last_seq &= drv->num_fences_mask;
+		ptr = &drv->fences[last_seq];
+
+		/* There is always exactly one thread signaling this fence slot */
+		fence = rcu_dereference_protected(*ptr, 1);
+		RCU_INIT_POINTER(*ptr, NULL);
+
+		if (!fence)
+			continue;
+
+		r = dma_fence_signal(fence);
+		if (!r)
+			DMA_FENCE_TRACE(fence, "signaled from irq context\n");
+		else
+			BUG();
+
+		dma_fence_put(fence);
+	} while (last_seq != seq);
+}
+
+/**
+ * amdgpu_fence_fallback - fallback for hardware interrupts
+ *
+ * @work: delayed work item
+ *
+ * Checks for fence activity.
+ */
+static void amdgpu_fence_fallback(void *arg)
+{
+	struct amdgpu_ring *ring = arg;
+
+	amdgpu_fence_process(ring);
+}
+
+/**
+ * amdgpu_fence_wait_empty - wait for all fences to signal
+ *
+ * @adev: amdgpu device pointer
+ * @ring: ring index the fence is associated with
+ *
+ * Wait for all fences on the requested ring to signal (all asics).
+ * Returns 0 if the fences have passed, error for all other cases.
+ */
+int amdgpu_fence_wait_empty(struct amdgpu_ring *ring)
+{
+	uint64_t seq = READ_ONCE(ring->fence_drv.sync_seq);
+	struct dma_fence *fence, **ptr;
+	int r;
+
+	if (!seq)
+		return 0;
+
+	ptr = &ring->fence_drv.fences[seq & ring->fence_drv.num_fences_mask];
+	rcu_read_lock();
+	fence = rcu_dereference(*ptr);
+	if (!fence || !dma_fence_get_rcu(fence)) {
+		rcu_read_unlock();
+		return 0;
+	}
+	rcu_read_unlock();
+
+	r = dma_fence_wait(fence, false);
+	dma_fence_put(fence);
+	return r;
+}
+
+/**
+ * amdgpu_fence_wait_polling - busy wait for givn sequence number
+ *
+ * @ring: ring index the fence is associated with
+ * @wait_seq: sequence number to wait
+ * @timeout: the timeout for waiting in usecs
+ *
+ * Wait for all fences on the requested ring to signal (all asics).
+ * Returns left time if no timeout, 0 or minus if timeout.
+ */
+signed long amdgpu_fence_wait_polling(struct amdgpu_ring *ring,
+				      uint32_t wait_seq,
+				      signed long timeout)
+{
+	uint32_t seq;
+
+	do {
+		seq = amdgpu_fence_read(ring);
+		udelay(5);
+		timeout -= 5;
+	} while ((int32_t)(wait_seq - seq) > 0 && timeout > 0);
+
+	return timeout > 0 ? timeout : 0;
+}
+/**
+ * amdgpu_fence_count_emitted - get the count of emitted fences
+ *
+ * @ring: ring the fence is associated with
+ *
+ * Get the number of fences emitted on the requested ring (all asics).
+ * Returns the number of emitted fences on the ring.  Used by the
+ * dynpm code to ring track activity.
+ */
+unsigned amdgpu_fence_count_emitted(struct amdgpu_ring *ring)
+{
+	uint64_t emitted;
+
+	/* We are not protected by ring lock when reading the last sequence
+	 * but it's ok to report slightly wrong fence count here.
+	 */
+	amdgpu_fence_process(ring);
+	emitted = 0x100000000ull;
+	emitted -= atomic_read(&ring->fence_drv.last_seq);
+	emitted += READ_ONCE(ring->fence_drv.sync_seq);
+	return lower_32_bits(emitted);
+}
+
+/**
+ * amdgpu_fence_driver_start_ring - make the fence driver
+ * ready for use on the requested ring.
+ *
+ * @ring: ring to start the fence driver on
+ * @irq_src: interrupt source to use for this ring
+ * @irq_type: interrupt type to use for this ring
+ *
+ * Make the fence driver ready for processing (all asics).
+ * Not all asics have all rings, so each asic will only
+ * start the fence driver on the rings it has.
+ * Returns 0 for success, errors for failure.
+ */
+int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring,
+				   struct amdgpu_irq_src *irq_src,
+				   unsigned irq_type)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint64_t index;
+
+	if (ring->funcs->type != AMDGPU_RING_TYPE_UVD) {
+		ring->fence_drv.cpu_addr = &adev->wb.wb[ring->fence_offs];
+		ring->fence_drv.gpu_addr = adev->wb.gpu_addr + (ring->fence_offs * 4);
+	} else {
+		/* put fence directly behind firmware */
+		index = roundup2(adev->uvd.fw->size, 8);
+		ring->fence_drv.cpu_addr = adev->uvd.inst[ring->me].cpu_addr + index;
+		ring->fence_drv.gpu_addr = adev->uvd.inst[ring->me].gpu_addr + index;
+	}
+	amdgpu_fence_write(ring, atomic_read(&ring->fence_drv.last_seq));
+	amdgpu_irq_get(adev, irq_src, irq_type);
+
+	ring->fence_drv.irq_src = irq_src;
+	ring->fence_drv.irq_type = irq_type;
+	ring->fence_drv.initialized = true;
+
+	dev_dbg(adev->dev, "fence driver on ring %d use gpu addr 0x%016llx, "
+		"cpu addr 0x%p\n", ring->idx,
+		ring->fence_drv.gpu_addr, ring->fence_drv.cpu_addr);
+	return 0;
+}
+
+/**
+ * amdgpu_fence_driver_init_ring - init the fence driver
+ * for the requested ring.
+ *
+ * @ring: ring to init the fence driver on
+ * @num_hw_submission: number of entries on the hardware queue
+ *
+ * Init the fence driver for the requested ring (all asics).
+ * Helper function for amdgpu_fence_driver_init().
+ */
+int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring,
+				  unsigned num_hw_submission)
+{
+	long timeout;
+	int r;
+
+	/* Check that num_hw_submission is a power of two */
+	if ((num_hw_submission & (num_hw_submission - 1)) != 0)
+		return -EINVAL;
+
+	ring->fence_drv.cpu_addr = NULL;
+	ring->fence_drv.gpu_addr = 0;
+	ring->fence_drv.sync_seq = 0;
+	atomic_set(&ring->fence_drv.last_seq, 0);
+	ring->fence_drv.initialized = false;
+
+#ifdef __linux__
+	timer_setup(&ring->fence_drv.fallback_timer, amdgpu_fence_fallback, 0);
+#else
+	timeout_set(&ring->fence_drv.fallback_timer, amdgpu_fence_fallback,
+	    ring);
+#endif
+
+	ring->fence_drv.num_fences_mask = num_hw_submission * 2 - 1;
+	mtx_init(&ring->fence_drv.lock, IPL_TTY);
+	ring->fence_drv.fences = kcalloc(num_hw_submission * 2, sizeof(void *),
+					 GFP_KERNEL);
+	if (!ring->fence_drv.fences)
+		return -ENOMEM;
+
+	/* No need to setup the GPU scheduler for KIQ ring */
+	if (ring->funcs->type != AMDGPU_RING_TYPE_KIQ) {
+		/* for non-sriov case, no timeout enforce on compute ring */
+		if ((ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE)
+				&& !amdgpu_sriov_vf(ring->adev))
+			timeout = MAX_SCHEDULE_TIMEOUT;
+		else
+			timeout = msecs_to_jiffies(amdgpu_lockup_timeout);
+
+		r = drm_sched_init(&ring->sched, &amdgpu_sched_ops,
+				   num_hw_submission, amdgpu_job_hang_limit,
+				   timeout, ring->name);
+		if (r) {
+			DRM_ERROR("Failed to create scheduler on ring %s.\n",
+				  ring->name);
+			return r;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_fence_driver_init - init the fence driver
+ * for all possible rings.
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Init the fence driver for all possible rings (all asics).
+ * Not all asics have all rings, so each asic will only
+ * start the fence driver on the rings it has using
+ * amdgpu_fence_driver_start_ring().
+ * Returns 0 for success.
+ */
+int amdgpu_fence_driver_init(struct amdgpu_device *adev)
+{
+	if (amdgpu_debugfs_fence_init(adev))
+		dev_err(adev->dev, "fence debugfs file creation failed\n");
+
+	return 0;
+}
+
+/**
+ * amdgpu_fence_driver_fini - tear down the fence driver
+ * for all possible rings.
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Tear down the fence driver for all possible rings (all asics).
+ */
+void amdgpu_fence_driver_fini(struct amdgpu_device *adev)
+{
+	unsigned i, j;
+	int r;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!ring || !ring->fence_drv.initialized)
+			continue;
+		r = amdgpu_fence_wait_empty(ring);
+		if (r) {
+			/* no need to trigger GPU reset as we are unloading */
+			amdgpu_fence_driver_force_completion(ring);
+		}
+		amdgpu_irq_put(adev, ring->fence_drv.irq_src,
+			       ring->fence_drv.irq_type);
+		drm_sched_fini(&ring->sched);
+		del_timer_sync(&ring->fence_drv.fallback_timer);
+		for (j = 0; j <= ring->fence_drv.num_fences_mask; ++j)
+			dma_fence_put(ring->fence_drv.fences[j]);
+		kfree(ring->fence_drv.fences);
+		ring->fence_drv.fences = NULL;
+		ring->fence_drv.initialized = false;
+	}
+}
+
+/**
+ * amdgpu_fence_driver_suspend - suspend the fence driver
+ * for all possible rings.
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Suspend the fence driver for all possible rings (all asics).
+ */
+void amdgpu_fence_driver_suspend(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		if (!ring || !ring->fence_drv.initialized)
+			continue;
+
+		/* wait for gpu to finish processing current batch */
+		r = amdgpu_fence_wait_empty(ring);
+		if (r) {
+			/* delay GPU reset to resume */
+			amdgpu_fence_driver_force_completion(ring);
+		}
+
+		/* disable the interrupt */
+		amdgpu_irq_put(adev, ring->fence_drv.irq_src,
+			       ring->fence_drv.irq_type);
+	}
+}
+
+/**
+ * amdgpu_fence_driver_resume - resume the fence driver
+ * for all possible rings.
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Resume the fence driver for all possible rings (all asics).
+ * Not all asics have all rings, so each asic will only
+ * start the fence driver on the rings it has using
+ * amdgpu_fence_driver_start_ring().
+ * Returns 0 for success.
+ */
+void amdgpu_fence_driver_resume(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		if (!ring || !ring->fence_drv.initialized)
+			continue;
+
+		/* enable the interrupt */
+		amdgpu_irq_get(adev, ring->fence_drv.irq_src,
+			       ring->fence_drv.irq_type);
+	}
+}
+
+/**
+ * amdgpu_fence_driver_force_completion - force signal latest fence of ring
+ *
+ * @ring: fence of the ring to signal
+ *
+ */
+void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring)
+{
+	amdgpu_fence_write(ring, ring->fence_drv.sync_seq);
+	amdgpu_fence_process(ring);
+}
+
+/*
+ * Common fence implementation
+ */
+
+static const char *amdgpu_fence_get_driver_name(struct dma_fence *fence)
+{
+	return "amdgpu";
+}
+
+static const char *amdgpu_fence_get_timeline_name(struct dma_fence *f)
+{
+	struct amdgpu_fence *fence = to_amdgpu_fence(f);
+	return (const char *)fence->ring->name;
+}
+
+/**
+ * amdgpu_fence_enable_signaling - enable signalling on fence
+ * @fence: fence
+ *
+ * This function is called with fence_queue lock held, and adds a callback
+ * to fence_queue that checks if this fence is signaled, and if so it
+ * signals the fence and removes itself.
+ */
+static bool amdgpu_fence_enable_signaling(struct dma_fence *f)
+{
+	struct amdgpu_fence *fence = to_amdgpu_fence(f);
+	struct amdgpu_ring *ring = fence->ring;
+
+	if (!timer_pending(&ring->fence_drv.fallback_timer))
+		amdgpu_fence_schedule_fallback(ring);
+
+	DMA_FENCE_TRACE(&fence->base, "armed on ring %i!\n", ring->idx);
+
+	return true;
+}
+
+/**
+ * amdgpu_fence_free - free up the fence memory
+ *
+ * @rcu: RCU callback head
+ *
+ * Free up the fence memory after the RCU grace period.
+ */
+static void amdgpu_fence_free(struct rcu_head *rcu)
+{
+	struct dma_fence *f = container_of(rcu, struct dma_fence, rcu);
+	struct amdgpu_fence *fence = to_amdgpu_fence(f);
+#ifdef __linux__
+	kmem_cache_free(amdgpu_fence_slab, fence);
+#else
+	pool_put(&amdgpu_fence_slab, fence);
+#endif
+}
+
+/**
+ * amdgpu_fence_release - callback that fence can be freed
+ *
+ * @fence: fence
+ *
+ * This function is called when the reference count becomes zero.
+ * It just RCU schedules freeing up the fence.
+ */
+static void amdgpu_fence_release(struct dma_fence *f)
+{
+	call_rcu(&f->rcu, amdgpu_fence_free);
+}
+
+static const struct dma_fence_ops amdgpu_fence_ops = {
+	.get_driver_name = amdgpu_fence_get_driver_name,
+	.get_timeline_name = amdgpu_fence_get_timeline_name,
+	.enable_signaling = amdgpu_fence_enable_signaling,
+	.release = amdgpu_fence_release,
+};
+
+/*
+ * Fence debugfs
+ */
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_debugfs_fence_info(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *)m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int i;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		if (!ring || !ring->fence_drv.initialized)
+			continue;
+
+		amdgpu_fence_process(ring);
+
+		seq_printf(m, "--- ring %d (%s) ---\n", i, ring->name);
+		seq_printf(m, "Last signaled fence 0x%08x\n",
+			   atomic_read(&ring->fence_drv.last_seq));
+		seq_printf(m, "Last emitted        0x%08x\n",
+			   ring->fence_drv.sync_seq);
+
+		if (ring->funcs->type != AMDGPU_RING_TYPE_GFX)
+			continue;
+
+		/* set in CP_VMID_PREEMPT and preemption occurred */
+		seq_printf(m, "Last preempted      0x%08x\n",
+			   le32_to_cpu(*(ring->fence_drv.cpu_addr + 2)));
+		/* set in CP_VMID_RESET and reset occurred */
+		seq_printf(m, "Last reset          0x%08x\n",
+			   le32_to_cpu(*(ring->fence_drv.cpu_addr + 4)));
+		/* Both preemption and reset occurred */
+		seq_printf(m, "Last both           0x%08x\n",
+			   le32_to_cpu(*(ring->fence_drv.cpu_addr + 6)));
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_debugfs_gpu_recover - manually trigger a gpu reset & recover
+ *
+ * Manually trigger a gpu reset at the next fence wait.
+ */
+static int amdgpu_debugfs_gpu_recover(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	seq_printf(m, "gpu recover\n");
+	amdgpu_device_gpu_recover(adev, NULL, true);
+
+	return 0;
+}
+
+static const struct drm_info_list amdgpu_debugfs_fence_list[] = {
+	{"amdgpu_fence_info", &amdgpu_debugfs_fence_info, 0, NULL},
+	{"amdgpu_gpu_recover", &amdgpu_debugfs_gpu_recover, 0, NULL}
+};
+
+static const struct drm_info_list amdgpu_debugfs_fence_list_sriov[] = {
+	{"amdgpu_fence_info", &amdgpu_debugfs_fence_info, 0, NULL},
+};
+#endif
+
+int amdgpu_debugfs_fence_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	if (amdgpu_sriov_vf(adev))
+		return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_fence_list_sriov, 1);
+	return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_fence_list, 2);
+#else
+	return 0;
+#endif
+}
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_gart.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gart.c
new file mode 100644
index 00000000000..762a51193f1
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gart.c
@@ -0,0 +1,393 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#ifdef CONFIG_X86
+#include <asm/set_memory.h>
+#endif
+#include "amdgpu.h"
+
+/*
+ * GART
+ * The GART (Graphics Aperture Remapping Table) is an aperture
+ * in the GPU's address space.  System pages can be mapped into
+ * the aperture and look like contiguous pages from the GPU's
+ * perspective.  A page table maps the pages in the aperture
+ * to the actual backing pages in system memory.
+ *
+ * Radeon GPUs support both an internal GART, as described above,
+ * and AGP.  AGP works similarly, but the GART table is configured
+ * and maintained by the northbridge rather than the driver.
+ * Radeon hw has a separate AGP aperture that is programmed to
+ * point to the AGP aperture provided by the northbridge and the
+ * requests are passed through to the northbridge aperture.
+ * Both AGP and internal GART can be used at the same time, however
+ * that is not currently supported by the driver.
+ *
+ * This file handles the common internal GART management.
+ */
+
+/*
+ * Common GART table functions.
+ */
+
+/**
+ * amdgpu_dummy_page_init - init dummy page used by the driver
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate the dummy page used by the driver (all asics).
+ * This dummy page is used by the driver as a filler for gart entries
+ * when pages are taken out of the GART
+ * Returns 0 on sucess, -ENOMEM on failure.
+ */
+static int amdgpu_gart_dummy_page_init(struct amdgpu_device *adev)
+{
+	struct vm_page *dummy_page = adev->mman.bdev.glob->dummy_read_page;
+
+	if (adev->dummy_page_addr)
+		return 0;
+	adev->dummy_page_addr = pci_map_page(adev->pdev, dummy_page, 0,
+					     PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
+	if (pci_dma_mapping_error(adev->pdev, adev->dummy_page_addr)) {
+		dev_err(&adev->pdev->dev, "Failed to DMA MAP the dummy page\n");
+		adev->dummy_page_addr = 0;
+		return -ENOMEM;
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_dummy_page_fini - free dummy page used by the driver
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Frees the dummy page used by the driver (all asics).
+ */
+static void amdgpu_gart_dummy_page_fini(struct amdgpu_device *adev)
+{
+	if (!adev->dummy_page_addr)
+		return;
+	pci_unmap_page(adev->pdev, adev->dummy_page_addr,
+		       PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
+	adev->dummy_page_addr = 0;
+}
+
+/**
+ * amdgpu_gart_table_vram_alloc - allocate vram for gart page table
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate video memory for GART page table
+ * (pcie r4xx, r5xx+).  These asics require the
+ * gart table to be in video memory.
+ * Returns 0 for success, error for failure.
+ */
+int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.robj == NULL) {
+		struct amdgpu_bo_param bp;
+
+		memset(&bp, 0, sizeof(bp));
+		bp.size = adev->gart.table_size;
+		bp.byte_align = PAGE_SIZE;
+		bp.domain = AMDGPU_GEM_DOMAIN_VRAM;
+		bp.flags = AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
+			AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+		bp.type = ttm_bo_type_kernel;
+		bp.resv = NULL;
+		r = amdgpu_bo_create(adev, &bp, &adev->gart.robj);
+		if (r) {
+			return r;
+		}
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_gart_table_vram_pin - pin gart page table in vram
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Pin the GART page table in vram so it will not be moved
+ * by the memory manager (pcie r4xx, r5xx+).  These asics require the
+ * gart table to be in video memory.
+ * Returns 0 for success, error for failure.
+ */
+int amdgpu_gart_table_vram_pin(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_bo_reserve(adev->gart.robj, false);
+	if (unlikely(r != 0))
+		return r;
+	r = amdgpu_bo_pin(adev->gart.robj, AMDGPU_GEM_DOMAIN_VRAM);
+	if (r) {
+		amdgpu_bo_unreserve(adev->gart.robj);
+		return r;
+	}
+	r = amdgpu_bo_kmap(adev->gart.robj, &adev->gart.ptr);
+	if (r)
+		amdgpu_bo_unpin(adev->gart.robj);
+	amdgpu_bo_unreserve(adev->gart.robj);
+	adev->gart.table_addr = amdgpu_bo_gpu_offset(adev->gart.robj);
+	return r;
+}
+
+/**
+ * amdgpu_gart_table_vram_unpin - unpin gart page table in vram
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Unpin the GART page table in vram (pcie r4xx, r5xx+).
+ * These asics require the gart table to be in video memory.
+ */
+void amdgpu_gart_table_vram_unpin(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.robj == NULL) {
+		return;
+	}
+	r = amdgpu_bo_reserve(adev->gart.robj, true);
+	if (likely(r == 0)) {
+		amdgpu_bo_kunmap(adev->gart.robj);
+		amdgpu_bo_unpin(adev->gart.robj);
+		amdgpu_bo_unreserve(adev->gart.robj);
+		adev->gart.ptr = NULL;
+	}
+}
+
+/**
+ * amdgpu_gart_table_vram_free - free gart page table vram
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Free the video memory used for the GART page table
+ * (pcie r4xx, r5xx+).  These asics require the gart table to
+ * be in video memory.
+ */
+void amdgpu_gart_table_vram_free(struct amdgpu_device *adev)
+{
+	if (adev->gart.robj == NULL) {
+		return;
+	}
+	amdgpu_bo_unref(&adev->gart.robj);
+}
+
+/*
+ * Common gart functions.
+ */
+/**
+ * amdgpu_gart_unbind - unbind pages from the gart page table
+ *
+ * @adev: amdgpu_device pointer
+ * @offset: offset into the GPU's gart aperture
+ * @pages: number of pages to unbind
+ *
+ * Unbinds the requested pages from the gart page table and
+ * replaces them with the dummy page (all asics).
+ * Returns 0 for success, -EINVAL for failure.
+ */
+int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
+			int pages)
+{
+	unsigned t;
+	unsigned p;
+	int i, j;
+	u64 page_base;
+	/* Starting from VEGA10, system bit must be 0 to mean invalid. */
+	uint64_t flags = 0;
+
+	if (!adev->gart.ready) {
+		WARN(1, "trying to unbind memory from uninitialized GART !\n");
+		return -EINVAL;
+	}
+
+	t = offset / AMDGPU_GPU_PAGE_SIZE;
+	p = t / AMDGPU_GPU_PAGES_IN_CPU_PAGE;
+	for (i = 0; i < pages; i++, p++) {
+#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
+		adev->gart.pages[p] = NULL;
+#endif
+		page_base = adev->dummy_page_addr;
+		if (!adev->gart.ptr)
+			continue;
+
+		for (j = 0; j < AMDGPU_GPU_PAGES_IN_CPU_PAGE; j++, t++) {
+			amdgpu_gmc_set_pte_pde(adev, adev->gart.ptr,
+					       t, page_base, flags);
+			page_base += AMDGPU_GPU_PAGE_SIZE;
+		}
+	}
+	mb();
+	amdgpu_asic_flush_hdp(adev, NULL);
+	amdgpu_gmc_flush_gpu_tlb(adev, 0);
+	return 0;
+}
+
+/**
+ * amdgpu_gart_map - map dma_addresses into GART entries
+ *
+ * @adev: amdgpu_device pointer
+ * @offset: offset into the GPU's gart aperture
+ * @pages: number of pages to bind
+ * @dma_addr: DMA addresses of pages
+ *
+ * Map the dma_addresses into GART entries (all asics).
+ * Returns 0 for success, -EINVAL for failure.
+ */
+int amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset,
+		    int pages, dma_addr_t *dma_addr, uint64_t flags,
+		    void *dst)
+{
+	uint64_t page_base;
+	unsigned i, j, t;
+
+	if (!adev->gart.ready) {
+		WARN(1, "trying to bind memory to uninitialized GART !\n");
+		return -EINVAL;
+	}
+
+	t = offset / AMDGPU_GPU_PAGE_SIZE;
+
+	for (i = 0; i < pages; i++) {
+		page_base = dma_addr[i];
+		for (j = 0; j < AMDGPU_GPU_PAGES_IN_CPU_PAGE; j++, t++) {
+			amdgpu_gmc_set_pte_pde(adev, dst, t, page_base, flags);
+			page_base += AMDGPU_GPU_PAGE_SIZE;
+		}
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_gart_bind - bind pages into the gart page table
+ *
+ * @adev: amdgpu_device pointer
+ * @offset: offset into the GPU's gart aperture
+ * @pages: number of pages to bind
+ * @pagelist: pages to bind
+ * @dma_addr: DMA addresses of pages
+ *
+ * Binds the requested pages to the gart page table
+ * (all asics).
+ * Returns 0 for success, -EINVAL for failure.
+ */
+int amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset,
+		     int pages, struct vm_page **pagelist, dma_addr_t *dma_addr,
+		     uint64_t flags)
+{
+#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
+	unsigned i,t,p;
+#endif
+	int r;
+
+	if (!adev->gart.ready) {
+		WARN(1, "trying to bind memory to uninitialized GART !\n");
+		return -EINVAL;
+	}
+
+#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
+	t = offset / AMDGPU_GPU_PAGE_SIZE;
+	p = t / AMDGPU_GPU_PAGES_IN_CPU_PAGE;
+	for (i = 0; i < pages; i++, p++)
+		adev->gart.pages[p] = pagelist ? pagelist[i] : NULL;
+#endif
+
+	if (!adev->gart.ptr)
+		return 0;
+
+	r = amdgpu_gart_map(adev, offset, pages, dma_addr, flags,
+		    adev->gart.ptr);
+	if (r)
+		return r;
+
+	mb();
+	amdgpu_asic_flush_hdp(adev, NULL);
+	amdgpu_gmc_flush_gpu_tlb(adev, 0);
+	return 0;
+}
+
+/**
+ * amdgpu_gart_init - init the driver info for managing the gart
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate the dummy page and init the gart driver info (all asics).
+ * Returns 0 for success, error for failure.
+ */
+int amdgpu_gart_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->dummy_page_addr)
+		return 0;
+
+	/* We need PAGE_SIZE >= AMDGPU_GPU_PAGE_SIZE */
+	if (PAGE_SIZE < AMDGPU_GPU_PAGE_SIZE) {
+		DRM_ERROR("Page size is smaller than GPU page size!\n");
+		return -EINVAL;
+	}
+	r = amdgpu_gart_dummy_page_init(adev);
+	if (r)
+		return r;
+	/* Compute table size */
+	adev->gart.num_cpu_pages = adev->gmc.gart_size / PAGE_SIZE;
+	adev->gart.num_gpu_pages = adev->gmc.gart_size / AMDGPU_GPU_PAGE_SIZE;
+	DRM_INFO("GART: num cpu pages %u, num gpu pages %u\n",
+		 adev->gart.num_cpu_pages, adev->gart.num_gpu_pages);
+
+#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
+	/* Allocate pages table */
+	adev->gart.pages = vzalloc(array_size(sizeof(void *),
+					      adev->gart.num_cpu_pages));
+	if (adev->gart.pages == NULL)
+		return -ENOMEM;
+#endif
+
+	return 0;
+}
+
+/**
+ * amdgpu_gart_fini - tear down the driver info for managing the gart
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the gart driver info and free the dummy page (all asics).
+ */
+void amdgpu_gart_fini(struct amdgpu_device *adev)
+{
+#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
+	vfree(adev->gart.pages);
+	adev->gart.pages = NULL;
+#endif
+	amdgpu_gart_dummy_page_fini(adev);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_gart.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gart.h
new file mode 100644
index 00000000000..927044062b6
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gart.h
@@ -0,0 +1,73 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_GART_H__
+#define __AMDGPU_GART_H__
+
+#include <linux/types.h>
+
+/*
+ * GART structures, functions & helpers
+ */
+struct amdgpu_device;
+struct amdgpu_bo;
+
+#define AMDGPU_GPU_PAGE_SIZE 4096
+#define AMDGPU_GPU_PAGE_MASK (AMDGPU_GPU_PAGE_SIZE - 1)
+#define AMDGPU_GPU_PAGE_SHIFT 12
+#define AMDGPU_GPU_PAGE_ALIGN(a) (((a) + AMDGPU_GPU_PAGE_MASK) & ~AMDGPU_GPU_PAGE_MASK)
+
+#define AMDGPU_GPU_PAGES_IN_CPU_PAGE (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE)
+
+struct amdgpu_gart {
+	u64				table_addr;
+	struct amdgpu_bo		*robj;
+	void				*ptr;
+	unsigned			num_gpu_pages;
+	unsigned			num_cpu_pages;
+	unsigned			table_size;
+#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
+	struct vm_page			**pages;
+#endif
+	bool				ready;
+
+	/* Asic default pte flags */
+	uint64_t			gart_pte_flags;
+};
+
+int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev);
+void amdgpu_gart_table_vram_free(struct amdgpu_device *adev);
+int amdgpu_gart_table_vram_pin(struct amdgpu_device *adev);
+void amdgpu_gart_table_vram_unpin(struct amdgpu_device *adev);
+int amdgpu_gart_init(struct amdgpu_device *adev);
+void amdgpu_gart_fini(struct amdgpu_device *adev);
+int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
+		       int pages);
+int amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset,
+		    int pages, dma_addr_t *dma_addr, uint64_t flags,
+		    void *dst);
+int amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset,
+		     int pages, struct vm_page **pagelist,
+		     dma_addr_t *dma_addr, uint64_t flags);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_gds.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gds.h
new file mode 100644
index 00000000000..e73728d9038
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gds.h
@@ -0,0 +1,64 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_GDS_H__
+#define __AMDGPU_GDS_H__
+
+/* Because TTM request that alloacted buffer should be PAGE_SIZE aligned,
+ * we should report GDS/GWS/OA size as PAGE_SIZE aligned
+ * */
+#define AMDGPU_GDS_SHIFT	2
+#define AMDGPU_GWS_SHIFT	PAGE_SHIFT
+#define AMDGPU_OA_SHIFT		PAGE_SHIFT
+
+struct amdgpu_ring;
+struct amdgpu_bo;
+
+struct amdgpu_gds_asic_info {
+	uint32_t	total_size;
+	uint32_t	gfx_partition_size;
+	uint32_t	cs_partition_size;
+};
+
+struct amdgpu_gds {
+	struct amdgpu_gds_asic_info	mem;
+	struct amdgpu_gds_asic_info	gws;
+	struct amdgpu_gds_asic_info	oa;
+	/* At present, GDS, GWS and OA resources for gfx (graphics)
+	 * is always pre-allocated and available for graphics operation.
+	 * Such resource is shared between all gfx clients.
+	 * TODO: move this operation to user space
+	 * */
+	struct amdgpu_bo*		gds_gfx_bo;
+	struct amdgpu_bo*		gws_gfx_bo;
+	struct amdgpu_bo*		oa_gfx_bo;
+};
+
+struct amdgpu_gds_reg_offset {
+	uint32_t	mem_base;
+	uint32_t	mem_size;
+	uint32_t	gws;
+	uint32_t	oa;
+};
+
+#endif /* __AMDGPU_GDS_H__ */
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_gem.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gem.c
new file mode 100644
index 00000000000..c8189306b59
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gem.c
@@ -0,0 +1,901 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <linux/ktime.h>
+#include <linux/pagemap.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_display.h"
+
+void amdgpu_gem_object_free(struct drm_gem_object *gobj)
+{
+	struct amdgpu_bo *robj = gem_to_amdgpu_bo(gobj);
+
+	if (robj) {
+		amdgpu_mn_unregister(robj);
+		amdgpu_bo_unref(&robj);
+	}
+}
+
+int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
+			     int alignment, u32 initial_domain,
+			     u64 flags, enum ttm_bo_type type,
+			     struct reservation_object *resv,
+			     struct drm_gem_object **obj)
+{
+	struct amdgpu_bo *bo;
+	struct amdgpu_bo_param bp;
+	int r;
+
+	memset(&bp, 0, sizeof(bp));
+	*obj = NULL;
+	/* At least align on page size */
+	if (alignment < PAGE_SIZE) {
+		alignment = PAGE_SIZE;
+	}
+
+	bp.size = size;
+	bp.byte_align = alignment;
+	bp.type = type;
+	bp.resv = resv;
+	bp.preferred_domain = initial_domain;
+retry:
+	bp.flags = flags;
+	bp.domain = initial_domain;
+	r = amdgpu_bo_create(adev, &bp, &bo);
+	if (r) {
+		if (r != -ERESTARTSYS) {
+			if (flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) {
+				flags &= ~AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+				goto retry;
+			}
+
+			if (initial_domain == AMDGPU_GEM_DOMAIN_VRAM) {
+				initial_domain |= AMDGPU_GEM_DOMAIN_GTT;
+				goto retry;
+			}
+			DRM_DEBUG("Failed to allocate GEM object (%ld, %d, %u, %d)\n",
+				  size, initial_domain, alignment, r);
+		}
+		return r;
+	}
+	*obj = &bo->gem_base;
+
+	return 0;
+}
+
+void amdgpu_gem_force_release(struct amdgpu_device *adev)
+{
+	STUB();
+#if 0
+	struct drm_device *ddev = adev->ddev;
+	struct drm_file *file;
+
+	mutex_lock(&ddev->filelist_mutex);
+
+	list_for_each_entry(file, &ddev->filelist, lhead) {
+		struct drm_gem_object *gobj;
+		int handle;
+
+		WARN_ONCE(1, "Still active user space clients!\n");
+		spin_lock(&file->table_lock);
+		idr_for_each_entry(&file->object_idr, gobj, handle) {
+			WARN_ONCE(1, "And also active allocations!\n");
+			drm_gem_object_put_unlocked(gobj);
+		}
+		idr_destroy(&file->object_idr);
+		spin_unlock(&file->table_lock);
+	}
+
+	mutex_unlock(&ddev->filelist_mutex);
+#endif
+}
+
+/*
+ * Call from drm_gem_handle_create which appear in both new and open ioctl
+ * case.
+ */
+int amdgpu_gem_object_open(struct drm_gem_object *obj,
+			   struct drm_file *file_priv)
+{
+	struct amdgpu_bo *abo = gem_to_amdgpu_bo(obj);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev);
+	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_bo_va *bo_va;
+#ifdef notyet
+	struct mm_struct *mm;
+#endif
+	int r;
+
+#ifdef notyet
+	mm = amdgpu_ttm_tt_get_usermm(abo->tbo.ttm);
+	if (mm && mm != current->mm)
+		return -EPERM;
+#endif
+
+	if (abo->flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID &&
+	    abo->tbo.resv != vm->root.base.bo->tbo.resv)
+		return -EPERM;
+
+	r = amdgpu_bo_reserve(abo, false);
+	if (r)
+		return r;
+
+	bo_va = amdgpu_vm_bo_find(vm, abo);
+	if (!bo_va) {
+		bo_va = amdgpu_vm_bo_add(adev, vm, abo);
+	} else {
+		++bo_va->ref_count;
+	}
+	amdgpu_bo_unreserve(abo);
+	return 0;
+}
+
+void amdgpu_gem_object_close(struct drm_gem_object *obj,
+			     struct drm_file *file_priv)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+
+	struct amdgpu_bo_list_entry vm_pd;
+	struct list_head list, duplicates;
+	struct ttm_validate_buffer tv;
+	struct ww_acquire_ctx ticket;
+	struct amdgpu_bo_va *bo_va;
+	int r;
+
+	INIT_LIST_HEAD(&list);
+	INIT_LIST_HEAD(&duplicates);
+
+	tv.bo = &bo->tbo;
+	tv.shared = true;
+	list_add(&tv.head, &list);
+
+	amdgpu_vm_get_pd_bo(vm, &list, &vm_pd);
+
+	r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates);
+	if (r) {
+		dev_err(adev->dev, "leaking bo va because "
+			"we fail to reserve bo (%d)\n", r);
+		return;
+	}
+	bo_va = amdgpu_vm_bo_find(vm, bo);
+	if (bo_va && --bo_va->ref_count == 0) {
+		amdgpu_vm_bo_rmv(adev, bo_va);
+
+		if (amdgpu_vm_ready(vm)) {
+			struct dma_fence *fence = NULL;
+
+			r = amdgpu_vm_clear_freed(adev, vm, &fence);
+			if (unlikely(r)) {
+				dev_err(adev->dev, "failed to clear page "
+					"tables on GEM object close (%d)\n", r);
+			}
+
+			if (fence) {
+				amdgpu_bo_fence(bo, fence, true);
+				dma_fence_put(fence);
+			}
+		}
+	}
+	ttm_eu_backoff_reservation(&ticket, &list);
+}
+
+/*
+ * GEM ioctls.
+ */
+int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
+			    struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	union drm_amdgpu_gem_create *args = data;
+	uint64_t flags = args->in.domain_flags;
+	uint64_t size = args->in.bo_size;
+	struct reservation_object *resv = NULL;
+	struct drm_gem_object *gobj;
+	uint32_t handle;
+	int r;
+
+	/* reject invalid gem flags */
+	if (flags & ~(AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
+		      AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
+		      AMDGPU_GEM_CREATE_CPU_GTT_USWC |
+		      AMDGPU_GEM_CREATE_VRAM_CLEARED |
+		      AMDGPU_GEM_CREATE_VM_ALWAYS_VALID |
+		      AMDGPU_GEM_CREATE_EXPLICIT_SYNC))
+
+		return -EINVAL;
+
+	/* reject invalid gem domains */
+	if (args->in.domains & ~AMDGPU_GEM_DOMAIN_MASK)
+		return -EINVAL;
+
+	/* create a gem object to contain this object in */
+	if (args->in.domains & (AMDGPU_GEM_DOMAIN_GDS |
+	    AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA)) {
+		if (flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID) {
+			/* if gds bo is created from user space, it must be
+			 * passed to bo list
+			 */
+			DRM_ERROR("GDS bo cannot be per-vm-bo\n");
+			return -EINVAL;
+		}
+		flags |= AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
+		if (args->in.domains == AMDGPU_GEM_DOMAIN_GDS)
+			size = size << AMDGPU_GDS_SHIFT;
+		else if (args->in.domains == AMDGPU_GEM_DOMAIN_GWS)
+			size = size << AMDGPU_GWS_SHIFT;
+		else if (args->in.domains == AMDGPU_GEM_DOMAIN_OA)
+			size = size << AMDGPU_OA_SHIFT;
+		else
+			return -EINVAL;
+	}
+	size = roundup(size, PAGE_SIZE);
+
+	if (flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID) {
+		r = amdgpu_bo_reserve(vm->root.base.bo, false);
+		if (r)
+			return r;
+
+		resv = vm->root.base.bo->tbo.resv;
+	}
+
+	r = amdgpu_gem_object_create(adev, size, args->in.alignment,
+				     (u32)(0xffffffff & args->in.domains),
+				     flags, ttm_bo_type_device, resv, &gobj);
+	if (flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID) {
+		if (!r) {
+			struct amdgpu_bo *abo = gem_to_amdgpu_bo(gobj);
+
+			abo->parent = amdgpu_bo_ref(vm->root.base.bo);
+		}
+		amdgpu_bo_unreserve(vm->root.base.bo);
+	}
+	if (r)
+		return r;
+
+	r = drm_gem_handle_create(filp, gobj, &handle);
+	/* drop reference from allocate - handle holds it now */
+	drm_gem_object_put_unlocked(gobj);
+	if (r)
+		return r;
+
+	memset(args, 0, sizeof(*args));
+	args->out.handle = handle;
+	return 0;
+}
+
+int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
+			     struct drm_file *filp)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct ttm_operation_ctx ctx = { true, false };
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_amdgpu_gem_userptr *args = data;
+	struct drm_gem_object *gobj;
+	struct amdgpu_bo *bo;
+	uint32_t handle;
+	int r;
+
+	if (offset_in_page(args->addr | args->size))
+		return -EINVAL;
+
+	/* reject unknown flag values */
+	if (args->flags & ~(AMDGPU_GEM_USERPTR_READONLY |
+	    AMDGPU_GEM_USERPTR_ANONONLY | AMDGPU_GEM_USERPTR_VALIDATE |
+	    AMDGPU_GEM_USERPTR_REGISTER))
+		return -EINVAL;
+
+	if (!(args->flags & AMDGPU_GEM_USERPTR_READONLY) &&
+	     !(args->flags & AMDGPU_GEM_USERPTR_REGISTER)) {
+
+		/* if we want to write to it we must install a MMU notifier */
+		return -EACCES;
+	}
+
+	/* create a gem object to contain this object in */
+	r = amdgpu_gem_object_create(adev, args->size, 0, AMDGPU_GEM_DOMAIN_CPU,
+				     0, ttm_bo_type_device, NULL, &gobj);
+	if (r)
+		return r;
+
+	bo = gem_to_amdgpu_bo(gobj);
+	bo->preferred_domains = AMDGPU_GEM_DOMAIN_GTT;
+	bo->allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
+	r = amdgpu_ttm_tt_set_userptr(bo->tbo.ttm, args->addr, args->flags);
+	if (r)
+		goto release_object;
+
+	if (args->flags & AMDGPU_GEM_USERPTR_REGISTER) {
+		r = amdgpu_mn_register(bo, args->addr);
+		if (r)
+			goto release_object;
+	}
+
+	if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE) {
+		r = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm,
+						 bo->tbo.ttm->pages);
+		if (r)
+			goto release_object;
+
+		r = amdgpu_bo_reserve(bo, true);
+		if (r)
+			goto free_pages;
+
+		amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
+		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+		amdgpu_bo_unreserve(bo);
+		if (r)
+			goto free_pages;
+	}
+
+	r = drm_gem_handle_create(filp, gobj, &handle);
+	/* drop reference from allocate - handle holds it now */
+	drm_gem_object_put_unlocked(gobj);
+	if (r)
+		return r;
+
+	args->handle = handle;
+	return 0;
+
+free_pages:
+	release_pages(bo->tbo.ttm->pages, bo->tbo.ttm->num_pages);
+
+release_object:
+	drm_gem_object_put_unlocked(gobj);
+
+	return r;
+#endif
+}
+
+int amdgpu_mode_dumb_mmap(struct drm_file *filp,
+			  struct drm_device *dev,
+			  uint32_t handle, uint64_t *offset_p)
+{
+	struct drm_gem_object *gobj;
+	struct amdgpu_bo *robj;
+
+	gobj = drm_gem_object_lookup(filp, handle);
+	if (gobj == NULL) {
+		return -ENOENT;
+	}
+	robj = gem_to_amdgpu_bo(gobj);
+	if (amdgpu_ttm_tt_get_usermm(robj->tbo.ttm) ||
+	    (robj->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)) {
+		drm_gem_object_put_unlocked(gobj);
+		return -EPERM;
+	}
+	*offset_p = amdgpu_bo_mmap_offset(robj);
+	drm_gem_object_put_unlocked(gobj);
+	return 0;
+}
+
+int amdgpu_gem_mmap_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp)
+{
+	union drm_amdgpu_gem_mmap *args = data;
+	uint32_t handle = args->in.handle;
+	memset(args, 0, sizeof(*args));
+	return amdgpu_mode_dumb_mmap(filp, dev, handle, &args->out.addr_ptr);
+}
+
+/**
+ * amdgpu_gem_timeout - calculate jiffies timeout from absolute value
+ *
+ * @timeout_ns: timeout in ns
+ *
+ * Calculate the timeout in jiffies from an absolute timeout in ns.
+ */
+unsigned long amdgpu_gem_timeout(uint64_t timeout_ns)
+{
+	unsigned long timeout_jiffies;
+	ktime_t timeout;
+
+	/* clamp timeout if it's to large */
+	if (((int64_t)timeout_ns) < 0)
+		return MAX_SCHEDULE_TIMEOUT;
+
+	timeout = ktime_sub(ns_to_ktime(timeout_ns), ktime_get());
+	if (ktime_to_ns(timeout) < 0)
+		return 0;
+
+	timeout_jiffies = nsecs_to_jiffies(ktime_to_ns(timeout));
+	/*  clamp timeout to avoid unsigned-> signed overflow */
+	if (timeout_jiffies > MAX_SCHEDULE_TIMEOUT )
+		return MAX_SCHEDULE_TIMEOUT - 1;
+
+	return timeout_jiffies;
+}
+
+int amdgpu_gem_wait_idle_ioctl(struct drm_device *dev, void *data,
+			      struct drm_file *filp)
+{
+	union drm_amdgpu_gem_wait_idle *args = data;
+	struct drm_gem_object *gobj;
+	struct amdgpu_bo *robj;
+	uint32_t handle = args->in.handle;
+	unsigned long timeout = amdgpu_gem_timeout(args->in.timeout);
+	int r = 0;
+	long ret;
+
+	gobj = drm_gem_object_lookup(filp, handle);
+	if (gobj == NULL) {
+		return -ENOENT;
+	}
+	robj = gem_to_amdgpu_bo(gobj);
+	ret = reservation_object_wait_timeout_rcu(robj->tbo.resv, true, true,
+						  timeout);
+
+	/* ret == 0 means not signaled,
+	 * ret > 0 means signaled
+	 * ret < 0 means interrupted before timeout
+	 */
+	if (ret >= 0) {
+		memset(args, 0, sizeof(*args));
+		args->out.status = (ret == 0);
+	} else
+		r = ret;
+
+	drm_gem_object_put_unlocked(gobj);
+	return r;
+}
+
+int amdgpu_gem_metadata_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *filp)
+{
+	struct drm_amdgpu_gem_metadata *args = data;
+	struct drm_gem_object *gobj;
+	struct amdgpu_bo *robj;
+	int r = -1;
+
+	DRM_DEBUG("%d \n", args->handle);
+	gobj = drm_gem_object_lookup(filp, args->handle);
+	if (gobj == NULL)
+		return -ENOENT;
+	robj = gem_to_amdgpu_bo(gobj);
+
+	r = amdgpu_bo_reserve(robj, false);
+	if (unlikely(r != 0))
+		goto out;
+
+	if (args->op == AMDGPU_GEM_METADATA_OP_GET_METADATA) {
+		amdgpu_bo_get_tiling_flags(robj, &args->data.tiling_info);
+		r = amdgpu_bo_get_metadata(robj, args->data.data,
+					   sizeof(args->data.data),
+					   &args->data.data_size_bytes,
+					   &args->data.flags);
+	} else if (args->op == AMDGPU_GEM_METADATA_OP_SET_METADATA) {
+		if (args->data.data_size_bytes > sizeof(args->data.data)) {
+			r = -EINVAL;
+			goto unreserve;
+		}
+		r = amdgpu_bo_set_tiling_flags(robj, args->data.tiling_info);
+		if (!r)
+			r = amdgpu_bo_set_metadata(robj, args->data.data,
+						   args->data.data_size_bytes,
+						   args->data.flags);
+	}
+
+unreserve:
+	amdgpu_bo_unreserve(robj);
+out:
+	drm_gem_object_put_unlocked(gobj);
+	return r;
+}
+
+/**
+ * amdgpu_gem_va_update_vm -update the bo_va in its VM
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: vm to update
+ * @bo_va: bo_va to update
+ * @operation: map, unmap or clear
+ *
+ * Update the bo_va directly after setting its address. Errors are not
+ * vital here, so they are not reported back to userspace.
+ */
+static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
+				    struct amdgpu_vm *vm,
+				    struct amdgpu_bo_va *bo_va,
+				    uint32_t operation)
+{
+	int r;
+
+	if (!amdgpu_vm_ready(vm))
+		return;
+
+	r = amdgpu_vm_clear_freed(adev, vm, NULL);
+	if (r)
+		goto error;
+
+	if (operation == AMDGPU_VA_OP_MAP ||
+	    operation == AMDGPU_VA_OP_REPLACE) {
+		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		if (r)
+			goto error;
+	}
+
+	r = amdgpu_vm_update_directories(adev, vm);
+
+error:
+	if (r && r != -ERESTARTSYS)
+		DRM_ERROR("Couldn't update BO_VA (%d)\n", r);
+}
+
+int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp)
+{
+	const uint32_t valid_flags = AMDGPU_VM_DELAY_UPDATE |
+		AMDGPU_VM_PAGE_READABLE | AMDGPU_VM_PAGE_WRITEABLE |
+		AMDGPU_VM_PAGE_EXECUTABLE | AMDGPU_VM_MTYPE_MASK;
+	const uint32_t prt_flags = AMDGPU_VM_DELAY_UPDATE |
+		AMDGPU_VM_PAGE_PRT;
+
+	struct drm_amdgpu_gem_va *args = data;
+	struct drm_gem_object *gobj;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_bo *abo;
+	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_bo_list_entry vm_pd;
+	struct ttm_validate_buffer tv;
+	struct ww_acquire_ctx ticket;
+	struct list_head list, duplicates;
+	uint64_t va_flags;
+	int r = 0;
+
+	if (args->va_address < AMDGPU_VA_RESERVED_SIZE) {
+		dev_dbg(&dev->pdev->dev,
+			"va_address 0x%LX is in reserved area 0x%LX\n",
+			args->va_address, AMDGPU_VA_RESERVED_SIZE);
+		return -EINVAL;
+	}
+
+	if (args->va_address >= AMDGPU_VA_HOLE_START &&
+	    args->va_address < AMDGPU_VA_HOLE_END) {
+		dev_dbg(&dev->pdev->dev,
+			"va_address 0x%LX is in VA hole 0x%LX-0x%LX\n",
+			args->va_address, AMDGPU_VA_HOLE_START,
+			AMDGPU_VA_HOLE_END);
+		return -EINVAL;
+	}
+
+	args->va_address &= AMDGPU_VA_HOLE_MASK;
+
+	if ((args->flags & ~valid_flags) && (args->flags & ~prt_flags)) {
+		dev_dbg(&dev->pdev->dev, "invalid flags combination 0x%08X\n",
+			args->flags);
+		return -EINVAL;
+	}
+
+	switch (args->operation) {
+	case AMDGPU_VA_OP_MAP:
+	case AMDGPU_VA_OP_UNMAP:
+	case AMDGPU_VA_OP_CLEAR:
+	case AMDGPU_VA_OP_REPLACE:
+		break;
+	default:
+		dev_dbg(&dev->pdev->dev, "unsupported operation %d\n",
+			args->operation);
+		return -EINVAL;
+	}
+
+	INIT_LIST_HEAD(&list);
+	INIT_LIST_HEAD(&duplicates);
+	if ((args->operation != AMDGPU_VA_OP_CLEAR) &&
+	    !(args->flags & AMDGPU_VM_PAGE_PRT)) {
+		gobj = drm_gem_object_lookup(filp, args->handle);
+		if (gobj == NULL)
+			return -ENOENT;
+		abo = gem_to_amdgpu_bo(gobj);
+		tv.bo = &abo->tbo;
+		tv.shared = !!(abo->flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID);
+		list_add(&tv.head, &list);
+	} else {
+		gobj = NULL;
+		abo = NULL;
+	}
+
+	amdgpu_vm_get_pd_bo(&fpriv->vm, &list, &vm_pd);
+
+	r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates);
+	if (r)
+		goto error_unref;
+
+	if (abo) {
+		bo_va = amdgpu_vm_bo_find(&fpriv->vm, abo);
+		if (!bo_va) {
+			r = -ENOENT;
+			goto error_backoff;
+		}
+	} else if (args->operation != AMDGPU_VA_OP_CLEAR) {
+		bo_va = fpriv->prt_va;
+	} else {
+		bo_va = NULL;
+	}
+
+	switch (args->operation) {
+	case AMDGPU_VA_OP_MAP:
+		r = amdgpu_vm_alloc_pts(adev, bo_va->base.vm, args->va_address,
+					args->map_size);
+		if (r)
+			goto error_backoff;
+
+		va_flags = amdgpu_gmc_get_pte_flags(adev, args->flags);
+		r = amdgpu_vm_bo_map(adev, bo_va, args->va_address,
+				     args->offset_in_bo, args->map_size,
+				     va_flags);
+		break;
+	case AMDGPU_VA_OP_UNMAP:
+		r = amdgpu_vm_bo_unmap(adev, bo_va, args->va_address);
+		break;
+
+	case AMDGPU_VA_OP_CLEAR:
+		r = amdgpu_vm_bo_clear_mappings(adev, &fpriv->vm,
+						args->va_address,
+						args->map_size);
+		break;
+	case AMDGPU_VA_OP_REPLACE:
+		r = amdgpu_vm_alloc_pts(adev, bo_va->base.vm, args->va_address,
+					args->map_size);
+		if (r)
+			goto error_backoff;
+
+		va_flags = amdgpu_gmc_get_pte_flags(adev, args->flags);
+		r = amdgpu_vm_bo_replace_map(adev, bo_va, args->va_address,
+					     args->offset_in_bo, args->map_size,
+					     va_flags);
+		break;
+	default:
+		break;
+	}
+	if (!r && !(args->flags & AMDGPU_VM_DELAY_UPDATE) && !amdgpu_vm_debug)
+		amdgpu_gem_va_update_vm(adev, &fpriv->vm, bo_va,
+					args->operation);
+
+error_backoff:
+	ttm_eu_backoff_reservation(&ticket, &list);
+
+error_unref:
+	drm_gem_object_put_unlocked(gobj);
+	return r;
+}
+
+int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
+			struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_amdgpu_gem_op *args = data;
+	struct drm_gem_object *gobj;
+	struct amdgpu_bo *robj;
+	int r;
+
+	gobj = drm_gem_object_lookup(filp, args->handle);
+	if (gobj == NULL) {
+		return -ENOENT;
+	}
+	robj = gem_to_amdgpu_bo(gobj);
+
+	r = amdgpu_bo_reserve(robj, false);
+	if (unlikely(r))
+		goto out;
+
+	switch (args->op) {
+	case AMDGPU_GEM_OP_GET_GEM_CREATE_INFO: {
+		struct drm_amdgpu_gem_create_in info;
+		void __user *out = u64_to_user_ptr(args->value);
+
+		info.bo_size = robj->gem_base.size;
+		info.alignment = robj->tbo.mem.page_alignment << PAGE_SHIFT;
+		info.domains = robj->preferred_domains;
+		info.domain_flags = robj->flags;
+		amdgpu_bo_unreserve(robj);
+		if (copy_to_user(out, &info, sizeof(info)))
+			r = -EFAULT;
+		break;
+	}
+	case AMDGPU_GEM_OP_SET_PLACEMENT:
+		if (robj->prime_shared_count && (args->value & AMDGPU_GEM_DOMAIN_VRAM)) {
+			r = -EINVAL;
+			amdgpu_bo_unreserve(robj);
+			break;
+		}
+		if (amdgpu_ttm_tt_get_usermm(robj->tbo.ttm)) {
+			r = -EPERM;
+			amdgpu_bo_unreserve(robj);
+			break;
+		}
+		robj->preferred_domains = args->value & (AMDGPU_GEM_DOMAIN_VRAM |
+							AMDGPU_GEM_DOMAIN_GTT |
+							AMDGPU_GEM_DOMAIN_CPU);
+		robj->allowed_domains = robj->preferred_domains;
+		if (robj->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
+			robj->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
+
+		if (robj->flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID)
+			amdgpu_vm_bo_invalidate(adev, robj, true);
+
+		amdgpu_bo_unreserve(robj);
+		break;
+	default:
+		amdgpu_bo_unreserve(robj);
+		r = -EINVAL;
+	}
+
+out:
+	drm_gem_object_put_unlocked(gobj);
+	return r;
+}
+
+int amdgpu_mode_dumb_create(struct drm_file *file_priv,
+			    struct drm_device *dev,
+			    struct drm_mode_create_dumb *args)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_gem_object *gobj;
+	uint32_t handle;
+	u32 domain;
+	int r;
+
+	args->pitch = amdgpu_align_pitch(adev, args->width,
+					 DIV_ROUND_UP(args->bpp, 8), 0);
+	args->size = (u64)args->pitch * args->height;
+	args->size = roundup2(args->size, PAGE_SIZE);
+	domain = amdgpu_bo_get_preferred_pin_domain(adev,
+				amdgpu_display_supported_domains(adev));
+	r = amdgpu_gem_object_create(adev, args->size, 0, domain,
+				     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+				     ttm_bo_type_device, NULL, &gobj);
+	if (r)
+		return -ENOMEM;
+
+	r = drm_gem_handle_create(file_priv, gobj, &handle);
+	/* drop reference from allocate - handle holds it now */
+	drm_gem_object_put_unlocked(gobj);
+	if (r) {
+		return r;
+	}
+	args->handle = handle;
+	return 0;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+#define amdgpu_debugfs_gem_bo_print_flag(m, bo, flag)	\
+	if (bo->flags & (AMDGPU_GEM_CREATE_ ## flag)) {	\
+		seq_printf((m), " " #flag);		\
+	}
+
+static int amdgpu_debugfs_gem_bo_info(int id, void *ptr, void *data)
+{
+	struct drm_gem_object *gobj = ptr;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(gobj);
+	struct seq_file *m = data;
+
+	struct dma_buf_attachment *attachment;
+	struct dma_buf *dma_buf;
+	unsigned domain;
+	const char *placement;
+	unsigned pin_count;
+
+	domain = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type);
+	switch (domain) {
+	case AMDGPU_GEM_DOMAIN_VRAM:
+		placement = "VRAM";
+		break;
+	case AMDGPU_GEM_DOMAIN_GTT:
+		placement = " GTT";
+		break;
+	case AMDGPU_GEM_DOMAIN_CPU:
+	default:
+		placement = " CPU";
+		break;
+	}
+	seq_printf(m, "\t0x%08x: %12ld byte %s",
+		   id, amdgpu_bo_size(bo), placement);
+
+	pin_count = READ_ONCE(bo->pin_count);
+	if (pin_count)
+		seq_printf(m, " pin count %d", pin_count);
+
+	dma_buf = READ_ONCE(bo->gem_base.dma_buf);
+	attachment = READ_ONCE(bo->gem_base.import_attach);
+
+	if (attachment)
+		seq_printf(m, " imported from %p", dma_buf);
+	else if (dma_buf)
+		seq_printf(m, " exported as %p", dma_buf);
+
+	amdgpu_debugfs_gem_bo_print_flag(m, bo, CPU_ACCESS_REQUIRED);
+	amdgpu_debugfs_gem_bo_print_flag(m, bo, NO_CPU_ACCESS);
+	amdgpu_debugfs_gem_bo_print_flag(m, bo, CPU_GTT_USWC);
+	amdgpu_debugfs_gem_bo_print_flag(m, bo, VRAM_CLEARED);
+	amdgpu_debugfs_gem_bo_print_flag(m, bo, SHADOW);
+	amdgpu_debugfs_gem_bo_print_flag(m, bo, VRAM_CONTIGUOUS);
+	amdgpu_debugfs_gem_bo_print_flag(m, bo, VM_ALWAYS_VALID);
+	amdgpu_debugfs_gem_bo_print_flag(m, bo, EXPLICIT_SYNC);
+
+	seq_printf(m, "\n");
+
+	return 0;
+}
+
+static int amdgpu_debugfs_gem_info(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *)m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct drm_file *file;
+	int r;
+
+	r = mutex_lock_interruptible(&dev->filelist_mutex);
+	if (r)
+		return r;
+
+	list_for_each_entry(file, &dev->filelist, lhead) {
+		struct task_struct *task;
+
+		/*
+		 * Although we have a valid reference on file->pid, that does
+		 * not guarantee that the task_struct who called get_pid() is
+		 * still alive (e.g. get_pid(current) => fork() => exit()).
+		 * Therefore, we need to protect this ->comm access using RCU.
+		 */
+		rcu_read_lock();
+		task = pid_task(file->pid, PIDTYPE_PID);
+		seq_printf(m, "pid %8d command %s:\n", pid_nr(file->pid),
+			   task ? task->comm : "<unknown>");
+		rcu_read_unlock();
+
+		spin_lock(&file->table_lock);
+		idr_for_each(&file->object_idr, amdgpu_debugfs_gem_bo_info, m);
+		spin_unlock(&file->table_lock);
+	}
+
+	mutex_unlock(&dev->filelist_mutex);
+	return 0;
+}
+
+static const struct drm_info_list amdgpu_debugfs_gem_list[] = {
+	{"amdgpu_gem_info", &amdgpu_debugfs_gem_info, 0, NULL},
+};
+#endif
+
+int amdgpu_debugfs_gem_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_gem_list, 1);
+#endif
+	return 0;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_gfx.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gfx.c
new file mode 100644
index 00000000000..35e7da3a443
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gfx.c
@@ -0,0 +1,344 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+
+/*
+ * GPU scratch registers helpers function.
+ */
+/**
+ * amdgpu_gfx_scratch_get - Allocate a scratch register
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: scratch register mmio offset
+ *
+ * Allocate a CP scratch register for use by the driver (all asics).
+ * Returns 0 on success or -EINVAL on failure.
+ */
+int amdgpu_gfx_scratch_get(struct amdgpu_device *adev, uint32_t *reg)
+{
+	int i;
+
+	i = ffs(adev->gfx.scratch.free_mask);
+	if (i != 0 && i <= adev->gfx.scratch.num_reg) {
+		i--;
+		adev->gfx.scratch.free_mask &= ~(1u << i);
+		*reg = adev->gfx.scratch.reg_base + i;
+		return 0;
+	}
+	return -EINVAL;
+}
+
+/**
+ * amdgpu_gfx_scratch_free - Free a scratch register
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: scratch register mmio offset
+ *
+ * Free a CP scratch register allocated for use by the driver (all asics)
+ */
+void amdgpu_gfx_scratch_free(struct amdgpu_device *adev, uint32_t reg)
+{
+	adev->gfx.scratch.free_mask |= 1u << (reg - adev->gfx.scratch.reg_base);
+}
+
+/**
+ * amdgpu_gfx_parse_disable_cu - Parse the disable_cu module parameter
+ *
+ * @mask: array in which the per-shader array disable masks will be stored
+ * @max_se: number of SEs
+ * @max_sh: number of SHs
+ *
+ * The bitmask of CUs to be disabled in the shader array determined by se and
+ * sh is stored in mask[se * max_sh + sh].
+ */
+void amdgpu_gfx_parse_disable_cu(unsigned *mask, unsigned max_se, unsigned max_sh)
+{
+	unsigned se, sh, cu;
+	const char *p;
+
+	memset(mask, 0, sizeof(*mask) * max_se * max_sh);
+
+	if (!amdgpu_disable_cu || !*amdgpu_disable_cu)
+		return;
+
+#ifdef notyet
+	p = amdgpu_disable_cu;
+	for (;;) {
+		char *next;
+		int ret = sscanf(p, "%u.%u.%u", &se, &sh, &cu);
+		if (ret < 3) {
+			DRM_ERROR("amdgpu: could not parse disable_cu\n");
+			return;
+		}
+
+		if (se < max_se && sh < max_sh && cu < 16) {
+			DRM_INFO("amdgpu: disabling CU %u.%u.%u\n", se, sh, cu);
+			mask[se * max_sh + sh] |= 1u << cu;
+		} else {
+			DRM_ERROR("amdgpu: disable_cu %u.%u.%u is out of range\n",
+				  se, sh, cu);
+		}
+
+		next = strchr(p, ',');
+		if (!next)
+			break;
+		p = next + 1;
+	}
+#endif
+}
+
+static bool amdgpu_gfx_is_multipipe_capable(struct amdgpu_device *adev)
+{
+	if (amdgpu_compute_multipipe != -1) {
+		DRM_INFO("amdgpu: forcing compute pipe policy %d\n",
+			 amdgpu_compute_multipipe);
+		return amdgpu_compute_multipipe == 1;
+	}
+
+	/* FIXME: spreading the queues across pipes causes perf regressions
+	 * on POLARIS11 compute workloads */
+	if (adev->asic_type == CHIP_POLARIS11)
+		return false;
+
+	return adev->gfx.mec.num_mec > 1;
+}
+
+void amdgpu_gfx_compute_queue_acquire(struct amdgpu_device *adev)
+{
+	int i, queue, pipe, mec;
+	bool multipipe_policy = amdgpu_gfx_is_multipipe_capable(adev);
+
+	/* policy for amdgpu compute queue ownership */
+	for (i = 0; i < AMDGPU_MAX_COMPUTE_QUEUES; ++i) {
+		queue = i % adev->gfx.mec.num_queue_per_pipe;
+		pipe = (i / adev->gfx.mec.num_queue_per_pipe)
+			% adev->gfx.mec.num_pipe_per_mec;
+		mec = (i / adev->gfx.mec.num_queue_per_pipe)
+			/ adev->gfx.mec.num_pipe_per_mec;
+
+		/* we've run out of HW */
+		if (mec >= adev->gfx.mec.num_mec)
+			break;
+
+		if (multipipe_policy) {
+			/* policy: amdgpu owns the first two queues of the first MEC */
+			if (mec == 0 && queue < 2)
+				set_bit(i, adev->gfx.mec.queue_bitmap);
+		} else {
+			/* policy: amdgpu owns all queues in the first pipe */
+			if (mec == 0 && pipe == 0)
+				set_bit(i, adev->gfx.mec.queue_bitmap);
+		}
+	}
+
+	/* update the number of active compute rings */
+	adev->gfx.num_compute_rings =
+		bitmap_weight(adev->gfx.mec.queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+
+	/* If you hit this case and edited the policy, you probably just
+	 * need to increase AMDGPU_MAX_COMPUTE_RINGS */
+	if (WARN_ON(adev->gfx.num_compute_rings > AMDGPU_MAX_COMPUTE_RINGS))
+		adev->gfx.num_compute_rings = AMDGPU_MAX_COMPUTE_RINGS;
+}
+
+static int amdgpu_gfx_kiq_acquire(struct amdgpu_device *adev,
+				  struct amdgpu_ring *ring)
+{
+	int queue_bit;
+	int mec, pipe, queue;
+
+	queue_bit = adev->gfx.mec.num_mec
+		    * adev->gfx.mec.num_pipe_per_mec
+		    * adev->gfx.mec.num_queue_per_pipe;
+
+	while (queue_bit-- >= 0) {
+		if (test_bit(queue_bit, adev->gfx.mec.queue_bitmap))
+			continue;
+
+		amdgpu_gfx_bit_to_queue(adev, queue_bit, &mec, &pipe, &queue);
+
+		/*
+		 * 1. Using pipes 2/3 from MEC 2 seems cause problems.
+		 * 2. It must use queue id 0, because CGPG_IDLE/SAVE/LOAD/RUN
+		 * only can be issued on queue 0.
+		 */
+		if ((mec == 1 && pipe > 1) || queue != 0)
+			continue;
+
+		ring->me = mec + 1;
+		ring->pipe = pipe;
+		ring->queue = queue;
+
+		return 0;
+	}
+
+	dev_err(adev->dev, "Failed to find a queue for KIQ\n");
+	return -EINVAL;
+}
+
+int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev,
+			     struct amdgpu_ring *ring,
+			     struct amdgpu_irq_src *irq)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
+	int r = 0;
+
+	mtx_init(&kiq->ring_lock, IPL_TTY);
+
+	r = amdgpu_device_wb_get(adev, &adev->virt.reg_val_offs);
+	if (r)
+		return r;
+
+	ring->adev = NULL;
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->doorbell_index = AMDGPU_DOORBELL_KIQ;
+
+	r = amdgpu_gfx_kiq_acquire(adev, ring);
+	if (r)
+		return r;
+
+	ring->eop_gpu_addr = kiq->eop_gpu_addr;
+	snprintf(ring->name, sizeof(ring->name), "kiq_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+	r = amdgpu_ring_init(adev, ring, 1024,
+			     irq, AMDGPU_CP_KIQ_IRQ_DRIVER0);
+	if (r)
+		dev_warn(adev->dev, "(%d) failed to init kiq ring\n", r);
+
+	return r;
+}
+
+void amdgpu_gfx_kiq_free_ring(struct amdgpu_ring *ring,
+			      struct amdgpu_irq_src *irq)
+{
+	amdgpu_device_wb_free(ring->adev, ring->adev->virt.reg_val_offs);
+	amdgpu_ring_fini(ring);
+}
+
+void amdgpu_gfx_kiq_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
+
+	amdgpu_bo_free_kernel(&kiq->eop_obj, &kiq->eop_gpu_addr, NULL);
+}
+
+int amdgpu_gfx_kiq_init(struct amdgpu_device *adev,
+			unsigned hpd_size)
+{
+	int r;
+	u32 *hpd;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
+
+	r = amdgpu_bo_create_kernel(adev, hpd_size, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_GTT, &kiq->eop_obj,
+				    &kiq->eop_gpu_addr, (void **)&hpd);
+	if (r) {
+		dev_warn(adev->dev, "failed to create KIQ bo (%d).\n", r);
+		return r;
+	}
+
+	memset(hpd, 0, hpd_size);
+
+	r = amdgpu_bo_reserve(kiq->eop_obj, true);
+	if (unlikely(r != 0))
+		dev_warn(adev->dev, "(%d) reserve kiq eop bo failed\n", r);
+	amdgpu_bo_kunmap(kiq->eop_obj);
+	amdgpu_bo_unreserve(kiq->eop_obj);
+
+	return 0;
+}
+
+/* create MQD for each compute queue */
+int amdgpu_gfx_compute_mqd_sw_init(struct amdgpu_device *adev,
+				   unsigned mqd_size)
+{
+	struct amdgpu_ring *ring = NULL;
+	int r, i;
+
+	/* create MQD for KIQ */
+	ring = &adev->gfx.kiq.ring;
+	if (!ring->mqd_obj) {
+		/* originaly the KIQ MQD is put in GTT domain, but for SRIOV VRAM domain is a must
+		 * otherwise hypervisor trigger SAVE_VF fail after driver unloaded which mean MQD
+		 * deallocated and gart_unbind, to strict diverage we decide to use VRAM domain for
+		 * KIQ MQD no matter SRIOV or Bare-metal
+		 */
+		r = amdgpu_bo_create_kernel(adev, mqd_size, PAGE_SIZE,
+					    AMDGPU_GEM_DOMAIN_VRAM, &ring->mqd_obj,
+					    &ring->mqd_gpu_addr, &ring->mqd_ptr);
+		if (r) {
+			dev_warn(adev->dev, "failed to create ring mqd ob (%d)", r);
+			return r;
+		}
+
+		/* prepare MQD backup */
+		adev->gfx.mec.mqd_backup[AMDGPU_MAX_COMPUTE_RINGS] = kmalloc(mqd_size, GFP_KERNEL);
+		if (!adev->gfx.mec.mqd_backup[AMDGPU_MAX_COMPUTE_RINGS])
+				dev_warn(adev->dev, "no memory to create MQD backup for ring %s\n", ring->name);
+	}
+
+	/* create MQD for each KCQ */
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+		if (!ring->mqd_obj) {
+			r = amdgpu_bo_create_kernel(adev, mqd_size, PAGE_SIZE,
+						    AMDGPU_GEM_DOMAIN_GTT, &ring->mqd_obj,
+						    &ring->mqd_gpu_addr, &ring->mqd_ptr);
+			if (r) {
+				dev_warn(adev->dev, "failed to create ring mqd ob (%d)", r);
+				return r;
+			}
+
+			/* prepare MQD backup */
+			adev->gfx.mec.mqd_backup[i] = kmalloc(mqd_size, GFP_KERNEL);
+			if (!adev->gfx.mec.mqd_backup[i])
+				dev_warn(adev->dev, "no memory to create MQD backup for ring %s\n", ring->name);
+		}
+	}
+
+	return 0;
+}
+
+void amdgpu_gfx_compute_mqd_sw_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = NULL;
+	int i;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+		kfree(adev->gfx.mec.mqd_backup[i]);
+		amdgpu_bo_free_kernel(&ring->mqd_obj,
+				      &ring->mqd_gpu_addr,
+				      &ring->mqd_ptr);
+	}
+
+	ring = &adev->gfx.kiq.ring;
+	kfree(adev->gfx.mec.mqd_backup[AMDGPU_MAX_COMPUTE_RINGS]);
+	amdgpu_bo_free_kernel(&ring->mqd_obj,
+			      &ring->mqd_gpu_addr,
+			      &ring->mqd_ptr);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_gfx.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gfx.h
new file mode 100644
index 00000000000..1f279050d33
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gfx.h
@@ -0,0 +1,93 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_GFX_H__
+#define __AMDGPU_GFX_H__
+
+int amdgpu_gfx_scratch_get(struct amdgpu_device *adev, uint32_t *reg);
+void amdgpu_gfx_scratch_free(struct amdgpu_device *adev, uint32_t reg);
+
+void amdgpu_gfx_parse_disable_cu(unsigned *mask, unsigned max_se,
+		unsigned max_sh);
+
+void amdgpu_gfx_compute_queue_acquire(struct amdgpu_device *adev);
+
+int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev,
+			     struct amdgpu_ring *ring,
+			     struct amdgpu_irq_src *irq);
+
+void amdgpu_gfx_kiq_free_ring(struct amdgpu_ring *ring,
+			      struct amdgpu_irq_src *irq);
+
+void amdgpu_gfx_kiq_fini(struct amdgpu_device *adev);
+int amdgpu_gfx_kiq_init(struct amdgpu_device *adev,
+			unsigned hpd_size);
+
+int amdgpu_gfx_compute_mqd_sw_init(struct amdgpu_device *adev,
+				   unsigned mqd_size);
+void amdgpu_gfx_compute_mqd_sw_fini(struct amdgpu_device *adev);
+
+/**
+ * amdgpu_gfx_create_bitmask - create a bitmask
+ *
+ * @bit_width: length of the mask
+ *
+ * create a variable length bit mask.
+ * Returns the bitmask.
+ */
+static inline u32 amdgpu_gfx_create_bitmask(u32 bit_width)
+{
+	return (u32)((1ULL << bit_width) - 1);
+}
+
+static inline int amdgpu_gfx_queue_to_bit(struct amdgpu_device *adev,
+					  int mec, int pipe, int queue)
+{
+	int bit = 0;
+
+	bit += mec * adev->gfx.mec.num_pipe_per_mec
+		* adev->gfx.mec.num_queue_per_pipe;
+	bit += pipe * adev->gfx.mec.num_queue_per_pipe;
+	bit += queue;
+
+	return bit;
+}
+
+static inline void amdgpu_gfx_bit_to_queue(struct amdgpu_device *adev, int bit,
+					   int *mec, int *pipe, int *queue)
+{
+	*queue = bit % adev->gfx.mec.num_queue_per_pipe;
+	*pipe = (bit / adev->gfx.mec.num_queue_per_pipe)
+		% adev->gfx.mec.num_pipe_per_mec;
+	*mec = (bit / adev->gfx.mec.num_queue_per_pipe)
+	       / adev->gfx.mec.num_pipe_per_mec;
+
+}
+static inline bool amdgpu_gfx_is_mec_queue_enabled(struct amdgpu_device *adev,
+						   int mec, int pipe, int queue)
+{
+	return test_bit(amdgpu_gfx_queue_to_bit(adev, mec, pipe, queue),
+			adev->gfx.mec.queue_bitmap);
+}
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_gmc.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gmc.h
new file mode 100644
index 00000000000..bb5a47a4579
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gmc.h
@@ -0,0 +1,129 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+#ifndef __AMDGPU_GMC_H__
+#define __AMDGPU_GMC_H__
+
+#include <linux/types.h>
+
+#include "amdgpu_irq.h"
+
+struct firmware;
+
+/*
+ * VMHUB structures, functions & helpers
+ */
+struct amdgpu_vmhub {
+	uint32_t	ctx0_ptb_addr_lo32;
+	uint32_t	ctx0_ptb_addr_hi32;
+	uint32_t	vm_inv_eng0_req;
+	uint32_t	vm_inv_eng0_ack;
+	uint32_t	vm_context0_cntl;
+	uint32_t	vm_l2_pro_fault_status;
+	uint32_t	vm_l2_pro_fault_cntl;
+};
+
+/*
+ * GPU MC structures, functions & helpers
+ */
+struct amdgpu_gmc_funcs {
+	/* flush the vm tlb via mmio */
+	void (*flush_gpu_tlb)(struct amdgpu_device *adev,
+			      uint32_t vmid);
+	/* flush the vm tlb via ring */
+	uint64_t (*emit_flush_gpu_tlb)(struct amdgpu_ring *ring, unsigned vmid,
+				       uint64_t pd_addr);
+	/* Change the VMID -> PASID mapping */
+	void (*emit_pasid_mapping)(struct amdgpu_ring *ring, unsigned vmid,
+				   unsigned pasid);
+	/* write pte/pde updates using the cpu */
+	int (*set_pte_pde)(struct amdgpu_device *adev,
+			   void *cpu_pt_addr, /* cpu addr of page table */
+			   uint32_t gpu_page_idx, /* pte/pde to update */
+			   uint64_t addr, /* addr to write into pte/pde */
+			   uint64_t flags); /* access flags */
+	/* enable/disable PRT support */
+	void (*set_prt)(struct amdgpu_device *adev, bool enable);
+	/* set pte flags based per asic */
+	uint64_t (*get_vm_pte_flags)(struct amdgpu_device *adev,
+				     uint32_t flags);
+	/* get the pde for a given mc addr */
+	void (*get_vm_pde)(struct amdgpu_device *adev, int level,
+			   u64 *dst, u64 *flags);
+};
+
+struct amdgpu_gmc {
+	resource_size_t		aper_size;
+	resource_size_t		aper_base;
+	/* for some chips with <= 32MB we need to lie
+	 * about vram size near mc fb location */
+	u64			mc_vram_size;
+	u64			visible_vram_size;
+	u64			gart_size;
+	u64			gart_start;
+	u64			gart_end;
+	u64			vram_start;
+	u64			vram_end;
+	unsigned		vram_width;
+	u64			real_vram_size;
+	int			vram_mtrr;
+	u64                     mc_mask;
+	const struct firmware   *fw;	/* MC firmware */
+	uint32_t                fw_version;
+	struct amdgpu_irq_src	vm_fault;
+	uint32_t		vram_type;
+	uint32_t                srbm_soft_reset;
+	bool			prt_warning;
+	uint64_t		stolen_size;
+	/* apertures */
+	u64			shared_aperture_start;
+	u64			shared_aperture_end;
+	u64			private_aperture_start;
+	u64			private_aperture_end;
+	/* protects concurrent invalidation */
+	spinlock_t		invalidate_lock;
+	bool			translate_further;
+	struct kfd_vm_fault_info *vm_fault_info;
+	atomic_t		vm_fault_info_updated;
+
+	const struct amdgpu_gmc_funcs	*gmc_funcs;
+};
+
+/**
+ * amdgpu_gmc_vram_full_visible - Check if full VRAM is visible through the BAR
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns:
+ * True if full VRAM is visible through the BAR
+ */
+static inline bool amdgpu_gmc_vram_full_visible(struct amdgpu_gmc *gmc)
+{
+	WARN_ON(gmc->real_vram_size < gmc->visible_vram_size);
+
+	return (gmc->real_vram_size == gmc->visible_vram_size);
+}
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_gtt_mgr.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gtt_mgr.c
new file mode 100644
index 00000000000..c879fff5b22
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_gtt_mgr.c
@@ -0,0 +1,300 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+
+struct amdgpu_gtt_mgr {
+	struct drm_mm mm;
+	spinlock_t lock;
+	atomic64_t available;
+};
+
+struct amdgpu_gtt_node {
+	struct drm_mm_node node;
+	struct ttm_buffer_object *tbo;
+};
+
+/**
+ * amdgpu_gtt_mgr_init - init GTT manager and DRM MM
+ *
+ * @man: TTM memory type manager
+ * @p_size: maximum size of GTT
+ *
+ * Allocate and initialize the GTT manager.
+ */
+static int amdgpu_gtt_mgr_init(struct ttm_mem_type_manager *man,
+			       unsigned long p_size)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
+	struct amdgpu_gtt_mgr *mgr;
+	uint64_t start, size;
+
+	mgr = kzalloc(sizeof(*mgr), GFP_KERNEL);
+	if (!mgr)
+		return -ENOMEM;
+
+	start = AMDGPU_GTT_MAX_TRANSFER_SIZE * AMDGPU_GTT_NUM_TRANSFER_WINDOWS;
+	size = (adev->gmc.gart_size >> PAGE_SHIFT) - start;
+	drm_mm_init(&mgr->mm, start, size);
+	mtx_init(&mgr->lock, IPL_TTY);
+	atomic64_set(&mgr->available, p_size);
+	man->priv = mgr;
+	return 0;
+}
+
+/**
+ * amdgpu_gtt_mgr_fini - free and destroy GTT manager
+ *
+ * @man: TTM memory type manager
+ *
+ * Destroy and free the GTT manager, returns -EBUSY if ranges are still
+ * allocated inside it.
+ */
+static int amdgpu_gtt_mgr_fini(struct ttm_mem_type_manager *man)
+{
+	struct amdgpu_gtt_mgr *mgr = man->priv;
+	spin_lock(&mgr->lock);
+	drm_mm_takedown(&mgr->mm);
+	spin_unlock(&mgr->lock);
+	kfree(mgr);
+	man->priv = NULL;
+	return 0;
+}
+
+/**
+ * amdgpu_gtt_mgr_has_gart_addr - Check if mem has address space
+ *
+ * @mem: the mem object to check
+ *
+ * Check if a mem object has already address space allocated.
+ */
+bool amdgpu_gtt_mgr_has_gart_addr(struct ttm_mem_reg *mem)
+{
+	struct amdgpu_gtt_node *node = mem->mm_node;
+
+	return (node->node.start != AMDGPU_BO_INVALID_OFFSET);
+}
+
+/**
+ * amdgpu_gtt_mgr_alloc - allocate new ranges
+ *
+ * @man: TTM memory type manager
+ * @tbo: TTM BO we need this range for
+ * @place: placement flags and restrictions
+ * @mem: the resulting mem object
+ *
+ * Allocate the address space for a node.
+ */
+static int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
+				struct ttm_buffer_object *tbo,
+				const struct ttm_place *place,
+				struct ttm_mem_reg *mem)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
+	struct amdgpu_gtt_mgr *mgr = man->priv;
+	struct amdgpu_gtt_node *node = mem->mm_node;
+	enum drm_mm_insert_mode mode;
+	unsigned long fpfn, lpfn;
+	int r;
+
+	if (amdgpu_gtt_mgr_has_gart_addr(mem))
+		return 0;
+
+	if (place)
+		fpfn = place->fpfn;
+	else
+		fpfn = 0;
+
+	if (place && place->lpfn)
+		lpfn = place->lpfn;
+	else
+		lpfn = adev->gart.num_cpu_pages;
+
+	mode = DRM_MM_INSERT_BEST;
+	if (place && place->flags & TTM_PL_FLAG_TOPDOWN)
+		mode = DRM_MM_INSERT_HIGH;
+
+	spin_lock(&mgr->lock);
+	r = drm_mm_insert_node_in_range(&mgr->mm, &node->node, mem->num_pages,
+					mem->page_alignment, 0, fpfn, lpfn,
+					mode);
+	spin_unlock(&mgr->lock);
+
+	if (!r)
+		mem->start = node->node.start;
+
+	return r;
+}
+
+/**
+ * amdgpu_gtt_mgr_new - allocate a new node
+ *
+ * @man: TTM memory type manager
+ * @tbo: TTM BO we need this range for
+ * @place: placement flags and restrictions
+ * @mem: the resulting mem object
+ *
+ * Dummy, allocate the node but no space for it yet.
+ */
+static int amdgpu_gtt_mgr_new(struct ttm_mem_type_manager *man,
+			      struct ttm_buffer_object *tbo,
+			      const struct ttm_place *place,
+			      struct ttm_mem_reg *mem)
+{
+	struct amdgpu_gtt_mgr *mgr = man->priv;
+	struct amdgpu_gtt_node *node;
+	int r;
+
+	spin_lock(&mgr->lock);
+	if ((&tbo->mem == mem || tbo->mem.mem_type != TTM_PL_TT) &&
+	    atomic64_read(&mgr->available) < mem->num_pages) {
+		spin_unlock(&mgr->lock);
+		return 0;
+	}
+	atomic64_sub(mem->num_pages, &mgr->available);
+	spin_unlock(&mgr->lock);
+
+	node = kzalloc(sizeof(*node), GFP_KERNEL);
+	if (!node) {
+		r = -ENOMEM;
+		goto err_out;
+	}
+
+	node->node.start = AMDGPU_BO_INVALID_OFFSET;
+	node->node.size = mem->num_pages;
+	node->tbo = tbo;
+	mem->mm_node = node;
+
+	if (place->fpfn || place->lpfn || place->flags & TTM_PL_FLAG_TOPDOWN) {
+		r = amdgpu_gtt_mgr_alloc(man, tbo, place, mem);
+		if (unlikely(r)) {
+			kfree(node);
+			mem->mm_node = NULL;
+			r = 0;
+			goto err_out;
+		}
+	} else {
+		mem->start = node->node.start;
+	}
+
+	return 0;
+err_out:
+	atomic64_add(mem->num_pages, &mgr->available);
+
+	return r;
+}
+
+/**
+ * amdgpu_gtt_mgr_del - free ranges
+ *
+ * @man: TTM memory type manager
+ * @tbo: TTM BO we need this range for
+ * @place: placement flags and restrictions
+ * @mem: TTM memory object
+ *
+ * Free the allocated GTT again.
+ */
+static void amdgpu_gtt_mgr_del(struct ttm_mem_type_manager *man,
+			       struct ttm_mem_reg *mem)
+{
+	struct amdgpu_gtt_mgr *mgr = man->priv;
+	struct amdgpu_gtt_node *node = mem->mm_node;
+
+	if (!node)
+		return;
+
+	spin_lock(&mgr->lock);
+	if (node->node.start != AMDGPU_BO_INVALID_OFFSET)
+		drm_mm_remove_node(&node->node);
+	spin_unlock(&mgr->lock);
+	atomic64_add(mem->num_pages, &mgr->available);
+
+	kfree(node);
+	mem->mm_node = NULL;
+}
+
+/**
+ * amdgpu_gtt_mgr_usage - return usage of GTT domain
+ *
+ * @man: TTM memory type manager
+ *
+ * Return how many bytes are used in the GTT domain
+ */
+uint64_t amdgpu_gtt_mgr_usage(struct ttm_mem_type_manager *man)
+{
+	struct amdgpu_gtt_mgr *mgr = man->priv;
+	s64 result = man->size - atomic64_read(&mgr->available);
+
+	return (result > 0 ? result : 0) * PAGE_SIZE;
+}
+
+int amdgpu_gtt_mgr_recover(struct ttm_mem_type_manager *man)
+{
+	struct amdgpu_gtt_mgr *mgr = man->priv;
+	struct amdgpu_gtt_node *node;
+	struct drm_mm_node *mm_node;
+	int r = 0;
+
+	spin_lock(&mgr->lock);
+	drm_mm_for_each_node(mm_node, &mgr->mm) {
+		node = container_of(mm_node, struct amdgpu_gtt_node, node);
+		r = amdgpu_ttm_recover_gart(node->tbo);
+		if (r)
+			break;
+	}
+	spin_unlock(&mgr->lock);
+
+	return r;
+}
+
+/**
+ * amdgpu_gtt_mgr_debug - dump VRAM table
+ *
+ * @man: TTM memory type manager
+ * @printer: DRM printer to use
+ *
+ * Dump the table content using printk.
+ */
+static void amdgpu_gtt_mgr_debug(struct ttm_mem_type_manager *man,
+				 struct drm_printer *printer)
+{
+	struct amdgpu_gtt_mgr *mgr = man->priv;
+
+	spin_lock(&mgr->lock);
+	drm_mm_print(&mgr->mm, printer);
+	spin_unlock(&mgr->lock);
+
+	drm_printf(printer, "man size:%llu pages, gtt available:%lld pages, usage:%lluMB\n",
+		   man->size, (u64)atomic64_read(&mgr->available),
+		   amdgpu_gtt_mgr_usage(man) >> 20);
+}
+
+const struct ttm_mem_type_manager_func amdgpu_gtt_mgr_func = {
+	.init = amdgpu_gtt_mgr_init,
+	.takedown = amdgpu_gtt_mgr_fini,
+	.get_node = amdgpu_gtt_mgr_new,
+	.put_node = amdgpu_gtt_mgr_del,
+	.debug = amdgpu_gtt_mgr_debug
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_i2c.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_i2c.c
new file mode 100644
index 00000000000..6bf61f8cd22
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_i2c.c
@@ -0,0 +1,501 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <linux/export.h>
+
+#include <drm/drmP.h>
+#include <drm/drm_edid.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_i2c.h"
+#include "amdgpu_atombios.h"
+#include "atom.h"
+#include "atombios_dp.h"
+#include "atombios_i2c.h"
+
+#include <dev/i2c/i2cvar.h>
+#include <dev/i2c/i2c_bitbang.h>
+
+/* bit banging i2c */
+static int amdgpu_i2c_pre_xfer(struct i2c_adapter *i2c_adap)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_get_adapdata(i2c_adap);
+	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
+	uint32_t temp;
+
+	mutex_lock(&i2c->mutex);
+
+	/* switch the pads to ddc mode */
+	if (rec->hw_capable) {
+		temp = RREG32(rec->mask_clk_reg);
+		temp &= ~(1 << 16);
+		WREG32(rec->mask_clk_reg, temp);
+	}
+
+	/* clear the output pin values */
+	temp = RREG32(rec->a_clk_reg) & ~rec->a_clk_mask;
+	WREG32(rec->a_clk_reg, temp);
+
+	temp = RREG32(rec->a_data_reg) & ~rec->a_data_mask;
+	WREG32(rec->a_data_reg, temp);
+
+	/* set the pins to input */
+	temp = RREG32(rec->en_clk_reg) & ~rec->en_clk_mask;
+	WREG32(rec->en_clk_reg, temp);
+
+	temp = RREG32(rec->en_data_reg) & ~rec->en_data_mask;
+	WREG32(rec->en_data_reg, temp);
+
+	/* mask the gpio pins for software use */
+	temp = RREG32(rec->mask_clk_reg) | rec->mask_clk_mask;
+	WREG32(rec->mask_clk_reg, temp);
+	temp = RREG32(rec->mask_clk_reg);
+
+	temp = RREG32(rec->mask_data_reg) | rec->mask_data_mask;
+	WREG32(rec->mask_data_reg, temp);
+	temp = RREG32(rec->mask_data_reg);
+
+	return 0;
+}
+
+static void amdgpu_i2c_post_xfer(struct i2c_adapter *i2c_adap)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_get_adapdata(i2c_adap);
+	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
+	uint32_t temp;
+
+	/* unmask the gpio pins for software use */
+	temp = RREG32(rec->mask_clk_reg) & ~rec->mask_clk_mask;
+	WREG32(rec->mask_clk_reg, temp);
+	temp = RREG32(rec->mask_clk_reg);
+
+	temp = RREG32(rec->mask_data_reg) & ~rec->mask_data_mask;
+	WREG32(rec->mask_data_reg, temp);
+	temp = RREG32(rec->mask_data_reg);
+
+	mutex_unlock(&i2c->mutex);
+}
+
+static int amdgpu_i2c_get_clock(void *i2c_priv)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_priv;
+	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
+	uint32_t val;
+
+	/* read the value off the pin */
+	val = RREG32(rec->y_clk_reg);
+	val &= rec->y_clk_mask;
+
+	return (val != 0);
+}
+
+
+static int amdgpu_i2c_get_data(void *i2c_priv)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_priv;
+	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
+	uint32_t val;
+
+	/* read the value off the pin */
+	val = RREG32(rec->y_data_reg);
+	val &= rec->y_data_mask;
+
+	return (val != 0);
+}
+
+static void amdgpu_i2c_set_clock(void *i2c_priv, int clock)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_priv;
+	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
+	uint32_t val;
+
+	/* set pin direction */
+	val = RREG32(rec->en_clk_reg) & ~rec->en_clk_mask;
+	val |= clock ? 0 : rec->en_clk_mask;
+	WREG32(rec->en_clk_reg, val);
+}
+
+static void amdgpu_i2c_set_data(void *i2c_priv, int data)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_priv;
+	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
+	uint32_t val;
+
+	/* set pin direction */
+	val = RREG32(rec->en_data_reg) & ~rec->en_data_mask;
+	val |= data ? 0 : rec->en_data_mask;
+	WREG32(rec->en_data_reg, val);
+}
+
+void	amdgpu_bb_set_bits(void *, uint32_t);
+void	amdgpu_bb_set_dir(void *, uint32_t);
+uint32_t amdgpu_bb_read_bits(void *);
+
+int	amdgpu_acquire_bus(void *, int);
+void	amdgpu_release_bus(void *, int);
+int	amdgpu_send_start(void *, int);
+int	amdgpu_send_stop(void *, int);
+int	amdgpu_initiate_xfer(void *, i2c_addr_t, int);
+int	amdgpu_read_byte(void *, u_int8_t *, int);
+int	amdgpu_write_byte(void *, u_int8_t, int);
+
+#define AMDGPU_BB_SDA		(1 << I2C_BIT_SDA)
+#define AMDGPU_BB_SCL		(1 << I2C_BIT_SCL)
+
+struct i2c_bitbang_ops amdgpu_bbops = {
+	amdgpu_bb_set_bits,
+	amdgpu_bb_set_dir,
+	amdgpu_bb_read_bits,
+	{ AMDGPU_BB_SDA, AMDGPU_BB_SCL, 0, 0 }
+};
+
+void
+amdgpu_bb_set_bits(void *cookie, uint32_t bits)
+{
+	amdgpu_i2c_set_clock(cookie, bits & AMDGPU_BB_SCL);
+	amdgpu_i2c_set_data(cookie, bits & AMDGPU_BB_SDA);
+}
+
+void
+amdgpu_bb_set_dir(void *cookie, uint32_t bits)
+{
+}
+
+uint32_t
+amdgpu_bb_read_bits(void *cookie)
+{
+	uint32_t bits = 0;
+
+	if (amdgpu_i2c_get_clock(cookie))
+		bits |= AMDGPU_BB_SCL;
+	if (amdgpu_i2c_get_data(cookie))
+		bits |= AMDGPU_BB_SDA;
+
+	return bits;
+}
+
+int
+amdgpu_acquire_bus(void *cookie, int flags)
+{
+	struct amdgpu_i2c_chan *i2c = cookie;
+	amdgpu_i2c_pre_xfer(&i2c->adapter);
+	return (0);
+}
+
+void
+amdgpu_release_bus(void *cookie, int flags)
+{
+	struct amdgpu_i2c_chan *i2c = cookie;
+	amdgpu_i2c_post_xfer(&i2c->adapter);
+}
+
+int
+amdgpu_send_start(void *cookie, int flags)
+{
+	return (i2c_bitbang_send_start(cookie, flags, &amdgpu_bbops));
+}
+
+int
+amdgpu_send_stop(void *cookie, int flags)
+{
+	return (i2c_bitbang_send_stop(cookie, flags, &amdgpu_bbops));
+}
+
+int
+amdgpu_initiate_xfer(void *cookie, i2c_addr_t addr, int flags)
+{
+	return (i2c_bitbang_initiate_xfer(cookie, addr, flags, &amdgpu_bbops));
+}
+
+int
+amdgpu_read_byte(void *cookie, u_int8_t *bytep, int flags)
+{
+	return (i2c_bitbang_read_byte(cookie, bytep, flags, &amdgpu_bbops));
+}
+
+int
+amdgpu_write_byte(void *cookie, u_int8_t byte, int flags)
+{
+	return (i2c_bitbang_write_byte(cookie, byte, flags, &amdgpu_bbops));
+}
+
+static const struct i2c_algorithm amdgpu_atombios_i2c_algo = {
+	.master_xfer = amdgpu_atombios_i2c_xfer,
+	.functionality = amdgpu_atombios_i2c_func,
+};
+
+struct amdgpu_i2c_chan *amdgpu_i2c_create(struct drm_device *dev,
+					  const struct amdgpu_i2c_bus_rec *rec,
+					  const char *name)
+{
+	struct amdgpu_i2c_chan *i2c;
+	int ret;
+
+	/* don't add the mm_i2c bus unless hw_i2c is enabled */
+	if (rec->mm_i2c && (amdgpu_hw_i2c == 0))
+		return NULL;
+
+	i2c = kzalloc(sizeof(struct amdgpu_i2c_chan), GFP_KERNEL);
+	if (i2c == NULL)
+		return NULL;
+
+	i2c->rec = *rec;
+#ifdef __linux__
+	i2c->adapter.owner = THIS_MODULE;
+	i2c->adapter.class = I2C_CLASS_DDC;
+	i2c->adapter.dev.parent = &dev->pdev->dev;
+#endif
+	i2c->dev = dev;
+	i2c_set_adapdata(&i2c->adapter, i2c);
+	rw_init(&i2c->mutex, "agiic");
+	if (rec->hw_capable &&
+	    amdgpu_hw_i2c) {
+		/* hw i2c using atom */
+		snprintf(i2c->adapter.name, sizeof(i2c->adapter.name),
+			 "AMDGPU i2c hw bus %s", name);
+		i2c->adapter.algo = &amdgpu_atombios_i2c_algo;
+		ret = i2c_add_adapter(&i2c->adapter);
+		if (ret)
+			goto out_free;
+	} else {
+		/* set the amdgpu bit adapter */
+		snprintf(i2c->adapter.name, sizeof(i2c->adapter.name),
+			 "AMDGPU i2c bit bus %s", name);
+		i2c->adapter.algo_data = &i2c->bit;
+#ifdef notyet
+		i2c->bit.pre_xfer = amdgpu_i2c_pre_xfer;
+		i2c->bit.post_xfer = amdgpu_i2c_post_xfer;
+		i2c->bit.setsda = amdgpu_i2c_set_data;
+		i2c->bit.setscl = amdgpu_i2c_set_clock;
+		i2c->bit.getsda = amdgpu_i2c_get_data;
+		i2c->bit.getscl = amdgpu_i2c_get_clock;
+		i2c->bit.udelay = 10;
+		i2c->bit.timeout = usecs_to_jiffies(2200);	/* from VESA */
+		i2c->bit.data = i2c;
+#else
+		i2c->bit.ic.ic_cookie = i2c;
+		i2c->bit.ic.ic_acquire_bus = amdgpu_acquire_bus;
+		i2c->bit.ic.ic_release_bus = amdgpu_release_bus;
+		i2c->bit.ic.ic_send_start = amdgpu_send_start;
+		i2c->bit.ic.ic_send_stop = amdgpu_send_stop;
+		i2c->bit.ic.ic_initiate_xfer = amdgpu_initiate_xfer;
+		i2c->bit.ic.ic_read_byte = amdgpu_read_byte;
+		i2c->bit.ic.ic_write_byte = amdgpu_write_byte;
+#endif
+		ret = i2c_bit_add_bus(&i2c->adapter);
+		if (ret) {
+			DRM_ERROR("Failed to register bit i2c %s\n", name);
+			goto out_free;
+		}
+	}
+
+	return i2c;
+out_free:
+	kfree(i2c);
+	return NULL;
+
+}
+
+void amdgpu_i2c_destroy(struct amdgpu_i2c_chan *i2c)
+{
+	if (!i2c)
+		return;
+	WARN_ON(i2c->has_aux);
+	i2c_del_adapter(&i2c->adapter);
+	kfree(i2c);
+}
+
+/* Add the default buses */
+void amdgpu_i2c_init(struct amdgpu_device *adev)
+{
+	if (amdgpu_hw_i2c)
+		DRM_INFO("hw_i2c forced on, you may experience display detection problems!\n");
+
+	amdgpu_atombios_i2c_init(adev);
+}
+
+/* remove all the buses */
+void amdgpu_i2c_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < AMDGPU_MAX_I2C_BUS; i++) {
+		if (adev->i2c_bus[i]) {
+			amdgpu_i2c_destroy(adev->i2c_bus[i]);
+			adev->i2c_bus[i] = NULL;
+		}
+	}
+}
+
+/* Add additional buses */
+void amdgpu_i2c_add(struct amdgpu_device *adev,
+		    const struct amdgpu_i2c_bus_rec *rec,
+		    const char *name)
+{
+	struct drm_device *dev = adev->ddev;
+	int i;
+
+	for (i = 0; i < AMDGPU_MAX_I2C_BUS; i++) {
+		if (!adev->i2c_bus[i]) {
+			adev->i2c_bus[i] = amdgpu_i2c_create(dev, rec, name);
+			return;
+		}
+	}
+}
+
+/* looks up bus based on id */
+struct amdgpu_i2c_chan *
+amdgpu_i2c_lookup(struct amdgpu_device *adev,
+		  const struct amdgpu_i2c_bus_rec *i2c_bus)
+{
+	int i;
+
+	for (i = 0; i < AMDGPU_MAX_I2C_BUS; i++) {
+		if (adev->i2c_bus[i] &&
+		    (adev->i2c_bus[i]->rec.i2c_id == i2c_bus->i2c_id)) {
+			return adev->i2c_bus[i];
+		}
+	}
+	return NULL;
+}
+
+static void amdgpu_i2c_get_byte(struct amdgpu_i2c_chan *i2c_bus,
+				 u8 slave_addr,
+				 u8 addr,
+				 u8 *val)
+{
+	u8 out_buf[2];
+	u8 in_buf[2];
+	struct i2c_msg msgs[] = {
+		{
+			.addr = slave_addr,
+			.flags = 0,
+			.len = 1,
+			.buf = out_buf,
+		},
+		{
+			.addr = slave_addr,
+			.flags = I2C_M_RD,
+			.len = 1,
+			.buf = in_buf,
+		}
+	};
+
+	out_buf[0] = addr;
+	out_buf[1] = 0;
+
+	if (i2c_transfer(&i2c_bus->adapter, msgs, 2) == 2) {
+		*val = in_buf[0];
+		DRM_DEBUG("val = 0x%02x\n", *val);
+	} else {
+		DRM_DEBUG("i2c 0x%02x 0x%02x read failed\n",
+			  addr, *val);
+	}
+}
+
+static void amdgpu_i2c_put_byte(struct amdgpu_i2c_chan *i2c_bus,
+				 u8 slave_addr,
+				 u8 addr,
+				 u8 val)
+{
+	uint8_t out_buf[2];
+	struct i2c_msg msg = {
+		.addr = slave_addr,
+		.flags = 0,
+		.len = 2,
+		.buf = out_buf,
+	};
+
+	out_buf[0] = addr;
+	out_buf[1] = val;
+
+	if (i2c_transfer(&i2c_bus->adapter, &msg, 1) != 1)
+		DRM_DEBUG("i2c 0x%02x 0x%02x write failed\n",
+			  addr, val);
+}
+
+/* ddc router switching */
+void
+amdgpu_i2c_router_select_ddc_port(const struct amdgpu_connector *amdgpu_connector)
+{
+	u8 val;
+
+	if (!amdgpu_connector->router.ddc_valid)
+		return;
+
+	if (!amdgpu_connector->router_bus)
+		return;
+
+	amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x3, &val);
+	val &= ~amdgpu_connector->router.ddc_mux_control_pin;
+	amdgpu_i2c_put_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x3, val);
+	amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x1, &val);
+	val &= ~amdgpu_connector->router.ddc_mux_control_pin;
+	val |= amdgpu_connector->router.ddc_mux_state;
+	amdgpu_i2c_put_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x1, val);
+}
+
+/* clock/data router switching */
+void
+amdgpu_i2c_router_select_cd_port(const struct amdgpu_connector *amdgpu_connector)
+{
+	u8 val;
+
+	if (!amdgpu_connector->router.cd_valid)
+		return;
+
+	if (!amdgpu_connector->router_bus)
+		return;
+
+	amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x3, &val);
+	val &= ~amdgpu_connector->router.cd_mux_control_pin;
+	amdgpu_i2c_put_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x3, val);
+	amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x1, &val);
+	val &= ~amdgpu_connector->router.cd_mux_control_pin;
+	val |= amdgpu_connector->router.cd_mux_state;
+	amdgpu_i2c_put_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x1, val);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_i2c.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_i2c.h
new file mode 100644
index 00000000000..63c2ff7499e
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_i2c.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_I2C_H__
+#define __AMDGPU_I2C_H__
+
+struct amdgpu_i2c_chan *amdgpu_i2c_create(struct drm_device *dev,
+					  const struct amdgpu_i2c_bus_rec *rec,
+					  const char *name);
+void amdgpu_i2c_destroy(struct amdgpu_i2c_chan *i2c);
+void amdgpu_i2c_init(struct amdgpu_device *adev);
+void amdgpu_i2c_fini(struct amdgpu_device *adev);
+void amdgpu_i2c_add(struct amdgpu_device *adev,
+		    const struct amdgpu_i2c_bus_rec *rec,
+		    const char *name);
+struct amdgpu_i2c_chan *
+amdgpu_i2c_lookup(struct amdgpu_device *adev,
+		  const struct amdgpu_i2c_bus_rec *i2c_bus);
+void
+amdgpu_i2c_router_select_ddc_port(const struct amdgpu_connector *connector);
+void
+amdgpu_i2c_router_select_cd_port(const struct amdgpu_connector *connector);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ib.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ib.c
new file mode 100644
index 00000000000..85326b2aaa1
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ib.c
@@ -0,0 +1,413 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ *          Christian König
+ */
+#include <linux/seq_file.h>
+#include <linux/slab.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "atom.h"
+
+#define AMDGPU_IB_TEST_TIMEOUT	msecs_to_jiffies(1000)
+
+/*
+ * IB
+ * IBs (Indirect Buffers) and areas of GPU accessible memory where
+ * commands are stored.  You can put a pointer to the IB in the
+ * command ring and the hw will fetch the commands from the IB
+ * and execute them.  Generally userspace acceleration drivers
+ * produce command buffers which are send to the kernel and
+ * put in IBs for execution by the requested ring.
+ */
+static int amdgpu_debugfs_sa_init(struct amdgpu_device *adev);
+
+/**
+ * amdgpu_ib_get - request an IB (Indirect Buffer)
+ *
+ * @ring: ring index the IB is associated with
+ * @size: requested IB size
+ * @ib: IB object returned
+ *
+ * Request an IB (all asics).  IBs are allocated using the
+ * suballocator.
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		  unsigned size, struct amdgpu_ib *ib)
+{
+	int r;
+
+	if (size) {
+		r = amdgpu_sa_bo_new(&adev->ring_tmp_bo,
+				      &ib->sa_bo, size, 256);
+		if (r) {
+			dev_err(adev->dev, "failed to get a new IB (%d)\n", r);
+			return r;
+		}
+
+		ib->ptr = amdgpu_sa_bo_cpu_addr(ib->sa_bo);
+
+		if (!vm)
+			ib->gpu_addr = amdgpu_sa_bo_gpu_addr(ib->sa_bo);
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_ib_free - free an IB (Indirect Buffer)
+ *
+ * @adev: amdgpu_device pointer
+ * @ib: IB object to free
+ * @f: the fence SA bo need wait on for the ib alloation
+ *
+ * Free an IB (all asics).
+ */
+void amdgpu_ib_free(struct amdgpu_device *adev, struct amdgpu_ib *ib,
+		    struct dma_fence *f)
+{
+	amdgpu_sa_bo_free(adev, &ib->sa_bo, f);
+}
+
+/**
+ * amdgpu_ib_schedule - schedule an IB (Indirect Buffer) on the ring
+ *
+ * @adev: amdgpu_device pointer
+ * @num_ibs: number of IBs to schedule
+ * @ibs: IB objects to schedule
+ * @f: fence created during this submission
+ *
+ * Schedule an IB on the associated ring (all asics).
+ * Returns 0 on success, error on failure.
+ *
+ * On SI, there are two parallel engines fed from the primary ring,
+ * the CE (Constant Engine) and the DE (Drawing Engine).  Since
+ * resource descriptors have moved to memory, the CE allows you to
+ * prime the caches while the DE is updating register state so that
+ * the resource descriptors will be already in cache when the draw is
+ * processed.  To accomplish this, the userspace driver submits two
+ * IBs, one for the CE and one for the DE.  If there is a CE IB (called
+ * a CONST_IB), it will be put on the ring prior to the DE IB.  Prior
+ * to SI there was just a DE IB.
+ */
+int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
+		       struct amdgpu_ib *ibs, struct amdgpu_job *job,
+		       struct dma_fence **f)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib *ib = &ibs[0];
+	struct dma_fence *tmp = NULL;
+	bool skip_preamble, need_ctx_switch;
+	unsigned patch_offset = ~0;
+	struct amdgpu_vm *vm;
+	uint64_t fence_ctx;
+	uint32_t status = 0, alloc_size;
+	unsigned fence_flags = 0;
+
+	unsigned i;
+	int r = 0;
+	bool need_pipe_sync = false;
+
+	if (num_ibs == 0)
+		return -EINVAL;
+
+	/* ring tests don't use a job */
+	if (job) {
+		vm = job->vm;
+		fence_ctx = job->base.s_fence->scheduled.context;
+	} else {
+		vm = NULL;
+		fence_ctx = 0;
+	}
+
+	if (!ring->ready) {
+		dev_err(adev->dev, "couldn't schedule ib on ring <%s>\n", ring->name);
+		return -EINVAL;
+	}
+
+	if (vm && !job->vmid) {
+		dev_err(adev->dev, "VM IB without ID\n");
+		return -EINVAL;
+	}
+
+	alloc_size = ring->funcs->emit_frame_size + num_ibs *
+		ring->funcs->emit_ib_size;
+
+	r = amdgpu_ring_alloc(ring, alloc_size);
+	if (r) {
+		dev_err(adev->dev, "scheduling IB failed (%d).\n", r);
+		return r;
+	}
+
+	need_ctx_switch = ring->current_ctx != fence_ctx;
+	if (ring->funcs->emit_pipeline_sync && job &&
+	    ((tmp = amdgpu_sync_get_fence(&job->sched_sync, NULL)) ||
+	     (amdgpu_sriov_vf(adev) && need_ctx_switch) ||
+	     amdgpu_vm_need_pipeline_sync(ring, job))) {
+		need_pipe_sync = true;
+		dma_fence_put(tmp);
+	}
+
+	if (ring->funcs->insert_start)
+		ring->funcs->insert_start(ring);
+
+	if (job) {
+		r = amdgpu_vm_flush(ring, job, need_pipe_sync);
+		if (r) {
+			amdgpu_ring_undo(ring);
+			return r;
+		}
+	}
+
+	if (job && ring->funcs->init_cond_exec)
+		patch_offset = amdgpu_ring_init_cond_exec(ring);
+
+#ifdef __amd64__
+	if (!(adev->flags & AMD_IS_APU))
+#endif
+	{
+		if (ring->funcs->emit_hdp_flush)
+			amdgpu_ring_emit_hdp_flush(ring);
+		else
+			amdgpu_asic_flush_hdp(adev, ring);
+	}
+
+	skip_preamble = ring->current_ctx == fence_ctx;
+	if (job && ring->funcs->emit_cntxcntl) {
+		if (need_ctx_switch)
+			status |= AMDGPU_HAVE_CTX_SWITCH;
+		status |= job->preamble_status;
+
+		amdgpu_ring_emit_cntxcntl(ring, status);
+	}
+
+	for (i = 0; i < num_ibs; ++i) {
+		ib = &ibs[i];
+
+		/* drop preamble IBs if we don't have a context switch */
+		if ((ib->flags & AMDGPU_IB_FLAG_PREAMBLE) &&
+			skip_preamble &&
+			!(status & AMDGPU_PREAMBLE_IB_PRESENT_FIRST) &&
+			!amdgpu_sriov_vf(adev)) /* for SRIOV preemption, Preamble CE ib must be inserted anyway */
+			continue;
+
+		amdgpu_ring_emit_ib(ring, ib, job ? job->vmid : 0,
+				    need_ctx_switch);
+		need_ctx_switch = false;
+	}
+
+	if (ring->funcs->emit_tmz)
+		amdgpu_ring_emit_tmz(ring, false);
+
+#ifdef __amd64__
+	if (!(adev->flags & AMD_IS_APU))
+#endif
+		amdgpu_asic_invalidate_hdp(adev, ring);
+
+	if (ib->flags & AMDGPU_IB_FLAG_TC_WB_NOT_INVALIDATE)
+		fence_flags |= AMDGPU_FENCE_FLAG_TC_WB_ONLY;
+
+	/* wrap the last IB with fence */
+	if (job && job->uf_addr) {
+		amdgpu_ring_emit_fence(ring, job->uf_addr, job->uf_sequence,
+				       fence_flags | AMDGPU_FENCE_FLAG_64BIT);
+	}
+
+	r = amdgpu_fence_emit(ring, f, fence_flags);
+	if (r) {
+		dev_err(adev->dev, "failed to emit fence (%d)\n", r);
+		if (job && job->vmid)
+			amdgpu_vmid_reset(adev, ring->funcs->vmhub, job->vmid);
+		amdgpu_ring_undo(ring);
+		return r;
+	}
+
+	if (ring->funcs->insert_end)
+		ring->funcs->insert_end(ring);
+
+	if (patch_offset != ~0 && ring->funcs->patch_cond_exec)
+		amdgpu_ring_patch_cond_exec(ring, patch_offset);
+
+	ring->current_ctx = fence_ctx;
+	if (vm && ring->funcs->emit_switch_buffer)
+		amdgpu_ring_emit_switch_buffer(ring);
+	amdgpu_ring_commit(ring);
+	return 0;
+}
+
+/**
+ * amdgpu_ib_pool_init - Init the IB (Indirect Buffer) pool
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the suballocator to manage a pool of memory
+ * for use as IBs (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_ib_pool_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->ib_pool_ready) {
+		return 0;
+	}
+	r = amdgpu_sa_bo_manager_init(adev, &adev->ring_tmp_bo,
+				      AMDGPU_IB_POOL_SIZE*64*1024,
+				      AMDGPU_GPU_PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT);
+	if (r) {
+		return r;
+	}
+
+	adev->ib_pool_ready = true;
+	if (amdgpu_debugfs_sa_init(adev)) {
+		dev_err(adev->dev, "failed to register debugfs file for SA\n");
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_ib_pool_fini - Free the IB (Indirect Buffer) pool
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the suballocator managing the pool of memory
+ * for use as IBs (all asics).
+ */
+void amdgpu_ib_pool_fini(struct amdgpu_device *adev)
+{
+	if (adev->ib_pool_ready) {
+		amdgpu_sa_bo_manager_fini(adev, &adev->ring_tmp_bo);
+		adev->ib_pool_ready = false;
+	}
+}
+
+/**
+ * amdgpu_ib_ring_tests - test IBs on the rings
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Test an IB (Indirect Buffer) on each ring.
+ * If the test fails, disable the ring.
+ * Returns 0 on success, error if the primary GFX ring
+ * IB test fails.
+ */
+int amdgpu_ib_ring_tests(struct amdgpu_device *adev)
+{
+	unsigned i;
+	int r, ret = 0;
+	long tmo_gfx, tmo_mm;
+
+	tmo_mm = tmo_gfx = AMDGPU_IB_TEST_TIMEOUT;
+	if (amdgpu_sriov_vf(adev)) {
+		/* for MM engines in hypervisor side they are not scheduled together
+		 * with CP and SDMA engines, so even in exclusive mode MM engine could
+		 * still running on other VF thus the IB TEST TIMEOUT for MM engines
+		 * under SR-IOV should be set to a long time. 8 sec should be enough
+		 * for the MM comes back to this VF.
+		 */
+		tmo_mm = 8 * AMDGPU_IB_TEST_TIMEOUT;
+	}
+
+	if (amdgpu_sriov_runtime(adev)) {
+		/* for CP & SDMA engines since they are scheduled together so
+		 * need to make the timeout width enough to cover the time
+		 * cost waiting for it coming back under RUNTIME only
+		*/
+		tmo_gfx = 8 * AMDGPU_IB_TEST_TIMEOUT;
+	}
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		long tmo;
+
+		if (!ring || !ring->ready)
+			continue;
+
+		/* MM engine need more time */
+		if (ring->funcs->type == AMDGPU_RING_TYPE_UVD ||
+			ring->funcs->type == AMDGPU_RING_TYPE_VCE ||
+			ring->funcs->type == AMDGPU_RING_TYPE_UVD_ENC ||
+			ring->funcs->type == AMDGPU_RING_TYPE_VCN_DEC ||
+			ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC ||
+			ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG)
+			tmo = tmo_mm;
+		else
+			tmo = tmo_gfx;
+
+		r = amdgpu_ring_test_ib(ring, tmo);
+		if (r) {
+			ring->ready = false;
+
+			if (ring == &adev->gfx.gfx_ring[0]) {
+				/* oh, oh, that's really bad */
+				DRM_ERROR("amdgpu: failed testing IB on GFX ring (%d).\n", r);
+				adev->accel_working = false;
+				return r;
+
+			} else {
+				/* still not good, but we can live with it */
+				DRM_ERROR("amdgpu: failed testing IB on ring %d (%d).\n", i, r);
+				ret = r;
+			}
+		}
+	}
+	return ret;
+}
+
+/*
+ * Debugfs info
+ */
+#if defined(CONFIG_DEBUG_FS)
+
+static int amdgpu_debugfs_sa_info(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	amdgpu_sa_bo_dump_debug_info(&adev->ring_tmp_bo, m);
+
+	return 0;
+
+}
+
+static const struct drm_info_list amdgpu_debugfs_sa_list[] = {
+	{"amdgpu_sa_info", &amdgpu_debugfs_sa_info, 0, NULL},
+};
+
+#endif
+
+static int amdgpu_debugfs_sa_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_sa_list, 1);
+#else
+	return 0;
+#endif
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ids.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ids.c
new file mode 100644
index 00000000000..8bd81538bf4
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ids.c
@@ -0,0 +1,608 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu_ids.h"
+
+#include <linux/idr.h>
+#include <linux/dma-fence-array.h>
+#include <drm/drmP.h>
+
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+/*
+ * PASID manager
+ *
+ * PASIDs are global address space identifiers that can be shared
+ * between the GPU, an IOMMU and the driver. VMs on different devices
+ * may use the same PASID if they share the same address
+ * space. Therefore PASIDs are allocated using a global IDA. VMs are
+ * looked up from the PASID per amdgpu_device.
+ */
+static DEFINE_IDA(amdgpu_pasid_ida);
+
+/* Helper to free pasid from a fence callback */
+struct amdgpu_pasid_cb {
+	struct dma_fence_cb cb;
+	unsigned int pasid;
+};
+
+/**
+ * amdgpu_pasid_alloc - Allocate a PASID
+ * @bits: Maximum width of the PASID in bits, must be at least 1
+ *
+ * Allocates a PASID of the given width while keeping smaller PASIDs
+ * available if possible.
+ *
+ * Returns a positive integer on success. Returns %-EINVAL if bits==0.
+ * Returns %-ENOSPC if no PASID was available. Returns %-ENOMEM on
+ * memory allocation failure.
+ */
+int amdgpu_pasid_alloc(unsigned int bits)
+{
+	int pasid = -EINVAL;
+
+	for (bits = min(bits, 31U); bits > 0; bits--) {
+		pasid = ida_simple_get(&amdgpu_pasid_ida,
+				       1U << (bits - 1), 1U << bits,
+				       GFP_KERNEL);
+		if (pasid != -ENOSPC)
+			break;
+	}
+
+	if (pasid >= 0)
+		trace_amdgpu_pasid_allocated(pasid);
+
+	return pasid;
+}
+
+/**
+ * amdgpu_pasid_free - Free a PASID
+ * @pasid: PASID to free
+ */
+void amdgpu_pasid_free(unsigned int pasid)
+{
+	trace_amdgpu_pasid_freed(pasid);
+	ida_simple_remove(&amdgpu_pasid_ida, pasid);
+}
+
+static void amdgpu_pasid_free_cb(struct dma_fence *fence,
+				 struct dma_fence_cb *_cb)
+{
+	struct amdgpu_pasid_cb *cb =
+		container_of(_cb, struct amdgpu_pasid_cb, cb);
+
+	amdgpu_pasid_free(cb->pasid);
+	dma_fence_put(fence);
+	kfree(cb);
+}
+
+/**
+ * amdgpu_pasid_free_delayed - free pasid when fences signal
+ *
+ * @resv: reservation object with the fences to wait for
+ * @pasid: pasid to free
+ *
+ * Free the pasid only after all the fences in resv are signaled.
+ */
+void amdgpu_pasid_free_delayed(struct reservation_object *resv,
+			       unsigned int pasid)
+{
+	struct dma_fence *fence, **fences;
+	struct amdgpu_pasid_cb *cb;
+	unsigned count;
+	int r;
+
+	r = reservation_object_get_fences_rcu(resv, NULL, &count, &fences);
+	if (r)
+		goto fallback;
+
+	if (count == 0) {
+		amdgpu_pasid_free(pasid);
+		return;
+	}
+
+	if (count == 1) {
+		fence = fences[0];
+		kfree(fences);
+	} else {
+		uint64_t context = dma_fence_context_alloc(1);
+		struct dma_fence_array *array;
+
+		array = dma_fence_array_create(count, fences, context,
+					       1, false);
+		if (!array) {
+			kfree(fences);
+			goto fallback;
+		}
+		fence = &array->base;
+	}
+
+	cb = kmalloc(sizeof(*cb), GFP_KERNEL);
+	if (!cb) {
+		/* Last resort when we are OOM */
+		dma_fence_wait(fence, false);
+		dma_fence_put(fence);
+		amdgpu_pasid_free(pasid);
+	} else {
+		cb->pasid = pasid;
+		if (dma_fence_add_callback(fence, &cb->cb,
+					   amdgpu_pasid_free_cb))
+			amdgpu_pasid_free_cb(fence, &cb->cb);
+	}
+
+	return;
+
+fallback:
+	/* Not enough memory for the delayed delete, as last resort
+	 * block for all the fences to complete.
+	 */
+	reservation_object_wait_timeout_rcu(resv, true, false,
+					    MAX_SCHEDULE_TIMEOUT);
+	amdgpu_pasid_free(pasid);
+}
+
+/*
+ * VMID manager
+ *
+ * VMIDs are a per VMHUB identifier for page tables handling.
+ */
+
+/**
+ * amdgpu_vmid_had_gpu_reset - check if reset occured since last use
+ *
+ * @adev: amdgpu_device pointer
+ * @id: VMID structure
+ *
+ * Check if GPU reset occured since last use of the VMID.
+ */
+bool amdgpu_vmid_had_gpu_reset(struct amdgpu_device *adev,
+			       struct amdgpu_vmid *id)
+{
+	return id->current_gpu_reset_count !=
+		atomic_read(&adev->gpu_reset_counter);
+}
+
+/**
+ * amdgpu_vm_grab_idle - grab idle VMID
+ *
+ * @vm: vm to allocate id for
+ * @ring: ring we want to submit job to
+ * @sync: sync object where we add dependencies
+ * @idle: resulting idle VMID
+ *
+ * Try to find an idle VMID, if none is idle add a fence to wait to the sync
+ * object. Returns -ENOMEM when we are out of memory.
+ */
+static int amdgpu_vmid_grab_idle(struct amdgpu_vm *vm,
+				 struct amdgpu_ring *ring,
+				 struct amdgpu_sync *sync,
+				 struct amdgpu_vmid **idle)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmhub = ring->funcs->vmhub;
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	struct dma_fence **fences;
+	unsigned i;
+	int r;
+
+	if (ring->vmid_wait && !dma_fence_is_signaled(ring->vmid_wait))
+		return amdgpu_sync_fence(adev, sync, ring->vmid_wait, false);
+
+	fences = kmalloc_array(sizeof(void *), id_mgr->num_ids, GFP_KERNEL);
+	if (!fences)
+		return -ENOMEM;
+
+	/* Check if we have an idle VMID */
+	i = 0;
+	list_for_each_entry((*idle), &id_mgr->ids_lru, list) {
+		fences[i] = amdgpu_sync_peek_fence(&(*idle)->active, ring);
+		if (!fences[i])
+			break;
+		++i;
+	}
+
+	/* If we can't find a idle VMID to use, wait till one becomes available */
+	if (&(*idle)->list == &id_mgr->ids_lru) {
+		u64 fence_context = adev->vm_manager.fence_context + ring->idx;
+		unsigned seqno = ++adev->vm_manager.seqno[ring->idx];
+		struct dma_fence_array *array;
+		unsigned j;
+
+		*idle = NULL;
+		for (j = 0; j < i; ++j)
+			dma_fence_get(fences[j]);
+
+		array = dma_fence_array_create(i, fences, fence_context,
+					       seqno, true);
+		if (!array) {
+			for (j = 0; j < i; ++j)
+				dma_fence_put(fences[j]);
+			kfree(fences);
+			return -ENOMEM;
+		}
+
+		r = amdgpu_sync_fence(adev, sync, &array->base, false);
+		dma_fence_put(ring->vmid_wait);
+		ring->vmid_wait = &array->base;
+		return r;
+	}
+	kfree(fences);
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_grab_reserved - try to assign reserved VMID
+ *
+ * @vm: vm to allocate id for
+ * @ring: ring we want to submit job to
+ * @sync: sync object where we add dependencies
+ * @fence: fence protecting ID from reuse
+ * @job: job who wants to use the VMID
+ *
+ * Try to assign a reserved VMID.
+ */
+static int amdgpu_vmid_grab_reserved(struct amdgpu_vm *vm,
+				     struct amdgpu_ring *ring,
+				     struct amdgpu_sync *sync,
+				     struct dma_fence *fence,
+				     struct amdgpu_job *job,
+				     struct amdgpu_vmid **id)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmhub = ring->funcs->vmhub;
+	uint64_t fence_context = adev->fence_context + ring->idx;
+	struct dma_fence *updates = sync->last_vm_update;
+	bool needs_flush = vm->use_cpu_for_update;
+	int r = 0;
+
+	*id = vm->reserved_vmid[vmhub];
+	if (updates && (*id)->flushed_updates &&
+	    updates->context == (*id)->flushed_updates->context &&
+	    !dma_fence_is_later(updates, (*id)->flushed_updates))
+	    updates = NULL;
+
+	if ((*id)->owner != vm->entity.fence_context ||
+	    job->vm_pd_addr != (*id)->pd_gpu_addr ||
+	    updates || !(*id)->last_flush ||
+	    ((*id)->last_flush->context != fence_context &&
+	     !dma_fence_is_signaled((*id)->last_flush))) {
+		struct dma_fence *tmp;
+
+		/* to prevent one context starved by another context */
+		(*id)->pd_gpu_addr = 0;
+		tmp = amdgpu_sync_peek_fence(&(*id)->active, ring);
+		if (tmp) {
+			*id = NULL;
+			r = amdgpu_sync_fence(adev, sync, tmp, false);
+			return r;
+		}
+		needs_flush = true;
+	}
+
+	/* Good we can use this VMID. Remember this submission as
+	* user of the VMID.
+	*/
+	r = amdgpu_sync_fence(ring->adev, &(*id)->active, fence, false);
+	if (r)
+		return r;
+
+	if (updates) {
+		dma_fence_put((*id)->flushed_updates);
+		(*id)->flushed_updates = dma_fence_get(updates);
+	}
+	job->vm_needs_flush = needs_flush;
+	return 0;
+}
+
+/**
+ * amdgpu_vm_grab_used - try to reuse a VMID
+ *
+ * @vm: vm to allocate id for
+ * @ring: ring we want to submit job to
+ * @sync: sync object where we add dependencies
+ * @fence: fence protecting ID from reuse
+ * @job: job who wants to use the VMID
+ * @id: resulting VMID
+ *
+ * Try to reuse a VMID for this submission.
+ */
+static int amdgpu_vmid_grab_used(struct amdgpu_vm *vm,
+				 struct amdgpu_ring *ring,
+				 struct amdgpu_sync *sync,
+				 struct dma_fence *fence,
+				 struct amdgpu_job *job,
+				 struct amdgpu_vmid **id)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmhub = ring->funcs->vmhub;
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	uint64_t fence_context = adev->fence_context + ring->idx;
+	struct dma_fence *updates = sync->last_vm_update;
+	int r;
+
+	job->vm_needs_flush = vm->use_cpu_for_update;
+
+	/* Check if we can use a VMID already assigned to this VM */
+	list_for_each_entry_reverse((*id), &id_mgr->ids_lru, list) {
+		bool needs_flush = vm->use_cpu_for_update;
+		struct dma_fence *flushed;
+
+		/* Check all the prerequisites to using this VMID */
+		if ((*id)->owner != vm->entity.fence_context)
+			continue;
+
+		if ((*id)->pd_gpu_addr != job->vm_pd_addr)
+			continue;
+
+		if (!(*id)->last_flush ||
+		    ((*id)->last_flush->context != fence_context &&
+		     !dma_fence_is_signaled((*id)->last_flush)))
+			needs_flush = true;
+
+		flushed  = (*id)->flushed_updates;
+		if (updates && (!flushed || dma_fence_is_later(updates, flushed)))
+			needs_flush = true;
+
+		/* Concurrent flushes are only possible starting with Vega10 */
+		if (adev->asic_type < CHIP_VEGA10 && needs_flush)
+			continue;
+
+		/* Good, we can use this VMID. Remember this submission as
+		 * user of the VMID.
+		 */
+		r = amdgpu_sync_fence(ring->adev, &(*id)->active, fence, false);
+		if (r)
+			return r;
+
+		if (updates && (!flushed || dma_fence_is_later(updates, flushed))) {
+			dma_fence_put((*id)->flushed_updates);
+			(*id)->flushed_updates = dma_fence_get(updates);
+		}
+
+		job->vm_needs_flush |= needs_flush;
+		return 0;
+	}
+
+	*id = NULL;
+	return 0;
+}
+
+/**
+ * amdgpu_vm_grab_id - allocate the next free VMID
+ *
+ * @vm: vm to allocate id for
+ * @ring: ring we want to submit job to
+ * @sync: sync object where we add dependencies
+ * @fence: fence protecting ID from reuse
+ * @job: job who wants to use the VMID
+ *
+ * Allocate an id for the vm, adding fences to the sync obj as necessary.
+ */
+int amdgpu_vmid_grab(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
+		     struct amdgpu_sync *sync, struct dma_fence *fence,
+		     struct amdgpu_job *job)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmhub = ring->funcs->vmhub;
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	struct amdgpu_vmid *idle = NULL;
+	struct amdgpu_vmid *id = NULL;
+	int r = 0;
+
+	mutex_lock(&id_mgr->lock);
+	r = amdgpu_vmid_grab_idle(vm, ring, sync, &idle);
+	if (r || !idle)
+		goto error;
+
+	if (vm->reserved_vmid[vmhub]) {
+		r = amdgpu_vmid_grab_reserved(vm, ring, sync, fence, job, &id);
+		if (r || !id)
+			goto error;
+	} else {
+		r = amdgpu_vmid_grab_used(vm, ring, sync, fence, job, &id);
+		if (r)
+			goto error;
+
+		if (!id) {
+			struct dma_fence *updates = sync->last_vm_update;
+
+			/* Still no ID to use? Then use the idle one found earlier */
+			id = idle;
+
+			/* Remember this submission as user of the VMID */
+			r = amdgpu_sync_fence(ring->adev, &id->active,
+					      fence, false);
+			if (r)
+				goto error;
+
+			dma_fence_put(id->flushed_updates);
+			id->flushed_updates = dma_fence_get(updates);
+			job->vm_needs_flush = true;
+		}
+
+		list_move_tail(&id->list, &id_mgr->ids_lru);
+	}
+
+	id->pd_gpu_addr = job->vm_pd_addr;
+	id->owner = vm->entity.fence_context;
+
+	if (job->vm_needs_flush) {
+		dma_fence_put(id->last_flush);
+		id->last_flush = NULL;
+	}
+	job->vmid = id - id_mgr->ids;
+	job->pasid = vm->pasid;
+	trace_amdgpu_vm_grab_id(vm, ring, job);
+
+error:
+	mutex_unlock(&id_mgr->lock);
+	return r;
+}
+
+int amdgpu_vmid_alloc_reserved(struct amdgpu_device *adev,
+			       struct amdgpu_vm *vm,
+			       unsigned vmhub)
+{
+	struct amdgpu_vmid_mgr *id_mgr;
+	struct amdgpu_vmid *idle;
+	int r = 0;
+
+	id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	mutex_lock(&id_mgr->lock);
+	if (vm->reserved_vmid[vmhub])
+		goto unlock;
+	if (atomic_inc_return(&id_mgr->reserved_vmid_num) >
+	    AMDGPU_VM_MAX_RESERVED_VMID) {
+		DRM_ERROR("Over limitation of reserved vmid\n");
+		atomic_dec(&id_mgr->reserved_vmid_num);
+		r = -EINVAL;
+		goto unlock;
+	}
+	/* Select the first entry VMID */
+	idle = list_first_entry(&id_mgr->ids_lru, struct amdgpu_vmid, list);
+	list_del_init(&idle->list);
+	vm->reserved_vmid[vmhub] = idle;
+	mutex_unlock(&id_mgr->lock);
+
+	return 0;
+unlock:
+	mutex_unlock(&id_mgr->lock);
+	return r;
+}
+
+void amdgpu_vmid_free_reserved(struct amdgpu_device *adev,
+			       struct amdgpu_vm *vm,
+			       unsigned vmhub)
+{
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+
+	mutex_lock(&id_mgr->lock);
+	if (vm->reserved_vmid[vmhub]) {
+		list_add(&vm->reserved_vmid[vmhub]->list,
+			&id_mgr->ids_lru);
+		vm->reserved_vmid[vmhub] = NULL;
+		atomic_dec(&id_mgr->reserved_vmid_num);
+	}
+	mutex_unlock(&id_mgr->lock);
+}
+
+/**
+ * amdgpu_vmid_reset - reset VMID to zero
+ *
+ * @adev: amdgpu device structure
+ * @vmid: vmid number to use
+ *
+ * Reset saved GDW, GWS and OA to force switch on next flush.
+ */
+void amdgpu_vmid_reset(struct amdgpu_device *adev, unsigned vmhub,
+		       unsigned vmid)
+{
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	struct amdgpu_vmid *id = &id_mgr->ids[vmid];
+
+	mutex_lock(&id_mgr->lock);
+	id->owner = 0;
+	id->gds_base = 0;
+	id->gds_size = 0;
+	id->gws_base = 0;
+	id->gws_size = 0;
+	id->oa_base = 0;
+	id->oa_size = 0;
+	mutex_unlock(&id_mgr->lock);
+}
+
+/**
+ * amdgpu_vmid_reset_all - reset VMID to zero
+ *
+ * @adev: amdgpu device structure
+ *
+ * Reset VMID to force flush on next use
+ */
+void amdgpu_vmid_reset_all(struct amdgpu_device *adev)
+{
+	unsigned i, j;
+
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) {
+		struct amdgpu_vmid_mgr *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		for (j = 1; j < id_mgr->num_ids; ++j)
+			amdgpu_vmid_reset(adev, i, j);
+	}
+}
+
+/**
+ * amdgpu_vmid_mgr_init - init the VMID manager
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the VM manager structures
+ */
+void amdgpu_vmid_mgr_init(struct amdgpu_device *adev)
+{
+	unsigned i, j;
+
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) {
+		struct amdgpu_vmid_mgr *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		rw_init(&id_mgr->lock, "idmgr");
+		INIT_LIST_HEAD(&id_mgr->ids_lru);
+		atomic_set(&id_mgr->reserved_vmid_num, 0);
+
+		/* skip over VMID 0, since it is the system VM */
+		for (j = 1; j < id_mgr->num_ids; ++j) {
+			amdgpu_vmid_reset(adev, i, j);
+			amdgpu_sync_create(&id_mgr->ids[j].active);
+			list_add_tail(&id_mgr->ids[j].list, &id_mgr->ids_lru);
+		}
+	}
+}
+
+/**
+ * amdgpu_vmid_mgr_fini - cleanup VM manager
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Cleanup the VM manager and free resources.
+ */
+void amdgpu_vmid_mgr_fini(struct amdgpu_device *adev)
+{
+	unsigned i, j;
+
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) {
+		struct amdgpu_vmid_mgr *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		mutex_destroy(&id_mgr->lock);
+		for (j = 0; j < AMDGPU_NUM_VMID; ++j) {
+			struct amdgpu_vmid *id = &id_mgr->ids[j];
+
+			amdgpu_sync_free(&id->active);
+			dma_fence_put(id->flushed_updates);
+			dma_fence_put(id->last_flush);
+			dma_fence_put(id->pasid_mapping);
+		}
+	}
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ids.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ids.h
new file mode 100644
index 00000000000..9b5b84d4950
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ids.h
@@ -0,0 +1,96 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_IDS_H__
+#define __AMDGPU_IDS_H__
+
+#include <linux/types.h>
+#include <linux/mutex.h>
+#include <linux/list.h>
+#include <linux/dma-fence.h>
+
+#include "amdgpu_sync.h"
+
+/* maximum number of VMIDs */
+#define AMDGPU_NUM_VMID	16
+
+struct amdgpu_device;
+struct amdgpu_vm;
+struct amdgpu_ring;
+struct amdgpu_sync;
+struct amdgpu_job;
+
+struct amdgpu_vmid {
+	struct list_head	list;
+	struct amdgpu_sync	active;
+	struct dma_fence	*last_flush;
+	uint64_t		owner;
+
+	uint64_t		pd_gpu_addr;
+	/* last flushed PD/PT update */
+	struct dma_fence	*flushed_updates;
+
+	uint32_t                current_gpu_reset_count;
+
+	uint32_t		gds_base;
+	uint32_t		gds_size;
+	uint32_t		gws_base;
+	uint32_t		gws_size;
+	uint32_t		oa_base;
+	uint32_t		oa_size;
+
+	unsigned		pasid;
+	struct dma_fence	*pasid_mapping;
+};
+
+struct amdgpu_vmid_mgr {
+	struct rwlock		lock;
+	unsigned		num_ids;
+	struct list_head	ids_lru;
+	struct amdgpu_vmid	ids[AMDGPU_NUM_VMID];
+	atomic_t		reserved_vmid_num;
+};
+
+int amdgpu_pasid_alloc(unsigned int bits);
+void amdgpu_pasid_free(unsigned int pasid);
+void amdgpu_pasid_free_delayed(struct reservation_object *resv,
+			       unsigned int pasid);
+
+bool amdgpu_vmid_had_gpu_reset(struct amdgpu_device *adev,
+			       struct amdgpu_vmid *id);
+int amdgpu_vmid_alloc_reserved(struct amdgpu_device *adev,
+			       struct amdgpu_vm *vm,
+			       unsigned vmhub);
+void amdgpu_vmid_free_reserved(struct amdgpu_device *adev,
+			       struct amdgpu_vm *vm,
+			       unsigned vmhub);
+int amdgpu_vmid_grab(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
+		     struct amdgpu_sync *sync, struct dma_fence *fence,
+		     struct amdgpu_job *job);
+void amdgpu_vmid_reset(struct amdgpu_device *adev, unsigned vmhub,
+		       unsigned vmid);
+void amdgpu_vmid_reset_all(struct amdgpu_device *adev);
+
+void amdgpu_vmid_mgr_init(struct amdgpu_device *adev);
+void amdgpu_vmid_mgr_fini(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ih.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ih.c
new file mode 100644
index 00000000000..589d1767365
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ih.c
@@ -0,0 +1,292 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_amdkfd.h"
+
+/**
+ * amdgpu_ih_ring_alloc - allocate memory for the IH ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller.
+ * Returns 0 for success, errors for failure.
+ */
+static int amdgpu_ih_ring_alloc(struct amdgpu_device *adev)
+{
+	int r;
+
+	/* Allocate ring buffer */
+	if (adev->irq.ih.ring_obj == NULL) {
+		r = amdgpu_bo_create_kernel(adev, adev->irq.ih.ring_size,
+					    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+					    &adev->irq.ih.ring_obj,
+					    &adev->irq.ih.gpu_addr,
+					    (void **)&adev->irq.ih.ring);
+		if (r) {
+			DRM_ERROR("amdgpu: failed to create ih ring buffer (%d).\n", r);
+			return r;
+		}
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_ih_ring_init - initialize the IH state
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initializes the IH state and allocates a buffer
+ * for the IH ring buffer.
+ * Returns 0 for success, errors for failure.
+ */
+int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
+			bool use_bus_addr)
+{
+	u32 rb_bufsz;
+	int r;
+	struct drm_dmamem *dmah;
+	int flags = 0;
+
+	/* Align ring size */
+	rb_bufsz = order_base_2(ring_size / 4);
+	ring_size = (1 << rb_bufsz) * 4;
+	adev->irq.ih.ring_size = ring_size;
+	adev->irq.ih.ptr_mask = adev->irq.ih.ring_size - 1;
+	adev->irq.ih.rptr = 0;
+	adev->irq.ih.use_bus_addr = use_bus_addr;
+
+	if (adev->irq.ih.use_bus_addr) {
+		if (!adev->irq.ih.ring) {
+			/* add 8 bytes for the rptr/wptr shadows and
+			 * add them to the end of the ring allocation.
+			 */
+#ifdef __linux__
+			adev->irq.ih.ring = pci_alloc_consistent(adev->pdev,
+								 adev->irq.ih.ring_size + 8,
+								 &adev->irq.ih.rb_dma_addr);
+			if (adev->irq.ih.ring == NULL)
+				return -ENOMEM;
+#else
+			dmah = drm_dmamem_alloc(adev->dmat,
+			    adev->irq.ih.ring_size + 8,
+			    PAGE_SIZE, 1,
+			    adev->irq.ih.ring_size + 8, flags, 0);
+			if (dmah == NULL)
+				return -ENOMEM;
+			adev->irq.ih.dmah = dmah;
+			adev->irq.ih.rb_dma_addr = dmah->map->dm_segs[0].ds_addr;
+			adev->irq.ih.ring = (volatile uint32_t *)dmah->kva;
+#endif
+			memset((void *)adev->irq.ih.ring, 0, adev->irq.ih.ring_size + 8);
+			adev->irq.ih.wptr_offs = (adev->irq.ih.ring_size / 4) + 0;
+			adev->irq.ih.rptr_offs = (adev->irq.ih.ring_size / 4) + 1;
+		}
+		return 0;
+	} else {
+		r = amdgpu_device_wb_get(adev, &adev->irq.ih.wptr_offs);
+		if (r) {
+			dev_err(adev->dev, "(%d) ih wptr_offs wb alloc failed\n", r);
+			return r;
+		}
+
+		r = amdgpu_device_wb_get(adev, &adev->irq.ih.rptr_offs);
+		if (r) {
+			amdgpu_device_wb_free(adev, adev->irq.ih.wptr_offs);
+			dev_err(adev->dev, "(%d) ih rptr_offs wb alloc failed\n", r);
+			return r;
+		}
+
+		return amdgpu_ih_ring_alloc(adev);
+	}
+}
+
+/**
+ * amdgpu_ih_ring_fini - tear down the IH state
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tears down the IH state and frees buffer
+ * used for the IH ring buffer.
+ */
+void amdgpu_ih_ring_fini(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih.use_bus_addr) {
+		if (adev->irq.ih.ring) {
+			/* add 8 bytes for the rptr/wptr shadows and
+			 * add them to the end of the ring allocation.
+			 */
+#ifdef __linux__
+			pci_free_consistent(adev->pdev, adev->irq.ih.ring_size + 8,
+					    (void *)adev->irq.ih.ring,
+					    adev->irq.ih.rb_dma_addr);
+#else
+			drm_dmamem_free(adev->dmat, adev->irq.ih.dmah);
+#endif
+			adev->irq.ih.ring = NULL;
+		}
+	} else {
+		amdgpu_bo_free_kernel(&adev->irq.ih.ring_obj,
+				      &adev->irq.ih.gpu_addr,
+				      (void **)&adev->irq.ih.ring);
+		amdgpu_device_wb_free(adev, adev->irq.ih.wptr_offs);
+		amdgpu_device_wb_free(adev, adev->irq.ih.rptr_offs);
+	}
+}
+
+/**
+ * amdgpu_ih_process - interrupt handler
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Interrupt hander (VI), walk the IH ring.
+ * Returns irq process return code.
+ */
+int amdgpu_ih_process(struct amdgpu_device *adev)
+{
+	struct amdgpu_iv_entry entry;
+	u32 wptr;
+
+	if (!adev->irq.ih.enabled || adev->shutdown)
+		return IRQ_NONE;
+
+	wptr = amdgpu_ih_get_wptr(adev);
+
+restart_ih:
+	/* is somebody else already processing irqs? */
+	if (atomic_xchg(&adev->irq.ih.lock, 1))
+		return IRQ_NONE;
+
+	DRM_DEBUG("%s: rptr %d, wptr %d\n", __func__, adev->irq.ih.rptr, wptr);
+
+	/* Order reading of wptr vs. reading of IH ring data */
+	rmb();
+
+	while (adev->irq.ih.rptr != wptr) {
+		u32 ring_index = adev->irq.ih.rptr >> 2;
+
+		/* Prescreening of high-frequency interrupts */
+		if (!amdgpu_ih_prescreen_iv(adev)) {
+			adev->irq.ih.rptr &= adev->irq.ih.ptr_mask;
+			continue;
+		}
+
+		/* Before dispatching irq to IP blocks, send it to amdkfd */
+		amdgpu_amdkfd_interrupt(adev,
+				(const void *) &adev->irq.ih.ring[ring_index]);
+
+		entry.iv_entry = (const uint32_t *)
+			&adev->irq.ih.ring[ring_index];
+		amdgpu_ih_decode_iv(adev, &entry);
+		adev->irq.ih.rptr &= adev->irq.ih.ptr_mask;
+
+		amdgpu_irq_dispatch(adev, &entry);
+	}
+	amdgpu_ih_set_rptr(adev);
+	atomic_set(&adev->irq.ih.lock, 0);
+
+	/* make sure wptr hasn't changed while processing */
+	wptr = amdgpu_ih_get_wptr(adev);
+	if (wptr != adev->irq.ih.rptr)
+		goto restart_ih;
+
+	return IRQ_HANDLED;
+}
+
+/**
+ * amdgpu_ih_add_fault - Add a page fault record
+ *
+ * @adev: amdgpu device pointer
+ * @key: 64-bit encoding of PASID and address
+ *
+ * This should be called when a retry page fault interrupt is
+ * received. If this is a new page fault, it will be added to a hash
+ * table. The return value indicates whether this is a new fault, or
+ * a fault that was already known and is already being handled.
+ *
+ * If there are too many pending page faults, this will fail. Retry
+ * interrupts should be ignored in this case until there is enough
+ * free space.
+ *
+ * Returns 0 if the fault was added, 1 if the fault was already known,
+ * -ENOSPC if there are too many pending faults.
+ */
+int amdgpu_ih_add_fault(struct amdgpu_device *adev, u64 key)
+{
+	unsigned long flags;
+	int r = -ENOSPC;
+
+	if (WARN_ON_ONCE(!adev->irq.ih.faults))
+		/* Should be allocated in <IP>_ih_sw_init on GPUs that
+		 * support retry faults and require retry filtering.
+		 */
+		return r;
+
+	spin_lock_irqsave(&adev->irq.ih.faults->lock, flags);
+
+	/* Only let the hash table fill up to 50% for best performance */
+	if (adev->irq.ih.faults->count >= (1 << (AMDGPU_PAGEFAULT_HASH_BITS-1)))
+		goto unlock_out;
+
+	r = chash_table_copy_in(&adev->irq.ih.faults->hash, key, NULL);
+	if (!r)
+		adev->irq.ih.faults->count++;
+
+	/* chash_table_copy_in should never fail unless we're losing count */
+	WARN_ON_ONCE(r < 0);
+
+unlock_out:
+	spin_unlock_irqrestore(&adev->irq.ih.faults->lock, flags);
+	return r;
+}
+
+/**
+ * amdgpu_ih_clear_fault - Remove a page fault record
+ *
+ * @adev: amdgpu device pointer
+ * @key: 64-bit encoding of PASID and address
+ *
+ * This should be called when a page fault has been handled. Any
+ * future interrupt with this key will be processed as a new
+ * page fault.
+ */
+void amdgpu_ih_clear_fault(struct amdgpu_device *adev, u64 key)
+{
+	unsigned long flags;
+	int r;
+
+	if (!adev->irq.ih.faults)
+		return;
+
+	spin_lock_irqsave(&adev->irq.ih.faults->lock, flags);
+
+	r = chash_table_remove(&adev->irq.ih.faults->hash, key, NULL);
+	if (!WARN_ON_ONCE(r < 0)) {
+		adev->irq.ih.faults->count--;
+		WARN_ON_ONCE(adev->irq.ih.faults->count < 0);
+	}
+
+	spin_unlock_irqrestore(&adev->irq.ih.faults->lock, flags);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ih.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ih.h
new file mode 100644
index 00000000000..726485b3ece
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ih.h
@@ -0,0 +1,87 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_IH_H__
+#define __AMDGPU_IH_H__
+
+#include <linux/chash.h>
+#include "soc15_ih_clientid.h"
+
+struct amdgpu_device;
+
+#define AMDGPU_IH_CLIENTID_LEGACY 0
+#define AMDGPU_IH_CLIENTID_MAX SOC15_IH_CLIENTID_MAX
+
+#define AMDGPU_PAGEFAULT_HASH_BITS 8
+struct amdgpu_retryfault_hashtable {
+	DECLARE_CHASH_TABLE(hash, AMDGPU_PAGEFAULT_HASH_BITS, 8, 0);
+	spinlock_t	lock;
+	int		count;
+};
+
+/*
+ * R6xx+ IH ring
+ */
+struct amdgpu_ih_ring {
+	struct amdgpu_bo	*ring_obj;
+	volatile uint32_t	*ring;
+	struct drm_dmamem	*dmah;
+	unsigned		rptr;
+	unsigned		ring_size;
+	uint64_t		gpu_addr;
+	uint32_t		ptr_mask;
+	atomic_t		lock;
+	bool                    enabled;
+	unsigned		wptr_offs;
+	unsigned		rptr_offs;
+	u32			doorbell_index;
+	bool			use_doorbell;
+	bool			use_bus_addr;
+	dma_addr_t		rb_dma_addr; /* only used when use_bus_addr = true */
+	struct amdgpu_retryfault_hashtable *faults;
+};
+
+#define AMDGPU_IH_SRC_DATA_MAX_SIZE_DW 4
+
+struct amdgpu_iv_entry {
+	unsigned client_id;
+	unsigned src_id;
+	unsigned ring_id;
+	unsigned vmid;
+	unsigned vmid_src;
+	uint64_t timestamp;
+	unsigned timestamp_src;
+	unsigned pasid;
+	unsigned pasid_src;
+	unsigned src_data[AMDGPU_IH_SRC_DATA_MAX_SIZE_DW];
+	const uint32_t *iv_entry;
+};
+
+int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
+			bool use_bus_addr);
+void amdgpu_ih_ring_fini(struct amdgpu_device *adev);
+int amdgpu_ih_process(struct amdgpu_device *adev);
+int amdgpu_ih_add_fault(struct amdgpu_device *adev, u64 key);
+void amdgpu_ih_clear_fault(struct amdgpu_device *adev, u64 key);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ioc32.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ioc32.c
new file mode 100644
index 00000000000..26482914dc4
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ioc32.c
@@ -0,0 +1,47 @@
+/**
+ * \file amdgpu_ioc32.c
+ *
+ * 32-bit ioctl compatibility routines for the AMDGPU DRM.
+ *
+ * \author Paul Mackerras <paulus@samba.org>
+ *
+ * Copyright (C) Paul Mackerras 2005
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+#include <linux/compat.h>
+
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu_drv.h"
+
+long amdgpu_kms_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
+{
+	unsigned int nr = DRM_IOCTL_NR(cmd);
+	int ret;
+
+	if (nr < DRM_COMMAND_BASE)
+		return drm_compat_ioctl(filp, cmd, arg);
+
+	ret = amdgpu_drm_ioctl(filp, cmd, arg);
+
+	return ret;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_irq.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_irq.c
new file mode 100644
index 00000000000..3d8fc4b4abd
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_irq.c
@@ -0,0 +1,673 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+
+/**
+ * DOC: Interrupt Handling
+ *
+ * Interrupts generated within GPU hardware raise interrupt requests that are
+ * passed to amdgpu IRQ handler which is responsible for detecting source and
+ * type of the interrupt and dispatching matching handlers. If handling an
+ * interrupt requires calling kernel functions that may sleep processing is
+ * dispatched to work handlers.
+ *
+ * If MSI functionality is not disabled by module parameter then MSI
+ * support will be enabled.
+ *
+ * For GPU interrupt sources that may be driven by another driver, IRQ domain
+ * support is used (with mapping between virtual and hardware IRQs).
+ */
+
+#include <linux/irq.h>
+#include <drm/drmP.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "atom.h"
+#include "amdgpu_connectors.h"
+#include "amdgpu_trace.h"
+
+#include <linux/pm_runtime.h>
+
+#ifdef CONFIG_DRM_AMD_DC
+#include "amdgpu_dm_irq.h"
+#endif
+
+#define AMDGPU_WAIT_IDLE_TIMEOUT 200
+
+/**
+ * amdgpu_hotplug_work_func - work handler for display hotplug event
+ *
+ * @work: work struct pointer
+ *
+ * This is the hotplug event work handler (all ASICs).
+ * The work gets scheduled from the IRQ handler if there
+ * was a hotplug interrupt.  It walks through the connector table
+ * and calls hotplug handler for each connector. After this, it sends
+ * a DRM hotplug event to alert userspace.
+ *
+ * This design approach is required in order to defer hotplug event handling
+ * from the IRQ handler to a work handler because hotplug handler has to use
+ * mutexes which cannot be locked in an IRQ handler (since &mutex_lock may
+ * sleep).
+ */
+static void amdgpu_hotplug_work_func(struct work_struct *work)
+{
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
+						  hotplug_work);
+	struct drm_device *dev = adev->ddev;
+	struct drm_mode_config *mode_config = &dev->mode_config;
+	struct drm_connector *connector;
+
+	mutex_lock(&mode_config->mutex);
+	list_for_each_entry(connector, &mode_config->connector_list, head)
+		amdgpu_connector_hotplug(connector);
+	mutex_unlock(&mode_config->mutex);
+	/* Just fire off a uevent and let userspace tell us what to do */
+	drm_helper_hpd_irq_event(dev);
+}
+
+/**
+ * amdgpu_irq_reset_work_func - execute GPU reset
+ *
+ * @work: work struct pointer
+ *
+ * Execute scheduled GPU reset (Cayman+).
+ * This function is called when the IRQ handler thinks we need a GPU reset.
+ */
+static void amdgpu_irq_reset_work_func(struct work_struct *work)
+{
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
+						  reset_work);
+
+	if (!amdgpu_sriov_vf(adev))
+		amdgpu_device_gpu_recover(adev, NULL, false);
+}
+
+/**
+ * amdgpu_irq_disable_all - disable *all* interrupts
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Disable all types of interrupts from all sources.
+ */
+void amdgpu_irq_disable_all(struct amdgpu_device *adev)
+{
+	unsigned long irqflags;
+	unsigned i, j, k;
+	int r;
+
+	spin_lock_irqsave(&adev->irq.lock, irqflags);
+	for (i = 0; i < AMDGPU_IH_CLIENTID_MAX; ++i) {
+		if (!adev->irq.client[i].sources)
+			continue;
+
+		for (j = 0; j < AMDGPU_MAX_IRQ_SRC_ID; ++j) {
+			struct amdgpu_irq_src *src = adev->irq.client[i].sources[j];
+
+			if (!src || !src->funcs->set || !src->num_types)
+				continue;
+
+			for (k = 0; k < src->num_types; ++k) {
+				atomic_set(&src->enabled_types[k], 0);
+				r = src->funcs->set(adev, src, k,
+						    AMDGPU_IRQ_STATE_DISABLE);
+				if (r)
+					DRM_ERROR("error disabling interrupt (%d)\n",
+						  r);
+			}
+		}
+	}
+	spin_unlock_irqrestore(&adev->irq.lock, irqflags);
+}
+
+/**
+ * amdgpu_irq_handler - IRQ handler
+ *
+ * @irq: IRQ number (unused)
+ * @arg: pointer to DRM device
+ *
+ * IRQ handler for amdgpu driver (all ASICs).
+ *
+ * Returns:
+ * result of handling the IRQ, as defined by &irqreturn_t
+ */
+irqreturn_t amdgpu_irq_handler(void *arg)
+{
+	struct drm_device *dev = (struct drm_device *) arg;
+	struct amdgpu_device *adev = dev->dev_private;
+	irqreturn_t ret;
+
+	if (!adev->irq.installed)
+		return 0;
+
+	ret = amdgpu_ih_process(adev);
+	if (ret == IRQ_HANDLED)
+		pm_runtime_mark_last_busy(dev->dev);
+	return ret;
+}
+
+/**
+ * amdgpu_msi_ok - check whether MSI functionality is enabled
+ *
+ * @adev: amdgpu device pointer (unused)
+ *
+ * Checks whether MSI functionality has been disabled via module parameter
+ * (all ASICs).
+ *
+ * Returns:
+ * *true* if MSIs are allowed to be enabled or *false* otherwise
+ */
+bool amdgpu_msi_ok(struct amdgpu_device *adev)
+{
+	if (amdgpu_msi == 1)
+		return true;
+	else if (amdgpu_msi == 0)
+		return false;
+
+	return true;
+}
+
+/**
+ * amdgpu_irq_init - initialize interrupt handling
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Sets up work functions for hotplug and reset interrupts, enables MSI
+ * functionality, initializes vblank, hotplug and reset interrupt handling.
+ *
+ * Returns:
+ * 0 on success or error code on failure
+ */
+int amdgpu_irq_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	mtx_init(&adev->irq.lock, IPL_TTY);
+
+#ifdef notyet
+	/* Enable MSI if not disabled by module parameter */
+	adev->irq.msi_enabled = false;
+
+	if (amdgpu_msi_ok(adev)) {
+		int ret = pci_enable_msi(adev->pdev);
+		if (!ret) {
+			adev->irq.msi_enabled = true;
+			dev_dbg(adev->dev, "amdgpu: using MSI.\n");
+		}
+	}
+#endif
+
+	if (!amdgpu_device_has_dc_support(adev)) {
+		if (!adev->enable_virtual_display)
+			/* Disable vblank IRQs aggressively for power-saving */
+			/* XXX: can this be enabled for DC? */
+			adev->ddev->vblank_disable_immediate = true;
+
+		r = drm_vblank_init(adev->ddev, adev->mode_info.num_crtc);
+		if (r)
+			return r;
+
+		/* Pre-DCE11 */
+		INIT_WORK(&adev->hotplug_work,
+				amdgpu_hotplug_work_func);
+	}
+
+	INIT_WORK(&adev->reset_work, amdgpu_irq_reset_work_func);
+
+	adev->irq.installed = true;
+	r = drm_irq_install(adev->ddev, adev->ddev->pdev->irq);
+	if (r) {
+		adev->irq.installed = false;
+		if (!amdgpu_device_has_dc_support(adev))
+			flush_work(&adev->hotplug_work);
+		cancel_work_sync(&adev->reset_work);
+		return r;
+	}
+	adev->ddev->max_vblank_count = 0x00ffffff;
+
+	DRM_DEBUG("amdgpu: irq initialized.\n");
+	return 0;
+}
+
+/**
+ * amdgpu_irq_fini - shut down interrupt handling
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Tears down work functions for hotplug and reset interrupts, disables MSI
+ * functionality, shuts down vblank, hotplug and reset interrupt handling,
+ * turns off interrupts from all sources (all ASICs).
+ */
+void amdgpu_irq_fini(struct amdgpu_device *adev)
+{
+	unsigned i, j;
+
+	if (adev->irq.installed) {
+		drm_irq_uninstall(adev->ddev);
+		adev->irq.installed = false;
+		if (adev->irq.msi_enabled)
+			pci_disable_msi(adev->pdev);
+		if (!amdgpu_device_has_dc_support(adev))
+			flush_work(&adev->hotplug_work);
+		cancel_work_sync(&adev->reset_work);
+	}
+
+	for (i = 0; i < AMDGPU_IH_CLIENTID_MAX; ++i) {
+		if (!adev->irq.client[i].sources)
+			continue;
+
+		for (j = 0; j < AMDGPU_MAX_IRQ_SRC_ID; ++j) {
+			struct amdgpu_irq_src *src = adev->irq.client[i].sources[j];
+
+			if (!src)
+				continue;
+
+			kfree(src->enabled_types);
+			src->enabled_types = NULL;
+			if (src->data) {
+				kfree(src->data);
+				kfree(src);
+				adev->irq.client[i].sources[j] = NULL;
+			}
+		}
+		kfree(adev->irq.client[i].sources);
+		adev->irq.client[i].sources = NULL;
+	}
+}
+
+/**
+ * amdgpu_irq_add_id - register IRQ source
+ *
+ * @adev: amdgpu device pointer
+ * @client_id: client id
+ * @src_id: source id
+ * @source: IRQ source pointer
+ *
+ * Registers IRQ source on a client.
+ *
+ * Returns:
+ * 0 on success or error code otherwise
+ */
+int amdgpu_irq_add_id(struct amdgpu_device *adev,
+		      unsigned client_id, unsigned src_id,
+		      struct amdgpu_irq_src *source)
+{
+	if (client_id >= AMDGPU_IH_CLIENTID_MAX)
+		return -EINVAL;
+
+	if (src_id >= AMDGPU_MAX_IRQ_SRC_ID)
+		return -EINVAL;
+
+	if (!source->funcs)
+		return -EINVAL;
+
+	if (!adev->irq.client[client_id].sources) {
+		adev->irq.client[client_id].sources =
+			kcalloc(AMDGPU_MAX_IRQ_SRC_ID,
+				sizeof(struct amdgpu_irq_src *),
+				GFP_KERNEL);
+		if (!adev->irq.client[client_id].sources)
+			return -ENOMEM;
+	}
+
+	if (adev->irq.client[client_id].sources[src_id] != NULL)
+		return -EINVAL;
+
+	if (source->num_types && !source->enabled_types) {
+		atomic_t *types;
+
+		types = kcalloc(source->num_types, sizeof(atomic_t),
+				GFP_KERNEL);
+		if (!types)
+			return -ENOMEM;
+
+		source->enabled_types = types;
+	}
+
+	adev->irq.client[client_id].sources[src_id] = source;
+	return 0;
+}
+
+/**
+ * amdgpu_irq_dispatch - dispatch IRQ to IP blocks
+ *
+ * @adev: amdgpu device pointer
+ * @entry: interrupt vector pointer
+ *
+ * Dispatches IRQ to IP blocks.
+ */
+void amdgpu_irq_dispatch(struct amdgpu_device *adev,
+			 struct amdgpu_iv_entry *entry)
+{
+	unsigned client_id = entry->client_id;
+	unsigned src_id = entry->src_id;
+	struct amdgpu_irq_src *src;
+	int r;
+
+	trace_amdgpu_iv(entry);
+
+	if (client_id >= AMDGPU_IH_CLIENTID_MAX) {
+		DRM_DEBUG("Invalid client_id in IV: %d\n", client_id);
+		return;
+	}
+
+	if (src_id >= AMDGPU_MAX_IRQ_SRC_ID) {
+		DRM_DEBUG("Invalid src_id in IV: %d\n", src_id);
+		return;
+	}
+
+	if (adev->irq.virq[src_id]) {
+		STUB();
+#ifdef notyet
+		generic_handle_irq(irq_find_mapping(adev->irq.domain, src_id));
+#endif
+	} else {
+		if (!adev->irq.client[client_id].sources) {
+			DRM_DEBUG("Unregistered interrupt client_id: %d src_id: %d\n",
+				  client_id, src_id);
+			return;
+		}
+
+		src = adev->irq.client[client_id].sources[src_id];
+		if (!src) {
+			DRM_DEBUG("Unhandled interrupt src_id: %d\n", src_id);
+			return;
+		}
+
+		r = src->funcs->process(adev, src, entry);
+		if (r)
+			DRM_ERROR("error processing interrupt (%d)\n", r);
+	}
+}
+
+/**
+ * amdgpu_irq_update - update hardware interrupt state
+ *
+ * @adev: amdgpu device pointer
+ * @src: interrupt source pointer
+ * @type: type of interrupt
+ *
+ * Updates interrupt state for the specific source (all ASICs).
+ */
+int amdgpu_irq_update(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *src, unsigned type)
+{
+	unsigned long irqflags;
+	enum amdgpu_interrupt_state state;
+	int r;
+
+	spin_lock_irqsave(&adev->irq.lock, irqflags);
+
+	/* We need to determine after taking the lock, otherwise
+	   we might disable just enabled interrupts again */
+	if (amdgpu_irq_enabled(adev, src, type))
+		state = AMDGPU_IRQ_STATE_ENABLE;
+	else
+		state = AMDGPU_IRQ_STATE_DISABLE;
+
+	r = src->funcs->set(adev, src, type, state);
+	spin_unlock_irqrestore(&adev->irq.lock, irqflags);
+	return r;
+}
+
+/**
+ * amdgpu_irq_gpu_reset_resume_helper - update interrupt states on all sources
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Updates state of all types of interrupts on all sources on resume after
+ * reset.
+ */
+void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev)
+{
+	int i, j, k;
+
+	for (i = 0; i < AMDGPU_IH_CLIENTID_MAX; ++i) {
+		if (!adev->irq.client[i].sources)
+			continue;
+
+		for (j = 0; j < AMDGPU_MAX_IRQ_SRC_ID; ++j) {
+			struct amdgpu_irq_src *src = adev->irq.client[i].sources[j];
+
+			if (!src)
+				continue;
+			for (k = 0; k < src->num_types; k++)
+				amdgpu_irq_update(adev, src, k);
+		}
+	}
+}
+
+/**
+ * amdgpu_irq_get - enable interrupt
+ *
+ * @adev: amdgpu device pointer
+ * @src: interrupt source pointer
+ * @type: type of interrupt
+ *
+ * Enables specified type of interrupt on the specified source (all ASICs).
+ *
+ * Returns:
+ * 0 on success or error code otherwise
+ */
+int amdgpu_irq_get(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+		   unsigned type)
+{
+	if (!adev->ddev->irq_enabled)
+		return -ENOENT;
+
+	if (type >= src->num_types)
+		return -EINVAL;
+
+	if (!src->enabled_types || !src->funcs->set)
+		return -EINVAL;
+
+	if (atomic_inc_return(&src->enabled_types[type]) == 1)
+		return amdgpu_irq_update(adev, src, type);
+
+	return 0;
+}
+
+/**
+ * amdgpu_irq_put - disable interrupt
+ *
+ * @adev: amdgpu device pointer
+ * @src: interrupt source pointer
+ * @type: type of interrupt
+ *
+ * Enables specified type of interrupt on the specified source (all ASICs).
+ *
+ * Returns:
+ * 0 on success or error code otherwise
+ */
+int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+		   unsigned type)
+{
+	if (!adev->ddev->irq_enabled)
+		return -ENOENT;
+
+	if (type >= src->num_types)
+		return -EINVAL;
+
+	if (!src->enabled_types || !src->funcs->set)
+		return -EINVAL;
+
+	if (atomic_dec_and_test(&src->enabled_types[type]))
+		return amdgpu_irq_update(adev, src, type);
+
+	return 0;
+}
+
+/**
+ * amdgpu_irq_enabled - check whether interrupt is enabled or not
+ *
+ * @adev: amdgpu device pointer
+ * @src: interrupt source pointer
+ * @type: type of interrupt
+ *
+ * Checks whether the given type of interrupt is enabled on the given source.
+ *
+ * Returns:
+ * *true* if interrupt is enabled, *false* if interrupt is disabled or on
+ * invalid parameters
+ */
+bool amdgpu_irq_enabled(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+			unsigned type)
+{
+	if (!adev->ddev->irq_enabled)
+		return false;
+
+	if (type >= src->num_types)
+		return false;
+
+	if (!src->enabled_types || !src->funcs->set)
+		return false;
+
+	return !!atomic_read(&src->enabled_types[type]);
+}
+
+#ifdef __linux__
+/* XXX: Generic IRQ handling */
+static void amdgpu_irq_mask(struct irq_data *irqd)
+{
+	/* XXX */
+}
+
+static void amdgpu_irq_unmask(struct irq_data *irqd)
+{
+	/* XXX */
+}
+
+/* amdgpu hardware interrupt chip descriptor */
+static struct irq_chip amdgpu_irq_chip = {
+	.name = "amdgpu-ih",
+	.irq_mask = amdgpu_irq_mask,
+	.irq_unmask = amdgpu_irq_unmask,
+};
+#endif
+
+#ifdef __linux__
+/**
+ * amdgpu_irqdomain_map - create mapping between virtual and hardware IRQ numbers
+ *
+ * @d: amdgpu IRQ domain pointer (unused)
+ * @irq: virtual IRQ number
+ * @hwirq: hardware irq number
+ *
+ * Current implementation assigns simple interrupt handler to the given virtual
+ * IRQ.
+ *
+ * Returns:
+ * 0 on success or error code otherwise
+ */
+static int amdgpu_irqdomain_map(struct irq_domain *d,
+				unsigned int irq, irq_hw_number_t hwirq)
+{
+	if (hwirq >= AMDGPU_MAX_IRQ_SRC_ID)
+		return -EPERM;
+
+	irq_set_chip_and_handler(irq,
+				 &amdgpu_irq_chip, handle_simple_irq);
+	return 0;
+}
+
+/* Implementation of methods for amdgpu IRQ domain */
+static const struct irq_domain_ops amdgpu_hw_irqdomain_ops = {
+	.map = amdgpu_irqdomain_map,
+};
+#endif
+
+/**
+ * amdgpu_irq_add_domain - create a linear IRQ domain
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Creates an IRQ domain for GPU interrupt sources
+ * that may be driven by another driver (e.g., ACP).
+ *
+ * Returns:
+ * 0 on success or error code otherwise
+ */
+int amdgpu_irq_add_domain(struct amdgpu_device *adev)
+{
+	STUB();
+	return 0;
+#if 0
+	adev->irq.domain = irq_domain_add_linear(NULL, AMDGPU_MAX_IRQ_SRC_ID,
+						 &amdgpu_hw_irqdomain_ops, adev);
+	if (!adev->irq.domain) {
+		DRM_ERROR("GPU irq add domain failed\n");
+		return -ENODEV;
+	}
+
+	return 0;
+#endif
+}
+
+/**
+ * amdgpu_irq_remove_domain - remove the IRQ domain
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Removes the IRQ domain for GPU interrupt sources
+ * that may be driven by another driver (e.g., ACP).
+ */
+void amdgpu_irq_remove_domain(struct amdgpu_device *adev)
+{
+	STUB();
+#if 0
+	if (adev->irq.domain) {
+		irq_domain_remove(adev->irq.domain);
+		adev->irq.domain = NULL;
+	}
+#endif
+}
+
+/**
+ * amdgpu_irq_create_mapping - create mapping between domain Linux IRQs
+ *
+ * @adev: amdgpu device pointer
+ * @src_id: IH source id
+ *
+ * Creates mapping between a domain IRQ (GPU IH src id) and a Linux IRQ
+ * Use this for components that generate a GPU interrupt, but are driven
+ * by a different driver (e.g., ACP).
+ *
+ * Returns:
+ * Linux IRQ
+ */
+unsigned amdgpu_irq_create_mapping(struct amdgpu_device *adev, unsigned src_id)
+{
+	STUB();
+	return 0;
+#if 0
+	adev->irq.virq[src_id] = irq_create_mapping(adev->irq.domain, src_id);
+
+	return adev->irq.virq[src_id];
+#endif
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_irq.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_irq.h
new file mode 100644
index 00000000000..e6cb4ebd954
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_irq.h
@@ -0,0 +1,105 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_IRQ_H__
+#define __AMDGPU_IRQ_H__
+
+#include <linux/irqdomain.h>
+#include "amdgpu_ih.h"
+
+#define AMDGPU_MAX_IRQ_SRC_ID	0x100
+#define AMDGPU_MAX_IRQ_CLIENT_ID	0x100
+
+struct amdgpu_device;
+struct amdgpu_iv_entry;
+
+enum amdgpu_interrupt_state {
+	AMDGPU_IRQ_STATE_DISABLE,
+	AMDGPU_IRQ_STATE_ENABLE,
+};
+
+struct amdgpu_irq_src {
+	unsigned				num_types;
+	atomic_t				*enabled_types;
+	const struct amdgpu_irq_src_funcs	*funcs;
+	void *data;
+};
+
+struct amdgpu_irq_client {
+	struct amdgpu_irq_src **sources;
+};
+
+/* provided by interrupt generating IP blocks */
+struct amdgpu_irq_src_funcs {
+	int (*set)(struct amdgpu_device *adev, struct amdgpu_irq_src *source,
+		   unsigned type, enum amdgpu_interrupt_state state);
+
+	int (*process)(struct amdgpu_device *adev,
+		       struct amdgpu_irq_src *source,
+		       struct amdgpu_iv_entry *entry);
+};
+
+struct amdgpu_irq {
+	bool				installed;
+	spinlock_t			lock;
+	/* interrupt sources */
+	struct amdgpu_irq_client	client[AMDGPU_IH_CLIENTID_MAX];
+
+	/* status, etc. */
+	bool				msi_enabled; /* msi enabled */
+
+	/* interrupt ring */
+	struct amdgpu_ih_ring		ih;
+	const struct amdgpu_ih_funcs	*ih_funcs;
+
+	/* gen irq stuff */
+	struct irq_domain		*domain; /* GPU irq controller domain */
+	unsigned			virq[AMDGPU_MAX_IRQ_SRC_ID];
+	uint32_t                        srbm_soft_reset;
+};
+
+void amdgpu_irq_disable_all(struct amdgpu_device *adev);
+irqreturn_t amdgpu_irq_handler(void *arg);
+
+int amdgpu_irq_init(struct amdgpu_device *adev);
+void amdgpu_irq_fini(struct amdgpu_device *adev);
+int amdgpu_irq_add_id(struct amdgpu_device *adev,
+		      unsigned client_id, unsigned src_id,
+		      struct amdgpu_irq_src *source);
+void amdgpu_irq_dispatch(struct amdgpu_device *adev,
+			 struct amdgpu_iv_entry *entry);
+int amdgpu_irq_update(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+		      unsigned type);
+int amdgpu_irq_get(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+		   unsigned type);
+int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+		   unsigned type);
+bool amdgpu_irq_enabled(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+			unsigned type);
+void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev);
+
+int amdgpu_irq_add_domain(struct amdgpu_device *adev);
+void amdgpu_irq_remove_domain(struct amdgpu_device *adev);
+unsigned amdgpu_irq_create_mapping(struct amdgpu_device *adev, unsigned src_id);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_job.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_job.c
new file mode 100644
index 00000000000..f823d4baf04
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_job.c
@@ -0,0 +1,239 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ */
+#include <linux/kthread.h>
+#include <linux/wait.h>
+#include <linux/sched.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+static void amdgpu_job_timedout(struct drm_sched_job *s_job)
+{
+	struct amdgpu_ring *ring = to_amdgpu_ring(s_job->sched);
+	struct amdgpu_job *job = to_amdgpu_job(s_job);
+
+	DRM_ERROR("ring %s timeout, signaled seq=%u, emitted seq=%u\n",
+		  job->base.sched->name, atomic_read(&ring->fence_drv.last_seq),
+		  ring->fence_drv.sync_seq);
+
+	amdgpu_device_gpu_recover(ring->adev, job, false);
+}
+
+int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs,
+		     struct amdgpu_job **job, struct amdgpu_vm *vm)
+{
+	size_t size = sizeof(struct amdgpu_job);
+
+	if (num_ibs == 0)
+		return -EINVAL;
+
+	size += sizeof(struct amdgpu_ib) * num_ibs;
+
+	*job = kzalloc(size, GFP_KERNEL);
+	if (!*job)
+		return -ENOMEM;
+
+	/*
+	 * Initialize the scheduler to at least some ring so that we always
+	 * have a pointer to adev.
+	 */
+	(*job)->base.sched = &adev->rings[0]->sched;
+	(*job)->vm = vm;
+	(*job)->ibs = (void *)&(*job)[1];
+	(*job)->num_ibs = num_ibs;
+
+	amdgpu_sync_create(&(*job)->sync);
+	amdgpu_sync_create(&(*job)->sched_sync);
+	(*job)->vram_lost_counter = atomic_read(&adev->vram_lost_counter);
+	(*job)->vm_pd_addr = AMDGPU_BO_INVALID_OFFSET;
+
+	return 0;
+}
+
+int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, unsigned size,
+			     struct amdgpu_job **job)
+{
+	int r;
+
+	r = amdgpu_job_alloc(adev, 1, job, NULL);
+	if (r)
+		return r;
+
+	r = amdgpu_ib_get(adev, NULL, size, &(*job)->ibs[0]);
+	if (r)
+		kfree(*job);
+	else
+		(*job)->vm_pd_addr = adev->gart.table_addr;
+
+	return r;
+}
+
+void amdgpu_job_free_resources(struct amdgpu_job *job)
+{
+	struct amdgpu_ring *ring = to_amdgpu_ring(job->base.sched);
+	struct dma_fence *f;
+	unsigned i;
+
+	/* use sched fence if available */
+	f = job->base.s_fence ? &job->base.s_fence->finished : job->fence;
+
+	for (i = 0; i < job->num_ibs; ++i)
+		amdgpu_ib_free(ring->adev, &job->ibs[i], f);
+}
+
+static void amdgpu_job_free_cb(struct drm_sched_job *s_job)
+{
+	struct amdgpu_ring *ring = to_amdgpu_ring(s_job->sched);
+	struct amdgpu_job *job = to_amdgpu_job(s_job);
+
+	amdgpu_ring_priority_put(ring, s_job->s_priority);
+	dma_fence_put(job->fence);
+	amdgpu_sync_free(&job->sync);
+	amdgpu_sync_free(&job->sched_sync);
+	kfree(job);
+}
+
+void amdgpu_job_free(struct amdgpu_job *job)
+{
+	amdgpu_job_free_resources(job);
+
+	dma_fence_put(job->fence);
+	amdgpu_sync_free(&job->sync);
+	amdgpu_sync_free(&job->sched_sync);
+	kfree(job);
+}
+
+int amdgpu_job_submit(struct amdgpu_job *job, struct drm_sched_entity *entity,
+		      void *owner, struct dma_fence **f)
+{
+	enum drm_sched_priority priority;
+	struct amdgpu_ring *ring;
+	int r;
+
+	if (!f)
+		return -EINVAL;
+
+	r = drm_sched_job_init(&job->base, entity, owner);
+	if (r)
+		return r;
+
+	job->owner = owner;
+	*f = dma_fence_get(&job->base.s_fence->finished);
+	amdgpu_job_free_resources(job);
+	priority = job->base.s_priority;
+	drm_sched_entity_push_job(&job->base, entity);
+
+	ring = to_amdgpu_ring(entity->rq->sched);
+	amdgpu_ring_priority_get(ring, priority);
+
+	return 0;
+}
+
+int amdgpu_job_submit_direct(struct amdgpu_job *job, struct amdgpu_ring *ring,
+			     struct dma_fence **fence)
+{
+	int r;
+
+	job->base.sched = &ring->sched;
+	r = amdgpu_ib_schedule(ring, job->num_ibs, job->ibs, NULL, fence);
+	job->fence = dma_fence_get(*fence);
+	if (r)
+		return r;
+
+	amdgpu_job_free(job);
+	return 0;
+}
+
+static struct dma_fence *amdgpu_job_dependency(struct drm_sched_job *sched_job,
+					       struct drm_sched_entity *s_entity)
+{
+	struct amdgpu_ring *ring = to_amdgpu_ring(s_entity->rq->sched);
+	struct amdgpu_job *job = to_amdgpu_job(sched_job);
+	struct amdgpu_vm *vm = job->vm;
+	struct dma_fence *fence;
+	bool explicit = false;
+	int r;
+
+	fence = amdgpu_sync_get_fence(&job->sync, &explicit);
+	if (fence && explicit) {
+		if (drm_sched_dependency_optimized(fence, s_entity)) {
+			r = amdgpu_sync_fence(ring->adev, &job->sched_sync,
+					      fence, false);
+			if (r)
+				DRM_ERROR("Error adding fence (%d)\n", r);
+		}
+	}
+
+	while (fence == NULL && vm && !job->vmid) {
+		r = amdgpu_vmid_grab(vm, ring, &job->sync,
+				     &job->base.s_fence->finished,
+				     job);
+		if (r)
+			DRM_ERROR("Error getting VM ID (%d)\n", r);
+
+		fence = amdgpu_sync_get_fence(&job->sync, NULL);
+	}
+
+	return fence;
+}
+
+static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job)
+{
+	struct amdgpu_ring *ring = to_amdgpu_ring(sched_job->sched);
+	struct dma_fence *fence = NULL, *finished;
+	struct amdgpu_job *job;
+	int r;
+
+	job = to_amdgpu_job(sched_job);
+	finished = &job->base.s_fence->finished;
+
+	BUG_ON(amdgpu_sync_peek_fence(&job->sync, NULL));
+
+	trace_amdgpu_sched_run_job(job);
+
+	if (job->vram_lost_counter != atomic_read(&ring->adev->vram_lost_counter))
+		dma_fence_set_error(finished, -ECANCELED);/* skip IB as well if VRAM lost */
+
+	if (finished->error < 0) {
+		DRM_INFO("Skip scheduling IBs!\n");
+	} else {
+		r = amdgpu_ib_schedule(ring, job->num_ibs, job->ibs, job,
+				       &fence);
+		if (r)
+			DRM_ERROR("Error scheduling IBs (%d)\n", r);
+	}
+	/* if gpu reset, hw fence will be replaced here */
+	dma_fence_put(job->fence);
+	job->fence = dma_fence_get(fence);
+
+	amdgpu_job_free_resources(job);
+	return fence;
+}
+
+const struct drm_sched_backend_ops amdgpu_sched_ops = {
+	.dependency = amdgpu_job_dependency,
+	.run_job = amdgpu_job_run,
+	.timedout_job = amdgpu_job_timedout,
+	.free_job = amdgpu_job_free_cb
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_job.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_job.h
new file mode 100644
index 00000000000..57cfe78a262
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_job.h
@@ -0,0 +1,74 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_JOB_H__
+#define __AMDGPU_JOB_H__
+
+/* bit set means command submit involves a preamble IB */
+#define AMDGPU_PREAMBLE_IB_PRESENT          (1 << 0)
+/* bit set means preamble IB is first presented in belonging context */
+#define AMDGPU_PREAMBLE_IB_PRESENT_FIRST    (1 << 1)
+/* bit set means context switch occured */
+#define AMDGPU_HAVE_CTX_SWITCH              (1 << 2)
+
+#define to_amdgpu_job(sched_job)		\
+		container_of((sched_job), struct amdgpu_job, base)
+
+struct amdgpu_fence;
+
+struct amdgpu_job {
+	struct drm_sched_job    base;
+	struct amdgpu_vm	*vm;
+	struct amdgpu_sync	sync;
+	struct amdgpu_sync	sched_sync;
+	struct amdgpu_ib	*ibs;
+	struct dma_fence	*fence; /* the hw fence */
+	uint32_t		preamble_status;
+	uint32_t		num_ibs;
+	void			*owner;
+	bool                    vm_needs_flush;
+	uint64_t		vm_pd_addr;
+	unsigned		vmid;
+	unsigned		pasid;
+	uint32_t		gds_base, gds_size;
+	uint32_t		gws_base, gws_size;
+	uint32_t		oa_base, oa_size;
+	uint32_t		vram_lost_counter;
+
+	/* user fence handling */
+	uint64_t		uf_addr;
+	uint64_t		uf_sequence;
+
+};
+
+int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs,
+		     struct amdgpu_job **job, struct amdgpu_vm *vm);
+int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, unsigned size,
+			     struct amdgpu_job **job);
+
+void amdgpu_job_free_resources(struct amdgpu_job *job);
+void amdgpu_job_free(struct amdgpu_job *job);
+int amdgpu_job_submit(struct amdgpu_job *job, struct drm_sched_entity *entity,
+		      void *owner, struct dma_fence **f);
+int amdgpu_job_submit_direct(struct amdgpu_job *job, struct amdgpu_ring *ring,
+			     struct dma_fence **fence);
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_kms.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_kms.c
new file mode 100644
index 00000000000..21a4020313f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_kms.c
@@ -0,0 +1,1928 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include <drm/amdgpu_drm.h>
+#include "amdgpu_sched.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+#include "atom.h"
+
+#include <linux/vga_switcheroo.h>
+#include <linux/slab.h>
+#include <linux/pm_runtime.h>
+#include "amdgpu_amdkfd.h"
+
+#include "vga.h"
+
+#if NVGA > 0
+#include <dev/ic/mc6845reg.h>
+#include <dev/ic/pcdisplayvar.h>
+#include <dev/ic/vgareg.h>
+#include <dev/ic/vgavar.h>
+
+extern int vga_console_attached;
+#endif
+
+#ifdef __amd64__
+#include "efifb.h"
+#include <machine/biosvar.h>
+#endif
+
+#if NEFIFB > 0
+#include <machine/efifbvar.h>
+#endif
+
+int     amdgpu_probe(struct device *, void *, void *);
+void    amdgpu_attach(struct device *, struct device *, void *);
+int     amdgpu_detach(struct device *, int);
+int     amdgpu_activate(struct device *, int);
+void    amdgpu_attachhook(struct device *);
+int     amdgpu_forcedetach(struct amdgpu_device *);
+
+bool	amdgpu_msi_ok(struct amdgpu_device *);
+
+extern const struct drm_pcidev amdgpu_pciidlist[];
+extern struct drm_driver amdgpu_kms_driver;
+extern int amdgpu_exp_hw_support;
+
+/*
+ * set if the mountroot hook has a fatal error
+ * such as not being able to find the firmware
+ */
+int amdgpu_fatal_error;
+
+struct cfattach amdgpu_ca = {
+        sizeof (struct amdgpu_device), amdgpu_probe, amdgpu_attach,
+        amdgpu_detach, amdgpu_activate
+};
+
+struct cfdriver amdgpu_cd = {
+        NULL, "amdgpu", DV_DULL
+};
+
+#ifdef __linux__
+/**
+ * amdgpu_driver_unload_kms - Main unload function for KMS.
+ *
+ * @dev: drm dev pointer
+ *
+ * This is the main unload function for KMS (all asics).
+ * Returns 0 on success.
+ */
+void amdgpu_driver_unload_kms(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (adev == NULL)
+		return;
+
+	if (adev->rmmio == NULL)
+		goto done_free;
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_request_full_gpu(adev, false);
+
+	if (amdgpu_device_is_px(dev)) {
+		pm_runtime_get_sync(dev->dev);
+		pm_runtime_forbid(dev->dev);
+	}
+
+	amdgpu_acpi_fini(adev);
+
+	amdgpu_device_fini(adev);
+
+done_free:
+	kfree(adev);
+	dev->dev_private = NULL;
+}
+
+/**
+ * amdgpu_driver_load_kms - Main load function for KMS.
+ *
+ * @dev: drm dev pointer
+ * @flags: device flags
+ *
+ * This is the main load function for KMS (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags)
+{
+	struct amdgpu_device *adev;
+	int r, acpi_status;
+
+#ifdef CONFIG_DRM_AMDGPU_SI
+	if (!amdgpu_si_support) {
+		switch (flags & AMD_ASIC_MASK) {
+		case CHIP_TAHITI:
+		case CHIP_PITCAIRN:
+		case CHIP_VERDE:
+		case CHIP_OLAND:
+		case CHIP_HAINAN:
+			dev_info(dev->dev,
+				 "SI support provided by radeon.\n");
+			dev_info(dev->dev,
+				 "Use radeon.si_support=0 amdgpu.si_support=1 to override.\n"
+				);
+			return -ENODEV;
+		}
+	}
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	if (!amdgpu_cik_support) {
+		switch (flags & AMD_ASIC_MASK) {
+		case CHIP_KAVERI:
+		case CHIP_BONAIRE:
+		case CHIP_HAWAII:
+		case CHIP_KABINI:
+		case CHIP_MULLINS:
+			dev_info(dev->dev,
+				 "CIK support provided by radeon.\n");
+			dev_info(dev->dev,
+				 "Use radeon.cik_support=0 amdgpu.cik_support=1 to override.\n"
+				);
+			return -ENODEV;
+		}
+	}
+#endif
+
+	adev = kzalloc(sizeof(struct amdgpu_device), GFP_KERNEL);
+	if (adev == NULL) {
+		return -ENOMEM;
+	}
+	dev->dev_private = (void *)adev;
+
+	if ((amdgpu_runtime_pm != 0) &&
+	    amdgpu_has_atpx() &&
+	    (amdgpu_is_atpx_hybrid() ||
+	     amdgpu_has_atpx_dgpu_power_cntl()) &&
+	    ((flags & AMD_IS_APU) == 0) &&
+	    !pci_is_thunderbolt_attached(dev->pdev))
+		flags |= AMD_IS_PX;
+
+	/* amdgpu_device_init should report only fatal error
+	 * like memory allocation failure or iomapping failure,
+	 * or memory manager initialization failure, it must
+	 * properly initialize the GPU MC controller and permit
+	 * VRAM allocation
+	 */
+	r = amdgpu_device_init(adev, dev, dev->pdev, flags);
+	if (r) {
+		dev_err(&dev->pdev->dev, "Fatal error during GPU init\n");
+		goto out;
+	}
+
+	/* Call ACPI methods: require modeset init
+	 * but failure is not fatal
+	 */
+	if (!r) {
+		acpi_status = amdgpu_acpi_init(adev);
+		if (acpi_status)
+		dev_dbg(&dev->pdev->dev,
+				"Error during ACPI methods call\n");
+	}
+
+	if (amdgpu_device_is_px(dev)) {
+		dev_pm_set_driver_flags(dev->dev, DPM_FLAG_NEVER_SKIP);
+		pm_runtime_use_autosuspend(dev->dev);
+		pm_runtime_set_autosuspend_delay(dev->dev, 5000);
+		pm_runtime_set_active(dev->dev);
+		pm_runtime_allow(dev->dev);
+		pm_runtime_mark_last_busy(dev->dev);
+		pm_runtime_put_autosuspend(dev->dev);
+	}
+
+out:
+	if (r) {
+		/* balance pm_runtime_get_sync in amdgpu_driver_unload_kms */
+		if (adev->rmmio && amdgpu_device_is_px(dev))
+			pm_runtime_put_noidle(dev->dev);
+		amdgpu_driver_unload_kms(dev);
+	}
+
+	return r;
+}
+#endif /* __linux__ */
+
+static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
+				struct drm_amdgpu_query_fw *query_fw,
+				struct amdgpu_device *adev)
+{
+	switch (query_fw->fw_type) {
+	case AMDGPU_INFO_FW_VCE:
+		fw_info->ver = adev->vce.fw_version;
+		fw_info->feature = adev->vce.fb_version;
+		break;
+	case AMDGPU_INFO_FW_UVD:
+		fw_info->ver = adev->uvd.fw_version;
+		fw_info->feature = 0;
+		break;
+	case AMDGPU_INFO_FW_VCN:
+		fw_info->ver = adev->vcn.fw_version;
+		fw_info->feature = 0;
+		break;
+	case AMDGPU_INFO_FW_GMC:
+		fw_info->ver = adev->gmc.fw_version;
+		fw_info->feature = 0;
+		break;
+	case AMDGPU_INFO_FW_GFX_ME:
+		fw_info->ver = adev->gfx.me_fw_version;
+		fw_info->feature = adev->gfx.me_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_PFP:
+		fw_info->ver = adev->gfx.pfp_fw_version;
+		fw_info->feature = adev->gfx.pfp_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_CE:
+		fw_info->ver = adev->gfx.ce_fw_version;
+		fw_info->feature = adev->gfx.ce_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_RLC:
+		fw_info->ver = adev->gfx.rlc_fw_version;
+		fw_info->feature = adev->gfx.rlc_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_RLC_RESTORE_LIST_CNTL:
+		fw_info->ver = adev->gfx.rlc_srlc_fw_version;
+		fw_info->feature = adev->gfx.rlc_srlc_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_RLC_RESTORE_LIST_GPM_MEM:
+		fw_info->ver = adev->gfx.rlc_srlg_fw_version;
+		fw_info->feature = adev->gfx.rlc_srlg_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_RLC_RESTORE_LIST_SRM_MEM:
+		fw_info->ver = adev->gfx.rlc_srls_fw_version;
+		fw_info->feature = adev->gfx.rlc_srls_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_MEC:
+		if (query_fw->index == 0) {
+			fw_info->ver = adev->gfx.mec_fw_version;
+			fw_info->feature = adev->gfx.mec_feature_version;
+		} else if (query_fw->index == 1) {
+			fw_info->ver = adev->gfx.mec2_fw_version;
+			fw_info->feature = adev->gfx.mec2_feature_version;
+		} else
+			return -EINVAL;
+		break;
+	case AMDGPU_INFO_FW_SMC:
+		fw_info->ver = adev->pm.fw_version;
+		fw_info->feature = 0;
+		break;
+	case AMDGPU_INFO_FW_SDMA:
+		if (query_fw->index >= adev->sdma.num_instances)
+			return -EINVAL;
+		fw_info->ver = adev->sdma.instance[query_fw->index].fw_version;
+		fw_info->feature = adev->sdma.instance[query_fw->index].feature_version;
+		break;
+	case AMDGPU_INFO_FW_SOS:
+		fw_info->ver = adev->psp.sos_fw_version;
+		fw_info->feature = adev->psp.sos_feature_version;
+		break;
+	case AMDGPU_INFO_FW_ASD:
+		fw_info->ver = adev->psp.asd_fw_version;
+		fw_info->feature = adev->psp.asd_feature_version;
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
+/*
+ * Userspace get information ioctl
+ */
+/**
+ * amdgpu_info_ioctl - answer a device specific request.
+ *
+ * @adev: amdgpu device pointer
+ * @data: request object
+ * @filp: drm filp
+ *
+ * This function is used to pass device specific parameters to the userspace
+ * drivers.  Examples include: pci device id, pipeline parms, tiling params,
+ * etc. (all asics).
+ * Returns 0 on success, -EINVAL on failure.
+ */
+static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_amdgpu_info *info = data;
+	struct amdgpu_mode_info *minfo = &adev->mode_info;
+	void __user *out = (void __user *)(uintptr_t)info->return_pointer;
+	uint32_t size = info->return_size;
+	struct drm_crtc *crtc;
+	uint32_t ui32 = 0;
+	uint64_t ui64 = 0;
+	int i, j, found;
+	int ui32_size = sizeof(ui32);
+
+	if (!info->return_size || !info->return_pointer)
+		return -EINVAL;
+
+	switch (info->query) {
+	case AMDGPU_INFO_ACCEL_WORKING:
+		ui32 = adev->accel_working;
+		return copy_to_user(out, &ui32, min(size, 4u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_CRTC_FROM_ID:
+		for (i = 0, found = 0; i < adev->mode_info.num_crtc; i++) {
+			crtc = (struct drm_crtc *)minfo->crtcs[i];
+			if (crtc && crtc->base.id == info->mode_crtc.id) {
+				struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+				ui32 = amdgpu_crtc->crtc_id;
+				found = 1;
+				break;
+			}
+		}
+		if (!found) {
+			DRM_DEBUG_KMS("unknown crtc id %d\n", info->mode_crtc.id);
+			return -EINVAL;
+		}
+		return copy_to_user(out, &ui32, min(size, 4u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_HW_IP_INFO: {
+		struct drm_amdgpu_info_hw_ip ip = {};
+		enum amd_ip_block_type type;
+		uint32_t ring_mask = 0;
+		uint32_t ib_start_alignment = 0;
+		uint32_t ib_size_alignment = 0;
+
+		if (info->query_hw_ip.ip_instance >= AMDGPU_HW_IP_INSTANCE_MAX_COUNT)
+			return -EINVAL;
+
+		switch (info->query_hw_ip.type) {
+		case AMDGPU_HW_IP_GFX:
+			type = AMD_IP_BLOCK_TYPE_GFX;
+			for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+				ring_mask |= adev->gfx.gfx_ring[i].ready << i;
+			ib_start_alignment = 32;
+			ib_size_alignment = 32;
+			break;
+		case AMDGPU_HW_IP_COMPUTE:
+			type = AMD_IP_BLOCK_TYPE_GFX;
+			for (i = 0; i < adev->gfx.num_compute_rings; i++)
+				ring_mask |= adev->gfx.compute_ring[i].ready << i;
+			ib_start_alignment = 32;
+			ib_size_alignment = 32;
+			break;
+		case AMDGPU_HW_IP_DMA:
+			type = AMD_IP_BLOCK_TYPE_SDMA;
+			for (i = 0; i < adev->sdma.num_instances; i++)
+				ring_mask |= adev->sdma.instance[i].ring.ready << i;
+			ib_start_alignment = 256;
+			ib_size_alignment = 4;
+			break;
+		case AMDGPU_HW_IP_UVD:
+			type = AMD_IP_BLOCK_TYPE_UVD;
+			for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+				if (adev->uvd.harvest_config & (1 << i))
+					continue;
+				ring_mask |= adev->uvd.inst[i].ring.ready;
+			}
+			ib_start_alignment = 64;
+			ib_size_alignment = 64;
+			break;
+		case AMDGPU_HW_IP_VCE:
+			type = AMD_IP_BLOCK_TYPE_VCE;
+			for (i = 0; i < adev->vce.num_rings; i++)
+				ring_mask |= adev->vce.ring[i].ready << i;
+			ib_start_alignment = 4;
+			ib_size_alignment = 1;
+			break;
+		case AMDGPU_HW_IP_UVD_ENC:
+			type = AMD_IP_BLOCK_TYPE_UVD;
+			for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+				if (adev->uvd.harvest_config & (1 << i))
+					continue;
+				for (j = 0; j < adev->uvd.num_enc_rings; j++)
+					ring_mask |= adev->uvd.inst[i].ring_enc[j].ready << j;
+			}
+			ib_start_alignment = 64;
+			ib_size_alignment = 64;
+			break;
+		case AMDGPU_HW_IP_VCN_DEC:
+			type = AMD_IP_BLOCK_TYPE_VCN;
+			ring_mask = adev->vcn.ring_dec.ready;
+			ib_start_alignment = 16;
+			ib_size_alignment = 16;
+			break;
+		case AMDGPU_HW_IP_VCN_ENC:
+			type = AMD_IP_BLOCK_TYPE_VCN;
+			for (i = 0; i < adev->vcn.num_enc_rings; i++)
+				ring_mask |= adev->vcn.ring_enc[i].ready << i;
+			ib_start_alignment = 64;
+			ib_size_alignment = 1;
+			break;
+		case AMDGPU_HW_IP_VCN_JPEG:
+			type = AMD_IP_BLOCK_TYPE_VCN;
+			ring_mask = adev->vcn.ring_jpeg.ready;
+			ib_start_alignment = 16;
+			ib_size_alignment = 16;
+			break;
+		default:
+			return -EINVAL;
+		}
+
+		for (i = 0; i < adev->num_ip_blocks; i++) {
+			if (adev->ip_blocks[i].version->type == type &&
+			    adev->ip_blocks[i].status.valid) {
+				ip.hw_ip_version_major = adev->ip_blocks[i].version->major;
+				ip.hw_ip_version_minor = adev->ip_blocks[i].version->minor;
+				ip.capabilities_flags = 0;
+				ip.available_rings = ring_mask;
+				ip.ib_start_alignment = ib_start_alignment;
+				ip.ib_size_alignment = ib_size_alignment;
+				break;
+			}
+		}
+		return copy_to_user(out, &ip,
+				    min((size_t)size, sizeof(ip))) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_HW_IP_COUNT: {
+		enum amd_ip_block_type type;
+		uint32_t count = 0;
+
+		switch (info->query_hw_ip.type) {
+		case AMDGPU_HW_IP_GFX:
+			type = AMD_IP_BLOCK_TYPE_GFX;
+			break;
+		case AMDGPU_HW_IP_COMPUTE:
+			type = AMD_IP_BLOCK_TYPE_GFX;
+			break;
+		case AMDGPU_HW_IP_DMA:
+			type = AMD_IP_BLOCK_TYPE_SDMA;
+			break;
+		case AMDGPU_HW_IP_UVD:
+			type = AMD_IP_BLOCK_TYPE_UVD;
+			break;
+		case AMDGPU_HW_IP_VCE:
+			type = AMD_IP_BLOCK_TYPE_VCE;
+			break;
+		case AMDGPU_HW_IP_UVD_ENC:
+			type = AMD_IP_BLOCK_TYPE_UVD;
+			break;
+		case AMDGPU_HW_IP_VCN_DEC:
+		case AMDGPU_HW_IP_VCN_ENC:
+		case AMDGPU_HW_IP_VCN_JPEG:
+			type = AMD_IP_BLOCK_TYPE_VCN;
+			break;
+		default:
+			return -EINVAL;
+		}
+
+		for (i = 0; i < adev->num_ip_blocks; i++)
+			if (adev->ip_blocks[i].version->type == type &&
+			    adev->ip_blocks[i].status.valid &&
+			    count < AMDGPU_HW_IP_INSTANCE_MAX_COUNT)
+				count++;
+
+		return copy_to_user(out, &count, min(size, 4u)) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_TIMESTAMP:
+		ui64 = amdgpu_gfx_get_gpu_clock_counter(adev);
+		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_FW_VERSION: {
+		struct drm_amdgpu_info_firmware fw_info;
+		int ret;
+
+		/* We only support one instance of each IP block right now. */
+		if (info->query_fw.ip_instance != 0)
+			return -EINVAL;
+
+		ret = amdgpu_firmware_info(&fw_info, &info->query_fw, adev);
+		if (ret)
+			return ret;
+
+		return copy_to_user(out, &fw_info,
+				    min((size_t)size, sizeof(fw_info))) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_NUM_BYTES_MOVED:
+		ui64 = atomic64_read(&adev->num_bytes_moved);
+		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_NUM_EVICTIONS:
+		ui64 = atomic64_read(&adev->num_evictions);
+		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_NUM_VRAM_CPU_PAGE_FAULTS:
+		ui64 = atomic64_read(&adev->num_vram_cpu_page_faults);
+		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_VRAM_USAGE:
+		ui64 = amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_VIS_VRAM_USAGE:
+		ui64 = amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_GTT_USAGE:
+		ui64 = amdgpu_gtt_mgr_usage(&adev->mman.bdev.man[TTM_PL_TT]);
+		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_GDS_CONFIG: {
+		struct drm_amdgpu_info_gds gds_info;
+
+		memset(&gds_info, 0, sizeof(gds_info));
+		gds_info.gds_gfx_partition_size = adev->gds.mem.gfx_partition_size >> AMDGPU_GDS_SHIFT;
+		gds_info.compute_partition_size = adev->gds.mem.cs_partition_size >> AMDGPU_GDS_SHIFT;
+		gds_info.gds_total_size = adev->gds.mem.total_size >> AMDGPU_GDS_SHIFT;
+		gds_info.gws_per_gfx_partition = adev->gds.gws.gfx_partition_size >> AMDGPU_GWS_SHIFT;
+		gds_info.gws_per_compute_partition = adev->gds.gws.cs_partition_size >> AMDGPU_GWS_SHIFT;
+		gds_info.oa_per_gfx_partition = adev->gds.oa.gfx_partition_size >> AMDGPU_OA_SHIFT;
+		gds_info.oa_per_compute_partition = adev->gds.oa.cs_partition_size >> AMDGPU_OA_SHIFT;
+		return copy_to_user(out, &gds_info,
+				    min((size_t)size, sizeof(gds_info))) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_VRAM_GTT: {
+		struct drm_amdgpu_info_vram_gtt vram_gtt;
+
+		vram_gtt.vram_size = adev->gmc.real_vram_size -
+			atomic64_read(&adev->vram_pin_size);
+		vram_gtt.vram_cpu_accessible_size = adev->gmc.visible_vram_size -
+			atomic64_read(&adev->visible_pin_size);
+		vram_gtt.gtt_size = adev->mman.bdev.man[TTM_PL_TT].size;
+		vram_gtt.gtt_size *= PAGE_SIZE;
+		vram_gtt.gtt_size -= atomic64_read(&adev->gart_pin_size);
+		return copy_to_user(out, &vram_gtt,
+				    min((size_t)size, sizeof(vram_gtt))) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_MEMORY: {
+		struct drm_amdgpu_memory_info mem;
+
+		memset(&mem, 0, sizeof(mem));
+		mem.vram.total_heap_size = adev->gmc.real_vram_size;
+		mem.vram.usable_heap_size = adev->gmc.real_vram_size -
+			atomic64_read(&adev->vram_pin_size);
+		mem.vram.heap_usage =
+			amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+		mem.vram.max_allocation = mem.vram.usable_heap_size * 3 / 4;
+
+		mem.cpu_accessible_vram.total_heap_size =
+			adev->gmc.visible_vram_size;
+		mem.cpu_accessible_vram.usable_heap_size = adev->gmc.visible_vram_size -
+			atomic64_read(&adev->visible_pin_size);
+		mem.cpu_accessible_vram.heap_usage =
+			amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+		mem.cpu_accessible_vram.max_allocation =
+			mem.cpu_accessible_vram.usable_heap_size * 3 / 4;
+
+		mem.gtt.total_heap_size = adev->mman.bdev.man[TTM_PL_TT].size;
+		mem.gtt.total_heap_size *= PAGE_SIZE;
+		mem.gtt.usable_heap_size = mem.gtt.total_heap_size -
+			atomic64_read(&adev->gart_pin_size);
+		mem.gtt.heap_usage =
+			amdgpu_gtt_mgr_usage(&adev->mman.bdev.man[TTM_PL_TT]);
+		mem.gtt.max_allocation = mem.gtt.usable_heap_size * 3 / 4;
+
+		return copy_to_user(out, &mem,
+				    min((size_t)size, sizeof(mem)))
+				    ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_READ_MMR_REG: {
+		unsigned n, alloc_size;
+		uint32_t *regs;
+		unsigned se_num = (info->read_mmr_reg.instance >>
+				   AMDGPU_INFO_MMR_SE_INDEX_SHIFT) &
+				  AMDGPU_INFO_MMR_SE_INDEX_MASK;
+		unsigned sh_num = (info->read_mmr_reg.instance >>
+				   AMDGPU_INFO_MMR_SH_INDEX_SHIFT) &
+				  AMDGPU_INFO_MMR_SH_INDEX_MASK;
+
+		/* set full masks if the userspace set all bits
+		 * in the bitfields */
+		if (se_num == AMDGPU_INFO_MMR_SE_INDEX_MASK)
+			se_num = 0xffffffff;
+		if (sh_num == AMDGPU_INFO_MMR_SH_INDEX_MASK)
+			sh_num = 0xffffffff;
+
+		regs = kmalloc_array(info->read_mmr_reg.count, sizeof(*regs), GFP_KERNEL);
+		if (!regs)
+			return -ENOMEM;
+		alloc_size = info->read_mmr_reg.count * sizeof(*regs);
+
+		for (i = 0; i < info->read_mmr_reg.count; i++)
+			if (amdgpu_asic_read_register(adev, se_num, sh_num,
+						      info->read_mmr_reg.dword_offset + i,
+						      &regs[i])) {
+				DRM_DEBUG_KMS("unallowed offset %#x\n",
+					      info->read_mmr_reg.dword_offset + i);
+				kfree(regs);
+				return -EFAULT;
+			}
+		n = copy_to_user(out, regs, min(size, alloc_size));
+		kfree(regs);
+		return n ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_DEV_INFO: {
+		struct drm_amdgpu_info_device dev_info = {};
+		uint64_t vm_size;
+
+		dev_info.device_id = dev->pdev->device;
+		dev_info.chip_rev = adev->rev_id;
+		dev_info.external_rev = adev->external_rev_id;
+		dev_info.pci_rev = dev->pdev->revision;
+		dev_info.family = adev->family;
+		dev_info.num_shader_engines = adev->gfx.config.max_shader_engines;
+		dev_info.num_shader_arrays_per_engine = adev->gfx.config.max_sh_per_se;
+		/* return all clocks in KHz */
+		dev_info.gpu_counter_freq = amdgpu_asic_get_xclk(adev) * 10;
+		if (adev->pm.dpm_enabled) {
+			dev_info.max_engine_clock = amdgpu_dpm_get_sclk(adev, false) * 10;
+			dev_info.max_memory_clock = amdgpu_dpm_get_mclk(adev, false) * 10;
+		} else {
+			dev_info.max_engine_clock = adev->clock.default_sclk * 10;
+			dev_info.max_memory_clock = adev->clock.default_mclk * 10;
+		}
+		dev_info.enabled_rb_pipes_mask = adev->gfx.config.backend_enable_mask;
+		dev_info.num_rb_pipes = adev->gfx.config.max_backends_per_se *
+			adev->gfx.config.max_shader_engines;
+		dev_info.num_hw_gfx_contexts = adev->gfx.config.max_hw_contexts;
+		dev_info._pad = 0;
+		dev_info.ids_flags = 0;
+		if (adev->flags & AMD_IS_APU)
+			dev_info.ids_flags |= AMDGPU_IDS_FLAGS_FUSION;
+		if (amdgpu_sriov_vf(adev))
+			dev_info.ids_flags |= AMDGPU_IDS_FLAGS_PREEMPTION;
+
+		vm_size = adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
+		vm_size -= AMDGPU_VA_RESERVED_SIZE;
+
+		/* Older VCE FW versions are buggy and can handle only 40bits */
+		if (adev->vce.fw_version < AMDGPU_VCE_FW_53_45)
+			vm_size = min(vm_size, 1ULL << 40);
+
+		dev_info.virtual_address_offset = AMDGPU_VA_RESERVED_SIZE;
+		dev_info.virtual_address_max =
+			min(vm_size, AMDGPU_VA_HOLE_START);
+
+		if (vm_size > AMDGPU_VA_HOLE_START) {
+			dev_info.high_va_offset = AMDGPU_VA_HOLE_END;
+			dev_info.high_va_max = AMDGPU_VA_HOLE_END | vm_size;
+		}
+		dev_info.virtual_address_alignment = max((int)PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE);
+		dev_info.pte_fragment_size = (1 << adev->vm_manager.fragment_size) * AMDGPU_GPU_PAGE_SIZE;
+		dev_info.gart_page_size = AMDGPU_GPU_PAGE_SIZE;
+		dev_info.cu_active_number = adev->gfx.cu_info.number;
+		dev_info.cu_ao_mask = adev->gfx.cu_info.ao_cu_mask;
+		dev_info.ce_ram_size = adev->gfx.ce_ram_size;
+		memcpy(&dev_info.cu_ao_bitmap[0], &adev->gfx.cu_info.ao_cu_bitmap[0],
+		       sizeof(adev->gfx.cu_info.ao_cu_bitmap));
+		memcpy(&dev_info.cu_bitmap[0], &adev->gfx.cu_info.bitmap[0],
+		       sizeof(adev->gfx.cu_info.bitmap));
+		dev_info.vram_type = adev->gmc.vram_type;
+		dev_info.vram_bit_width = adev->gmc.vram_width;
+		dev_info.vce_harvest_config = adev->vce.harvest_config;
+		dev_info.gc_double_offchip_lds_buf =
+			adev->gfx.config.double_offchip_lds_buf;
+
+		if (amdgpu_ngg) {
+			dev_info.prim_buf_gpu_addr = adev->gfx.ngg.buf[NGG_PRIM].gpu_addr;
+			dev_info.prim_buf_size = adev->gfx.ngg.buf[NGG_PRIM].size;
+			dev_info.pos_buf_gpu_addr = adev->gfx.ngg.buf[NGG_POS].gpu_addr;
+			dev_info.pos_buf_size = adev->gfx.ngg.buf[NGG_POS].size;
+			dev_info.cntl_sb_buf_gpu_addr = adev->gfx.ngg.buf[NGG_CNTL].gpu_addr;
+			dev_info.cntl_sb_buf_size = adev->gfx.ngg.buf[NGG_CNTL].size;
+			dev_info.param_buf_gpu_addr = adev->gfx.ngg.buf[NGG_PARAM].gpu_addr;
+			dev_info.param_buf_size = adev->gfx.ngg.buf[NGG_PARAM].size;
+		}
+		dev_info.wave_front_size = adev->gfx.cu_info.wave_front_size;
+		dev_info.num_shader_visible_vgprs = adev->gfx.config.max_gprs;
+		dev_info.num_cu_per_sh = adev->gfx.config.max_cu_per_sh;
+		dev_info.num_tcc_blocks = adev->gfx.config.max_texture_channel_caches;
+		dev_info.gs_vgt_table_depth = adev->gfx.config.gs_vgt_table_depth;
+		dev_info.gs_prim_buffer_depth = adev->gfx.config.gs_prim_buffer_depth;
+		dev_info.max_gs_waves_per_vgt = adev->gfx.config.max_gs_threads;
+
+		return copy_to_user(out, &dev_info,
+				    min((size_t)size, sizeof(dev_info))) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_VCE_CLOCK_TABLE: {
+		unsigned i;
+		struct drm_amdgpu_info_vce_clock_table vce_clk_table = {};
+		struct amd_vce_state *vce_state;
+
+		for (i = 0; i < AMDGPU_VCE_CLOCK_TABLE_ENTRIES; i++) {
+			vce_state = amdgpu_dpm_get_vce_clock_state(adev, i);
+			if (vce_state) {
+				vce_clk_table.entries[i].sclk = vce_state->sclk;
+				vce_clk_table.entries[i].mclk = vce_state->mclk;
+				vce_clk_table.entries[i].eclk = vce_state->evclk;
+				vce_clk_table.num_valid_entries++;
+			}
+		}
+
+		return copy_to_user(out, &vce_clk_table,
+				    min((size_t)size, sizeof(vce_clk_table))) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_VBIOS: {
+		uint32_t bios_size = adev->bios_size;
+
+		switch (info->vbios_info.type) {
+		case AMDGPU_INFO_VBIOS_SIZE:
+			return copy_to_user(out, &bios_size,
+					min((size_t)size, sizeof(bios_size)))
+					? -EFAULT : 0;
+		case AMDGPU_INFO_VBIOS_IMAGE: {
+			uint8_t *bios;
+			uint32_t bios_offset = info->vbios_info.offset;
+
+			if (bios_offset >= bios_size)
+				return -EINVAL;
+
+			bios = adev->bios + bios_offset;
+			return copy_to_user(out, bios,
+					    min((size_t)size, (size_t)(bios_size - bios_offset)))
+					? -EFAULT : 0;
+		}
+		default:
+			DRM_DEBUG_KMS("Invalid request %d\n",
+					info->vbios_info.type);
+			return -EINVAL;
+		}
+	}
+	case AMDGPU_INFO_NUM_HANDLES: {
+		struct drm_amdgpu_info_num_handles handle;
+
+		switch (info->query_hw_ip.type) {
+		case AMDGPU_HW_IP_UVD:
+			/* Starting Polaris, we support unlimited UVD handles */
+			if (adev->asic_type < CHIP_POLARIS10) {
+				handle.uvd_max_handles = adev->uvd.max_handles;
+				handle.uvd_used_handles = amdgpu_uvd_used_handles(adev);
+
+				return copy_to_user(out, &handle,
+					min((size_t)size, sizeof(handle))) ? -EFAULT : 0;
+			} else {
+				return -ENODATA;
+			}
+
+			break;
+		default:
+			return -EINVAL;
+		}
+	}
+	case AMDGPU_INFO_SENSOR: {
+		if (!adev->pm.dpm_enabled)
+			return -ENOENT;
+
+		switch (info->sensor_info.type) {
+		case AMDGPU_INFO_SENSOR_GFX_SCLK:
+			/* get sclk in Mhz */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_GFX_SCLK,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			ui32 /= 100;
+			break;
+		case AMDGPU_INFO_SENSOR_GFX_MCLK:
+			/* get mclk in Mhz */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_GFX_MCLK,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			ui32 /= 100;
+			break;
+		case AMDGPU_INFO_SENSOR_GPU_TEMP:
+			/* get temperature in millidegrees C */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_GPU_TEMP,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			break;
+		case AMDGPU_INFO_SENSOR_GPU_LOAD:
+			/* get GPU load */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_GPU_LOAD,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			break;
+		case AMDGPU_INFO_SENSOR_GPU_AVG_POWER:
+			/* get average GPU power */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_GPU_POWER,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			ui32 >>= 8;
+			break;
+		case AMDGPU_INFO_SENSOR_VDDNB:
+			/* get VDDNB in millivolts */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_VDDNB,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			break;
+		case AMDGPU_INFO_SENSOR_VDDGFX:
+			/* get VDDGFX in millivolts */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_VDDGFX,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			break;
+		case AMDGPU_INFO_SENSOR_STABLE_PSTATE_GFX_SCLK:
+			/* get stable pstate sclk in Mhz */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_STABLE_PSTATE_SCLK,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			ui32 /= 100;
+			break;
+		case AMDGPU_INFO_SENSOR_STABLE_PSTATE_GFX_MCLK:
+			/* get stable pstate mclk in Mhz */
+			if (amdgpu_dpm_read_sensor(adev,
+						   AMDGPU_PP_SENSOR_STABLE_PSTATE_MCLK,
+						   (void *)&ui32, &ui32_size)) {
+				return -EINVAL;
+			}
+			ui32 /= 100;
+			break;
+		default:
+			DRM_DEBUG_KMS("Invalid request %d\n",
+				      info->sensor_info.type);
+			return -EINVAL;
+		}
+		return copy_to_user(out, &ui32, min(size, 4u)) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_VRAM_LOST_COUNTER:
+		ui32 = atomic_read(&adev->vram_lost_counter);
+		return copy_to_user(out, &ui32, min(size, 4u)) ? -EFAULT : 0;
+	default:
+		DRM_DEBUG_KMS("Invalid request %d\n", info->query);
+		return -EINVAL;
+	}
+	return 0;
+}
+
+
+/*
+ * Outdated mess for old drm with Xorg being in charge (void function now).
+ */
+/**
+ * amdgpu_driver_lastclose_kms - drm callback for last close
+ *
+ * @dev: drm dev pointer
+ *
+ * Switch vga_switcheroo state after last close (all asics).
+ */
+void amdgpu_driver_lastclose_kms(struct drm_device *dev)
+{
+	drm_fb_helper_lastclose(dev);
+	vga_switcheroo_process_delayed_switch();
+}
+
+/**
+ * amdgpu_driver_open_kms - drm callback for open
+ *
+ * @dev: drm dev pointer
+ * @file_priv: drm file
+ *
+ * On device open, init vm on cayman+ (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv;
+	int r, pasid;
+
+	/* Ensure IB tests are run on ring */
+	flush_delayed_work(&adev->late_init_work);
+
+	file_priv->driver_priv = NULL;
+
+	r = pm_runtime_get_sync(dev->dev);
+	if (r < 0)
+		return r;
+
+	fpriv = kzalloc(sizeof(*fpriv), GFP_KERNEL);
+	if (unlikely(!fpriv)) {
+		r = -ENOMEM;
+		goto out_suspend;
+	}
+
+	pasid = amdgpu_pasid_alloc(16);
+	if (pasid < 0) {
+		dev_warn(adev->dev, "No more PASIDs available!");
+		pasid = 0;
+	}
+	r = amdgpu_vm_init(adev, &fpriv->vm, AMDGPU_VM_CONTEXT_GFX, pasid);
+	if (r)
+		goto error_pasid;
+
+	fpriv->prt_va = amdgpu_vm_bo_add(adev, &fpriv->vm, NULL);
+	if (!fpriv->prt_va) {
+		r = -ENOMEM;
+		goto error_vm;
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		r = amdgpu_map_static_csa(adev, &fpriv->vm, &fpriv->csa_va);
+		if (r)
+			goto error_vm;
+	}
+
+	rw_init(&fpriv->bo_list_lock, "agbo");
+	idr_init(&fpriv->bo_list_handles);
+
+	amdgpu_ctx_mgr_init(&fpriv->ctx_mgr);
+
+	file_priv->driver_priv = fpriv;
+	goto out_suspend;
+
+error_vm:
+	amdgpu_vm_fini(adev, &fpriv->vm);
+
+error_pasid:
+	if (pasid)
+		amdgpu_pasid_free(pasid);
+
+	kfree(fpriv);
+
+out_suspend:
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
+
+	return r;
+}
+
+/**
+ * amdgpu_driver_postclose_kms - drm callback for post close
+ *
+ * @dev: drm dev pointer
+ * @file_priv: drm file
+ *
+ * On device post close, tear down vm on cayman+ (all asics).
+ */
+void amdgpu_driver_postclose_kms(struct drm_device *dev,
+				 struct drm_file *file_priv)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
+	struct amdgpu_bo_list *list;
+	struct amdgpu_bo *pd;
+	unsigned int pasid;
+	int handle;
+
+	if (!fpriv)
+		return;
+
+	pm_runtime_get_sync(dev->dev);
+
+	if (adev->asic_type != CHIP_RAVEN) {
+		amdgpu_uvd_free_handles(adev, file_priv);
+		amdgpu_vce_free_handles(adev, file_priv);
+	}
+
+	amdgpu_vm_bo_rmv(adev, fpriv->prt_va);
+
+	if (amdgpu_sriov_vf(adev)) {
+		/* TODO: how to handle reserve failure */
+		BUG_ON(amdgpu_bo_reserve(adev->virt.csa_obj, true));
+		amdgpu_vm_bo_rmv(adev, fpriv->csa_va);
+		fpriv->csa_va = NULL;
+		amdgpu_bo_unreserve(adev->virt.csa_obj);
+	}
+
+	pasid = fpriv->vm.pasid;
+	pd = amdgpu_bo_ref(fpriv->vm.root.base.bo);
+
+	amdgpu_vm_fini(adev, &fpriv->vm);
+	amdgpu_ctx_mgr_fini(&fpriv->ctx_mgr);
+
+	if (pasid)
+		amdgpu_pasid_free_delayed(pd->tbo.resv, pasid);
+	amdgpu_bo_unref(&pd);
+
+	idr_for_each_entry(&fpriv->bo_list_handles, list, handle)
+		amdgpu_bo_list_put(list);
+
+	idr_destroy(&fpriv->bo_list_handles);
+	mutex_destroy(&fpriv->bo_list_lock);
+
+	kfree(fpriv);
+	file_priv->driver_priv = NULL;
+
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
+}
+
+/*
+ * VBlank related functions.
+ */
+/**
+ * amdgpu_get_vblank_counter_kms - get frame count
+ *
+ * @dev: drm dev pointer
+ * @pipe: crtc to get the frame count from
+ *
+ * Gets the frame count on the requested crtc (all asics).
+ * Returns frame count on success, -EINVAL on failure.
+ */
+u32 amdgpu_get_vblank_counter_kms(struct drm_device *dev, unsigned int pipe)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	int vpos, hpos, stat;
+	u32 count;
+
+	if (pipe >= adev->mode_info.num_crtc) {
+		DRM_ERROR("Invalid crtc %u\n", pipe);
+		return -EINVAL;
+	}
+
+	/* The hw increments its frame counter at start of vsync, not at start
+	 * of vblank, as is required by DRM core vblank counter handling.
+	 * Cook the hw count here to make it appear to the caller as if it
+	 * incremented at start of vblank. We measure distance to start of
+	 * vblank in vpos. vpos therefore will be >= 0 between start of vblank
+	 * and start of vsync, so vpos >= 0 means to bump the hw frame counter
+	 * result by 1 to give the proper appearance to caller.
+	 */
+	if (adev->mode_info.crtcs[pipe]) {
+		/* Repeat readout if needed to provide stable result if
+		 * we cross start of vsync during the queries.
+		 */
+		do {
+			count = amdgpu_display_vblank_get_counter(adev, pipe);
+			/* Ask amdgpu_display_get_crtc_scanoutpos to return
+			 * vpos as distance to start of vblank, instead of
+			 * regular vertical scanout pos.
+			 */
+			stat = amdgpu_display_get_crtc_scanoutpos(
+				dev, pipe, GET_DISTANCE_TO_VBLANKSTART,
+				&vpos, &hpos, NULL, NULL,
+				&adev->mode_info.crtcs[pipe]->base.hwmode);
+		} while (count != amdgpu_display_vblank_get_counter(adev, pipe));
+
+		if (((stat & (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE)) !=
+		    (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE))) {
+			DRM_DEBUG_VBL("Query failed! stat %d\n", stat);
+		} else {
+			DRM_DEBUG_VBL("crtc %d: dist from vblank start %d\n",
+				      pipe, vpos);
+
+			/* Bump counter if we are at >= leading edge of vblank,
+			 * but before vsync where vpos would turn negative and
+			 * the hw counter really increments.
+			 */
+			if (vpos >= 0)
+				count++;
+		}
+	} else {
+		/* Fallback to use value as is. */
+		count = amdgpu_display_vblank_get_counter(adev, pipe);
+		DRM_DEBUG_VBL("NULL mode info! Returned count may be wrong.\n");
+	}
+
+	return count;
+}
+
+/**
+ * amdgpu_enable_vblank_kms - enable vblank interrupt
+ *
+ * @dev: drm dev pointer
+ * @pipe: crtc to enable vblank interrupt for
+ *
+ * Enable the interrupt on the requested crtc (all asics).
+ * Returns 0 on success, -EINVAL on failure.
+ */
+int amdgpu_enable_vblank_kms(struct drm_device *dev, unsigned int pipe)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	int idx = amdgpu_display_crtc_idx_to_irq_type(adev, pipe);
+
+	return amdgpu_irq_get(adev, &adev->crtc_irq, idx);
+}
+
+/**
+ * amdgpu_disable_vblank_kms - disable vblank interrupt
+ *
+ * @dev: drm dev pointer
+ * @pipe: crtc to disable vblank interrupt for
+ *
+ * Disable the interrupt on the requested crtc (all asics).
+ */
+void amdgpu_disable_vblank_kms(struct drm_device *dev, unsigned int pipe)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	int idx = amdgpu_display_crtc_idx_to_irq_type(adev, pipe);
+
+	amdgpu_irq_put(adev, &adev->crtc_irq, idx);
+}
+
+const struct drm_ioctl_desc amdgpu_ioctls_kms[] = {
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_CREATE, amdgpu_gem_create_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_CTX, amdgpu_ctx_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_VM, amdgpu_vm_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_SCHED, amdgpu_sched_ioctl, DRM_MASTER),
+	DRM_IOCTL_DEF_DRV(AMDGPU_BO_LIST, amdgpu_bo_list_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_FENCE_TO_HANDLE, amdgpu_cs_fence_to_handle_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	/* KMS */
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_MMAP, amdgpu_gem_mmap_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_WAIT_IDLE, amdgpu_gem_wait_idle_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_CS, amdgpu_cs_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_INFO, amdgpu_info_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_WAIT_CS, amdgpu_cs_wait_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_WAIT_FENCES, amdgpu_cs_wait_fences_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_METADATA, amdgpu_gem_metadata_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_VA, amdgpu_gem_va_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_OP, amdgpu_gem_op_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_USERPTR, amdgpu_gem_userptr_ioctl, DRM_AUTH|DRM_RENDER_ALLOW)
+};
+const int amdgpu_max_kms_ioctl = ARRAY_SIZE(amdgpu_ioctls_kms);
+
+/*
+ * Debugfs info
+ */
+#if defined(CONFIG_DEBUG_FS)
+
+static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_amdgpu_info_firmware fw_info;
+	struct drm_amdgpu_query_fw query_fw;
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	int ret, i;
+
+	/* VCE */
+	query_fw.fw_type = AMDGPU_INFO_FW_VCE;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "VCE feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* UVD */
+	query_fw.fw_type = AMDGPU_INFO_FW_UVD;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "UVD feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* GMC */
+	query_fw.fw_type = AMDGPU_INFO_FW_GMC;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "MC feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* ME */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_ME;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "ME feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* PFP */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_PFP;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "PFP feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* CE */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_CE;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "CE feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* RLC */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_RLC;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "RLC feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* RLC SAVE RESTORE LIST CNTL */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_RLC_RESTORE_LIST_CNTL;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "RLC SRLC feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* RLC SAVE RESTORE LIST GPM MEM */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_RLC_RESTORE_LIST_GPM_MEM;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "RLC SRLG feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* RLC SAVE RESTORE LIST SRM MEM */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_RLC_RESTORE_LIST_SRM_MEM;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "RLC SRLS feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* MEC */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_MEC;
+	query_fw.index = 0;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "MEC feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* MEC2 */
+	if (adev->asic_type == CHIP_KAVERI ||
+	    (adev->asic_type > CHIP_TOPAZ && adev->asic_type != CHIP_STONEY)) {
+		query_fw.index = 1;
+		ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+		if (ret)
+			return ret;
+		seq_printf(m, "MEC2 feature version: %u, firmware version: 0x%08x\n",
+			   fw_info.feature, fw_info.ver);
+	}
+
+	/* PSP SOS */
+	query_fw.fw_type = AMDGPU_INFO_FW_SOS;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "SOS feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+
+	/* PSP ASD */
+	query_fw.fw_type = AMDGPU_INFO_FW_ASD;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "ASD feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* SMC */
+	query_fw.fw_type = AMDGPU_INFO_FW_SMC;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "SMC feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* SDMA */
+	query_fw.fw_type = AMDGPU_INFO_FW_SDMA;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		query_fw.index = i;
+		ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+		if (ret)
+			return ret;
+		seq_printf(m, "SDMA%d feature version: %u, firmware version: 0x%08x\n",
+			   i, fw_info.feature, fw_info.ver);
+	}
+
+	/* VCN */
+	query_fw.fw_type = AMDGPU_INFO_FW_VCN;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "VCN feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+
+	seq_printf(m, "VBIOS version: %s\n", ctx->vbios_version);
+
+	return 0;
+}
+
+static const struct drm_info_list amdgpu_firmware_info_list[] = {
+	{"amdgpu_firmware_info", amdgpu_debugfs_firmware_info, 0, NULL},
+};
+#endif
+
+int amdgpu_debugfs_firmware_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	return amdgpu_debugfs_add_files(adev, amdgpu_firmware_info_list,
+					ARRAY_SIZE(amdgpu_firmware_info_list));
+#else
+	return 0;
+#endif
+}
+
+int
+amdgpu_probe(struct device *parent, void *match, void *aux)
+{
+	if (amdgpu_fatal_error)
+		return 0;
+	if (drm_pciprobe(aux, amdgpu_pciidlist))
+		return 20;
+	return 0;
+}
+
+/*
+ * some functions are only called once on init regardless of how many times
+ * amdgpu attaches in linux this is handled via module_init()/module_exit()
+ */
+int amdgpu_refcnt;
+
+int __init drm_sched_fence_slab_init(void);
+void __exit drm_sched_fence_slab_fini(void);
+
+void
+amdgpu_attach(struct device *parent, struct device *self, void *aux)
+{
+	struct amdgpu_device	*adev = (struct amdgpu_device *)self;
+	struct drm_device	*dev;
+	struct pci_attach_args	*pa = aux;
+	const struct drm_pcidev	*id_entry;
+	pcireg_t		 type;
+	int			 i;
+	uint8_t			 rmmio_bar;
+	paddr_t			 fb_aper;
+	pcireg_t		 addr, mask;
+	int			 s;
+
+	id_entry = drm_find_description(PCI_VENDOR(pa->pa_id),
+	    PCI_PRODUCT(pa->pa_id), amdgpu_pciidlist);
+	adev->flags = id_entry->driver_data;
+	adev->family = adev->flags & AMD_ASIC_MASK;
+	adev->pc = pa->pa_pc;
+	adev->pa_tag = pa->pa_tag;
+	adev->iot = pa->pa_iot;
+	adev->memt = pa->pa_memt;
+	adev->dmat = pa->pa_dmat;
+
+	if (PCI_CLASS(pa->pa_class) == PCI_CLASS_DISPLAY &&
+	    PCI_SUBCLASS(pa->pa_class) == PCI_SUBCLASS_DISPLAY_VGA &&
+	    (pci_conf_read(pa->pa_pc, pa->pa_tag, PCI_COMMAND_STATUS_REG)
+	    & (PCI_COMMAND_IO_ENABLE | PCI_COMMAND_MEM_ENABLE))
+	    == (PCI_COMMAND_IO_ENABLE | PCI_COMMAND_MEM_ENABLE)) {
+		adev->primary = 1;
+#if NVGA > 0
+		adev->console = vga_is_console(pa->pa_iot, -1);
+		vga_console_attached = 1;
+#endif
+	}
+#if NEFIFB > 0
+	if (efifb_is_primary(pa)) {
+		adev->primary = 1;
+		adev->console = efifb_is_console(pa);
+		efifb_detach();
+	}
+#endif
+
+#define AMDGPU_PCI_MEM		0x10
+
+	type = pci_mapreg_type(pa->pa_pc, pa->pa_tag, AMDGPU_PCI_MEM);
+	if (PCI_MAPREG_TYPE(type) != PCI_MAPREG_TYPE_MEM ||
+	    pci_mapreg_info(pa->pa_pc, pa->pa_tag, AMDGPU_PCI_MEM,
+	    type, &adev->fb_aper_offset, &adev->fb_aper_size, NULL)) {
+		printf(": can't get frambuffer info\n");
+		return;
+	}
+
+	if (adev->fb_aper_offset == 0) {
+		bus_size_t start, end, pci_mem_end;
+		bus_addr_t base;
+
+		start = max(PCI_MEM_START, pa->pa_memex->ex_start);
+		if (PCI_MAPREG_MEM_TYPE(type) == PCI_MAPREG_MEM_TYPE_64BIT)
+			pci_mem_end = PCI_MEM64_END;
+		else
+			pci_mem_end = PCI_MEM_END;
+		end = min(pci_mem_end, pa->pa_memex->ex_end);
+		if (pa->pa_memex == NULL ||
+		    extent_alloc_subregion(pa->pa_memex, start, end,
+		    adev->fb_aper_size, adev->fb_aper_size, 0, 0, 0, &base)) {
+			printf(": can't reserve framebuffer space\n");
+			return;
+		}
+		pci_conf_write(pa->pa_pc, pa->pa_tag, AMDGPU_PCI_MEM, base);
+		if (PCI_MAPREG_MEM_TYPE(type) == PCI_MAPREG_MEM_TYPE_64BIT)
+			pci_conf_write(pa->pa_pc, pa->pa_tag,
+			    AMDGPU_PCI_MEM + 4, (uint64_t)base >> 32);
+		adev->fb_aper_offset = base;
+	}
+
+	for (i = PCI_MAPREG_START; i < PCI_MAPREG_END ;) {
+		type = pci_mapreg_type(pa->pa_pc, pa->pa_tag, i);
+		if (PCI_MAPREG_TYPE(type) != PCI_MAPREG_TYPE_IO) {
+			if (type & PCI_MAPREG_MEM_TYPE_64BIT)
+				i += 8;
+			else
+				i += 4;
+			continue;
+		}
+		if (pci_mapreg_map(pa, i, type, 0,
+		    &adev->rio_mem_bst, &adev->rio_mem_bsh, NULL,
+		    &adev->rio_mem_size, 0)) {
+			printf(": can't map rio space\n");
+			return;
+		}
+		break;
+	}
+
+	if (adev->family >= CHIP_BONAIRE) {
+		type = pci_mapreg_type(pa->pa_pc, pa->pa_tag, 0x18);
+		if (PCI_MAPREG_TYPE(type) != PCI_MAPREG_TYPE_MEM ||
+		    pci_mapreg_map(pa, 0x18, type, 0,
+		    &adev->doorbell.bst, &adev->doorbell.bsh,
+		    &adev->doorbell.base, &adev->doorbell.size, 0)) {
+			printf(": can't map doorbell space\n");
+			return;
+		}
+	}
+
+	if (adev->family >= CHIP_BONAIRE)
+		rmmio_bar = 0x24;
+	else
+		rmmio_bar = 0x18;
+
+	type = pci_mapreg_type(pa->pa_pc, pa->pa_tag, rmmio_bar);
+	if (PCI_MAPREG_TYPE(type) != PCI_MAPREG_TYPE_MEM ||
+	    pci_mapreg_map(pa, rmmio_bar, type, 0,
+	    &adev->rmmio_bst, &adev->rmmio_bsh, &adev->rmmio_base,
+	    &adev->rmmio_size, 0)) {
+		printf(": can't map rmmio space\n");
+		return;
+	}
+
+	/*
+	 * Make sure we have a base address for the ROM such that we
+	 * can map it later.
+	 */
+	s = splhigh();
+	addr = pci_conf_read(pa->pa_pc, pa->pa_tag, PCI_ROM_REG);
+	pci_conf_write(pa->pa_pc, pa->pa_tag, PCI_ROM_REG, ~PCI_ROM_ENABLE);
+	mask = pci_conf_read(pa->pa_pc, pa->pa_tag, PCI_ROM_REG);
+	pci_conf_write(pa->pa_pc, pa->pa_tag, PCI_ROM_REG, addr);
+	splx(s);
+
+	if (addr == 0 && PCI_ROM_SIZE(mask) != 0 && pa->pa_memex) {
+		bus_size_t size, start, end;
+		bus_addr_t base;
+
+		size = PCI_ROM_SIZE(mask);
+		start = max(PCI_MEM_START, pa->pa_memex->ex_start);
+		end = min(PCI_MEM_END, pa->pa_memex->ex_end);
+		if (extent_alloc_subregion(pa->pa_memex, start, end, size,
+		    size, 0, 0, 0, &base) == 0)
+			pci_conf_write(pa->pa_pc, pa->pa_tag, PCI_ROM_REG, base);
+	}
+
+	printf("\n");
+
+	/* from amdgpu_init() */
+	if (amdgpu_refcnt == 0) {
+		drm_sched_fence_slab_init();
+
+		if (amdgpu_sync_init()) {
+			printf(": amdgpu_sync_init failed\n");
+			return;
+		}
+
+		if (amdgpu_fence_slab_init()) {
+			amdgpu_sync_fini();
+			printf(": amdgpu_fence_slab_init failed\n");
+			return;
+		}
+
+		amdgpu_kms_driver.num_ioctls = amdgpu_max_kms_ioctl;
+		amdgpu_register_atpx_handler();
+	}
+	amdgpu_refcnt++;
+
+	/* from amdgpu_pci_probe() */
+{
+	int ret;
+	bool supports_atomic = false;
+
+	if (!amdgpu_virtual_display &&
+	     amdgpu_device_asic_has_dc_support(adev->family))
+		supports_atomic = true;
+
+	if ((adev->flags & AMD_EXP_HW_SUPPORT) && !amdgpu_exp_hw_support) {
+		DRM_INFO("This hardware requires experimental hardware support.\n");
+	}
+
+	/*
+	 * Initialize amdkfd before starting radeon.
+	 */
+	amdgpu_amdkfd_init();
+
+	/* warn the user if they mix atomic and non-atomic capable GPUs */
+	if ((amdgpu_kms_driver.driver_features & DRIVER_ATOMIC) && !supports_atomic)
+		DRM_ERROR("Mixing atomic and non-atomic capable GPUs!\n");
+	/* support atomic early so the atomic debugfs stuff gets created */
+	if (supports_atomic)
+		amdgpu_kms_driver.driver_features |= DRIVER_ATOMIC;
+}
+
+	dev = drm_attach_pci(&amdgpu_kms_driver, pa, 0, adev->primary,
+	    self, NULL);
+	adev->ddev = dev;
+	adev->pdev = dev->pdev;
+
+	if (!amdgpu_msi_ok(adev))
+		pa->pa_flags &= ~PCI_FLAGS_MSI_ENABLED;
+
+	adev->irq.msi_enabled = false;
+	if (pci_intr_map_msi(pa, &adev->intrh) == 0)
+		adev->irq.msi_enabled = true;
+	else if (pci_intr_map(pa, &adev->intrh) != 0) {
+		printf(": couldn't map interrupt\n");
+		return;
+	}
+	printf("%s: %s\n", adev->self.dv_xname,
+	    pci_intr_string(pa->pa_pc, adev->intrh));
+
+	adev->irqh = pci_intr_establish(pa->pa_pc, adev->intrh, IPL_TTY,
+	    amdgpu_irq_handler, adev->ddev, adev->self.dv_xname);
+	if (adev->irqh == NULL) {
+		printf("%s: couldn't establish interrupt\n",
+		    adev->self.dv_xname);
+		return;
+	}
+	adev->pdev->irq = -1;
+
+	fb_aper = bus_space_mmap(adev->memt, adev->fb_aper_offset, 0, 0, 0);
+	if (fb_aper != -1)
+		rasops_claim_framebuffer(fb_aper, adev->fb_aper_size, self);
+
+
+	adev->shutdown = true;
+	config_mountroot(self, amdgpu_attachhook);
+}
+
+int
+amdgpu_forcedetach(struct amdgpu_device *adev)
+{
+	struct pci_softc	*sc = (struct pci_softc *)adev->self.dv_parent;
+	pcitag_t		 tag = adev->pa_tag;
+
+#if NVGA > 0
+	if (adev->primary)
+		vga_console_attached = 0;
+#endif
+
+	/* reprobe pci device for non efi systems */
+#if NEFIFB > 0
+	if (bios_efiinfo == NULL && !efifb_cb_found()) {
+#endif
+		config_detach(&adev->self, 0);
+		return pci_probe_device(sc, tag, NULL, NULL);
+#if NEFIFB > 0
+	} else if (adev->primary) {
+		efifb_reattach();
+	}
+#endif
+
+	return 0;
+}
+
+void amdgpu_burner(void *, u_int, u_int);
+int amdgpu_wsioctl(void *, u_long, caddr_t, int, struct proc *);
+paddr_t amdgpu_wsmmap(void *, off_t, int);
+int amdgpu_alloc_screen(void *, const struct wsscreen_descr *,
+    void **, int *, int *, long *);
+void amdgpu_free_screen(void *, void *);
+int amdgpu_show_screen(void *, void *, int,
+    void (*)(void *, int, int), void *);
+void amdgpu_doswitch(void *);
+void amdgpu_enter_ddb(void *, void *);
+
+struct wsscreen_descr amdgpu_stdscreen = {
+	"std",
+	0, 0,
+	0,
+	0, 0,
+	WSSCREEN_UNDERLINE | WSSCREEN_HILIT |
+	WSSCREEN_REVERSE | WSSCREEN_WSCOLORS
+};
+
+const struct wsscreen_descr *amdgpu_scrlist[] = {
+	&amdgpu_stdscreen,
+};
+
+struct wsscreen_list amdgpu_screenlist = {
+	nitems(amdgpu_scrlist), amdgpu_scrlist
+};
+
+struct wsdisplay_accessops amdgpu_accessops = {
+	.ioctl = amdgpu_wsioctl,
+	.mmap = amdgpu_wsmmap,
+	.alloc_screen = amdgpu_alloc_screen,
+	.free_screen = amdgpu_free_screen,
+	.show_screen = amdgpu_show_screen,
+	.enter_ddb = amdgpu_enter_ddb,
+	.getchar = rasops_getchar,
+	.load_font = rasops_load_font,
+	.list_font = rasops_list_font,
+	.scrollback = rasops_scrollback,
+	.burn_screen = amdgpu_burner
+};
+
+int
+amdgpu_wsioctl(void *v, u_long cmd, caddr_t data, int flag, struct proc *p)
+{
+	struct rasops_info *ri = v;
+	struct wsdisplay_fbinfo *wdf;
+
+	switch (cmd) {
+	case WSDISPLAYIO_GTYPE:
+		*(int *)data = WSDISPLAY_TYPE_RADEONDRM;
+		return 0;
+	case WSDISPLAYIO_GINFO:
+		wdf = (struct wsdisplay_fbinfo *)data;
+		wdf->width = ri->ri_width;
+		wdf->height = ri->ri_height;
+		wdf->depth = ri->ri_depth;
+		wdf->cmsize = 0;
+		return 0;
+	default:
+		return -1;
+	}
+}
+
+paddr_t
+amdgpu_wsmmap(void *v, off_t off, int prot)
+{
+	return (-1);
+}
+
+int
+amdgpu_alloc_screen(void *v, const struct wsscreen_descr *type,
+    void **cookiep, int *curxp, int *curyp, long *attrp)
+{
+	return rasops_alloc_screen(v, cookiep, curxp, curyp, attrp);
+}
+
+void
+amdgpu_free_screen(void *v, void *cookie)
+{
+	return rasops_free_screen(v, cookie);
+}
+
+int
+amdgpu_show_screen(void *v, void *cookie, int waitok,
+    void (*cb)(void *, int, int), void *cbarg)
+{
+	struct rasops_info *ri = v;
+	struct amdgpu_device *adev = ri->ri_hw;
+
+	if (cookie == ri->ri_active)
+		return (0);
+
+	adev->switchcb = cb;
+	adev->switchcbarg = cbarg;
+	adev->switchcookie = cookie;
+	if (cb) {
+		task_add(systq, &adev->switchtask);
+		return (EAGAIN);
+	}
+
+	amdgpu_doswitch(v);
+
+	return (0);
+}
+
+void
+amdgpu_doswitch(void *v)
+{
+	struct rasops_info *ri = v;
+	struct amdgpu_device *adev = ri->ri_hw;
+	struct amdgpu_crtc *amdgpu_crtc;
+	int i, crtc;
+
+	rasops_show_screen(ri, adev->switchcookie, 0, NULL, NULL);
+	drm_fb_helper_restore_fbdev_mode_unlocked((void *)adev->mode_info.rfbdev);
+
+	if (adev->switchcb)
+		(adev->switchcb)(adev->switchcbarg, 0, 0);
+}
+
+void
+amdgpu_enter_ddb(void *v, void *cookie)
+{
+	struct rasops_info *ri = v;
+	struct amdgpu_device *adev = ri->ri_hw;
+	struct drm_fb_helper *fb_helper = (void *)adev->mode_info.rfbdev;
+
+	if (cookie == ri->ri_active)
+		return;
+
+	rasops_show_screen(ri, cookie, 0, NULL, NULL);
+	drm_fb_helper_debug_enter(fb_helper->fbdev);
+}
+
+
+void
+amdgpu_attachhook(struct device *self)
+{
+	struct amdgpu_device	*adev = (struct amdgpu_device *)self;
+	struct drm_device	*dev = adev->ddev;
+	int r, acpi_status;
+
+	if ((amdgpu_runtime_pm != 0) &&
+	    amdgpu_has_atpx() &&
+	    (amdgpu_is_atpx_hybrid() ||
+	     amdgpu_has_atpx_dgpu_power_cntl()) &&
+	    ((adev->flags & AMD_IS_APU) == 0) &&
+	    !pci_is_thunderbolt_attached(dev->pdev))
+		adev->flags |= AMD_IS_PX;
+
+	/* amdgpu_device_init should report only fatal error
+	 * like memory allocation failure or iomapping failure,
+	 * or memory manager initialization failure, it must
+	 * properly initialize the GPU MC controller and permit
+	 * VRAM allocation
+	 */
+	r = amdgpu_device_init(adev, dev, dev->pdev, adev->flags);
+	if (r) {
+		dev_err(&dev->pdev->dev, "Fatal error during GPU init\n");
+		goto out;
+	}
+
+	/* Call ACPI methods: require modeset init
+	 * but failure is not fatal
+	 */
+	if (!r) {
+		acpi_status = amdgpu_acpi_init(adev);
+		if (acpi_status)
+		dev_dbg(&dev->pdev->dev,
+				"Error during ACPI methods call\n");
+	}
+
+	if (amdgpu_device_is_px(dev)) {
+		pm_runtime_use_autosuspend(dev->dev);
+		pm_runtime_set_autosuspend_delay(dev->dev, 5000);
+		pm_runtime_set_active(dev->dev);
+		pm_runtime_allow(dev->dev);
+		pm_runtime_mark_last_busy(dev->dev);
+		pm_runtime_put_autosuspend(dev->dev);
+	}
+
+{
+	struct drm_fb_helper *fb_helper = (void *)adev->mode_info.rfbdev;
+	struct wsemuldisplaydev_attach_args aa;
+	struct rasops_info *ri = &adev->ro;
+
+	task_set(&adev->switchtask, amdgpu_doswitch, ri);
+
+	if (ri->ri_bits == NULL)
+		return;
+
+	drm_fb_helper_restore_fbdev_mode_unlocked(fb_helper);
+
+	ri->ri_flg = RI_CENTER | RI_VCONS | RI_WRONLY;
+	rasops_init(ri, 160, 160);
+
+	ri->ri_hw = adev;
+
+	amdgpu_stdscreen.capabilities = ri->ri_caps;
+	amdgpu_stdscreen.nrows = ri->ri_rows;
+	amdgpu_stdscreen.ncols = ri->ri_cols;
+	amdgpu_stdscreen.textops = &ri->ri_ops;
+	amdgpu_stdscreen.fontwidth = ri->ri_font->fontwidth;
+	amdgpu_stdscreen.fontheight = ri->ri_font->fontheight;
+
+	aa.console = adev->console;
+	aa.primary = adev->primary;
+	aa.scrdata = &amdgpu_screenlist;
+	aa.accessops = &amdgpu_accessops;
+	aa.accesscookie = ri;
+	aa.defaultscreens = 0;
+
+	if (adev->console) {
+		long defattr;
+
+		ri->ri_ops.alloc_attr(ri->ri_active, 0, 0, 0, &defattr);
+		wsdisplay_cnattach(&amdgpu_stdscreen, ri->ri_active,
+		    ri->ri_ccol, ri->ri_crow, defattr);
+	}
+
+	/*
+	 * Now that we've taken over the console, disable decoding of
+	 * VGA legacy addresses, and opt out of arbitration.
+	 */
+	amdgpu_asic_set_vga_state(adev, false);
+	pci_disable_legacy_vga(&adev->self);
+
+	printf("%s: %dx%d, %dbpp\n", adev->self.dv_xname,
+	    ri->ri_width, ri->ri_height, ri->ri_depth);
+
+	config_found_sm(&adev->self, &aa, wsemuldisplaydevprint,
+	    wsemuldisplaydevsubmatch);
+
+	/*
+	 * in linux via amdgpu_pci_probe -> drm_dev_register
+	 */
+	drm_dev_register(dev, adev->flags);
+}
+
+out:
+	if (r) {
+		/* balance pm_runtime_get_sync in amdgpu_driver_unload_kms */
+		if (amdgpu_device_is_px(dev))
+			pm_runtime_put_noidle(dev->dev);
+		amdgpu_fatal_error = 1;
+		amdgpu_forcedetach(adev);
+	}
+}
+
+/* from amdgpu_exit amdgpu_driver_unload_kms */
+int
+amdgpu_detach(struct device *self, int flags)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)self;
+	struct drm_device *dev = adev->ddev;
+
+	if (adev == NULL)
+		return 0;
+
+	amdgpu_refcnt--;
+
+	if (amdgpu_refcnt == 0)
+		amdgpu_amdkfd_fini();
+
+	pci_intr_disestablish(adev->pc, adev->irqh);
+
+	if (amdgpu_sriov_vf(adev))
+		amdgpu_virt_request_full_gpu(adev, false);
+
+	if (amdgpu_device_is_px(dev)) {
+		pm_runtime_get_sync(dev->dev);
+		pm_runtime_forbid(dev->dev);
+	}
+
+	amdgpu_acpi_fini(adev);
+
+	amdgpu_device_fini(adev);
+
+	if (amdgpu_refcnt == 0) {
+		amdgpu_unregister_atpx_handler();
+		amdgpu_sync_fini();
+		amdgpu_fence_slab_fini();
+
+		drm_sched_fence_slab_fini();
+	}
+	
+	if (adev->ddev != NULL) {
+		config_detach(adev->ddev->dev, flags);
+		adev->ddev = NULL;
+	}
+
+	return 0;
+}
+
+int     
+amdgpu_activate(struct device *self, int act)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)self;
+	int rv = 0;
+
+	if (adev->ddev == NULL)
+		return (0);
+
+	switch (act) {
+	case DVACT_QUIESCE:
+		rv = config_activate_children(self, act);
+		amdgpu_device_suspend(adev->ddev, true, true);
+		break;
+	case DVACT_SUSPEND:
+		break;
+	case DVACT_RESUME:
+		break;
+	case DVACT_WAKEUP:
+		amdgpu_device_resume(adev->ddev, true, true);
+		rv = config_activate_children(self, act);
+		break;
+	}
+
+	return (rv);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_kv_dpm.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_kv_dpm.c
new file mode 100644
index 00000000000..cb79a93c2eb
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_kv_dpm.c
@@ -0,0 +1,3383 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "cikd.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_dpm.h"
+#include "kv_dpm.h"
+#include "gfx_v7_0.h"
+#include <linux/seq_file.h>
+
+#include "smu/smu_7_0_0_d.h"
+#include "smu/smu_7_0_0_sh_mask.h"
+
+#include "gca/gfx_7_2_d.h"
+#include "gca/gfx_7_2_sh_mask.h"
+
+#define KV_MAX_DEEPSLEEP_DIVIDER_ID     5
+#define KV_MINIMUM_ENGINE_CLOCK         800
+#define SMC_RAM_END                     0x40000
+
+static const struct amd_pm_funcs kv_dpm_funcs;
+
+static void kv_dpm_set_irq_funcs(struct amdgpu_device *adev);
+static int kv_enable_nb_dpm(struct amdgpu_device *adev,
+			    bool enable);
+static void kv_init_graphics_levels(struct amdgpu_device *adev);
+static int kv_calculate_ds_divider(struct amdgpu_device *adev);
+static int kv_calculate_nbps_level_settings(struct amdgpu_device *adev);
+static int kv_calculate_dpm_settings(struct amdgpu_device *adev);
+static void kv_enable_new_levels(struct amdgpu_device *adev);
+static void kv_program_nbps_index_settings(struct amdgpu_device *adev,
+					   struct amdgpu_ps *new_rps);
+static int kv_set_enabled_level(struct amdgpu_device *adev, u32 level);
+static int kv_set_enabled_levels(struct amdgpu_device *adev);
+static int kv_force_dpm_highest(struct amdgpu_device *adev);
+static int kv_force_dpm_lowest(struct amdgpu_device *adev);
+static void kv_apply_state_adjust_rules(struct amdgpu_device *adev,
+					struct amdgpu_ps *new_rps,
+					struct amdgpu_ps *old_rps);
+static int kv_set_thermal_temperature_range(struct amdgpu_device *adev,
+					    int min_temp, int max_temp);
+static int kv_init_fps_limits(struct amdgpu_device *adev);
+
+static void kv_dpm_powergate_samu(struct amdgpu_device *adev, bool gate);
+static void kv_dpm_powergate_acp(struct amdgpu_device *adev, bool gate);
+
+
+static u32 kv_convert_vid2_to_vid7(struct amdgpu_device *adev,
+				   struct sumo_vid_mapping_table *vid_mapping_table,
+				   u32 vid_2bit)
+{
+	struct amdgpu_clock_voltage_dependency_table *vddc_sclk_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	u32 i;
+
+	if (vddc_sclk_table && vddc_sclk_table->count) {
+		if (vid_2bit < vddc_sclk_table->count)
+			return vddc_sclk_table->entries[vid_2bit].v;
+		else
+			return vddc_sclk_table->entries[vddc_sclk_table->count - 1].v;
+	} else {
+		for (i = 0; i < vid_mapping_table->num_entries; i++) {
+			if (vid_mapping_table->entries[i].vid_2bit == vid_2bit)
+				return vid_mapping_table->entries[i].vid_7bit;
+		}
+		return vid_mapping_table->entries[vid_mapping_table->num_entries - 1].vid_7bit;
+	}
+}
+
+static u32 kv_convert_vid7_to_vid2(struct amdgpu_device *adev,
+				   struct sumo_vid_mapping_table *vid_mapping_table,
+				   u32 vid_7bit)
+{
+	struct amdgpu_clock_voltage_dependency_table *vddc_sclk_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	u32 i;
+
+	if (vddc_sclk_table && vddc_sclk_table->count) {
+		for (i = 0; i < vddc_sclk_table->count; i++) {
+			if (vddc_sclk_table->entries[i].v == vid_7bit)
+				return i;
+		}
+		return vddc_sclk_table->count - 1;
+	} else {
+		for (i = 0; i < vid_mapping_table->num_entries; i++) {
+			if (vid_mapping_table->entries[i].vid_7bit == vid_7bit)
+				return vid_mapping_table->entries[i].vid_2bit;
+		}
+
+		return vid_mapping_table->entries[vid_mapping_table->num_entries - 1].vid_2bit;
+	}
+}
+
+static void sumo_take_smu_control(struct amdgpu_device *adev, bool enable)
+{
+/* This bit selects who handles display phy powergating.
+ * Clear the bit to let atom handle it.
+ * Set it to let the driver handle it.
+ * For now we just let atom handle it.
+ */
+#if 0
+	u32 v = RREG32(mmDOUT_SCRATCH3);
+
+	if (enable)
+		v |= 0x4;
+	else
+		v &= 0xFFFFFFFB;
+
+	WREG32(mmDOUT_SCRATCH3, v);
+#endif
+}
+
+static void sumo_construct_sclk_voltage_mapping_table(struct amdgpu_device *adev,
+						      struct sumo_sclk_voltage_mapping_table *sclk_voltage_mapping_table,
+						      ATOM_AVAILABLE_SCLK_LIST *table)
+{
+	u32 i;
+	u32 n = 0;
+	u32 prev_sclk = 0;
+
+	for (i = 0; i < SUMO_MAX_HARDWARE_POWERLEVELS; i++) {
+		if (table[i].ulSupportedSCLK > prev_sclk) {
+			sclk_voltage_mapping_table->entries[n].sclk_frequency =
+				table[i].ulSupportedSCLK;
+			sclk_voltage_mapping_table->entries[n].vid_2bit =
+				table[i].usVoltageIndex;
+			prev_sclk = table[i].ulSupportedSCLK;
+			n++;
+		}
+	}
+
+	sclk_voltage_mapping_table->num_max_dpm_entries = n;
+}
+
+static void sumo_construct_vid_mapping_table(struct amdgpu_device *adev,
+					     struct sumo_vid_mapping_table *vid_mapping_table,
+					     ATOM_AVAILABLE_SCLK_LIST *table)
+{
+	u32 i, j;
+
+	for (i = 0; i < SUMO_MAX_HARDWARE_POWERLEVELS; i++) {
+		if (table[i].ulSupportedSCLK != 0) {
+			vid_mapping_table->entries[table[i].usVoltageIndex].vid_7bit =
+				table[i].usVoltageID;
+			vid_mapping_table->entries[table[i].usVoltageIndex].vid_2bit =
+				table[i].usVoltageIndex;
+		}
+	}
+
+	for (i = 0; i < SUMO_MAX_NUMBER_VOLTAGES; i++) {
+		if (vid_mapping_table->entries[i].vid_7bit == 0) {
+			for (j = i + 1; j < SUMO_MAX_NUMBER_VOLTAGES; j++) {
+				if (vid_mapping_table->entries[j].vid_7bit != 0) {
+					vid_mapping_table->entries[i] =
+						vid_mapping_table->entries[j];
+					vid_mapping_table->entries[j].vid_7bit = 0;
+					break;
+				}
+			}
+
+			if (j == SUMO_MAX_NUMBER_VOLTAGES)
+				break;
+		}
+	}
+
+	vid_mapping_table->num_entries = i;
+}
+
+#if 0
+static const struct kv_lcac_config_values sx_local_cac_cfg_kv[] =
+{
+	{  0,       4,        1    },
+	{  1,       4,        1    },
+	{  2,       5,        1    },
+	{  3,       4,        2    },
+	{  4,       1,        1    },
+	{  5,       5,        2    },
+	{  6,       6,        1    },
+	{  7,       9,        2    },
+	{ 0xffffffff }
+};
+
+static const struct kv_lcac_config_values mc0_local_cac_cfg_kv[] =
+{
+	{  0,       4,        1    },
+	{ 0xffffffff }
+};
+
+static const struct kv_lcac_config_values mc1_local_cac_cfg_kv[] =
+{
+	{  0,       4,        1    },
+	{ 0xffffffff }
+};
+
+static const struct kv_lcac_config_values mc2_local_cac_cfg_kv[] =
+{
+	{  0,       4,        1    },
+	{ 0xffffffff }
+};
+
+static const struct kv_lcac_config_values mc3_local_cac_cfg_kv[] =
+{
+	{  0,       4,        1    },
+	{ 0xffffffff }
+};
+
+static const struct kv_lcac_config_values cpl_local_cac_cfg_kv[] =
+{
+	{  0,       4,        1    },
+	{  1,       4,        1    },
+	{  2,       5,        1    },
+	{  3,       4,        1    },
+	{  4,       1,        1    },
+	{  5,       5,        1    },
+	{  6,       6,        1    },
+	{  7,       9,        1    },
+	{  8,       4,        1    },
+	{  9,       2,        1    },
+	{  10,      3,        1    },
+	{  11,      6,        1    },
+	{  12,      8,        2    },
+	{  13,      1,        1    },
+	{  14,      2,        1    },
+	{  15,      3,        1    },
+	{  16,      1,        1    },
+	{  17,      4,        1    },
+	{  18,      3,        1    },
+	{  19,      1,        1    },
+	{  20,      8,        1    },
+	{  21,      5,        1    },
+	{  22,      1,        1    },
+	{  23,      1,        1    },
+	{  24,      4,        1    },
+	{  27,      6,        1    },
+	{  28,      1,        1    },
+	{ 0xffffffff }
+};
+
+static const struct kv_lcac_config_reg sx0_cac_config_reg[] =
+{
+	{ 0xc0400d00, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
+};
+
+static const struct kv_lcac_config_reg mc0_cac_config_reg[] =
+{
+	{ 0xc0400d30, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
+};
+
+static const struct kv_lcac_config_reg mc1_cac_config_reg[] =
+{
+	{ 0xc0400d3c, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
+};
+
+static const struct kv_lcac_config_reg mc2_cac_config_reg[] =
+{
+	{ 0xc0400d48, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
+};
+
+static const struct kv_lcac_config_reg mc3_cac_config_reg[] =
+{
+	{ 0xc0400d54, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
+};
+
+static const struct kv_lcac_config_reg cpl_cac_config_reg[] =
+{
+	{ 0xc0400d80, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
+};
+#endif
+
+static const struct kv_pt_config_reg didt_config_kv[] =
+{
+	{ 0x10, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x10, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x10, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x10, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x11, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x11, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x11, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x11, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x12, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x12, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x12, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x12, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x2, 0x00003fff, 0, 0x4, KV_CONFIGREG_DIDT_IND },
+	{ 0x2, 0x03ff0000, 16, 0x80, KV_CONFIGREG_DIDT_IND },
+	{ 0x2, 0x78000000, 27, 0x3, KV_CONFIGREG_DIDT_IND },
+	{ 0x1, 0x0000ffff, 0, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x1, 0xffff0000, 16, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x0, 0x00000001, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x30, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x30, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x30, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x30, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x31, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x31, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x31, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x31, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x32, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x32, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x32, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x32, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x22, 0x00003fff, 0, 0x4, KV_CONFIGREG_DIDT_IND },
+	{ 0x22, 0x03ff0000, 16, 0x80, KV_CONFIGREG_DIDT_IND },
+	{ 0x22, 0x78000000, 27, 0x3, KV_CONFIGREG_DIDT_IND },
+	{ 0x21, 0x0000ffff, 0, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x21, 0xffff0000, 16, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x20, 0x00000001, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x50, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x50, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x50, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x50, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x51, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x51, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x51, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x51, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x52, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x52, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x52, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x52, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x42, 0x00003fff, 0, 0x4, KV_CONFIGREG_DIDT_IND },
+	{ 0x42, 0x03ff0000, 16, 0x80, KV_CONFIGREG_DIDT_IND },
+	{ 0x42, 0x78000000, 27, 0x3, KV_CONFIGREG_DIDT_IND },
+	{ 0x41, 0x0000ffff, 0, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x41, 0xffff0000, 16, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x40, 0x00000001, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x70, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x70, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x70, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x70, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x71, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x71, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x71, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x71, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x72, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x72, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x72, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x72, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x62, 0x00003fff, 0, 0x4, KV_CONFIGREG_DIDT_IND },
+	{ 0x62, 0x03ff0000, 16, 0x80, KV_CONFIGREG_DIDT_IND },
+	{ 0x62, 0x78000000, 27, 0x3, KV_CONFIGREG_DIDT_IND },
+	{ 0x61, 0x0000ffff, 0, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x61, 0xffff0000, 16, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x60, 0x00000001, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0xFFFFFFFF }
+};
+
+static struct kv_ps *kv_get_ps(struct amdgpu_ps *rps)
+{
+	struct kv_ps *ps = rps->ps_priv;
+
+	return ps;
+}
+
+static struct kv_power_info *kv_get_pi(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = adev->pm.dpm.priv;
+
+	return pi;
+}
+
+#if 0
+static void kv_program_local_cac_table(struct amdgpu_device *adev,
+				       const struct kv_lcac_config_values *local_cac_table,
+				       const struct kv_lcac_config_reg *local_cac_reg)
+{
+	u32 i, count, data;
+	const struct kv_lcac_config_values *values = local_cac_table;
+
+	while (values->block_id != 0xffffffff) {
+		count = values->signal_id;
+		for (i = 0; i < count; i++) {
+			data = ((values->block_id << local_cac_reg->block_shift) &
+				local_cac_reg->block_mask);
+			data |= ((i << local_cac_reg->signal_shift) &
+				 local_cac_reg->signal_mask);
+			data |= ((values->t << local_cac_reg->t_shift) &
+				 local_cac_reg->t_mask);
+			data |= ((1 << local_cac_reg->enable_shift) &
+				 local_cac_reg->enable_mask);
+			WREG32_SMC(local_cac_reg->cntl, data);
+		}
+		values++;
+	}
+}
+#endif
+
+static int kv_program_pt_config_registers(struct amdgpu_device *adev,
+					  const struct kv_pt_config_reg *cac_config_regs)
+{
+	const struct kv_pt_config_reg *config_regs = cac_config_regs;
+	u32 data;
+	u32 cache = 0;
+
+	if (config_regs == NULL)
+		return -EINVAL;
+
+	while (config_regs->offset != 0xFFFFFFFF) {
+		if (config_regs->type == KV_CONFIGREG_CACHE) {
+			cache |= ((config_regs->value << config_regs->shift) & config_regs->mask);
+		} else {
+			switch (config_regs->type) {
+			case KV_CONFIGREG_SMC_IND:
+				data = RREG32_SMC(config_regs->offset);
+				break;
+			case KV_CONFIGREG_DIDT_IND:
+				data = RREG32_DIDT(config_regs->offset);
+				break;
+			default:
+				data = RREG32(config_regs->offset);
+				break;
+			}
+
+			data &= ~config_regs->mask;
+			data |= ((config_regs->value << config_regs->shift) & config_regs->mask);
+			data |= cache;
+			cache = 0;
+
+			switch (config_regs->type) {
+			case KV_CONFIGREG_SMC_IND:
+				WREG32_SMC(config_regs->offset, data);
+				break;
+			case KV_CONFIGREG_DIDT_IND:
+				WREG32_DIDT(config_regs->offset, data);
+				break;
+			default:
+				WREG32(config_regs->offset, data);
+				break;
+			}
+		}
+		config_regs++;
+	}
+
+	return 0;
+}
+
+static void kv_do_enable_didt(struct amdgpu_device *adev, bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 data;
+
+	if (pi->caps_sq_ramping) {
+		data = RREG32_DIDT(ixDIDT_SQ_CTRL0);
+		if (enable)
+			data |= DIDT_SQ_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_SQ_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_SQ_CTRL0, data);
+	}
+
+	if (pi->caps_db_ramping) {
+		data = RREG32_DIDT(ixDIDT_DB_CTRL0);
+		if (enable)
+			data |= DIDT_DB_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_DB_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_DB_CTRL0, data);
+	}
+
+	if (pi->caps_td_ramping) {
+		data = RREG32_DIDT(ixDIDT_TD_CTRL0);
+		if (enable)
+			data |= DIDT_TD_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_TD_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_TD_CTRL0, data);
+	}
+
+	if (pi->caps_tcp_ramping) {
+		data = RREG32_DIDT(ixDIDT_TCP_CTRL0);
+		if (enable)
+			data |= DIDT_TCP_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_TCP_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_TCP_CTRL0, data);
+	}
+}
+
+static int kv_enable_didt(struct amdgpu_device *adev, bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	if (pi->caps_sq_ramping ||
+	    pi->caps_db_ramping ||
+	    pi->caps_td_ramping ||
+	    pi->caps_tcp_ramping) {
+		adev->gfx.rlc.funcs->enter_safe_mode(adev);
+
+		if (enable) {
+			ret = kv_program_pt_config_registers(adev, didt_config_kv);
+			if (ret) {
+				adev->gfx.rlc.funcs->exit_safe_mode(adev);
+				return ret;
+			}
+		}
+
+		kv_do_enable_didt(adev, enable);
+
+		adev->gfx.rlc.funcs->exit_safe_mode(adev);
+	}
+
+	return 0;
+}
+
+#if 0
+static void kv_initialize_hardware_cac_manager(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	if (pi->caps_cac) {
+		WREG32_SMC(ixLCAC_SX0_OVR_SEL, 0);
+		WREG32_SMC(ixLCAC_SX0_OVR_VAL, 0);
+		kv_program_local_cac_table(adev, sx_local_cac_cfg_kv, sx0_cac_config_reg);
+
+		WREG32_SMC(ixLCAC_MC0_OVR_SEL, 0);
+		WREG32_SMC(ixLCAC_MC0_OVR_VAL, 0);
+		kv_program_local_cac_table(adev, mc0_local_cac_cfg_kv, mc0_cac_config_reg);
+
+		WREG32_SMC(ixLCAC_MC1_OVR_SEL, 0);
+		WREG32_SMC(ixLCAC_MC1_OVR_VAL, 0);
+		kv_program_local_cac_table(adev, mc1_local_cac_cfg_kv, mc1_cac_config_reg);
+
+		WREG32_SMC(ixLCAC_MC2_OVR_SEL, 0);
+		WREG32_SMC(ixLCAC_MC2_OVR_VAL, 0);
+		kv_program_local_cac_table(adev, mc2_local_cac_cfg_kv, mc2_cac_config_reg);
+
+		WREG32_SMC(ixLCAC_MC3_OVR_SEL, 0);
+		WREG32_SMC(ixLCAC_MC3_OVR_VAL, 0);
+		kv_program_local_cac_table(adev, mc3_local_cac_cfg_kv, mc3_cac_config_reg);
+
+		WREG32_SMC(ixLCAC_CPL_OVR_SEL, 0);
+		WREG32_SMC(ixLCAC_CPL_OVR_VAL, 0);
+		kv_program_local_cac_table(adev, cpl_local_cac_cfg_kv, cpl_cac_config_reg);
+	}
+}
+#endif
+
+static int kv_enable_smc_cac(struct amdgpu_device *adev, bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret = 0;
+
+	if (pi->caps_cac) {
+		if (enable) {
+			ret = amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_EnableCac);
+			if (ret)
+				pi->cac_enabled = false;
+			else
+				pi->cac_enabled = true;
+		} else if (pi->cac_enabled) {
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_DisableCac);
+			pi->cac_enabled = false;
+		}
+	}
+
+	return ret;
+}
+
+static int kv_process_firmware_header(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 tmp;
+	int ret;
+
+	ret = amdgpu_kv_read_smc_sram_dword(adev, SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, DpmTable),
+				     &tmp, pi->sram_end);
+
+	if (ret == 0)
+		pi->dpm_table_start = tmp;
+
+	ret = amdgpu_kv_read_smc_sram_dword(adev, SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, SoftRegisters),
+				     &tmp, pi->sram_end);
+
+	if (ret == 0)
+		pi->soft_regs_start = tmp;
+
+	return ret;
+}
+
+static int kv_enable_dpm_voltage_scaling(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	pi->graphics_voltage_change_enable = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, GraphicsVoltageChangeEnable),
+				   &pi->graphics_voltage_change_enable,
+				   sizeof(u8), pi->sram_end);
+
+	return ret;
+}
+
+static int kv_set_dpm_interval(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	pi->graphics_interval = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, GraphicsInterval),
+				   &pi->graphics_interval,
+				   sizeof(u8), pi->sram_end);
+
+	return ret;
+}
+
+static int kv_set_dpm_boot_state(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, GraphicsBootLevel),
+				   &pi->graphics_boot_level,
+				   sizeof(u8), pi->sram_end);
+
+	return ret;
+}
+
+static void kv_program_vc(struct amdgpu_device *adev)
+{
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_0, 0x3FFFC100);
+}
+
+static void kv_clear_vc(struct amdgpu_device *adev)
+{
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_0, 0);
+}
+
+static int kv_set_divider_value(struct amdgpu_device *adev,
+				u32 index, u32 sclk)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct atom_clock_dividers dividers;
+	int ret;
+
+	ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+						 sclk, false, &dividers);
+	if (ret)
+		return ret;
+
+	pi->graphics_level[index].SclkDid = (u8)dividers.post_div;
+	pi->graphics_level[index].SclkFrequency = cpu_to_be32(sclk);
+
+	return 0;
+}
+
+static u16 kv_convert_8bit_index_to_voltage(struct amdgpu_device *adev,
+					    u16 voltage)
+{
+	return 6200 - (voltage * 25);
+}
+
+static u16 kv_convert_2bit_index_to_voltage(struct amdgpu_device *adev,
+					    u32 vid_2bit)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 vid_8bit = kv_convert_vid2_to_vid7(adev,
+					       &pi->sys_info.vid_mapping_table,
+					       vid_2bit);
+
+	return kv_convert_8bit_index_to_voltage(adev, (u16)vid_8bit);
+}
+
+
+static int kv_set_vid(struct amdgpu_device *adev, u32 index, u32 vid)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->graphics_level[index].VoltageDownH = (u8)pi->voltage_drop_t;
+	pi->graphics_level[index].MinVddNb =
+		cpu_to_be32(kv_convert_2bit_index_to_voltage(adev, vid));
+
+	return 0;
+}
+
+static int kv_set_at(struct amdgpu_device *adev, u32 index, u32 at)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->graphics_level[index].AT = cpu_to_be16((u16)at);
+
+	return 0;
+}
+
+static void kv_dpm_power_level_enable(struct amdgpu_device *adev,
+				      u32 index, bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->graphics_level[index].EnabledForActivity = enable ? 1 : 0;
+}
+
+static void kv_start_dpm(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+
+	tmp |= GENERAL_PWRMGT__GLOBAL_PWRMGT_EN_MASK;
+	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+
+	amdgpu_kv_smc_dpm_enable(adev, true);
+}
+
+static void kv_stop_dpm(struct amdgpu_device *adev)
+{
+	amdgpu_kv_smc_dpm_enable(adev, false);
+}
+
+static void kv_start_am(struct amdgpu_device *adev)
+{
+	u32 sclk_pwrmgt_cntl = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+
+	sclk_pwrmgt_cntl &= ~(SCLK_PWRMGT_CNTL__RESET_SCLK_CNT_MASK |
+			SCLK_PWRMGT_CNTL__RESET_BUSY_CNT_MASK);
+	sclk_pwrmgt_cntl |= SCLK_PWRMGT_CNTL__DYNAMIC_PM_EN_MASK;
+
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, sclk_pwrmgt_cntl);
+}
+
+static void kv_reset_am(struct amdgpu_device *adev)
+{
+	u32 sclk_pwrmgt_cntl = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+
+	sclk_pwrmgt_cntl |= (SCLK_PWRMGT_CNTL__RESET_SCLK_CNT_MASK |
+			SCLK_PWRMGT_CNTL__RESET_BUSY_CNT_MASK);
+
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, sclk_pwrmgt_cntl);
+}
+
+static int kv_freeze_sclk_dpm(struct amdgpu_device *adev, bool freeze)
+{
+	return amdgpu_kv_notify_message_to_smu(adev, freeze ?
+					PPSMC_MSG_SCLKDPM_FreezeLevel : PPSMC_MSG_SCLKDPM_UnfreezeLevel);
+}
+
+static int kv_force_lowest_valid(struct amdgpu_device *adev)
+{
+	return kv_force_dpm_lowest(adev);
+}
+
+static int kv_unforce_levels(struct amdgpu_device *adev)
+{
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS)
+		return amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_NoForcedLevel);
+	else
+		return kv_set_enabled_levels(adev);
+}
+
+static int kv_update_sclk_t(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 low_sclk_interrupt_t = 0;
+	int ret = 0;
+
+	if (pi->caps_sclk_throttle_low_notification) {
+		low_sclk_interrupt_t = cpu_to_be32(pi->low_sclk_interrupt_t);
+
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, LowSclkInterruptT),
+					   (u8 *)&low_sclk_interrupt_t,
+					   sizeof(u32), pi->sram_end);
+	}
+	return ret;
+}
+
+static int kv_program_bootup_state(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+
+	if (table && table->count) {
+		for (i = pi->graphics_dpm_level_count - 1; i > 0; i--) {
+			if (table->entries[i].clk == pi->boot_pl.sclk)
+				break;
+		}
+
+		pi->graphics_boot_level = (u8)i;
+		kv_dpm_power_level_enable(adev, i, true);
+	} else {
+		struct sumo_sclk_voltage_mapping_table *table =
+			&pi->sys_info.sclk_voltage_mapping_table;
+
+		if (table->num_max_dpm_entries == 0)
+			return -EINVAL;
+
+		for (i = pi->graphics_dpm_level_count - 1; i > 0; i--) {
+			if (table->entries[i].sclk_frequency == pi->boot_pl.sclk)
+				break;
+		}
+
+		pi->graphics_boot_level = (u8)i;
+		kv_dpm_power_level_enable(adev, i, true);
+	}
+	return 0;
+}
+
+static int kv_enable_auto_thermal_throttling(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	pi->graphics_therm_throttle_enable = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, GraphicsThermThrottleEnable),
+				   &pi->graphics_therm_throttle_enable,
+				   sizeof(u8), pi->sram_end);
+
+	return ret;
+}
+
+static int kv_upload_dpm_settings(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, GraphicsLevel),
+				   (u8 *)&pi->graphics_level,
+				   sizeof(SMU7_Fusion_GraphicsLevel) * SMU7_MAX_LEVELS_GRAPHICS,
+				   pi->sram_end);
+
+	if (ret)
+		return ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, GraphicsDpmLevelCount),
+				   &pi->graphics_dpm_level_count,
+				   sizeof(u8), pi->sram_end);
+
+	return ret;
+}
+
+static u32 kv_get_clock_difference(u32 a, u32 b)
+{
+	return (a >= b) ? a - b : b - a;
+}
+
+static u32 kv_get_clk_bypass(struct amdgpu_device *adev, u32 clk)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 value;
+
+	if (pi->caps_enable_dfs_bypass) {
+		if (kv_get_clock_difference(clk, 40000) < 200)
+			value = 3;
+		else if (kv_get_clock_difference(clk, 30000) < 200)
+			value = 2;
+		else if (kv_get_clock_difference(clk, 20000) < 200)
+			value = 7;
+		else if (kv_get_clock_difference(clk, 15000) < 200)
+			value = 6;
+		else if (kv_get_clock_difference(clk, 10000) < 200)
+			value = 8;
+		else
+			value = 0;
+	} else {
+		value = 0;
+	}
+
+	return value;
+}
+
+static int kv_populate_uvd_table(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_uvd_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table;
+	struct atom_clock_dividers dividers;
+	int ret;
+	u32 i;
+
+	if (table == NULL || table->count == 0)
+		return 0;
+
+	pi->uvd_level_count = 0;
+	for (i = 0; i < table->count; i++) {
+		if (pi->high_voltage_t &&
+		    (pi->high_voltage_t < table->entries[i].v))
+			break;
+
+		pi->uvd_level[i].VclkFrequency = cpu_to_be32(table->entries[i].vclk);
+		pi->uvd_level[i].DclkFrequency = cpu_to_be32(table->entries[i].dclk);
+		pi->uvd_level[i].MinVddNb = cpu_to_be16(table->entries[i].v);
+
+		pi->uvd_level[i].VClkBypassCntl =
+			(u8)kv_get_clk_bypass(adev, table->entries[i].vclk);
+		pi->uvd_level[i].DClkBypassCntl =
+			(u8)kv_get_clk_bypass(adev, table->entries[i].dclk);
+
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+							 table->entries[i].vclk, false, &dividers);
+		if (ret)
+			return ret;
+		pi->uvd_level[i].VclkDivider = (u8)dividers.post_div;
+
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+							 table->entries[i].dclk, false, &dividers);
+		if (ret)
+			return ret;
+		pi->uvd_level[i].DclkDivider = (u8)dividers.post_div;
+
+		pi->uvd_level_count++;
+	}
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, UvdLevelCount),
+				   (u8 *)&pi->uvd_level_count,
+				   sizeof(u8), pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->uvd_interval = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, UVDInterval),
+				   &pi->uvd_interval,
+				   sizeof(u8), pi->sram_end);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, UvdLevel),
+				   (u8 *)&pi->uvd_level,
+				   sizeof(SMU7_Fusion_UvdLevel) * SMU7_MAX_LEVELS_UVD,
+				   pi->sram_end);
+
+	return ret;
+
+}
+
+static int kv_populate_vce_table(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+	u32 i;
+	struct amdgpu_vce_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+	struct atom_clock_dividers dividers;
+
+	if (table == NULL || table->count == 0)
+		return 0;
+
+	pi->vce_level_count = 0;
+	for (i = 0; i < table->count; i++) {
+		if (pi->high_voltage_t &&
+		    pi->high_voltage_t < table->entries[i].v)
+			break;
+
+		pi->vce_level[i].Frequency = cpu_to_be32(table->entries[i].evclk);
+		pi->vce_level[i].MinVoltage = cpu_to_be16(table->entries[i].v);
+
+		pi->vce_level[i].ClkBypassCntl =
+			(u8)kv_get_clk_bypass(adev, table->entries[i].evclk);
+
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+							 table->entries[i].evclk, false, &dividers);
+		if (ret)
+			return ret;
+		pi->vce_level[i].Divider = (u8)dividers.post_div;
+
+		pi->vce_level_count++;
+	}
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, VceLevelCount),
+				   (u8 *)&pi->vce_level_count,
+				   sizeof(u8),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->vce_interval = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, VCEInterval),
+				   (u8 *)&pi->vce_interval,
+				   sizeof(u8),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, VceLevel),
+				   (u8 *)&pi->vce_level,
+				   sizeof(SMU7_Fusion_ExtClkLevel) * SMU7_MAX_LEVELS_VCE,
+				   pi->sram_end);
+
+	return ret;
+}
+
+static int kv_populate_samu_table(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table;
+	struct atom_clock_dividers dividers;
+	int ret;
+	u32 i;
+
+	if (table == NULL || table->count == 0)
+		return 0;
+
+	pi->samu_level_count = 0;
+	for (i = 0; i < table->count; i++) {
+		if (pi->high_voltage_t &&
+		    pi->high_voltage_t < table->entries[i].v)
+			break;
+
+		pi->samu_level[i].Frequency = cpu_to_be32(table->entries[i].clk);
+		pi->samu_level[i].MinVoltage = cpu_to_be16(table->entries[i].v);
+
+		pi->samu_level[i].ClkBypassCntl =
+			(u8)kv_get_clk_bypass(adev, table->entries[i].clk);
+
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+							 table->entries[i].clk, false, &dividers);
+		if (ret)
+			return ret;
+		pi->samu_level[i].Divider = (u8)dividers.post_div;
+
+		pi->samu_level_count++;
+	}
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, SamuLevelCount),
+				   (u8 *)&pi->samu_level_count,
+				   sizeof(u8),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->samu_interval = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, SAMUInterval),
+				   (u8 *)&pi->samu_interval,
+				   sizeof(u8),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, SamuLevel),
+				   (u8 *)&pi->samu_level,
+				   sizeof(SMU7_Fusion_ExtClkLevel) * SMU7_MAX_LEVELS_SAMU,
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	return ret;
+}
+
+
+static int kv_populate_acp_table(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table;
+	struct atom_clock_dividers dividers;
+	int ret;
+	u32 i;
+
+	if (table == NULL || table->count == 0)
+		return 0;
+
+	pi->acp_level_count = 0;
+	for (i = 0; i < table->count; i++) {
+		pi->acp_level[i].Frequency = cpu_to_be32(table->entries[i].clk);
+		pi->acp_level[i].MinVoltage = cpu_to_be16(table->entries[i].v);
+
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+							 table->entries[i].clk, false, &dividers);
+		if (ret)
+			return ret;
+		pi->acp_level[i].Divider = (u8)dividers.post_div;
+
+		pi->acp_level_count++;
+	}
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, AcpLevelCount),
+				   (u8 *)&pi->acp_level_count,
+				   sizeof(u8),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->acp_interval = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, ACPInterval),
+				   (u8 *)&pi->acp_interval,
+				   sizeof(u8),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, AcpLevel),
+				   (u8 *)&pi->acp_level,
+				   sizeof(SMU7_Fusion_ExtClkLevel) * SMU7_MAX_LEVELS_ACP,
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	return ret;
+}
+
+static void kv_calculate_dfs_bypass_settings(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+
+	if (table && table->count) {
+		for (i = 0; i < pi->graphics_dpm_level_count; i++) {
+			if (pi->caps_enable_dfs_bypass) {
+				if (kv_get_clock_difference(table->entries[i].clk, 40000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 3;
+				else if (kv_get_clock_difference(table->entries[i].clk, 30000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 2;
+				else if (kv_get_clock_difference(table->entries[i].clk, 26600) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 7;
+				else if (kv_get_clock_difference(table->entries[i].clk , 20000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 6;
+				else if (kv_get_clock_difference(table->entries[i].clk , 10000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 8;
+				else
+					pi->graphics_level[i].ClkBypassCntl = 0;
+			} else {
+				pi->graphics_level[i].ClkBypassCntl = 0;
+			}
+		}
+	} else {
+		struct sumo_sclk_voltage_mapping_table *table =
+			&pi->sys_info.sclk_voltage_mapping_table;
+		for (i = 0; i < pi->graphics_dpm_level_count; i++) {
+			if (pi->caps_enable_dfs_bypass) {
+				if (kv_get_clock_difference(table->entries[i].sclk_frequency, 40000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 3;
+				else if (kv_get_clock_difference(table->entries[i].sclk_frequency, 30000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 2;
+				else if (kv_get_clock_difference(table->entries[i].sclk_frequency, 26600) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 7;
+				else if (kv_get_clock_difference(table->entries[i].sclk_frequency, 20000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 6;
+				else if (kv_get_clock_difference(table->entries[i].sclk_frequency, 10000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 8;
+				else
+					pi->graphics_level[i].ClkBypassCntl = 0;
+			} else {
+				pi->graphics_level[i].ClkBypassCntl = 0;
+			}
+		}
+	}
+}
+
+static int kv_enable_ulv(struct amdgpu_device *adev, bool enable)
+{
+	return amdgpu_kv_notify_message_to_smu(adev, enable ?
+					PPSMC_MSG_EnableULV : PPSMC_MSG_DisableULV);
+}
+
+static void kv_reset_acp_boot_level(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->acp_boot_level = 0xff;
+}
+
+static void kv_update_current_ps(struct amdgpu_device *adev,
+				 struct amdgpu_ps *rps)
+{
+	struct kv_ps *new_ps = kv_get_ps(rps);
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->current_rps = *rps;
+	pi->current_ps = *new_ps;
+	pi->current_rps.ps_priv = &pi->current_ps;
+	adev->pm.dpm.current_ps = &pi->current_rps;
+}
+
+static void kv_update_requested_ps(struct amdgpu_device *adev,
+				   struct amdgpu_ps *rps)
+{
+	struct kv_ps *new_ps = kv_get_ps(rps);
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->requested_rps = *rps;
+	pi->requested_ps = *new_ps;
+	pi->requested_rps.ps_priv = &pi->requested_ps;
+	adev->pm.dpm.requested_ps = &pi->requested_rps;
+}
+
+static void kv_dpm_enable_bapm(void *handle, bool enable)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	if (pi->bapm_enable) {
+		ret = amdgpu_kv_smc_bapm_enable(adev, enable);
+		if (ret)
+			DRM_ERROR("amdgpu_kv_smc_bapm_enable failed\n");
+	}
+}
+
+static int kv_dpm_enable(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	ret = kv_process_firmware_header(adev);
+	if (ret) {
+		DRM_ERROR("kv_process_firmware_header failed\n");
+		return ret;
+	}
+	kv_init_fps_limits(adev);
+	kv_init_graphics_levels(adev);
+	ret = kv_program_bootup_state(adev);
+	if (ret) {
+		DRM_ERROR("kv_program_bootup_state failed\n");
+		return ret;
+	}
+	kv_calculate_dfs_bypass_settings(adev);
+	ret = kv_upload_dpm_settings(adev);
+	if (ret) {
+		DRM_ERROR("kv_upload_dpm_settings failed\n");
+		return ret;
+	}
+	ret = kv_populate_uvd_table(adev);
+	if (ret) {
+		DRM_ERROR("kv_populate_uvd_table failed\n");
+		return ret;
+	}
+	ret = kv_populate_vce_table(adev);
+	if (ret) {
+		DRM_ERROR("kv_populate_vce_table failed\n");
+		return ret;
+	}
+	ret = kv_populate_samu_table(adev);
+	if (ret) {
+		DRM_ERROR("kv_populate_samu_table failed\n");
+		return ret;
+	}
+	ret = kv_populate_acp_table(adev);
+	if (ret) {
+		DRM_ERROR("kv_populate_acp_table failed\n");
+		return ret;
+	}
+	kv_program_vc(adev);
+#if 0
+	kv_initialize_hardware_cac_manager(adev);
+#endif
+	kv_start_am(adev);
+	if (pi->enable_auto_thermal_throttling) {
+		ret = kv_enable_auto_thermal_throttling(adev);
+		if (ret) {
+			DRM_ERROR("kv_enable_auto_thermal_throttling failed\n");
+			return ret;
+		}
+	}
+	ret = kv_enable_dpm_voltage_scaling(adev);
+	if (ret) {
+		DRM_ERROR("kv_enable_dpm_voltage_scaling failed\n");
+		return ret;
+	}
+	ret = kv_set_dpm_interval(adev);
+	if (ret) {
+		DRM_ERROR("kv_set_dpm_interval failed\n");
+		return ret;
+	}
+	ret = kv_set_dpm_boot_state(adev);
+	if (ret) {
+		DRM_ERROR("kv_set_dpm_boot_state failed\n");
+		return ret;
+	}
+	ret = kv_enable_ulv(adev, true);
+	if (ret) {
+		DRM_ERROR("kv_enable_ulv failed\n");
+		return ret;
+	}
+	kv_start_dpm(adev);
+	ret = kv_enable_didt(adev, true);
+	if (ret) {
+		DRM_ERROR("kv_enable_didt failed\n");
+		return ret;
+	}
+	ret = kv_enable_smc_cac(adev, true);
+	if (ret) {
+		DRM_ERROR("kv_enable_smc_cac failed\n");
+		return ret;
+	}
+
+	kv_reset_acp_boot_level(adev);
+
+	ret = amdgpu_kv_smc_bapm_enable(adev, false);
+	if (ret) {
+		DRM_ERROR("amdgpu_kv_smc_bapm_enable failed\n");
+		return ret;
+	}
+
+	if (adev->irq.installed &&
+	    amdgpu_is_internal_thermal_sensor(adev->pm.int_thermal_type)) {
+		ret = kv_set_thermal_temperature_range(adev, KV_TEMP_RANGE_MIN, KV_TEMP_RANGE_MAX);
+		if (ret) {
+			DRM_ERROR("kv_set_thermal_temperature_range failed\n");
+			return ret;
+		}
+		amdgpu_irq_get(adev, &adev->pm.dpm.thermal.irq,
+			       AMDGPU_THERMAL_IRQ_LOW_TO_HIGH);
+		amdgpu_irq_get(adev, &adev->pm.dpm.thermal.irq,
+			       AMDGPU_THERMAL_IRQ_HIGH_TO_LOW);
+	}
+
+	return ret;
+}
+
+static void kv_dpm_disable(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
+		       AMDGPU_THERMAL_IRQ_LOW_TO_HIGH);
+	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
+		       AMDGPU_THERMAL_IRQ_HIGH_TO_LOW);
+
+	amdgpu_kv_smc_bapm_enable(adev, false);
+
+	if (adev->asic_type == CHIP_MULLINS)
+		kv_enable_nb_dpm(adev, false);
+
+	/* powerup blocks */
+	kv_dpm_powergate_acp(adev, false);
+	kv_dpm_powergate_samu(adev, false);
+	if (pi->caps_vce_pg) /* power on the VCE block */
+		amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_VCEPowerON);
+	if (pi->caps_uvd_pg) /* power on the UVD block */
+		amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_UVDPowerON);
+
+	kv_enable_smc_cac(adev, false);
+	kv_enable_didt(adev, false);
+	kv_clear_vc(adev);
+	kv_stop_dpm(adev);
+	kv_enable_ulv(adev, false);
+	kv_reset_am(adev);
+
+	kv_update_current_ps(adev, adev->pm.dpm.boot_ps);
+}
+
+#if 0
+static int kv_write_smc_soft_register(struct amdgpu_device *adev,
+				      u16 reg_offset, u32 value)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	return amdgpu_kv_copy_bytes_to_smc(adev, pi->soft_regs_start + reg_offset,
+				    (u8 *)&value, sizeof(u16), pi->sram_end);
+}
+
+static int kv_read_smc_soft_register(struct amdgpu_device *adev,
+				     u16 reg_offset, u32 *value)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	return amdgpu_kv_read_smc_sram_dword(adev, pi->soft_regs_start + reg_offset,
+				      value, pi->sram_end);
+}
+#endif
+
+static void kv_init_sclk_t(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->low_sclk_interrupt_t = 0;
+}
+
+static int kv_init_fps_limits(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret = 0;
+
+	if (pi->caps_fps) {
+		u16 tmp;
+
+		tmp = 45;
+		pi->fps_high_t = cpu_to_be16(tmp);
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, FpsHighT),
+					   (u8 *)&pi->fps_high_t,
+					   sizeof(u16), pi->sram_end);
+
+		tmp = 30;
+		pi->fps_low_t = cpu_to_be16(tmp);
+
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, FpsLowT),
+					   (u8 *)&pi->fps_low_t,
+					   sizeof(u16), pi->sram_end);
+
+	}
+	return ret;
+}
+
+static void kv_init_powergate_state(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->uvd_power_gated = false;
+	pi->vce_power_gated = false;
+	pi->samu_power_gated = false;
+	pi->acp_power_gated = false;
+
+}
+
+static int kv_enable_uvd_dpm(struct amdgpu_device *adev, bool enable)
+{
+	return amdgpu_kv_notify_message_to_smu(adev, enable ?
+					PPSMC_MSG_UVDDPM_Enable : PPSMC_MSG_UVDDPM_Disable);
+}
+
+static int kv_enable_vce_dpm(struct amdgpu_device *adev, bool enable)
+{
+	return amdgpu_kv_notify_message_to_smu(adev, enable ?
+					PPSMC_MSG_VCEDPM_Enable : PPSMC_MSG_VCEDPM_Disable);
+}
+
+static int kv_enable_samu_dpm(struct amdgpu_device *adev, bool enable)
+{
+	return amdgpu_kv_notify_message_to_smu(adev, enable ?
+					PPSMC_MSG_SAMUDPM_Enable : PPSMC_MSG_SAMUDPM_Disable);
+}
+
+static int kv_enable_acp_dpm(struct amdgpu_device *adev, bool enable)
+{
+	return amdgpu_kv_notify_message_to_smu(adev, enable ?
+					PPSMC_MSG_ACPDPM_Enable : PPSMC_MSG_ACPDPM_Disable);
+}
+
+static int kv_update_uvd_dpm(struct amdgpu_device *adev, bool gate)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_uvd_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table;
+	int ret;
+	u32 mask;
+
+	if (!gate) {
+		if (table->count)
+			pi->uvd_boot_level = table->count - 1;
+		else
+			pi->uvd_boot_level = 0;
+
+		if (!pi->caps_uvd_dpm || pi->caps_stable_p_state) {
+			mask = 1 << pi->uvd_boot_level;
+		} else {
+			mask = 0x1f;
+		}
+
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, UvdBootLevel),
+					   (uint8_t *)&pi->uvd_boot_level,
+					   sizeof(u8), pi->sram_end);
+		if (ret)
+			return ret;
+
+		amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+						  PPSMC_MSG_UVDDPM_SetEnabledMask,
+						  mask);
+	}
+
+	return kv_enable_uvd_dpm(adev, !gate);
+}
+
+static u8 kv_get_vce_boot_level(struct amdgpu_device *adev, u32 evclk)
+{
+	u8 i;
+	struct amdgpu_vce_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+
+	for (i = 0; i < table->count; i++) {
+		if (table->entries[i].evclk >= evclk)
+			break;
+	}
+
+	return i;
+}
+
+static int kv_update_vce_dpm(struct amdgpu_device *adev,
+			     struct amdgpu_ps *amdgpu_new_state,
+			     struct amdgpu_ps *amdgpu_current_state)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_vce_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+	int ret;
+
+	if (amdgpu_new_state->evclk > 0 && amdgpu_current_state->evclk == 0) {
+		if (pi->caps_stable_p_state)
+			pi->vce_boot_level = table->count - 1;
+		else
+			pi->vce_boot_level = kv_get_vce_boot_level(adev, amdgpu_new_state->evclk);
+
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, VceBootLevel),
+					   (u8 *)&pi->vce_boot_level,
+					   sizeof(u8),
+					   pi->sram_end);
+		if (ret)
+			return ret;
+
+		if (pi->caps_stable_p_state)
+			amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+							  PPSMC_MSG_VCEDPM_SetEnabledMask,
+							  (1 << pi->vce_boot_level));
+		kv_enable_vce_dpm(adev, true);
+	} else if (amdgpu_new_state->evclk == 0 && amdgpu_current_state->evclk > 0) {
+		kv_enable_vce_dpm(adev, false);
+	}
+
+	return 0;
+}
+
+static int kv_update_samu_dpm(struct amdgpu_device *adev, bool gate)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table;
+	int ret;
+
+	if (!gate) {
+		if (pi->caps_stable_p_state)
+			pi->samu_boot_level = table->count - 1;
+		else
+			pi->samu_boot_level = 0;
+
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, SamuBootLevel),
+					   (u8 *)&pi->samu_boot_level,
+					   sizeof(u8),
+					   pi->sram_end);
+		if (ret)
+			return ret;
+
+		if (pi->caps_stable_p_state)
+			amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+							  PPSMC_MSG_SAMUDPM_SetEnabledMask,
+							  (1 << pi->samu_boot_level));
+	}
+
+	return kv_enable_samu_dpm(adev, !gate);
+}
+
+static u8 kv_get_acp_boot_level(struct amdgpu_device *adev)
+{
+	u8 i;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table;
+
+	for (i = 0; i < table->count; i++) {
+		if (table->entries[i].clk >= 0) /* XXX */
+			break;
+	}
+
+	if (i >= table->count)
+		i = table->count - 1;
+
+	return i;
+}
+
+static void kv_update_acp_boot_level(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u8 acp_boot_level;
+
+	if (!pi->caps_stable_p_state) {
+		acp_boot_level = kv_get_acp_boot_level(adev);
+		if (acp_boot_level != pi->acp_boot_level) {
+			pi->acp_boot_level = acp_boot_level;
+			amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+							  PPSMC_MSG_ACPDPM_SetEnabledMask,
+							  (1 << pi->acp_boot_level));
+		}
+	}
+}
+
+static int kv_update_acp_dpm(struct amdgpu_device *adev, bool gate)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table;
+	int ret;
+
+	if (!gate) {
+		if (pi->caps_stable_p_state)
+			pi->acp_boot_level = table->count - 1;
+		else
+			pi->acp_boot_level = kv_get_acp_boot_level(adev);
+
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, AcpBootLevel),
+					   (u8 *)&pi->acp_boot_level,
+					   sizeof(u8),
+					   pi->sram_end);
+		if (ret)
+			return ret;
+
+		if (pi->caps_stable_p_state)
+			amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+							  PPSMC_MSG_ACPDPM_SetEnabledMask,
+							  (1 << pi->acp_boot_level));
+	}
+
+	return kv_enable_acp_dpm(adev, !gate);
+}
+
+static void kv_dpm_powergate_uvd(void *handle, bool gate)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	pi->uvd_power_gated = gate;
+
+	if (gate) {
+		/* stop the UVD block */
+		ret = amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							     AMD_PG_STATE_GATE);
+		kv_update_uvd_dpm(adev, gate);
+		if (pi->caps_uvd_pg)
+			/* power off the UVD block */
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_UVDPowerOFF);
+	} else {
+		if (pi->caps_uvd_pg)
+			/* power on the UVD block */
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_UVDPowerON);
+			/* re-init the UVD block */
+		kv_update_uvd_dpm(adev, gate);
+
+		ret = amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							     AMD_PG_STATE_UNGATE);
+	}
+}
+
+static void kv_dpm_powergate_vce(void *handle, bool gate)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	pi->vce_power_gated = gate;
+
+	if (gate) {
+		/* stop the VCE block */
+		ret = amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							     AMD_PG_STATE_GATE);
+		kv_enable_vce_dpm(adev, false);
+		if (pi->caps_vce_pg) /* power off the VCE block */
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_VCEPowerOFF);
+	} else {
+		if (pi->caps_vce_pg) /* power on the VCE block */
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_VCEPowerON);
+		kv_enable_vce_dpm(adev, true);
+		/* re-init the VCE block */
+		ret = amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							     AMD_PG_STATE_UNGATE);
+	}
+}
+
+
+static void kv_dpm_powergate_samu(struct amdgpu_device *adev, bool gate)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	if (pi->samu_power_gated == gate)
+		return;
+
+	pi->samu_power_gated = gate;
+
+	if (gate) {
+		kv_update_samu_dpm(adev, true);
+		if (pi->caps_samu_pg)
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_SAMPowerOFF);
+	} else {
+		if (pi->caps_samu_pg)
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_SAMPowerON);
+		kv_update_samu_dpm(adev, false);
+	}
+}
+
+static void kv_dpm_powergate_acp(struct amdgpu_device *adev, bool gate)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	if (pi->acp_power_gated == gate)
+		return;
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS)
+		return;
+
+	pi->acp_power_gated = gate;
+
+	if (gate) {
+		kv_update_acp_dpm(adev, true);
+		if (pi->caps_acp_pg)
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_ACPPowerOFF);
+	} else {
+		if (pi->caps_acp_pg)
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_ACPPowerON);
+		kv_update_acp_dpm(adev, false);
+	}
+}
+
+static void kv_set_valid_clock_range(struct amdgpu_device *adev,
+				     struct amdgpu_ps *new_rps)
+{
+	struct kv_ps *new_ps = kv_get_ps(new_rps);
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+
+	if (table && table->count) {
+		for (i = 0; i < pi->graphics_dpm_level_count; i++) {
+			if ((table->entries[i].clk >= new_ps->levels[0].sclk) ||
+			    (i == (pi->graphics_dpm_level_count - 1))) {
+				pi->lowest_valid = i;
+				break;
+			}
+		}
+
+		for (i = pi->graphics_dpm_level_count - 1; i > 0; i--) {
+			if (table->entries[i].clk <= new_ps->levels[new_ps->num_levels - 1].sclk)
+				break;
+		}
+		pi->highest_valid = i;
+
+		if (pi->lowest_valid > pi->highest_valid) {
+			if ((new_ps->levels[0].sclk - table->entries[pi->highest_valid].clk) >
+			    (table->entries[pi->lowest_valid].clk - new_ps->levels[new_ps->num_levels - 1].sclk))
+				pi->highest_valid = pi->lowest_valid;
+			else
+				pi->lowest_valid =  pi->highest_valid;
+		}
+	} else {
+		struct sumo_sclk_voltage_mapping_table *table =
+			&pi->sys_info.sclk_voltage_mapping_table;
+
+		for (i = 0; i < (int)pi->graphics_dpm_level_count; i++) {
+			if (table->entries[i].sclk_frequency >= new_ps->levels[0].sclk ||
+			    i == (int)(pi->graphics_dpm_level_count - 1)) {
+				pi->lowest_valid = i;
+				break;
+			}
+		}
+
+		for (i = pi->graphics_dpm_level_count - 1; i > 0; i--) {
+			if (table->entries[i].sclk_frequency <=
+			    new_ps->levels[new_ps->num_levels - 1].sclk)
+				break;
+		}
+		pi->highest_valid = i;
+
+		if (pi->lowest_valid > pi->highest_valid) {
+			if ((new_ps->levels[0].sclk -
+			     table->entries[pi->highest_valid].sclk_frequency) >
+			    (table->entries[pi->lowest_valid].sclk_frequency -
+			     new_ps->levels[new_ps->num_levels -1].sclk))
+				pi->highest_valid = pi->lowest_valid;
+			else
+				pi->lowest_valid =  pi->highest_valid;
+		}
+	}
+}
+
+static int kv_update_dfs_bypass_settings(struct amdgpu_device *adev,
+					 struct amdgpu_ps *new_rps)
+{
+	struct kv_ps *new_ps = kv_get_ps(new_rps);
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret = 0;
+	u8 clk_bypass_cntl;
+
+	if (pi->caps_enable_dfs_bypass) {
+		clk_bypass_cntl = new_ps->need_dfs_bypass ?
+			pi->graphics_level[pi->graphics_boot_level].ClkBypassCntl : 0;
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   (pi->dpm_table_start +
+					    offsetof(SMU7_Fusion_DpmTable, GraphicsLevel) +
+					    (pi->graphics_boot_level * sizeof(SMU7_Fusion_GraphicsLevel)) +
+					    offsetof(SMU7_Fusion_GraphicsLevel, ClkBypassCntl)),
+					   &clk_bypass_cntl,
+					   sizeof(u8), pi->sram_end);
+	}
+
+	return ret;
+}
+
+static int kv_enable_nb_dpm(struct amdgpu_device *adev,
+			    bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret = 0;
+
+	if (enable) {
+		if (pi->enable_nb_dpm && !pi->nb_dpm_enabled) {
+			ret = amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_NBDPM_Enable);
+			if (ret == 0)
+				pi->nb_dpm_enabled = true;
+		}
+	} else {
+		if (pi->enable_nb_dpm && pi->nb_dpm_enabled) {
+			ret = amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_NBDPM_Disable);
+			if (ret == 0)
+				pi->nb_dpm_enabled = false;
+		}
+	}
+
+	return ret;
+}
+
+static int kv_dpm_force_performance_level(void *handle,
+					  enum amd_dpm_forced_level level)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (level == AMD_DPM_FORCED_LEVEL_HIGH) {
+		ret = kv_force_dpm_highest(adev);
+		if (ret)
+			return ret;
+	} else if (level == AMD_DPM_FORCED_LEVEL_LOW) {
+		ret = kv_force_dpm_lowest(adev);
+		if (ret)
+			return ret;
+	} else if (level == AMD_DPM_FORCED_LEVEL_AUTO) {
+		ret = kv_unforce_levels(adev);
+		if (ret)
+			return ret;
+	}
+
+	adev->pm.dpm.forced_level = level;
+
+	return 0;
+}
+
+static int kv_dpm_pre_set_power_state(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_ps requested_ps = *adev->pm.dpm.requested_ps;
+	struct amdgpu_ps *new_ps = &requested_ps;
+
+	kv_update_requested_ps(adev, new_ps);
+
+	kv_apply_state_adjust_rules(adev,
+				    &pi->requested_rps,
+				    &pi->current_rps);
+
+	return 0;
+}
+
+static int kv_dpm_set_power_state(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_ps *new_ps = &pi->requested_rps;
+	struct amdgpu_ps *old_ps = &pi->current_rps;
+	int ret;
+
+	if (pi->bapm_enable) {
+		ret = amdgpu_kv_smc_bapm_enable(adev, adev->pm.ac_power);
+		if (ret) {
+			DRM_ERROR("amdgpu_kv_smc_bapm_enable failed\n");
+			return ret;
+		}
+	}
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS) {
+		if (pi->enable_dpm) {
+			kv_set_valid_clock_range(adev, new_ps);
+			kv_update_dfs_bypass_settings(adev, new_ps);
+			ret = kv_calculate_ds_divider(adev);
+			if (ret) {
+				DRM_ERROR("kv_calculate_ds_divider failed\n");
+				return ret;
+			}
+			kv_calculate_nbps_level_settings(adev);
+			kv_calculate_dpm_settings(adev);
+			kv_force_lowest_valid(adev);
+			kv_enable_new_levels(adev);
+			kv_upload_dpm_settings(adev);
+			kv_program_nbps_index_settings(adev, new_ps);
+			kv_unforce_levels(adev);
+			kv_set_enabled_levels(adev);
+			kv_force_lowest_valid(adev);
+			kv_unforce_levels(adev);
+
+			ret = kv_update_vce_dpm(adev, new_ps, old_ps);
+			if (ret) {
+				DRM_ERROR("kv_update_vce_dpm failed\n");
+				return ret;
+			}
+			kv_update_sclk_t(adev);
+			if (adev->asic_type == CHIP_MULLINS)
+				kv_enable_nb_dpm(adev, true);
+		}
+	} else {
+		if (pi->enable_dpm) {
+			kv_set_valid_clock_range(adev, new_ps);
+			kv_update_dfs_bypass_settings(adev, new_ps);
+			ret = kv_calculate_ds_divider(adev);
+			if (ret) {
+				DRM_ERROR("kv_calculate_ds_divider failed\n");
+				return ret;
+			}
+			kv_calculate_nbps_level_settings(adev);
+			kv_calculate_dpm_settings(adev);
+			kv_freeze_sclk_dpm(adev, true);
+			kv_upload_dpm_settings(adev);
+			kv_program_nbps_index_settings(adev, new_ps);
+			kv_freeze_sclk_dpm(adev, false);
+			kv_set_enabled_levels(adev);
+			ret = kv_update_vce_dpm(adev, new_ps, old_ps);
+			if (ret) {
+				DRM_ERROR("kv_update_vce_dpm failed\n");
+				return ret;
+			}
+			kv_update_acp_boot_level(adev);
+			kv_update_sclk_t(adev);
+			kv_enable_nb_dpm(adev, true);
+		}
+	}
+
+	return 0;
+}
+
+static void kv_dpm_post_set_power_state(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_ps *new_ps = &pi->requested_rps;
+
+	kv_update_current_ps(adev, new_ps);
+}
+
+static void kv_dpm_setup_asic(struct amdgpu_device *adev)
+{
+	sumo_take_smu_control(adev, true);
+	kv_init_powergate_state(adev);
+	kv_init_sclk_t(adev);
+}
+
+#if 0
+static void kv_dpm_reset_asic(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS) {
+		kv_force_lowest_valid(adev);
+		kv_init_graphics_levels(adev);
+		kv_program_bootup_state(adev);
+		kv_upload_dpm_settings(adev);
+		kv_force_lowest_valid(adev);
+		kv_unforce_levels(adev);
+	} else {
+		kv_init_graphics_levels(adev);
+		kv_program_bootup_state(adev);
+		kv_freeze_sclk_dpm(adev, true);
+		kv_upload_dpm_settings(adev);
+		kv_freeze_sclk_dpm(adev, false);
+		kv_set_enabled_level(adev, pi->graphics_boot_level);
+	}
+}
+#endif
+
+static void kv_construct_max_power_limits_table(struct amdgpu_device *adev,
+						struct amdgpu_clock_and_voltage_limits *table)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	if (pi->sys_info.sclk_voltage_mapping_table.num_max_dpm_entries > 0) {
+		int idx = pi->sys_info.sclk_voltage_mapping_table.num_max_dpm_entries - 1;
+		table->sclk =
+			pi->sys_info.sclk_voltage_mapping_table.entries[idx].sclk_frequency;
+		table->vddc =
+			kv_convert_2bit_index_to_voltage(adev,
+							 pi->sys_info.sclk_voltage_mapping_table.entries[idx].vid_2bit);
+	}
+
+	table->mclk = pi->sys_info.nbp_memory_clock[0];
+}
+
+static void kv_patch_voltage_values(struct amdgpu_device *adev)
+{
+	int i;
+	struct amdgpu_uvd_clock_voltage_dependency_table *uvd_table =
+		&adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table;
+	struct amdgpu_vce_clock_voltage_dependency_table *vce_table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+	struct amdgpu_clock_voltage_dependency_table *samu_table =
+		&adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table;
+	struct amdgpu_clock_voltage_dependency_table *acp_table =
+		&adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table;
+
+	if (uvd_table->count) {
+		for (i = 0; i < uvd_table->count; i++)
+			uvd_table->entries[i].v =
+				kv_convert_8bit_index_to_voltage(adev,
+								 uvd_table->entries[i].v);
+	}
+
+	if (vce_table->count) {
+		for (i = 0; i < vce_table->count; i++)
+			vce_table->entries[i].v =
+				kv_convert_8bit_index_to_voltage(adev,
+								 vce_table->entries[i].v);
+	}
+
+	if (samu_table->count) {
+		for (i = 0; i < samu_table->count; i++)
+			samu_table->entries[i].v =
+				kv_convert_8bit_index_to_voltage(adev,
+								 samu_table->entries[i].v);
+	}
+
+	if (acp_table->count) {
+		for (i = 0; i < acp_table->count; i++)
+			acp_table->entries[i].v =
+				kv_convert_8bit_index_to_voltage(adev,
+								 acp_table->entries[i].v);
+	}
+
+}
+
+static void kv_construct_boot_state(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->boot_pl.sclk = pi->sys_info.bootup_sclk;
+	pi->boot_pl.vddc_index = pi->sys_info.bootup_nb_voltage_index;
+	pi->boot_pl.ds_divider_index = 0;
+	pi->boot_pl.ss_divider_index = 0;
+	pi->boot_pl.allow_gnb_slow = 1;
+	pi->boot_pl.force_nbp_state = 0;
+	pi->boot_pl.display_wm = 0;
+	pi->boot_pl.vce_wm = 0;
+}
+
+static int kv_force_dpm_highest(struct amdgpu_device *adev)
+{
+	int ret;
+	u32 enable_mask, i;
+
+	ret = amdgpu_kv_dpm_get_enable_mask(adev, &enable_mask);
+	if (ret)
+		return ret;
+
+	for (i = SMU7_MAX_LEVELS_GRAPHICS - 1; i > 0; i--) {
+		if (enable_mask & (1 << i))
+			break;
+	}
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS)
+		return amdgpu_kv_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_DPM_ForceState, i);
+	else
+		return kv_set_enabled_level(adev, i);
+}
+
+static int kv_force_dpm_lowest(struct amdgpu_device *adev)
+{
+	int ret;
+	u32 enable_mask, i;
+
+	ret = amdgpu_kv_dpm_get_enable_mask(adev, &enable_mask);
+	if (ret)
+		return ret;
+
+	for (i = 0; i < SMU7_MAX_LEVELS_GRAPHICS; i++) {
+		if (enable_mask & (1 << i))
+			break;
+	}
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS)
+		return amdgpu_kv_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_DPM_ForceState, i);
+	else
+		return kv_set_enabled_level(adev, i);
+}
+
+static u8 kv_get_sleep_divider_id_from_clock(struct amdgpu_device *adev,
+					     u32 sclk, u32 min_sclk_in_sr)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+	u32 temp;
+	u32 min = max(min_sclk_in_sr, (u32)KV_MINIMUM_ENGINE_CLOCK);
+
+	if (sclk < min)
+		return 0;
+
+	if (!pi->caps_sclk_ds)
+		return 0;
+
+	for (i = KV_MAX_DEEPSLEEP_DIVIDER_ID; i > 0; i--) {
+		temp = sclk >> i;
+		if (temp >= min)
+			break;
+	}
+
+	return (u8)i;
+}
+
+static int kv_get_high_voltage_limit(struct amdgpu_device *adev, int *limit)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	int i;
+
+	if (table && table->count) {
+		for (i = table->count - 1; i >= 0; i--) {
+			if (pi->high_voltage_t &&
+			    (kv_convert_8bit_index_to_voltage(adev, table->entries[i].v) <=
+			     pi->high_voltage_t)) {
+				*limit = i;
+				return 0;
+			}
+		}
+	} else {
+		struct sumo_sclk_voltage_mapping_table *table =
+			&pi->sys_info.sclk_voltage_mapping_table;
+
+		for (i = table->num_max_dpm_entries - 1; i >= 0; i--) {
+			if (pi->high_voltage_t &&
+			    (kv_convert_2bit_index_to_voltage(adev, table->entries[i].vid_2bit) <=
+			     pi->high_voltage_t)) {
+				*limit = i;
+				return 0;
+			}
+		}
+	}
+
+	*limit = 0;
+	return 0;
+}
+
+static void kv_apply_state_adjust_rules(struct amdgpu_device *adev,
+					struct amdgpu_ps *new_rps,
+					struct amdgpu_ps *old_rps)
+{
+	struct kv_ps *ps = kv_get_ps(new_rps);
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 min_sclk = 10000; /* ??? */
+	u32 sclk, mclk = 0;
+	int i, limit;
+	bool force_high;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	u32 stable_p_state_sclk = 0;
+	struct amdgpu_clock_and_voltage_limits *max_limits =
+		&adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+
+	if (new_rps->vce_active) {
+		new_rps->evclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].evclk;
+		new_rps->ecclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].ecclk;
+	} else {
+		new_rps->evclk = 0;
+		new_rps->ecclk = 0;
+	}
+
+	mclk = max_limits->mclk;
+	sclk = min_sclk;
+
+	if (pi->caps_stable_p_state) {
+		stable_p_state_sclk = (max_limits->sclk * 75) / 100;
+
+		for (i = table->count - 1; i >= 0; i--) {
+			if (stable_p_state_sclk >= table->entries[i].clk) {
+				stable_p_state_sclk = table->entries[i].clk;
+				break;
+			}
+		}
+
+		if (i > 0)
+			stable_p_state_sclk = table->entries[0].clk;
+
+		sclk = stable_p_state_sclk;
+	}
+
+	if (new_rps->vce_active) {
+		if (sclk < adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].sclk)
+			sclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].sclk;
+	}
+
+	ps->need_dfs_bypass = true;
+
+	for (i = 0; i < ps->num_levels; i++) {
+		if (ps->levels[i].sclk < sclk)
+			ps->levels[i].sclk = sclk;
+	}
+
+	if (table && table->count) {
+		for (i = 0; i < ps->num_levels; i++) {
+			if (pi->high_voltage_t &&
+			    (pi->high_voltage_t <
+			     kv_convert_8bit_index_to_voltage(adev, ps->levels[i].vddc_index))) {
+				kv_get_high_voltage_limit(adev, &limit);
+				ps->levels[i].sclk = table->entries[limit].clk;
+			}
+		}
+	} else {
+		struct sumo_sclk_voltage_mapping_table *table =
+			&pi->sys_info.sclk_voltage_mapping_table;
+
+		for (i = 0; i < ps->num_levels; i++) {
+			if (pi->high_voltage_t &&
+			    (pi->high_voltage_t <
+			     kv_convert_8bit_index_to_voltage(adev, ps->levels[i].vddc_index))) {
+				kv_get_high_voltage_limit(adev, &limit);
+				ps->levels[i].sclk = table->entries[limit].sclk_frequency;
+			}
+		}
+	}
+
+	if (pi->caps_stable_p_state) {
+		for (i = 0; i < ps->num_levels; i++) {
+			ps->levels[i].sclk = stable_p_state_sclk;
+		}
+	}
+
+	pi->video_start = new_rps->dclk || new_rps->vclk ||
+		new_rps->evclk || new_rps->ecclk;
+
+	if ((new_rps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) ==
+	    ATOM_PPLIB_CLASSIFICATION_UI_BATTERY)
+		pi->battery_state = true;
+	else
+		pi->battery_state = false;
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS) {
+		ps->dpm0_pg_nb_ps_lo = 0x1;
+		ps->dpm0_pg_nb_ps_hi = 0x0;
+		ps->dpmx_nb_ps_lo = 0x1;
+		ps->dpmx_nb_ps_hi = 0x0;
+	} else {
+		ps->dpm0_pg_nb_ps_lo = 0x3;
+		ps->dpm0_pg_nb_ps_hi = 0x0;
+		ps->dpmx_nb_ps_lo = 0x3;
+		ps->dpmx_nb_ps_hi = 0x0;
+
+		if (pi->sys_info.nb_dpm_enable) {
+			force_high = (mclk >= pi->sys_info.nbp_memory_clock[3]) ||
+				pi->video_start || (adev->pm.dpm.new_active_crtc_count >= 3) ||
+				pi->disable_nb_ps3_in_battery;
+			ps->dpm0_pg_nb_ps_lo = force_high ? 0x2 : 0x3;
+			ps->dpm0_pg_nb_ps_hi = 0x2;
+			ps->dpmx_nb_ps_lo = force_high ? 0x2 : 0x3;
+			ps->dpmx_nb_ps_hi = 0x2;
+		}
+	}
+}
+
+static void kv_dpm_power_level_enabled_for_throttle(struct amdgpu_device *adev,
+						    u32 index, bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->graphics_level[index].EnabledForThrottle = enable ? 1 : 0;
+}
+
+static int kv_calculate_ds_divider(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 sclk_in_sr = 10000; /* ??? */
+	u32 i;
+
+	if (pi->lowest_valid > pi->highest_valid)
+		return -EINVAL;
+
+	for (i = pi->lowest_valid; i <= pi->highest_valid; i++) {
+		pi->graphics_level[i].DeepSleepDivId =
+			kv_get_sleep_divider_id_from_clock(adev,
+							   be32_to_cpu(pi->graphics_level[i].SclkFrequency),
+							   sclk_in_sr);
+	}
+	return 0;
+}
+
+static int kv_calculate_nbps_level_settings(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+	bool force_high;
+	struct amdgpu_clock_and_voltage_limits *max_limits =
+		&adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	u32 mclk = max_limits->mclk;
+
+	if (pi->lowest_valid > pi->highest_valid)
+		return -EINVAL;
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS) {
+		for (i = pi->lowest_valid; i <= pi->highest_valid; i++) {
+			pi->graphics_level[i].GnbSlow = 1;
+			pi->graphics_level[i].ForceNbPs1 = 0;
+			pi->graphics_level[i].UpH = 0;
+		}
+
+		if (!pi->sys_info.nb_dpm_enable)
+			return 0;
+
+		force_high = ((mclk >= pi->sys_info.nbp_memory_clock[3]) ||
+			      (adev->pm.dpm.new_active_crtc_count >= 3) || pi->video_start);
+
+		if (force_high) {
+			for (i = pi->lowest_valid; i <= pi->highest_valid; i++)
+				pi->graphics_level[i].GnbSlow = 0;
+		} else {
+			if (pi->battery_state)
+				pi->graphics_level[0].ForceNbPs1 = 1;
+
+			pi->graphics_level[1].GnbSlow = 0;
+			pi->graphics_level[2].GnbSlow = 0;
+			pi->graphics_level[3].GnbSlow = 0;
+			pi->graphics_level[4].GnbSlow = 0;
+		}
+	} else {
+		for (i = pi->lowest_valid; i <= pi->highest_valid; i++) {
+			pi->graphics_level[i].GnbSlow = 1;
+			pi->graphics_level[i].ForceNbPs1 = 0;
+			pi->graphics_level[i].UpH = 0;
+		}
+
+		if (pi->sys_info.nb_dpm_enable && pi->battery_state) {
+			pi->graphics_level[pi->lowest_valid].UpH = 0x28;
+			pi->graphics_level[pi->lowest_valid].GnbSlow = 0;
+			if (pi->lowest_valid != pi->highest_valid)
+				pi->graphics_level[pi->lowest_valid].ForceNbPs1 = 1;
+		}
+	}
+	return 0;
+}
+
+static int kv_calculate_dpm_settings(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+
+	if (pi->lowest_valid > pi->highest_valid)
+		return -EINVAL;
+
+	for (i = pi->lowest_valid; i <= pi->highest_valid; i++)
+		pi->graphics_level[i].DisplayWatermark = (i == pi->highest_valid) ? 1 : 0;
+
+	return 0;
+}
+
+static void kv_init_graphics_levels(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+
+	if (table && table->count) {
+		u32 vid_2bit;
+
+		pi->graphics_dpm_level_count = 0;
+		for (i = 0; i < table->count; i++) {
+			if (pi->high_voltage_t &&
+			    (pi->high_voltage_t <
+			     kv_convert_8bit_index_to_voltage(adev, table->entries[i].v)))
+				break;
+
+			kv_set_divider_value(adev, i, table->entries[i].clk);
+			vid_2bit = kv_convert_vid7_to_vid2(adev,
+							   &pi->sys_info.vid_mapping_table,
+							   table->entries[i].v);
+			kv_set_vid(adev, i, vid_2bit);
+			kv_set_at(adev, i, pi->at[i]);
+			kv_dpm_power_level_enabled_for_throttle(adev, i, true);
+			pi->graphics_dpm_level_count++;
+		}
+	} else {
+		struct sumo_sclk_voltage_mapping_table *table =
+			&pi->sys_info.sclk_voltage_mapping_table;
+
+		pi->graphics_dpm_level_count = 0;
+		for (i = 0; i < table->num_max_dpm_entries; i++) {
+			if (pi->high_voltage_t &&
+			    pi->high_voltage_t <
+			    kv_convert_2bit_index_to_voltage(adev, table->entries[i].vid_2bit))
+				break;
+
+			kv_set_divider_value(adev, i, table->entries[i].sclk_frequency);
+			kv_set_vid(adev, i, table->entries[i].vid_2bit);
+			kv_set_at(adev, i, pi->at[i]);
+			kv_dpm_power_level_enabled_for_throttle(adev, i, true);
+			pi->graphics_dpm_level_count++;
+		}
+	}
+
+	for (i = 0; i < SMU7_MAX_LEVELS_GRAPHICS; i++)
+		kv_dpm_power_level_enable(adev, i, false);
+}
+
+static void kv_enable_new_levels(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+
+	for (i = 0; i < SMU7_MAX_LEVELS_GRAPHICS; i++) {
+		if (i >= pi->lowest_valid && i <= pi->highest_valid)
+			kv_dpm_power_level_enable(adev, i, true);
+	}
+}
+
+static int kv_set_enabled_level(struct amdgpu_device *adev, u32 level)
+{
+	u32 new_mask = (1 << level);
+
+	return amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+						 PPSMC_MSG_SCLKDPM_SetEnabledMask,
+						 new_mask);
+}
+
+static int kv_set_enabled_levels(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i, new_mask = 0;
+
+	for (i = pi->lowest_valid; i <= pi->highest_valid; i++)
+		new_mask |= (1 << i);
+
+	return amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+						 PPSMC_MSG_SCLKDPM_SetEnabledMask,
+						 new_mask);
+}
+
+static void kv_program_nbps_index_settings(struct amdgpu_device *adev,
+					   struct amdgpu_ps *new_rps)
+{
+	struct kv_ps *new_ps = kv_get_ps(new_rps);
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 nbdpmconfig1;
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS)
+		return;
+
+	if (pi->sys_info.nb_dpm_enable) {
+		nbdpmconfig1 = RREG32_SMC(ixNB_DPM_CONFIG_1);
+		nbdpmconfig1 &= ~(NB_DPM_CONFIG_1__Dpm0PgNbPsLo_MASK |
+				NB_DPM_CONFIG_1__Dpm0PgNbPsHi_MASK |
+				NB_DPM_CONFIG_1__DpmXNbPsLo_MASK |
+				NB_DPM_CONFIG_1__DpmXNbPsHi_MASK);
+		nbdpmconfig1 |= (new_ps->dpm0_pg_nb_ps_lo << NB_DPM_CONFIG_1__Dpm0PgNbPsLo__SHIFT) |
+				(new_ps->dpm0_pg_nb_ps_hi << NB_DPM_CONFIG_1__Dpm0PgNbPsHi__SHIFT) |
+				(new_ps->dpmx_nb_ps_lo << NB_DPM_CONFIG_1__DpmXNbPsLo__SHIFT) |
+				(new_ps->dpmx_nb_ps_hi << NB_DPM_CONFIG_1__DpmXNbPsHi__SHIFT);
+		WREG32_SMC(ixNB_DPM_CONFIG_1, nbdpmconfig1);
+	}
+}
+
+static int kv_set_thermal_temperature_range(struct amdgpu_device *adev,
+					    int min_temp, int max_temp)
+{
+	int low_temp = 0 * 1000;
+	int high_temp = 255 * 1000;
+	u32 tmp;
+
+	if (low_temp < min_temp)
+		low_temp = min_temp;
+	if (high_temp > max_temp)
+		high_temp = max_temp;
+	if (high_temp < low_temp) {
+		DRM_ERROR("invalid thermal range: %d - %d\n", low_temp, high_temp);
+		return -EINVAL;
+	}
+
+	tmp = RREG32_SMC(ixCG_THERMAL_INT_CTRL);
+	tmp &= ~(CG_THERMAL_INT_CTRL__DIG_THERM_INTH_MASK |
+		CG_THERMAL_INT_CTRL__DIG_THERM_INTL_MASK);
+	tmp |= ((49 + (high_temp / 1000)) << CG_THERMAL_INT_CTRL__DIG_THERM_INTH__SHIFT) |
+		((49 + (low_temp / 1000)) << CG_THERMAL_INT_CTRL__DIG_THERM_INTL__SHIFT);
+	WREG32_SMC(ixCG_THERMAL_INT_CTRL, tmp);
+
+	adev->pm.dpm.thermal.min_temp = low_temp;
+	adev->pm.dpm.thermal.max_temp = high_temp;
+
+	return 0;
+}
+
+union igp_info {
+	struct _ATOM_INTEGRATED_SYSTEM_INFO info;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V2 info_2;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V5 info_5;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V6 info_6;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_7 info_7;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_8 info_8;
+};
+
+static int kv_parse_sys_info_table(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, IntegratedSystemInfo);
+	union igp_info *igp_info;
+	u8 frev, crev;
+	u16 data_offset;
+	int i;
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		igp_info = (union igp_info *)(mode_info->atom_context->bios +
+					      data_offset);
+
+		if (crev != 8) {
+			DRM_ERROR("Unsupported IGP table: %d %d\n", frev, crev);
+			return -EINVAL;
+		}
+		pi->sys_info.bootup_sclk = le32_to_cpu(igp_info->info_8.ulBootUpEngineClock);
+		pi->sys_info.bootup_uma_clk = le32_to_cpu(igp_info->info_8.ulBootUpUMAClock);
+		pi->sys_info.bootup_nb_voltage_index =
+			le16_to_cpu(igp_info->info_8.usBootUpNBVoltage);
+		if (igp_info->info_8.ucHtcTmpLmt == 0)
+			pi->sys_info.htc_tmp_lmt = 203;
+		else
+			pi->sys_info.htc_tmp_lmt = igp_info->info_8.ucHtcTmpLmt;
+		if (igp_info->info_8.ucHtcHystLmt == 0)
+			pi->sys_info.htc_hyst_lmt = 5;
+		else
+			pi->sys_info.htc_hyst_lmt = igp_info->info_8.ucHtcHystLmt;
+		if (pi->sys_info.htc_tmp_lmt <= pi->sys_info.htc_hyst_lmt) {
+			DRM_ERROR("The htcTmpLmt should be larger than htcHystLmt.\n");
+		}
+
+		if (le32_to_cpu(igp_info->info_8.ulSystemConfig) & (1 << 3))
+			pi->sys_info.nb_dpm_enable = true;
+		else
+			pi->sys_info.nb_dpm_enable = false;
+
+		for (i = 0; i < KV_NUM_NBPSTATES; i++) {
+			pi->sys_info.nbp_memory_clock[i] =
+				le32_to_cpu(igp_info->info_8.ulNbpStateMemclkFreq[i]);
+			pi->sys_info.nbp_n_clock[i] =
+				le32_to_cpu(igp_info->info_8.ulNbpStateNClkFreq[i]);
+		}
+		if (le32_to_cpu(igp_info->info_8.ulGPUCapInfo) &
+		    SYS_INFO_GPUCAPS__ENABEL_DFS_BYPASS)
+			pi->caps_enable_dfs_bypass = true;
+
+		sumo_construct_sclk_voltage_mapping_table(adev,
+							  &pi->sys_info.sclk_voltage_mapping_table,
+							  igp_info->info_8.sAvail_SCLK);
+
+		sumo_construct_vid_mapping_table(adev,
+						 &pi->sys_info.vid_mapping_table,
+						 igp_info->info_8.sAvail_SCLK);
+
+		kv_construct_max_power_limits_table(adev,
+						    &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac);
+	}
+	return 0;
+}
+
+union power_info {
+	struct _ATOM_POWERPLAY_INFO info;
+	struct _ATOM_POWERPLAY_INFO_V2 info_2;
+	struct _ATOM_POWERPLAY_INFO_V3 info_3;
+	struct _ATOM_PPLIB_POWERPLAYTABLE pplib;
+	struct _ATOM_PPLIB_POWERPLAYTABLE2 pplib2;
+	struct _ATOM_PPLIB_POWERPLAYTABLE3 pplib3;
+};
+
+union pplib_clock_info {
+	struct _ATOM_PPLIB_R600_CLOCK_INFO r600;
+	struct _ATOM_PPLIB_RS780_CLOCK_INFO rs780;
+	struct _ATOM_PPLIB_EVERGREEN_CLOCK_INFO evergreen;
+	struct _ATOM_PPLIB_SUMO_CLOCK_INFO sumo;
+};
+
+union pplib_power_state {
+	struct _ATOM_PPLIB_STATE v1;
+	struct _ATOM_PPLIB_STATE_V2 v2;
+};
+
+static void kv_patch_boot_state(struct amdgpu_device *adev,
+				struct kv_ps *ps)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	ps->num_levels = 1;
+	ps->levels[0] = pi->boot_pl;
+}
+
+static void kv_parse_pplib_non_clock_info(struct amdgpu_device *adev,
+					  struct amdgpu_ps *rps,
+					  struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info,
+					  u8 table_rev)
+{
+	struct kv_ps *ps = kv_get_ps(rps);
+
+	rps->caps = le32_to_cpu(non_clock_info->ulCapsAndSettings);
+	rps->class = le16_to_cpu(non_clock_info->usClassification);
+	rps->class2 = le16_to_cpu(non_clock_info->usClassification2);
+
+	if (ATOM_PPLIB_NONCLOCKINFO_VER1 < table_rev) {
+		rps->vclk = le32_to_cpu(non_clock_info->ulVCLK);
+		rps->dclk = le32_to_cpu(non_clock_info->ulDCLK);
+	} else {
+		rps->vclk = 0;
+		rps->dclk = 0;
+	}
+
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_BOOT) {
+		adev->pm.dpm.boot_ps = rps;
+		kv_patch_boot_state(adev, ps);
+	}
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
+		adev->pm.dpm.uvd_ps = rps;
+}
+
+static void kv_parse_pplib_clock_info(struct amdgpu_device *adev,
+				      struct amdgpu_ps *rps, int index,
+					union pplib_clock_info *clock_info)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct kv_ps *ps = kv_get_ps(rps);
+	struct kv_pl *pl = &ps->levels[index];
+	u32 sclk;
+
+	sclk = le16_to_cpu(clock_info->sumo.usEngineClockLow);
+	sclk |= clock_info->sumo.ucEngineClockHigh << 16;
+	pl->sclk = sclk;
+	pl->vddc_index = clock_info->sumo.vddcIndex;
+
+	ps->num_levels = index + 1;
+
+	if (pi->caps_sclk_ds) {
+		pl->ds_divider_index = 5;
+		pl->ss_divider_index = 5;
+	}
+}
+
+static int kv_parse_power_table(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info;
+	union pplib_power_state *power_state;
+	int i, j, k, non_clock_array_index, clock_array_index;
+	union pplib_clock_info *clock_info;
+	struct _StateArray *state_array;
+	struct _ClockInfoArray *clock_info_array;
+	struct _NonClockInfoArray *non_clock_info_array;
+	union power_info *power_info;
+	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
+	u16 data_offset;
+	u8 frev, crev;
+	u8 *power_state_offset;
+	struct kv_ps *ps;
+
+	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset))
+		return -EINVAL;
+	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
+
+	amdgpu_add_thermal_controller(adev);
+
+	state_array = (struct _StateArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usStateArrayOffset));
+	clock_info_array = (struct _ClockInfoArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usClockInfoArrayOffset));
+	non_clock_info_array = (struct _NonClockInfoArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usNonClockInfoArrayOffset));
+
+	adev->pm.dpm.ps = kcalloc(state_array->ucNumEntries,
+				  sizeof(struct amdgpu_ps),
+				  GFP_KERNEL);
+	if (!adev->pm.dpm.ps)
+		return -ENOMEM;
+	power_state_offset = (u8 *)state_array->states;
+	for (i = 0; i < state_array->ucNumEntries; i++) {
+		u8 *idx;
+		power_state = (union pplib_power_state *)power_state_offset;
+		non_clock_array_index = power_state->v2.nonClockInfoIndex;
+		non_clock_info = (struct _ATOM_PPLIB_NONCLOCK_INFO *)
+			&non_clock_info_array->nonClockInfo[non_clock_array_index];
+		ps = kzalloc(sizeof(struct kv_ps), GFP_KERNEL);
+		if (ps == NULL) {
+			kfree(adev->pm.dpm.ps);
+			return -ENOMEM;
+		}
+		adev->pm.dpm.ps[i].ps_priv = ps;
+		k = 0;
+		idx = (u8 *)&power_state->v2.clockInfoIndex[0];
+		for (j = 0; j < power_state->v2.ucNumDPMLevels; j++) {
+			clock_array_index = idx[j];
+			if (clock_array_index >= clock_info_array->ucNumEntries)
+				continue;
+			if (k >= SUMO_MAX_HARDWARE_POWERLEVELS)
+				break;
+			clock_info = (union pplib_clock_info *)
+				((u8 *)&clock_info_array->clockInfo[0] +
+				 (clock_array_index * clock_info_array->ucEntrySize));
+			kv_parse_pplib_clock_info(adev,
+						  &adev->pm.dpm.ps[i], k,
+						  clock_info);
+			k++;
+		}
+		kv_parse_pplib_non_clock_info(adev, &adev->pm.dpm.ps[i],
+					      non_clock_info,
+					      non_clock_info_array->ucEntrySize);
+		power_state_offset += 2 + power_state->v2.ucNumDPMLevels;
+	}
+	adev->pm.dpm.num_ps = state_array->ucNumEntries;
+
+	/* fill in the vce power states */
+	for (i = 0; i < adev->pm.dpm.num_of_vce_states; i++) {
+		u32 sclk;
+		clock_array_index = adev->pm.dpm.vce_states[i].clk_idx;
+		clock_info = (union pplib_clock_info *)
+			&clock_info_array->clockInfo[clock_array_index * clock_info_array->ucEntrySize];
+		sclk = le16_to_cpu(clock_info->sumo.usEngineClockLow);
+		sclk |= clock_info->sumo.ucEngineClockHigh << 16;
+		adev->pm.dpm.vce_states[i].sclk = sclk;
+		adev->pm.dpm.vce_states[i].mclk = 0;
+	}
+
+	return 0;
+}
+
+static int kv_dpm_init(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi;
+	int ret, i;
+
+	pi = kzalloc(sizeof(struct kv_power_info), GFP_KERNEL);
+	if (pi == NULL)
+		return -ENOMEM;
+	adev->pm.dpm.priv = pi;
+
+	ret = amdgpu_get_platform_caps(adev);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_parse_extended_power_table(adev);
+	if (ret)
+		return ret;
+
+	for (i = 0; i < SUMO_MAX_HARDWARE_POWERLEVELS; i++)
+		pi->at[i] = TRINITY_AT_DFLT;
+
+	pi->sram_end = SMC_RAM_END;
+
+	pi->enable_nb_dpm = true;
+
+	pi->caps_power_containment = true;
+	pi->caps_cac = true;
+	pi->enable_didt = false;
+	if (pi->enable_didt) {
+		pi->caps_sq_ramping = true;
+		pi->caps_db_ramping = true;
+		pi->caps_td_ramping = true;
+		pi->caps_tcp_ramping = true;
+	}
+
+	if (adev->powerplay.pp_feature & PP_SCLK_DEEP_SLEEP_MASK)
+		pi->caps_sclk_ds = true;
+	else
+		pi->caps_sclk_ds = false;
+
+	pi->enable_auto_thermal_throttling = true;
+	pi->disable_nb_ps3_in_battery = false;
+	if (amdgpu_bapm == 0)
+		pi->bapm_enable = false;
+	else
+		pi->bapm_enable = true;
+	pi->voltage_drop_t = 0;
+	pi->caps_sclk_throttle_low_notification = false;
+	pi->caps_fps = false; /* true? */
+	pi->caps_uvd_pg = (adev->pg_flags & AMD_PG_SUPPORT_UVD) ? true : false;
+	pi->caps_uvd_dpm = true;
+	pi->caps_vce_pg = (adev->pg_flags & AMD_PG_SUPPORT_VCE) ? true : false;
+	pi->caps_samu_pg = (adev->pg_flags & AMD_PG_SUPPORT_SAMU) ? true : false;
+	pi->caps_acp_pg = (adev->pg_flags & AMD_PG_SUPPORT_ACP) ? true : false;
+	pi->caps_stable_p_state = false;
+
+	ret = kv_parse_sys_info_table(adev);
+	if (ret)
+		return ret;
+
+	kv_patch_voltage_values(adev);
+	kv_construct_boot_state(adev);
+
+	ret = kv_parse_power_table(adev);
+	if (ret)
+		return ret;
+
+	pi->enable_dpm = true;
+
+	return 0;
+}
+
+static void
+kv_dpm_debugfs_print_current_performance_level(void *handle,
+					       struct seq_file *m)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 current_index =
+		(RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
+		TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX_MASK) >>
+		TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX__SHIFT;
+	u32 sclk, tmp;
+	u16 vddc;
+
+	if (current_index >= SMU__NUM_SCLK_DPM_STATE) {
+		seq_printf(m, "invalid dpm profile %d\n", current_index);
+	} else {
+		sclk = be32_to_cpu(pi->graphics_level[current_index].SclkFrequency);
+		tmp = (RREG32_SMC(ixSMU_VOLTAGE_STATUS) &
+			SMU_VOLTAGE_STATUS__SMU_VOLTAGE_CURRENT_LEVEL_MASK) >>
+			SMU_VOLTAGE_STATUS__SMU_VOLTAGE_CURRENT_LEVEL__SHIFT;
+		vddc = kv_convert_8bit_index_to_voltage(adev, (u16)tmp);
+		seq_printf(m, "uvd    %sabled\n", pi->uvd_power_gated ? "dis" : "en");
+		seq_printf(m, "vce    %sabled\n", pi->vce_power_gated ? "dis" : "en");
+		seq_printf(m, "power level %d    sclk: %u vddc: %u\n",
+			   current_index, sclk, vddc);
+	}
+}
+
+static void
+kv_dpm_print_power_state(void *handle, void *request_ps)
+{
+	int i;
+	struct amdgpu_ps *rps = (struct amdgpu_ps *)request_ps;
+	struct kv_ps *ps = kv_get_ps(rps);
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_dpm_print_class_info(rps->class, rps->class2);
+	amdgpu_dpm_print_cap_info(rps->caps);
+	printk("\tuvd    vclk: %d dclk: %d\n", rps->vclk, rps->dclk);
+	for (i = 0; i < ps->num_levels; i++) {
+		struct kv_pl *pl = &ps->levels[i];
+		printk("\t\tpower level %d    sclk: %u vddc: %u\n",
+		       i, pl->sclk,
+		       kv_convert_8bit_index_to_voltage(adev, pl->vddc_index));
+	}
+	amdgpu_dpm_print_ps_status(adev, rps);
+}
+
+static void kv_dpm_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->pm.dpm.num_ps; i++) {
+		kfree(adev->pm.dpm.ps[i].ps_priv);
+	}
+	kfree(adev->pm.dpm.ps);
+	kfree(adev->pm.dpm.priv);
+	amdgpu_free_extended_power_table(adev);
+}
+
+static void kv_dpm_display_configuration_changed(void *handle)
+{
+
+}
+
+static u32 kv_dpm_get_sclk(void *handle, bool low)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct kv_ps *requested_state = kv_get_ps(&pi->requested_rps);
+
+	if (low)
+		return requested_state->levels[0].sclk;
+	else
+		return requested_state->levels[requested_state->num_levels - 1].sclk;
+}
+
+static u32 kv_dpm_get_mclk(void *handle, bool low)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	return pi->sys_info.bootup_uma_clk;
+}
+
+/* get temperature in millidegrees */
+static int kv_dpm_get_temp(void *handle)
+{
+	u32 temp;
+	int actual_temp = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	temp = RREG32_SMC(0xC0300E0C);
+
+	if (temp)
+		actual_temp = (temp / 8) - 49;
+	else
+		actual_temp = 0;
+
+	actual_temp = actual_temp * 1000;
+
+	return actual_temp;
+}
+
+static int kv_dpm_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->powerplay.pp_funcs = &kv_dpm_funcs;
+	adev->powerplay.pp_handle = adev;
+	kv_dpm_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int kv_dpm_late_init(void *handle)
+{
+	/* powerdown unused blocks for now */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->pm.dpm_enabled)
+		return 0;
+
+	kv_dpm_powergate_acp(adev, true);
+	kv_dpm_powergate_samu(adev, true);
+
+	return 0;
+}
+
+static int kv_dpm_sw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 230,
+				&adev->pm.dpm.thermal.irq);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 231,
+				&adev->pm.dpm.thermal.irq);
+	if (ret)
+		return ret;
+
+	/* default to balanced state */
+	adev->pm.dpm.state = POWER_STATE_TYPE_BALANCED;
+	adev->pm.dpm.user_state = POWER_STATE_TYPE_BALANCED;
+	adev->pm.dpm.forced_level = AMD_DPM_FORCED_LEVEL_AUTO;
+	adev->pm.default_sclk = adev->clock.default_sclk;
+	adev->pm.default_mclk = adev->clock.default_mclk;
+	adev->pm.current_sclk = adev->clock.default_sclk;
+	adev->pm.current_mclk = adev->clock.default_mclk;
+	adev->pm.int_thermal_type = THERMAL_TYPE_NONE;
+
+	if (amdgpu_dpm == 0)
+		return 0;
+
+	INIT_WORK(&adev->pm.dpm.thermal.work, amdgpu_dpm_thermal_work_handler);
+	mutex_lock(&adev->pm.mutex);
+	ret = kv_dpm_init(adev);
+	if (ret)
+		goto dpm_failed;
+	adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
+	if (amdgpu_dpm == 1)
+		amdgpu_pm_print_power_states(adev);
+	mutex_unlock(&adev->pm.mutex);
+	DRM_INFO("amdgpu: dpm initialized\n");
+
+	return 0;
+
+dpm_failed:
+	kv_dpm_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+	DRM_ERROR("amdgpu: dpm initialization failed\n");
+	return ret;
+}
+
+static int kv_dpm_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	flush_work(&adev->pm.dpm.thermal.work);
+
+	mutex_lock(&adev->pm.mutex);
+	kv_dpm_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+
+	return 0;
+}
+
+static int kv_dpm_hw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!amdgpu_dpm)
+		return 0;
+
+	mutex_lock(&adev->pm.mutex);
+	kv_dpm_setup_asic(adev);
+	ret = kv_dpm_enable(adev);
+	if (ret)
+		adev->pm.dpm_enabled = false;
+	else
+		adev->pm.dpm_enabled = true;
+	mutex_unlock(&adev->pm.mutex);
+	amdgpu_pm_compute_clocks(adev);
+	return ret;
+}
+
+static int kv_dpm_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		kv_dpm_disable(adev);
+		mutex_unlock(&adev->pm.mutex);
+	}
+
+	return 0;
+}
+
+static int kv_dpm_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		/* disable dpm */
+		kv_dpm_disable(adev);
+		/* reset the power state */
+		adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
+		mutex_unlock(&adev->pm.mutex);
+	}
+	return 0;
+}
+
+static int kv_dpm_resume(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		/* asic init will reset to the boot state */
+		mutex_lock(&adev->pm.mutex);
+		kv_dpm_setup_asic(adev);
+		ret = kv_dpm_enable(adev);
+		if (ret)
+			adev->pm.dpm_enabled = false;
+		else
+			adev->pm.dpm_enabled = true;
+		mutex_unlock(&adev->pm.mutex);
+		if (adev->pm.dpm_enabled)
+			amdgpu_pm_compute_clocks(adev);
+	}
+	return 0;
+}
+
+static bool kv_dpm_is_idle(void *handle)
+{
+	return true;
+}
+
+static int kv_dpm_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+
+static int kv_dpm_soft_reset(void *handle)
+{
+	return 0;
+}
+
+static int kv_dpm_set_interrupt_state(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *src,
+				      unsigned type,
+				      enum amdgpu_interrupt_state state)
+{
+	u32 cg_thermal_int;
+
+	switch (type) {
+	case AMDGPU_THERMAL_IRQ_LOW_TO_HIGH:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT_CTRL);
+			cg_thermal_int &= ~CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT_CTRL, cg_thermal_int);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT_CTRL);
+			cg_thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT_CTRL, cg_thermal_int);
+			break;
+		default:
+			break;
+		}
+		break;
+
+	case AMDGPU_THERMAL_IRQ_HIGH_TO_LOW:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT_CTRL);
+			cg_thermal_int &= ~CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT_CTRL, cg_thermal_int);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT_CTRL);
+			cg_thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT_CTRL, cg_thermal_int);
+			break;
+		default:
+			break;
+		}
+		break;
+
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int kv_dpm_process_interrupt(struct amdgpu_device *adev,
+				    struct amdgpu_irq_src *source,
+				    struct amdgpu_iv_entry *entry)
+{
+	bool queue_thermal = false;
+
+	if (entry == NULL)
+		return -EINVAL;
+
+	switch (entry->src_id) {
+	case 230: /* thermal low to high */
+		DRM_DEBUG("IH: thermal low to high\n");
+		adev->pm.dpm.thermal.high_to_low = false;
+		queue_thermal = true;
+		break;
+	case 231: /* thermal high to low */
+		DRM_DEBUG("IH: thermal high to low\n");
+		adev->pm.dpm.thermal.high_to_low = true;
+		queue_thermal = true;
+		break;
+	default:
+		break;
+	}
+
+	if (queue_thermal)
+		schedule_work(&adev->pm.dpm.thermal.work);
+
+	return 0;
+}
+
+static int kv_dpm_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int kv_dpm_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static inline bool kv_are_power_levels_equal(const struct kv_pl *kv_cpl1,
+						const struct kv_pl *kv_cpl2)
+{
+	return ((kv_cpl1->sclk == kv_cpl2->sclk) &&
+		  (kv_cpl1->vddc_index == kv_cpl2->vddc_index) &&
+		  (kv_cpl1->ds_divider_index == kv_cpl2->ds_divider_index) &&
+		  (kv_cpl1->force_nbp_state == kv_cpl2->force_nbp_state));
+}
+
+static int kv_check_state_equal(void *handle,
+				void *current_ps,
+				void *request_ps,
+				bool *equal)
+{
+	struct kv_ps *kv_cps;
+	struct kv_ps *kv_rps;
+	int i;
+	struct amdgpu_ps *cps = (struct amdgpu_ps *)current_ps;
+	struct amdgpu_ps *rps = (struct amdgpu_ps *)request_ps;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev == NULL || cps == NULL || rps == NULL || equal == NULL)
+		return -EINVAL;
+
+	kv_cps = kv_get_ps(cps);
+	kv_rps = kv_get_ps(rps);
+
+	if (kv_cps == NULL) {
+		*equal = false;
+		return 0;
+	}
+
+	if (kv_cps->num_levels != kv_rps->num_levels) {
+		*equal = false;
+		return 0;
+	}
+
+	for (i = 0; i < kv_cps->num_levels; i++) {
+		if (!kv_are_power_levels_equal(&(kv_cps->levels[i]),
+					&(kv_rps->levels[i]))) {
+			*equal = false;
+			return 0;
+		}
+	}
+
+	/* If all performance levels are the same try to use the UVD clocks to break the tie.*/
+	*equal = ((cps->vclk == rps->vclk) && (cps->dclk == rps->dclk));
+	*equal &= ((cps->evclk == rps->evclk) && (cps->ecclk == rps->ecclk));
+
+	return 0;
+}
+
+static int kv_dpm_read_sensor(void *handle, int idx,
+			      void *value, int *size)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct kv_power_info *pi = kv_get_pi(adev);
+	uint32_t sclk;
+	u32 pl_index =
+		(RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
+		TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX_MASK) >>
+		TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX__SHIFT;
+
+	/* size must be at least 4 bytes for all sensors */
+	if (*size < 4)
+		return -EINVAL;
+
+	switch (idx) {
+	case AMDGPU_PP_SENSOR_GFX_SCLK:
+		if (pl_index < SMU__NUM_SCLK_DPM_STATE) {
+			sclk = be32_to_cpu(
+				pi->graphics_level[pl_index].SclkFrequency);
+			*((uint32_t *)value) = sclk;
+			*size = 4;
+			return 0;
+		}
+		return -EINVAL;
+	case AMDGPU_PP_SENSOR_GPU_TEMP:
+		*((uint32_t *)value) = kv_dpm_get_temp(adev);
+		*size = 4;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
+static int kv_set_powergating_by_smu(void *handle,
+				uint32_t block_type, bool gate)
+{
+	switch (block_type) {
+	case AMD_IP_BLOCK_TYPE_UVD:
+		kv_dpm_powergate_uvd(handle, gate);
+		break;
+	case AMD_IP_BLOCK_TYPE_VCE:
+		kv_dpm_powergate_vce(handle, gate);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static const struct amd_ip_funcs kv_dpm_ip_funcs = {
+	.name = "kv_dpm",
+	.early_init = kv_dpm_early_init,
+	.late_init = kv_dpm_late_init,
+	.sw_init = kv_dpm_sw_init,
+	.sw_fini = kv_dpm_sw_fini,
+	.hw_init = kv_dpm_hw_init,
+	.hw_fini = kv_dpm_hw_fini,
+	.suspend = kv_dpm_suspend,
+	.resume = kv_dpm_resume,
+	.is_idle = kv_dpm_is_idle,
+	.wait_for_idle = kv_dpm_wait_for_idle,
+	.soft_reset = kv_dpm_soft_reset,
+	.set_clockgating_state = kv_dpm_set_clockgating_state,
+	.set_powergating_state = kv_dpm_set_powergating_state,
+};
+
+const struct amdgpu_ip_block_version kv_smu_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_SMC,
+	.major = 1,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &kv_dpm_ip_funcs,
+};
+
+static const struct amd_pm_funcs kv_dpm_funcs = {
+	.pre_set_power_state = &kv_dpm_pre_set_power_state,
+	.set_power_state = &kv_dpm_set_power_state,
+	.post_set_power_state = &kv_dpm_post_set_power_state,
+	.display_configuration_changed = &kv_dpm_display_configuration_changed,
+	.get_sclk = &kv_dpm_get_sclk,
+	.get_mclk = &kv_dpm_get_mclk,
+	.print_power_state = &kv_dpm_print_power_state,
+	.debugfs_print_current_performance_level = &kv_dpm_debugfs_print_current_performance_level,
+	.force_performance_level = &kv_dpm_force_performance_level,
+	.set_powergating_by_smu = kv_set_powergating_by_smu,
+	.enable_bapm = &kv_dpm_enable_bapm,
+	.get_vce_clock_state = amdgpu_get_vce_clock_state,
+	.check_state_equal = kv_check_state_equal,
+	.read_sensor = &kv_dpm_read_sensor,
+};
+
+static const struct amdgpu_irq_src_funcs kv_dpm_irq_funcs = {
+	.set = kv_dpm_set_interrupt_state,
+	.process = kv_dpm_process_interrupt,
+};
+
+static void kv_dpm_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->pm.dpm.thermal.irq.num_types = AMDGPU_THERMAL_IRQ_LAST;
+	adev->pm.dpm.thermal.irq.funcs = &kv_dpm_irq_funcs;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_kv_smc.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_kv_smc.c
new file mode 100644
index 00000000000..b82e33c0157
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_kv_smc.c
@@ -0,0 +1,219 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "cikd.h"
+#include "kv_dpm.h"
+
+#include "smu/smu_7_0_0_d.h"
+#include "smu/smu_7_0_0_sh_mask.h"
+
+int amdgpu_kv_notify_message_to_smu(struct amdgpu_device *adev, u32 id)
+{
+	u32 i;
+	u32 tmp = 0;
+
+	WREG32(mmSMC_MESSAGE_0, id & SMC_MESSAGE_0__SMC_MSG_MASK);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if ((RREG32(mmSMC_RESP_0) & SMC_RESP_0__SMC_RESP_MASK) != 0)
+			break;
+		udelay(1);
+	}
+	tmp = RREG32(mmSMC_RESP_0) & SMC_RESP_0__SMC_RESP_MASK;
+
+	if (tmp != 1) {
+		if (tmp == 0xFF)
+			return -EINVAL;
+		else if (tmp == 0xFE)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+int amdgpu_kv_dpm_get_enable_mask(struct amdgpu_device *adev, u32 *enable_mask)
+{
+	int ret;
+
+	ret = amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_SCLKDPM_GetEnabledMask);
+
+	if (ret == 0)
+		*enable_mask = RREG32_SMC(ixSMC_SYSCON_MSG_ARG_0);
+
+	return ret;
+}
+
+int amdgpu_kv_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+				      PPSMC_Msg msg, u32 parameter)
+{
+
+	WREG32(mmSMC_MSG_ARG_0, parameter);
+
+	return amdgpu_kv_notify_message_to_smu(adev, msg);
+}
+
+static int kv_set_smc_sram_address(struct amdgpu_device *adev,
+				   u32 smc_address, u32 limit)
+{
+	if (smc_address & 3)
+		return -EINVAL;
+	if ((smc_address + 3) > limit)
+		return -EINVAL;
+
+	WREG32(mmSMC_IND_INDEX_0, smc_address);
+	WREG32_P(mmSMC_IND_ACCESS_CNTL, 0,
+			~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
+
+	return 0;
+}
+
+int amdgpu_kv_read_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
+			   u32 *value, u32 limit)
+{
+	int ret;
+
+	ret = kv_set_smc_sram_address(adev, smc_address, limit);
+	if (ret)
+		return ret;
+
+	*value = RREG32(mmSMC_IND_DATA_0);
+	return 0;
+}
+
+int amdgpu_kv_smc_dpm_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		return amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_DPM_Enable);
+	else
+		return amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_DPM_Disable);
+}
+
+int amdgpu_kv_smc_bapm_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		return amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_EnableBAPM);
+	else
+		return amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_DisableBAPM);
+}
+
+int amdgpu_kv_copy_bytes_to_smc(struct amdgpu_device *adev,
+			 u32 smc_start_address,
+			 const u8 *src, u32 byte_count, u32 limit)
+{
+	int ret;
+	u32 data, original_data, addr, extra_shift, t_byte, count, mask;
+
+	if ((smc_start_address + byte_count) > limit)
+		return -EINVAL;
+
+	addr = smc_start_address;
+	t_byte = addr & 3;
+
+	/* RMW for the initial bytes */
+	if  (t_byte != 0) {
+		addr -= t_byte;
+
+		ret = kv_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			return ret;
+
+		original_data = RREG32(mmSMC_IND_DATA_0);
+
+		data = 0;
+		mask = 0;
+		count = 4;
+		while (count > 0) {
+			if (t_byte > 0) {
+				mask = (mask << 8) | 0xff;
+				t_byte--;
+			} else if (byte_count > 0) {
+				data = (data << 8) + *src++;
+				byte_count--;
+				mask <<= 8;
+			} else {
+				data <<= 8;
+				mask = (mask << 8) | 0xff;
+			}
+			count--;
+		}
+
+		data |= original_data & mask;
+
+		ret = kv_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			return ret;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+
+		addr += 4;
+	}
+
+	while (byte_count >= 4) {
+		/* SMC address space is BE */
+		data = (src[0] << 24) + (src[1] << 16) + (src[2] << 8) + src[3];
+
+		ret = kv_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			return ret;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+
+		src += 4;
+		byte_count -= 4;
+		addr += 4;
+	}
+
+	/* RMW for the final bytes */
+	if (byte_count > 0) {
+		data = 0;
+
+		ret = kv_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			return ret;
+
+		original_data = RREG32(mmSMC_IND_DATA_0);
+
+		extra_shift = 8 * (4 - byte_count);
+
+		while (byte_count > 0) {
+			/* SMC address space is BE */
+			data = (data << 8) + *src++;
+			byte_count--;
+		}
+
+		data <<= extra_shift;
+
+		data |= (original_data & ~((~0UL) << extra_shift));
+
+		ret = kv_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			return ret;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+	}
+	return 0;
+}
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_mn.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_mn.c
new file mode 100644
index 00000000000..cb70282b014
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_mn.c
@@ -0,0 +1,537 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Christian König <christian.koenig@amd.com>
+ */
+
+/**
+ * DOC: MMU Notifier
+ *
+ * For coherent userptr handling registers an MMU notifier to inform the driver
+ * about updates on the page tables of a process.
+ *
+ * When somebody tries to invalidate the page tables we block the update until
+ * all operations on the pages in question are completed, then those pages are
+ * marked as accessed and also dirty if it wasn't a read only access.
+ *
+ * New command submissions using the userptrs in question are delayed until all
+ * page table invalidation are completed and we once more see a coherent process
+ * address space.
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/mmu_notifier.h>
+#include <linux/interval_tree.h>
+#include <drm/drmP.h>
+#include <drm/drm.h>
+
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+
+/**
+ * struct amdgpu_mn
+ *
+ * @adev: amdgpu device pointer
+ * @mm: process address space
+ * @mn: MMU notifier structure
+ * @type: type of MMU notifier
+ * @work: destruction work item
+ * @node: hash table node to find structure by adev and mn
+ * @lock: rw semaphore protecting the notifier nodes
+ * @objects: interval tree containing amdgpu_mn_nodes
+ * @read_lock: mutex for recursive locking of @lock
+ * @recursion: depth of recursion
+ *
+ * Data for each amdgpu device and process address space.
+ */
+struct amdgpu_mn {
+	/* constant after initialisation */
+	struct amdgpu_device	*adev;
+	struct mm_struct	*mm;
+	struct mmu_notifier	mn;
+	enum amdgpu_mn_type	type;
+
+	/* only used on destruction */
+	struct work_struct	work;
+
+	/* protected by adev->mn_lock */
+	struct hlist_node	node;
+
+	/* objects protected by lock */
+	struct rw_semaphore	lock;
+	struct rb_root_cached	objects;
+	struct rwlock		read_lock;
+	atomic_t		recursion;
+};
+
+/**
+ * struct amdgpu_mn_node
+ *
+ * @it: interval node defining start-last of the affected address range
+ * @bos: list of all BOs in the affected address range
+ *
+ * Manages all BOs which are affected of a certain range of address space.
+ */
+struct amdgpu_mn_node {
+	struct interval_tree_node	it;
+	struct list_head		bos;
+};
+
+/**
+ * amdgpu_mn_destroy - destroy the MMU notifier
+ *
+ * @work: previously sheduled work item
+ *
+ * Lazy destroys the notifier from a work item
+ */
+static void amdgpu_mn_destroy(struct work_struct *work)
+{
+	struct amdgpu_mn *amn = container_of(work, struct amdgpu_mn, work);
+	struct amdgpu_device *adev = amn->adev;
+	struct amdgpu_mn_node *node, *next_node;
+	struct amdgpu_bo *bo, *next_bo;
+
+	mutex_lock(&adev->mn_lock);
+	down_write(&amn->lock);
+	hash_del(&amn->node);
+	rbtree_postorder_for_each_entry_safe(node, next_node,
+					     &amn->objects.rb_root, it.rb) {
+		list_for_each_entry_safe(bo, next_bo, &node->bos, mn_list) {
+			bo->mn = NULL;
+			list_del_init(&bo->mn_list);
+		}
+		kfree(node);
+	}
+	up_write(&amn->lock);
+	mutex_unlock(&adev->mn_lock);
+	mmu_notifier_unregister_no_release(&amn->mn, amn->mm);
+	kfree(amn);
+}
+
+/**
+ * amdgpu_mn_release - callback to notify about mm destruction
+ *
+ * @mn: our notifier
+ * @mm: the mm this callback is about
+ *
+ * Shedule a work item to lazy destroy our notifier.
+ */
+static void amdgpu_mn_release(struct mmu_notifier *mn,
+			      struct mm_struct *mm)
+{
+	struct amdgpu_mn *amn = container_of(mn, struct amdgpu_mn, mn);
+
+	INIT_WORK(&amn->work, amdgpu_mn_destroy);
+	schedule_work(&amn->work);
+}
+
+
+/**
+ * amdgpu_mn_lock - take the write side lock for this notifier
+ *
+ * @mn: our notifier
+ */
+void amdgpu_mn_lock(struct amdgpu_mn *mn)
+{
+	if (mn)
+		down_write(&mn->lock);
+}
+
+/**
+ * amdgpu_mn_unlock - drop the write side lock for this notifier
+ *
+ * @mn: our notifier
+ */
+void amdgpu_mn_unlock(struct amdgpu_mn *mn)
+{
+	if (mn)
+		up_write(&mn->lock);
+}
+
+/**
+ * amdgpu_mn_read_lock - take the read side lock for this notifier
+ *
+ * @amn: our notifier
+ */
+static int amdgpu_mn_read_lock(struct amdgpu_mn *amn, bool blockable)
+{
+	if (blockable)
+		mutex_lock(&amn->read_lock);
+	else if (!mutex_trylock(&amn->read_lock))
+		return -EAGAIN;
+
+	if (atomic_inc_return(&amn->recursion) == 1)
+		down_read_non_owner(&amn->lock);
+	mutex_unlock(&amn->read_lock);
+
+	return 0;
+}
+
+/**
+ * amdgpu_mn_read_unlock - drop the read side lock for this notifier
+ *
+ * @amn: our notifier
+ */
+static void amdgpu_mn_read_unlock(struct amdgpu_mn *amn)
+{
+	if (atomic_dec_return(&amn->recursion) == 0)
+		up_read_non_owner(&amn->lock);
+}
+
+/**
+ * amdgpu_mn_invalidate_node - unmap all BOs of a node
+ *
+ * @node: the node with the BOs to unmap
+ * @start: start of address range affected
+ * @end: end of address range affected
+ *
+ * Block for operations on BOs to finish and mark pages as accessed and
+ * potentially dirty.
+ */
+static void amdgpu_mn_invalidate_node(struct amdgpu_mn_node *node,
+				      unsigned long start,
+				      unsigned long end)
+{
+	struct amdgpu_bo *bo;
+	long r;
+
+	list_for_each_entry(bo, &node->bos, mn_list) {
+
+		if (!amdgpu_ttm_tt_affect_userptr(bo->tbo.ttm, start, end))
+			continue;
+
+		r = reservation_object_wait_timeout_rcu(bo->tbo.resv,
+			true, false, MAX_SCHEDULE_TIMEOUT);
+		if (r <= 0)
+			DRM_ERROR("(%ld) failed to wait for user bo\n", r);
+
+		amdgpu_ttm_tt_mark_user_pages(bo->tbo.ttm);
+	}
+}
+
+/**
+ * amdgpu_mn_invalidate_range_start_gfx - callback to notify about mm change
+ *
+ * @mn: our notifier
+ * @mm: the mm this callback is about
+ * @start: start of updated range
+ * @end: end of updated range
+ *
+ * Block for operations on BOs to finish and mark pages as accessed and
+ * potentially dirty.
+ */
+static int amdgpu_mn_invalidate_range_start_gfx(struct mmu_notifier *mn,
+						 struct mm_struct *mm,
+						 unsigned long start,
+						 unsigned long end,
+						 bool blockable)
+{
+	struct amdgpu_mn *amn = container_of(mn, struct amdgpu_mn, mn);
+	struct interval_tree_node *it;
+
+	/* notification is exclusive, but interval is inclusive */
+	end -= 1;
+
+	/* TODO we should be able to split locking for interval tree and
+	 * amdgpu_mn_invalidate_node
+	 */
+	if (amdgpu_mn_read_lock(amn, blockable))
+		return -EAGAIN;
+
+	it = interval_tree_iter_first(&amn->objects, start, end);
+	while (it) {
+		struct amdgpu_mn_node *node;
+
+		if (!blockable) {
+			amdgpu_mn_read_unlock(amn);
+			return -EAGAIN;
+		}
+
+		node = container_of(it, struct amdgpu_mn_node, it);
+		it = interval_tree_iter_next(it, start, end);
+
+		amdgpu_mn_invalidate_node(node, start, end);
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_mn_invalidate_range_start_hsa - callback to notify about mm change
+ *
+ * @mn: our notifier
+ * @mm: the mm this callback is about
+ * @start: start of updated range
+ * @end: end of updated range
+ *
+ * We temporarily evict all BOs between start and end. This
+ * necessitates evicting all user-mode queues of the process. The BOs
+ * are restorted in amdgpu_mn_invalidate_range_end_hsa.
+ */
+static int amdgpu_mn_invalidate_range_start_hsa(struct mmu_notifier *mn,
+						 struct mm_struct *mm,
+						 unsigned long start,
+						 unsigned long end,
+						 bool blockable)
+{
+	struct amdgpu_mn *amn = container_of(mn, struct amdgpu_mn, mn);
+	struct interval_tree_node *it;
+
+	/* notification is exclusive, but interval is inclusive */
+	end -= 1;
+
+	if (amdgpu_mn_read_lock(amn, blockable))
+		return -EAGAIN;
+
+	it = interval_tree_iter_first(&amn->objects, start, end);
+	while (it) {
+		struct amdgpu_mn_node *node;
+		struct amdgpu_bo *bo;
+
+		if (!blockable) {
+			amdgpu_mn_read_unlock(amn);
+			return -EAGAIN;
+		}
+
+		node = container_of(it, struct amdgpu_mn_node, it);
+		it = interval_tree_iter_next(it, start, end);
+
+		list_for_each_entry(bo, &node->bos, mn_list) {
+			struct kgd_mem *mem = bo->kfd_bo;
+
+			if (amdgpu_ttm_tt_affect_userptr(bo->tbo.ttm,
+							 start, end))
+				amdgpu_amdkfd_evict_userptr(mem, mm);
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_mn_invalidate_range_end - callback to notify about mm change
+ *
+ * @mn: our notifier
+ * @mm: the mm this callback is about
+ * @start: start of updated range
+ * @end: end of updated range
+ *
+ * Release the lock again to allow new command submissions.
+ */
+static void amdgpu_mn_invalidate_range_end(struct mmu_notifier *mn,
+					   struct mm_struct *mm,
+					   unsigned long start,
+					   unsigned long end)
+{
+	struct amdgpu_mn *amn = container_of(mn, struct amdgpu_mn, mn);
+
+	amdgpu_mn_read_unlock(amn);
+}
+
+static const struct mmu_notifier_ops amdgpu_mn_ops[] = {
+	[AMDGPU_MN_TYPE_GFX] = {
+		.release = amdgpu_mn_release,
+		.invalidate_range_start = amdgpu_mn_invalidate_range_start_gfx,
+		.invalidate_range_end = amdgpu_mn_invalidate_range_end,
+	},
+	[AMDGPU_MN_TYPE_HSA] = {
+		.release = amdgpu_mn_release,
+		.invalidate_range_start = amdgpu_mn_invalidate_range_start_hsa,
+		.invalidate_range_end = amdgpu_mn_invalidate_range_end,
+	},
+};
+
+/* Low bits of any reasonable mm pointer will be unused due to struct
+ * alignment. Use these bits to make a unique key from the mm pointer
+ * and notifier type.
+ */
+#define AMDGPU_MN_KEY(mm, type) ((unsigned long)(mm) + (type))
+
+/**
+ * amdgpu_mn_get - create notifier context
+ *
+ * @adev: amdgpu device pointer
+ * @type: type of MMU notifier context
+ *
+ * Creates a notifier context for current->mm.
+ */
+struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev,
+				enum amdgpu_mn_type type)
+{
+	struct mm_struct *mm = current->mm;
+	struct amdgpu_mn *amn;
+	unsigned long key = AMDGPU_MN_KEY(mm, type);
+	int r;
+
+	mutex_lock(&adev->mn_lock);
+	if (down_write_killable(&mm->mmap_sem)) {
+		mutex_unlock(&adev->mn_lock);
+		return ERR_PTR(-EINTR);
+	}
+
+	hash_for_each_possible(adev->mn_hash, amn, node, key)
+		if (AMDGPU_MN_KEY(amn->mm, amn->type) == key)
+			goto release_locks;
+
+	amn = kzalloc(sizeof(*amn), GFP_KERNEL);
+	if (!amn) {
+		amn = ERR_PTR(-ENOMEM);
+		goto release_locks;
+	}
+
+	amn->adev = adev;
+	amn->mm = mm;
+	init_rwsem(&amn->lock);
+	amn->type = type;
+	amn->mn.ops = &amdgpu_mn_ops[type];
+	amn->objects = RB_ROOT_CACHED;
+	rw_init(&amn->read_lock, "amnrd");
+	atomic_set(&amn->recursion, 0);
+
+	r = __mmu_notifier_register(&amn->mn, mm);
+	if (r)
+		goto free_amn;
+
+	hash_add(adev->mn_hash, &amn->node, AMDGPU_MN_KEY(mm, type));
+
+release_locks:
+	up_write(&mm->mmap_sem);
+	mutex_unlock(&adev->mn_lock);
+
+	return amn;
+
+free_amn:
+	up_write(&mm->mmap_sem);
+	mutex_unlock(&adev->mn_lock);
+	kfree(amn);
+
+	return ERR_PTR(r);
+}
+
+/**
+ * amdgpu_mn_register - register a BO for notifier updates
+ *
+ * @bo: amdgpu buffer object
+ * @addr: userptr addr we should monitor
+ *
+ * Registers an MMU notifier for the given BO at the specified address.
+ * Returns 0 on success, -ERRNO if anything goes wrong.
+ */
+int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr)
+{
+	unsigned long end = addr + amdgpu_bo_size(bo) - 1;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	enum amdgpu_mn_type type =
+		bo->kfd_bo ? AMDGPU_MN_TYPE_HSA : AMDGPU_MN_TYPE_GFX;
+	struct amdgpu_mn *amn;
+	struct amdgpu_mn_node *node = NULL, *new_node;
+	struct list_head bos;
+	struct interval_tree_node *it;
+
+	amn = amdgpu_mn_get(adev, type);
+	if (IS_ERR(amn))
+		return PTR_ERR(amn);
+
+	new_node = kmalloc(sizeof(*new_node), GFP_KERNEL);
+	if (!new_node)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&bos);
+
+	down_write(&amn->lock);
+
+	while ((it = interval_tree_iter_first(&amn->objects, addr, end))) {
+		kfree(node);
+		node = container_of(it, struct amdgpu_mn_node, it);
+		interval_tree_remove(&node->it, &amn->objects);
+		addr = min(it->start, addr);
+		end = max(it->last, end);
+		list_splice(&node->bos, &bos);
+	}
+
+	if (!node)
+		node = new_node;
+	else
+		kfree(new_node);
+
+	bo->mn = amn;
+
+	node->it.start = addr;
+	node->it.last = end;
+	INIT_LIST_HEAD(&node->bos);
+	list_splice(&bos, &node->bos);
+	list_add(&bo->mn_list, &node->bos);
+
+	interval_tree_insert(&node->it, &amn->objects);
+
+	up_write(&amn->lock);
+
+	return 0;
+}
+
+/**
+ * amdgpu_mn_unregister - unregister a BO for notifier updates
+ *
+ * @bo: amdgpu buffer object
+ *
+ * Remove any registration of MMU notifier updates from the buffer object.
+ */
+void amdgpu_mn_unregister(struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct amdgpu_mn *amn;
+	struct list_head *head;
+
+	mutex_lock(&adev->mn_lock);
+
+	amn = bo->mn;
+	if (amn == NULL) {
+		mutex_unlock(&adev->mn_lock);
+		return;
+	}
+
+	down_write(&amn->lock);
+
+	/* save the next list entry for later */
+	head = bo->mn_list.next;
+
+	bo->mn = NULL;
+	list_del_init(&bo->mn_list);
+
+	if (list_empty(head)) {
+		struct amdgpu_mn_node *node;
+
+		node = container_of(head, struct amdgpu_mn_node, bos);
+		interval_tree_remove(&node->it, &amn->objects);
+		kfree(node);
+	}
+
+	up_write(&amn->lock);
+	mutex_unlock(&adev->mn_lock);
+}
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_mn.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_mn.h
new file mode 100644
index 00000000000..eb0f432f78f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_mn.h
@@ -0,0 +1,59 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König
+ */
+#ifndef __AMDGPU_MN_H__
+#define __AMDGPU_MN_H__
+
+/*
+ * MMU Notifier
+ */
+struct amdgpu_mn;
+
+enum amdgpu_mn_type {
+	AMDGPU_MN_TYPE_GFX,
+	AMDGPU_MN_TYPE_HSA,
+};
+
+#if defined(CONFIG_MMU_NOTIFIER)
+void amdgpu_mn_lock(struct amdgpu_mn *mn);
+void amdgpu_mn_unlock(struct amdgpu_mn *mn);
+struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev,
+				enum amdgpu_mn_type type);
+int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr);
+void amdgpu_mn_unregister(struct amdgpu_bo *bo);
+#else
+static inline void amdgpu_mn_lock(struct amdgpu_mn *mn) {}
+static inline void amdgpu_mn_unlock(struct amdgpu_mn *mn) {}
+static inline struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev,
+					      enum amdgpu_mn_type type)
+{
+	return NULL;
+}
+static inline int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr)
+{
+	return -ENODEV;
+}
+static inline void amdgpu_mn_unregister(struct amdgpu_bo *bo) {}
+#endif
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_mode.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_mode.h
new file mode 100644
index 00000000000..65ae35e5283
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_mode.h
@@ -0,0 +1,652 @@
+/*
+ * Copyright 2000 ATI Technologies Inc., Markham, Ontario, and
+ *                VA Linux Systems Inc., Fremont, California.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Original Authors:
+ *   Kevin E. Martin, Rickard E. Faith, Alan Hourihane
+ *
+ * Kernel port Author: Dave Airlie
+ */
+
+#ifndef AMDGPU_MODE_H
+#define AMDGPU_MODE_H
+
+#include <drm/drm_crtc.h>
+#include <drm/drm_edid.h>
+#include <drm/drm_encoder.h>
+#include <drm/drm_dp_helper.h>
+#include <drm/drm_fixed.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/drm_fb_helper.h>
+#include <drm/drm_plane_helper.h>
+#include <drm/drm_fb_helper.h>
+#include <linux/i2c.h>
+#include <linux/i2c-algo-bit.h>
+#include <linux/hrtimer.h>
+#include "amdgpu_irq.h"
+
+#include <drm/drm_dp_mst_helper.h>
+#include "modules/inc/mod_freesync.h"
+
+struct amdgpu_bo;
+struct amdgpu_device;
+struct amdgpu_encoder;
+struct amdgpu_router;
+struct amdgpu_hpd;
+
+#define to_amdgpu_crtc(x) container_of(x, struct amdgpu_crtc, base)
+#define to_amdgpu_connector(x) container_of(x, struct amdgpu_connector, base)
+#define to_amdgpu_encoder(x) container_of(x, struct amdgpu_encoder, base)
+#define to_amdgpu_framebuffer(x) container_of(x, struct amdgpu_framebuffer, base)
+#define to_amdgpu_plane(x)	container_of(x, struct amdgpu_plane, base)
+
+#define to_dm_plane_state(x)	container_of(x, struct dm_plane_state, base);
+
+#define AMDGPU_MAX_HPD_PINS 6
+#define AMDGPU_MAX_CRTCS 6
+#define AMDGPU_MAX_PLANES 6
+#define AMDGPU_MAX_AFMT_BLOCKS 9
+
+enum amdgpu_rmx_type {
+	RMX_OFF,
+	RMX_FULL,
+	RMX_CENTER,
+	RMX_ASPECT
+};
+
+enum amdgpu_underscan_type {
+	UNDERSCAN_OFF,
+	UNDERSCAN_ON,
+	UNDERSCAN_AUTO,
+};
+
+#define AMDGPU_HPD_CONNECT_INT_DELAY_IN_MS 50
+#define AMDGPU_HPD_DISCONNECT_INT_DELAY_IN_MS 10
+
+enum amdgpu_hpd_id {
+	AMDGPU_HPD_1 = 0,
+	AMDGPU_HPD_2,
+	AMDGPU_HPD_3,
+	AMDGPU_HPD_4,
+	AMDGPU_HPD_5,
+	AMDGPU_HPD_6,
+	AMDGPU_HPD_NONE = 0xff,
+};
+
+enum amdgpu_crtc_irq {
+	AMDGPU_CRTC_IRQ_VBLANK1 = 0,
+	AMDGPU_CRTC_IRQ_VBLANK2,
+	AMDGPU_CRTC_IRQ_VBLANK3,
+	AMDGPU_CRTC_IRQ_VBLANK4,
+	AMDGPU_CRTC_IRQ_VBLANK5,
+	AMDGPU_CRTC_IRQ_VBLANK6,
+	AMDGPU_CRTC_IRQ_VLINE1,
+	AMDGPU_CRTC_IRQ_VLINE2,
+	AMDGPU_CRTC_IRQ_VLINE3,
+	AMDGPU_CRTC_IRQ_VLINE4,
+	AMDGPU_CRTC_IRQ_VLINE5,
+	AMDGPU_CRTC_IRQ_VLINE6,
+	AMDGPU_CRTC_IRQ_NONE = 0xff
+};
+
+enum amdgpu_pageflip_irq {
+	AMDGPU_PAGEFLIP_IRQ_D1 = 0,
+	AMDGPU_PAGEFLIP_IRQ_D2,
+	AMDGPU_PAGEFLIP_IRQ_D3,
+	AMDGPU_PAGEFLIP_IRQ_D4,
+	AMDGPU_PAGEFLIP_IRQ_D5,
+	AMDGPU_PAGEFLIP_IRQ_D6,
+	AMDGPU_PAGEFLIP_IRQ_NONE = 0xff
+};
+
+enum amdgpu_flip_status {
+	AMDGPU_FLIP_NONE,
+	AMDGPU_FLIP_PENDING,
+	AMDGPU_FLIP_SUBMITTED
+};
+
+#define AMDGPU_MAX_I2C_BUS 16
+
+/* amdgpu gpio-based i2c
+ * 1. "mask" reg and bits
+ *    grabs the gpio pins for software use
+ *    0=not held  1=held
+ * 2. "a" reg and bits
+ *    output pin value
+ *    0=low 1=high
+ * 3. "en" reg and bits
+ *    sets the pin direction
+ *    0=input 1=output
+ * 4. "y" reg and bits
+ *    input pin value
+ *    0=low 1=high
+ */
+struct amdgpu_i2c_bus_rec {
+	bool valid;
+	/* id used by atom */
+	uint8_t i2c_id;
+	/* id used by atom */
+	enum amdgpu_hpd_id hpd;
+	/* can be used with hw i2c engine */
+	bool hw_capable;
+	/* uses multi-media i2c engine */
+	bool mm_i2c;
+	/* regs and bits */
+	uint32_t mask_clk_reg;
+	uint32_t mask_data_reg;
+	uint32_t a_clk_reg;
+	uint32_t a_data_reg;
+	uint32_t en_clk_reg;
+	uint32_t en_data_reg;
+	uint32_t y_clk_reg;
+	uint32_t y_data_reg;
+	uint32_t mask_clk_mask;
+	uint32_t mask_data_mask;
+	uint32_t a_clk_mask;
+	uint32_t a_data_mask;
+	uint32_t en_clk_mask;
+	uint32_t en_data_mask;
+	uint32_t y_clk_mask;
+	uint32_t y_data_mask;
+};
+
+#define AMDGPU_MAX_BIOS_CONNECTOR 16
+
+/* pll flags */
+#define AMDGPU_PLL_USE_BIOS_DIVS        (1 << 0)
+#define AMDGPU_PLL_NO_ODD_POST_DIV      (1 << 1)
+#define AMDGPU_PLL_USE_REF_DIV          (1 << 2)
+#define AMDGPU_PLL_LEGACY               (1 << 3)
+#define AMDGPU_PLL_PREFER_LOW_REF_DIV   (1 << 4)
+#define AMDGPU_PLL_PREFER_HIGH_REF_DIV  (1 << 5)
+#define AMDGPU_PLL_PREFER_LOW_FB_DIV    (1 << 6)
+#define AMDGPU_PLL_PREFER_HIGH_FB_DIV   (1 << 7)
+#define AMDGPU_PLL_PREFER_LOW_POST_DIV  (1 << 8)
+#define AMDGPU_PLL_PREFER_HIGH_POST_DIV (1 << 9)
+#define AMDGPU_PLL_USE_FRAC_FB_DIV      (1 << 10)
+#define AMDGPU_PLL_PREFER_CLOSEST_LOWER (1 << 11)
+#define AMDGPU_PLL_USE_POST_DIV         (1 << 12)
+#define AMDGPU_PLL_IS_LCD               (1 << 13)
+#define AMDGPU_PLL_PREFER_MINM_OVER_MAXP (1 << 14)
+
+struct amdgpu_pll {
+	/* reference frequency */
+	uint32_t reference_freq;
+
+	/* fixed dividers */
+	uint32_t reference_div;
+	uint32_t post_div;
+
+	/* pll in/out limits */
+	uint32_t pll_in_min;
+	uint32_t pll_in_max;
+	uint32_t pll_out_min;
+	uint32_t pll_out_max;
+	uint32_t lcd_pll_out_min;
+	uint32_t lcd_pll_out_max;
+	uint32_t best_vco;
+
+	/* divider limits */
+	uint32_t min_ref_div;
+	uint32_t max_ref_div;
+	uint32_t min_post_div;
+	uint32_t max_post_div;
+	uint32_t min_feedback_div;
+	uint32_t max_feedback_div;
+	uint32_t min_frac_feedback_div;
+	uint32_t max_frac_feedback_div;
+
+	/* flags for the current clock */
+	uint32_t flags;
+
+	/* pll id */
+	uint32_t id;
+};
+
+struct amdgpu_i2c_chan {
+	struct i2c_adapter adapter;
+	struct drm_device *dev;
+	struct i2c_algo_bit_data bit;
+	struct amdgpu_i2c_bus_rec rec;
+	struct drm_dp_aux aux;
+	bool has_aux;
+	struct rwlock mutex;
+};
+
+struct amdgpu_fbdev;
+
+struct amdgpu_afmt {
+	bool enabled;
+	int offset;
+	bool last_buffer_filled_status;
+	int id;
+	struct amdgpu_audio_pin *pin;
+};
+
+/*
+ * Audio
+ */
+struct amdgpu_audio_pin {
+	int			channels;
+	int			rate;
+	int			bits_per_sample;
+	u8			status_bits;
+	u8			category_code;
+	u32			offset;
+	bool			connected;
+	u32			id;
+};
+
+struct amdgpu_audio {
+	bool enabled;
+	struct amdgpu_audio_pin pin[AMDGPU_MAX_AFMT_BLOCKS];
+	int num_pins;
+};
+
+struct amdgpu_display_funcs {
+	/* display watermarks */
+	void (*bandwidth_update)(struct amdgpu_device *adev);
+	/* get frame count */
+	u32 (*vblank_get_counter)(struct amdgpu_device *adev, int crtc);
+	/* set backlight level */
+	void (*backlight_set_level)(struct amdgpu_encoder *amdgpu_encoder,
+				    u8 level);
+	/* get backlight level */
+	u8 (*backlight_get_level)(struct amdgpu_encoder *amdgpu_encoder);
+	/* hotplug detect */
+	bool (*hpd_sense)(struct amdgpu_device *adev, enum amdgpu_hpd_id hpd);
+	void (*hpd_set_polarity)(struct amdgpu_device *adev,
+				 enum amdgpu_hpd_id hpd);
+	u32 (*hpd_get_gpio_reg)(struct amdgpu_device *adev);
+	/* pageflipping */
+	void (*page_flip)(struct amdgpu_device *adev,
+			  int crtc_id, u64 crtc_base, bool async);
+	int (*page_flip_get_scanoutpos)(struct amdgpu_device *adev, int crtc,
+					u32 *vbl, u32 *position);
+	/* display topology setup */
+	void (*add_encoder)(struct amdgpu_device *adev,
+			    uint32_t encoder_enum,
+			    uint32_t supported_device,
+			    u16 caps);
+	void (*add_connector)(struct amdgpu_device *adev,
+			      uint32_t connector_id,
+			      uint32_t supported_device,
+			      int connector_type,
+			      struct amdgpu_i2c_bus_rec *i2c_bus,
+			      uint16_t connector_object_id,
+			      struct amdgpu_hpd *hpd,
+			      struct amdgpu_router *router);
+	/* it is used to enter or exit into free sync mode */
+	int (*notify_freesync)(struct drm_device *dev, void *data,
+			       struct drm_file *filp);
+	/* it is used to allow enablement of freesync mode */
+	int (*set_freesync_property)(struct drm_connector *connector,
+				     struct drm_property *property,
+				     uint64_t val);
+
+
+};
+
+struct amdgpu_framebuffer {
+	struct drm_framebuffer base;
+
+	/* caching for later use */
+	uint64_t address;
+};
+
+struct amdgpu_fbdev {
+	struct drm_fb_helper helper;
+	struct amdgpu_framebuffer rfb;
+	struct list_head fbdev_list;
+	struct amdgpu_device *adev;
+};
+
+struct amdgpu_mode_info {
+	struct atom_context *atom_context;
+	struct card_info *atom_card_info;
+	bool mode_config_initialized;
+	struct amdgpu_crtc *crtcs[AMDGPU_MAX_CRTCS];
+	struct amdgpu_plane *planes[AMDGPU_MAX_PLANES];
+	struct amdgpu_afmt *afmt[AMDGPU_MAX_AFMT_BLOCKS];
+	/* DVI-I properties */
+	struct drm_property *coherent_mode_property;
+	/* DAC enable load detect */
+	struct drm_property *load_detect_property;
+	/* underscan */
+	struct drm_property *underscan_property;
+	struct drm_property *underscan_hborder_property;
+	struct drm_property *underscan_vborder_property;
+	/* audio */
+	struct drm_property *audio_property;
+	/* FMT dithering */
+	struct drm_property *dither_property;
+	/* maximum number of bits per channel for monitor color */
+	struct drm_property *max_bpc_property;
+	/* hardcoded DFP edid from BIOS */
+	struct edid *bios_hardcoded_edid;
+	int bios_hardcoded_edid_size;
+
+	/* pointer to fbdev info structure */
+	struct amdgpu_fbdev *rfbdev;
+	/* firmware flags */
+	u16 firmware_flags;
+	/* pointer to backlight encoder */
+	struct amdgpu_encoder *bl_encoder;
+	u8 bl_level; /* saved backlight level */
+	struct amdgpu_audio	audio; /* audio stuff */
+	int			num_crtc; /* number of crtcs */
+	int			num_hpd; /* number of hpd pins */
+	int			num_dig; /* number of dig blocks */
+	int			disp_priority;
+	const struct amdgpu_display_funcs *funcs;
+	const enum drm_plane_type *plane_type;
+};
+
+#define AMDGPU_MAX_BL_LEVEL 0xFF
+
+#if defined(CONFIG_BACKLIGHT_CLASS_DEVICE) || defined(CONFIG_BACKLIGHT_CLASS_DEVICE_MODULE)
+
+struct amdgpu_backlight_privdata {
+	struct amdgpu_encoder *encoder;
+	uint8_t negative;
+};
+
+#endif
+
+struct amdgpu_atom_ss {
+	uint16_t percentage;
+	uint16_t percentage_divider;
+	uint8_t type;
+	uint16_t step;
+	uint8_t delay;
+	uint8_t range;
+	uint8_t refdiv;
+	/* asic_ss */
+	uint16_t rate;
+	uint16_t amount;
+};
+
+struct amdgpu_crtc {
+	struct drm_crtc base;
+	int crtc_id;
+	bool enabled;
+	bool can_tile;
+	uint32_t crtc_offset;
+	struct drm_gem_object *cursor_bo;
+	uint64_t cursor_addr;
+	int cursor_x;
+	int cursor_y;
+	int cursor_hot_x;
+	int cursor_hot_y;
+	int cursor_width;
+	int cursor_height;
+	int max_cursor_width;
+	int max_cursor_height;
+	enum amdgpu_rmx_type rmx_type;
+	u8 h_border;
+	u8 v_border;
+	fixed20_12 vsc;
+	fixed20_12 hsc;
+	struct drm_display_mode native_mode;
+	u32 pll_id;
+	/* page flipping */
+	struct amdgpu_flip_work *pflip_works;
+	enum amdgpu_flip_status pflip_status;
+	int deferred_flip_completion;
+	/* pll sharing */
+	struct amdgpu_atom_ss ss;
+	bool ss_enabled;
+	u32 adjusted_clock;
+	int bpc;
+	u32 pll_reference_div;
+	u32 pll_post_div;
+	u32 pll_flags;
+	struct drm_encoder *encoder;
+	struct drm_connector *connector;
+	/* for dpm */
+	u32 line_time;
+	u32 wm_low;
+	u32 wm_high;
+	u32 lb_vblank_lead_lines;
+	struct drm_display_mode hw_mode;
+	/* for virtual dce */
+#ifdef __linux__
+	struct hrtimer vblank_timer;
+#else
+	struct timeout vblank_timer;
+#endif
+	enum amdgpu_interrupt_state vsync_timer_enabled;
+
+	int otg_inst;
+	struct drm_pending_vblank_event *event;
+};
+
+struct amdgpu_plane {
+	struct drm_plane base;
+	enum drm_plane_type plane_type;
+};
+
+struct amdgpu_encoder_atom_dig {
+	bool linkb;
+	/* atom dig */
+	bool coherent_mode;
+	int dig_encoder; /* -1 disabled, 0 DIGA, 1 DIGB, etc. */
+	/* atom lvds/edp */
+	uint32_t lcd_misc;
+	uint16_t panel_pwr_delay;
+	uint32_t lcd_ss_id;
+	/* panel mode */
+	struct drm_display_mode native_mode;
+	struct backlight_device *bl_dev;
+	int dpms_mode;
+	uint8_t backlight_level;
+	int panel_mode;
+	struct amdgpu_afmt *afmt;
+};
+
+struct amdgpu_encoder {
+	struct drm_encoder base;
+	uint32_t encoder_enum;
+	uint32_t encoder_id;
+	uint32_t devices;
+	uint32_t active_device;
+	uint32_t flags;
+	uint32_t pixel_clock;
+	enum amdgpu_rmx_type rmx_type;
+	enum amdgpu_underscan_type underscan_type;
+	uint32_t underscan_hborder;
+	uint32_t underscan_vborder;
+	struct drm_display_mode native_mode;
+	void *enc_priv;
+	int audio_polling_active;
+	bool is_ext_encoder;
+	u16 caps;
+};
+
+struct amdgpu_connector_atom_dig {
+	/* displayport */
+	u8 dpcd[DP_RECEIVER_CAP_SIZE];
+	u8 dp_sink_type;
+	int dp_clock;
+	int dp_lane_count;
+	bool edp_on;
+};
+
+struct amdgpu_gpio_rec {
+	bool valid;
+	u8 id;
+	u32 reg;
+	u32 mask;
+	u32 shift;
+};
+
+struct amdgpu_hpd {
+	enum amdgpu_hpd_id hpd;
+	u8 plugged_state;
+	struct amdgpu_gpio_rec gpio;
+};
+
+struct amdgpu_router {
+	u32 router_id;
+	struct amdgpu_i2c_bus_rec i2c_info;
+	u8 i2c_addr;
+	/* i2c mux */
+	bool ddc_valid;
+	u8 ddc_mux_type;
+	u8 ddc_mux_control_pin;
+	u8 ddc_mux_state;
+	/* clock/data mux */
+	bool cd_valid;
+	u8 cd_mux_type;
+	u8 cd_mux_control_pin;
+	u8 cd_mux_state;
+};
+
+enum amdgpu_connector_audio {
+	AMDGPU_AUDIO_DISABLE = 0,
+	AMDGPU_AUDIO_ENABLE = 1,
+	AMDGPU_AUDIO_AUTO = 2
+};
+
+enum amdgpu_connector_dither {
+	AMDGPU_FMT_DITHER_DISABLE = 0,
+	AMDGPU_FMT_DITHER_ENABLE = 1,
+};
+
+struct amdgpu_dm_dp_aux {
+	struct drm_dp_aux aux;
+	struct ddc_service *ddc_service;
+};
+
+struct amdgpu_i2c_adapter {
+	struct i2c_adapter base;
+
+	struct ddc_service *ddc_service;
+};
+
+#define TO_DM_AUX(x) container_of((x), struct amdgpu_dm_dp_aux, aux)
+
+struct amdgpu_connector {
+	struct drm_connector base;
+	uint32_t connector_id;
+	uint32_t devices;
+	struct amdgpu_i2c_chan *ddc_bus;
+	/* some systems have an hdmi and vga port with a shared ddc line */
+	bool shared_ddc;
+	bool use_digital;
+	/* we need to mind the EDID between detect
+	   and get modes due to analog/digital/tvencoder */
+	struct edid *edid;
+	void *con_priv;
+	bool dac_load_detect;
+	bool detected_by_load; /* if the connection status was determined by load */
+	uint16_t connector_object_id;
+	struct amdgpu_hpd hpd;
+	struct amdgpu_router router;
+	struct amdgpu_i2c_chan *router_bus;
+	enum amdgpu_connector_audio audio;
+	enum amdgpu_connector_dither dither;
+	unsigned pixelclock_for_modeset;
+};
+
+/* TODO: start to use this struct and remove same field from base one */
+struct amdgpu_mst_connector {
+	struct amdgpu_connector base;
+
+	struct drm_dp_mst_topology_mgr mst_mgr;
+	struct amdgpu_dm_dp_aux dm_dp_aux;
+	struct drm_dp_mst_port *port;
+	struct amdgpu_connector *mst_port;
+	bool is_mst_connector;
+	struct amdgpu_encoder *mst_encoder;
+};
+
+#define ENCODER_MODE_IS_DP(em) (((em) == ATOM_ENCODER_MODE_DP) || \
+				((em) == ATOM_ENCODER_MODE_DP_MST))
+
+/* Driver internal use only flags of amdgpu_display_get_crtc_scanoutpos() */
+#define DRM_SCANOUTPOS_VALID        (1 << 0)
+#define DRM_SCANOUTPOS_IN_VBLANK    (1 << 1)
+#define DRM_SCANOUTPOS_ACCURATE     (1 << 2)
+#define USE_REAL_VBLANKSTART		(1 << 30)
+#define GET_DISTANCE_TO_VBLANKSTART	(1 << 31)
+
+void amdgpu_link_encoder_connector(struct drm_device *dev);
+
+struct drm_connector *
+amdgpu_get_connector_for_encoder(struct drm_encoder *encoder);
+struct drm_connector *
+amdgpu_get_connector_for_encoder_init(struct drm_encoder *encoder);
+bool amdgpu_dig_monitor_is_duallink(struct drm_encoder *encoder,
+				    u32 pixel_clock);
+
+u16 amdgpu_encoder_get_dp_bridge_encoder_id(struct drm_encoder *encoder);
+struct drm_encoder *amdgpu_get_external_encoder(struct drm_encoder *encoder);
+
+bool amdgpu_display_ddc_probe(struct amdgpu_connector *amdgpu_connector,
+			      bool use_aux);
+
+void amdgpu_encoder_set_active_device(struct drm_encoder *encoder);
+
+int amdgpu_display_get_crtc_scanoutpos(struct drm_device *dev,
+			unsigned int pipe, unsigned int flags, int *vpos,
+			int *hpos, ktime_t *stime, ktime_t *etime,
+			const struct drm_display_mode *mode);
+
+int amdgpu_display_framebuffer_init(struct drm_device *dev,
+				    struct amdgpu_framebuffer *rfb,
+				    const struct drm_mode_fb_cmd2 *mode_cmd,
+				    struct drm_gem_object *obj);
+
+int amdgpufb_remove(struct drm_device *dev, struct drm_framebuffer *fb);
+
+void amdgpu_enc_destroy(struct drm_encoder *encoder);
+void amdgpu_copy_fb(struct drm_device *dev, struct drm_gem_object *dst_obj);
+bool amdgpu_display_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
+				const struct drm_display_mode *mode,
+				struct drm_display_mode *adjusted_mode);
+void amdgpu_panel_mode_fixup(struct drm_encoder *encoder,
+			     struct drm_display_mode *adjusted_mode);
+int amdgpu_display_crtc_idx_to_irq_type(struct amdgpu_device *adev, int crtc);
+
+/* fbdev layer */
+int amdgpu_fbdev_init(struct amdgpu_device *adev);
+void amdgpu_fbdev_fini(struct amdgpu_device *adev);
+void amdgpu_fbdev_set_suspend(struct amdgpu_device *adev, int state);
+int amdgpu_fbdev_total_size(struct amdgpu_device *adev);
+bool amdgpu_fbdev_robj_is_fb(struct amdgpu_device *adev, struct amdgpu_bo *robj);
+
+int amdgpu_align_pitch(struct amdgpu_device *adev, int width, int bpp, bool tiled);
+
+/* amdgpu_display.c */
+void amdgpu_display_print_display_setup(struct drm_device *dev);
+int amdgpu_display_modeset_create_props(struct amdgpu_device *adev);
+int amdgpu_display_crtc_set_config(struct drm_mode_set *set,
+				   struct drm_modeset_acquire_ctx *ctx);
+int amdgpu_display_crtc_page_flip_target(struct drm_crtc *crtc,
+				struct drm_framebuffer *fb,
+				struct drm_pending_vblank_event *event,
+				uint32_t page_flip_flags, uint32_t target,
+				struct drm_modeset_acquire_ctx *ctx);
+extern const struct drm_mode_config_funcs amdgpu_mode_funcs;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_object.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_object.c
new file mode 100644
index 00000000000..b2f93eb5ac9
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_object.c
@@ -0,0 +1,1420 @@
+/*
+ * Copyright 2009 Jerome Glisse.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Jerome Glisse <glisse@freedesktop.org>
+ *    Thomas Hellstrom <thomas-at-tungstengraphics-dot-com>
+ *    Dave Airlie
+ */
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include <drm/drm_cache.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+#include "amdgpu_amdkfd.h"
+
+/**
+ * DOC: amdgpu_object
+ *
+ * This defines the interfaces to operate on an &amdgpu_bo buffer object which
+ * represents memory used by driver (VRAM, system memory, etc.). The driver
+ * provides DRM/GEM APIs to userspace. DRM/GEM APIs then use these interfaces
+ * to create/destroy/set buffer object which are then managed by the kernel TTM
+ * memory manager.
+ * The interfaces are also used internally by kernel clients, including gfx,
+ * uvd, etc. for kernel managed allocations used by the GPU.
+ *
+ */
+
+static bool amdgpu_bo_need_backup(struct amdgpu_device *adev)
+{
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	if (amdgpu_gpu_recovery == 0 ||
+	    (amdgpu_gpu_recovery == -1  && !amdgpu_sriov_vf(adev)))
+		return false;
+
+	return true;
+}
+
+/**
+ * amdgpu_bo_subtract_pin_size - Remove BO from pin_size accounting
+ *
+ * @bo: &amdgpu_bo buffer object
+ *
+ * This function is called when a BO stops being pinned, and updates the
+ * &amdgpu_device pin_size values accordingly.
+ */
+static void amdgpu_bo_subtract_pin_size(struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+
+	if (bo->tbo.mem.mem_type == TTM_PL_VRAM) {
+		atomic64_sub(amdgpu_bo_size(bo), &adev->vram_pin_size);
+		atomic64_sub(amdgpu_vram_mgr_bo_visible_size(bo),
+			     &adev->visible_pin_size);
+	} else if (bo->tbo.mem.mem_type == TTM_PL_TT) {
+		atomic64_sub(amdgpu_bo_size(bo), &adev->gart_pin_size);
+	}
+}
+
+static void amdgpu_bo_destroy(struct ttm_buffer_object *tbo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(tbo->bdev);
+	struct amdgpu_bo *bo = ttm_to_amdgpu_bo(tbo);
+
+	if (bo->pin_count > 0)
+		amdgpu_bo_subtract_pin_size(bo);
+
+	if (bo->kfd_bo)
+		amdgpu_amdkfd_unreserve_system_memory_limit(bo);
+
+	amdgpu_bo_kunmap(bo);
+
+	if (bo->gem_base.import_attach)
+		drm_prime_gem_destroy(&bo->gem_base, bo->tbo.sg);
+	drm_gem_object_release(&bo->gem_base);
+	amdgpu_bo_unref(&bo->parent);
+	if (!list_empty(&bo->shadow_list)) {
+		mutex_lock(&adev->shadow_list_lock);
+		list_del_init(&bo->shadow_list);
+		mutex_unlock(&adev->shadow_list_lock);
+	}
+	kfree(bo->metadata);
+	pool_put(&bo->adev->ddev->objpl, bo);
+}
+
+/**
+ * amdgpu_bo_is_amdgpu_bo - check if the buffer object is an &amdgpu_bo
+ * @bo: buffer object to be checked
+ *
+ * Uses destroy function associated with the object to determine if this is
+ * an &amdgpu_bo.
+ *
+ * Returns:
+ * true if the object belongs to &amdgpu_bo, false if not.
+ */
+bool amdgpu_bo_is_amdgpu_bo(struct ttm_buffer_object *bo)
+{
+	if (bo->destroy == &amdgpu_bo_destroy)
+		return true;
+	return false;
+}
+
+/**
+ * amdgpu_bo_placement_from_domain - set buffer's placement
+ * @abo: &amdgpu_bo buffer object whose placement is to be set
+ * @domain: requested domain
+ *
+ * Sets buffer's placement according to requested domain and the buffer's
+ * flags.
+ */
+void amdgpu_bo_placement_from_domain(struct amdgpu_bo *abo, u32 domain)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev);
+	struct ttm_placement *placement = &abo->placement;
+	struct ttm_place *places = abo->placements;
+	u64 flags = abo->flags;
+	u32 c = 0;
+
+	if (domain & AMDGPU_GEM_DOMAIN_VRAM) {
+		unsigned visible_pfn = adev->gmc.visible_vram_size >> PAGE_SHIFT;
+
+		places[c].fpfn = 0;
+		places[c].lpfn = 0;
+		places[c].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED |
+			TTM_PL_FLAG_VRAM;
+
+		if (flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)
+			places[c].lpfn = visible_pfn;
+		else
+			places[c].flags |= TTM_PL_FLAG_TOPDOWN;
+
+		if (flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS)
+			places[c].flags |= TTM_PL_FLAG_CONTIGUOUS;
+		c++;
+	}
+
+	if (domain & AMDGPU_GEM_DOMAIN_GTT) {
+		places[c].fpfn = 0;
+		if (flags & AMDGPU_GEM_CREATE_SHADOW)
+			places[c].lpfn = adev->gmc.gart_size >> PAGE_SHIFT;
+		else
+			places[c].lpfn = 0;
+		places[c].flags = TTM_PL_FLAG_TT;
+		if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC)
+			places[c].flags |= TTM_PL_FLAG_WC |
+				TTM_PL_FLAG_UNCACHED;
+		else
+			places[c].flags |= TTM_PL_FLAG_CACHED;
+		c++;
+	}
+
+	if (domain & AMDGPU_GEM_DOMAIN_CPU) {
+		places[c].fpfn = 0;
+		places[c].lpfn = 0;
+		places[c].flags = TTM_PL_FLAG_SYSTEM;
+		if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC)
+			places[c].flags |= TTM_PL_FLAG_WC |
+				TTM_PL_FLAG_UNCACHED;
+		else
+			places[c].flags |= TTM_PL_FLAG_CACHED;
+		c++;
+	}
+
+	if (domain & AMDGPU_GEM_DOMAIN_GDS) {
+		places[c].fpfn = 0;
+		places[c].lpfn = 0;
+		places[c].flags = TTM_PL_FLAG_UNCACHED | AMDGPU_PL_FLAG_GDS;
+		c++;
+	}
+
+	if (domain & AMDGPU_GEM_DOMAIN_GWS) {
+		places[c].fpfn = 0;
+		places[c].lpfn = 0;
+		places[c].flags = TTM_PL_FLAG_UNCACHED | AMDGPU_PL_FLAG_GWS;
+		c++;
+	}
+
+	if (domain & AMDGPU_GEM_DOMAIN_OA) {
+		places[c].fpfn = 0;
+		places[c].lpfn = 0;
+		places[c].flags = TTM_PL_FLAG_UNCACHED | AMDGPU_PL_FLAG_OA;
+		c++;
+	}
+
+	if (!c) {
+		places[c].fpfn = 0;
+		places[c].lpfn = 0;
+		places[c].flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM;
+		c++;
+	}
+
+	BUG_ON(c >= AMDGPU_BO_MAX_PLACEMENTS);
+
+	placement->num_placement = c;
+	placement->placement = places;
+
+	placement->num_busy_placement = c;
+	placement->busy_placement = places;
+}
+
+/**
+ * amdgpu_bo_create_reserved - create reserved BO for kernel use
+ *
+ * @adev: amdgpu device object
+ * @size: size for the new BO
+ * @align: alignment for the new BO
+ * @domain: where to place it
+ * @bo_ptr: used to initialize BOs in structures
+ * @gpu_addr: GPU addr of the pinned BO
+ * @cpu_addr: optional CPU address mapping
+ *
+ * Allocates and pins a BO for kernel internal use, and returns it still
+ * reserved.
+ *
+ * Note: For bo_ptr new BO is only created if bo_ptr points to NULL.
+ *
+ * Returns:
+ * 0 on success, negative error code otherwise.
+ */
+int amdgpu_bo_create_reserved(struct amdgpu_device *adev,
+			      unsigned long size, int align,
+			      u32 domain, struct amdgpu_bo **bo_ptr,
+			      u64 *gpu_addr, void **cpu_addr)
+{
+	struct amdgpu_bo_param bp;
+	bool free = false;
+	int r;
+
+	memset(&bp, 0, sizeof(bp));
+	bp.size = size;
+	bp.byte_align = align;
+	bp.domain = domain;
+	bp.flags = AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
+		AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	bp.type = ttm_bo_type_kernel;
+	bp.resv = NULL;
+
+	if (!*bo_ptr) {
+		r = amdgpu_bo_create(adev, &bp, bo_ptr);
+		if (r) {
+			dev_err(adev->dev, "(%d) failed to allocate kernel bo\n",
+				r);
+			return r;
+		}
+		free = true;
+	}
+
+	r = amdgpu_bo_reserve(*bo_ptr, false);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to reserve kernel bo\n", r);
+		goto error_free;
+	}
+
+	r = amdgpu_bo_pin(*bo_ptr, domain);
+	if (r) {
+		dev_err(adev->dev, "(%d) kernel bo pin failed\n", r);
+		goto error_unreserve;
+	}
+
+	r = amdgpu_ttm_alloc_gart(&(*bo_ptr)->tbo);
+	if (r) {
+		dev_err(adev->dev, "%p bind failed\n", *bo_ptr);
+		goto error_unpin;
+	}
+
+	if (gpu_addr)
+		*gpu_addr = amdgpu_bo_gpu_offset(*bo_ptr);
+
+	if (cpu_addr) {
+		r = amdgpu_bo_kmap(*bo_ptr, cpu_addr);
+		if (r) {
+			dev_err(adev->dev, "(%d) kernel bo map failed\n", r);
+			goto error_unpin;
+		}
+	}
+
+	return 0;
+
+error_unpin:
+	amdgpu_bo_unpin(*bo_ptr);
+error_unreserve:
+	amdgpu_bo_unreserve(*bo_ptr);
+
+error_free:
+	if (free)
+		amdgpu_bo_unref(bo_ptr);
+
+	return r;
+}
+
+/**
+ * amdgpu_bo_create_kernel - create BO for kernel use
+ *
+ * @adev: amdgpu device object
+ * @size: size for the new BO
+ * @align: alignment for the new BO
+ * @domain: where to place it
+ * @bo_ptr:  used to initialize BOs in structures
+ * @gpu_addr: GPU addr of the pinned BO
+ * @cpu_addr: optional CPU address mapping
+ *
+ * Allocates and pins a BO for kernel internal use.
+ *
+ * Note: For bo_ptr new BO is only created if bo_ptr points to NULL.
+ *
+ * Returns:
+ * 0 on success, negative error code otherwise.
+ */
+int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
+			    unsigned long size, int align,
+			    u32 domain, struct amdgpu_bo **bo_ptr,
+			    u64 *gpu_addr, void **cpu_addr)
+{
+	int r;
+
+	r = amdgpu_bo_create_reserved(adev, size, align, domain, bo_ptr,
+				      gpu_addr, cpu_addr);
+
+	if (r)
+		return r;
+
+	amdgpu_bo_unreserve(*bo_ptr);
+
+	return 0;
+}
+
+/**
+ * amdgpu_bo_free_kernel - free BO for kernel use
+ *
+ * @bo: amdgpu BO to free
+ * @gpu_addr: pointer to where the BO's GPU memory space address was stored
+ * @cpu_addr: pointer to where the BO's CPU memory space address was stored
+ *
+ * unmaps and unpin a BO for kernel internal use.
+ */
+void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr,
+			   void **cpu_addr)
+{
+	if (*bo == NULL)
+		return;
+
+	if (likely(amdgpu_bo_reserve(*bo, true) == 0)) {
+		if (cpu_addr)
+			amdgpu_bo_kunmap(*bo);
+
+		amdgpu_bo_unpin(*bo);
+		amdgpu_bo_unreserve(*bo);
+	}
+	amdgpu_bo_unref(bo);
+
+	if (gpu_addr)
+		*gpu_addr = 0;
+
+	if (cpu_addr)
+		*cpu_addr = NULL;
+}
+
+/* Validate bo size is bit bigger then the request domain */
+static bool amdgpu_bo_validate_size(struct amdgpu_device *adev,
+					  unsigned long size, u32 domain)
+{
+	struct ttm_mem_type_manager *man = NULL;
+
+	/*
+	 * If GTT is part of requested domains the check must succeed to
+	 * allow fall back to GTT
+	 */
+	if (domain & AMDGPU_GEM_DOMAIN_GTT) {
+		man = &adev->mman.bdev.man[TTM_PL_TT];
+
+		if (size < (man->size << PAGE_SHIFT))
+			return true;
+		else
+			goto fail;
+	}
+
+	if (domain & AMDGPU_GEM_DOMAIN_VRAM) {
+		man = &adev->mman.bdev.man[TTM_PL_VRAM];
+
+		if (size < (man->size << PAGE_SHIFT))
+			return true;
+		else
+			goto fail;
+	}
+
+
+	/* TODO add more domains checks, such as AMDGPU_GEM_DOMAIN_CPU */
+	return true;
+
+fail:
+	DRM_DEBUG("BO size %lu > total memory in domain: %llu\n", size,
+		  man->size << PAGE_SHIFT);
+	return false;
+}
+
+#ifdef __amd64__
+#define CONFIG_X86	1
+#define CONFIG_X86_64	1
+#define CONFIG_X86_PAT	1
+#endif
+
+#ifdef __i386__
+#define CONFIG_X86	1
+#define CONFIG_X86_32	1
+#define CONFIG_X86_PAT	1
+#endif
+
+static int amdgpu_bo_do_create(struct amdgpu_device *adev,
+			       struct amdgpu_bo_param *bp,
+			       struct amdgpu_bo **bo_ptr)
+{
+	struct ttm_operation_ctx ctx = {
+		.interruptible = (bp->type != ttm_bo_type_kernel),
+		.no_wait_gpu = false,
+		.resv = bp->resv,
+		.flags = TTM_OPT_FLAG_ALLOW_RES_EVICT
+	};
+	struct amdgpu_bo *bo;
+	unsigned long page_align, size = bp->size;
+	size_t acc_size;
+	int r;
+
+	page_align = roundup(bp->byte_align, PAGE_SIZE) >> PAGE_SHIFT;
+	size = roundup2(size, PAGE_SIZE);
+
+	if (!amdgpu_bo_validate_size(adev, size, bp->domain))
+		return -ENOMEM;
+
+	*bo_ptr = NULL;
+
+	acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size,
+				       sizeof(struct amdgpu_bo));
+
+	bo = pool_get(&adev->ddev->objpl, PR_WAITOK | PR_ZERO);
+	if (bo == NULL)
+		return -ENOMEM;
+	drm_gem_private_object_init(adev->ddev, &bo->gem_base, size);
+	bo->adev = adev;
+	INIT_LIST_HEAD(&bo->shadow_list);
+	INIT_LIST_HEAD(&bo->va);
+	bo->preferred_domains = bp->preferred_domain ? bp->preferred_domain :
+		bp->domain;
+	bo->allowed_domains = bo->preferred_domains;
+	if (bp->type != ttm_bo_type_kernel &&
+	    bo->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
+		bo->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
+
+	bo->flags = bp->flags;
+
+#ifdef CONFIG_X86_32
+	/* XXX: Write-combined CPU mappings of GTT seem broken on 32-bit
+	 * See https://bugs.freedesktop.org/show_bug.cgi?id=84627
+	 */
+	bo->flags &= ~AMDGPU_GEM_CREATE_CPU_GTT_USWC;
+#elif defined(CONFIG_X86) && !defined(CONFIG_X86_PAT)
+	/* Don't try to enable write-combining when it can't work, or things
+	 * may be slow
+	 * See https://bugs.freedesktop.org/show_bug.cgi?id=88758
+	 */
+
+#ifndef CONFIG_COMPILE_TEST
+#warning Please enable CONFIG_MTRR and CONFIG_X86_PAT for better performance \
+	 thanks to write-combining
+#endif
+
+	if (bo->flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC)
+		DRM_INFO_ONCE("Please enable CONFIG_MTRR and CONFIG_X86_PAT for "
+			      "better performance thanks to write-combining\n");
+	bo->flags &= ~AMDGPU_GEM_CREATE_CPU_GTT_USWC;
+#else
+	/* For architectures that don't support WC memory,
+	 * mask out the WC flag from the BO
+	 */
+	if (!drm_arch_can_wc_memory())
+		bo->flags &= ~AMDGPU_GEM_CREATE_CPU_GTT_USWC;
+#endif
+
+	bo->tbo.bdev = &adev->mman.bdev;
+	amdgpu_bo_placement_from_domain(bo, bp->domain);
+	if (bp->type == ttm_bo_type_kernel)
+		bo->tbo.priority = 1;
+
+	r = ttm_bo_init_reserved(&adev->mman.bdev, &bo->tbo, size, bp->type,
+				 &bo->placement, page_align, &ctx, acc_size,
+				 NULL, bp->resv, &amdgpu_bo_destroy);
+	if (unlikely(r != 0))
+		return r;
+
+	if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+	    bo->tbo.mem.mem_type == TTM_PL_VRAM &&
+	    bo->tbo.mem.start < adev->gmc.visible_vram_size >> PAGE_SHIFT)
+		amdgpu_cs_report_moved_bytes(adev, ctx.bytes_moved,
+					     ctx.bytes_moved);
+	else
+		amdgpu_cs_report_moved_bytes(adev, ctx.bytes_moved, 0);
+
+	if (bp->flags & AMDGPU_GEM_CREATE_VRAM_CLEARED &&
+	    bo->tbo.mem.placement & TTM_PL_FLAG_VRAM) {
+		struct dma_fence *fence;
+
+		r = amdgpu_fill_buffer(bo, 0, bo->tbo.resv, &fence);
+		if (unlikely(r))
+			goto fail_unreserve;
+
+		amdgpu_bo_fence(bo, fence, false);
+		dma_fence_put(bo->tbo.moving);
+		bo->tbo.moving = dma_fence_get(fence);
+		dma_fence_put(fence);
+	}
+	if (!bp->resv)
+		amdgpu_bo_unreserve(bo);
+	*bo_ptr = bo;
+
+	trace_amdgpu_bo_create(bo);
+
+	/* Treat CPU_ACCESS_REQUIRED only as a hint if given by UMD */
+	if (bp->type == ttm_bo_type_device)
+		bo->flags &= ~AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+
+	return 0;
+
+fail_unreserve:
+	if (!bp->resv)
+		ww_mutex_unlock(&bo->tbo.resv->lock);
+	amdgpu_bo_unref(&bo);
+	return r;
+}
+
+static int amdgpu_bo_create_shadow(struct amdgpu_device *adev,
+				   unsigned long size, int byte_align,
+				   struct amdgpu_bo *bo)
+{
+	struct amdgpu_bo_param bp;
+	int r;
+
+	if (bo->shadow)
+		return 0;
+
+	memset(&bp, 0, sizeof(bp));
+	bp.size = size;
+	bp.byte_align = byte_align;
+	bp.domain = AMDGPU_GEM_DOMAIN_GTT;
+	bp.flags = AMDGPU_GEM_CREATE_CPU_GTT_USWC |
+		AMDGPU_GEM_CREATE_SHADOW;
+	bp.type = ttm_bo_type_kernel;
+	bp.resv = bo->tbo.resv;
+
+	r = amdgpu_bo_do_create(adev, &bp, &bo->shadow);
+	if (!r) {
+		bo->shadow->parent = amdgpu_bo_ref(bo);
+		mutex_lock(&adev->shadow_list_lock);
+		list_add_tail(&bo->shadow_list, &adev->shadow_list);
+		mutex_unlock(&adev->shadow_list_lock);
+	}
+
+	return r;
+}
+
+/**
+ * amdgpu_bo_create - create an &amdgpu_bo buffer object
+ * @adev: amdgpu device object
+ * @bp: parameters to be used for the buffer object
+ * @bo_ptr: pointer to the buffer object pointer
+ *
+ * Creates an &amdgpu_bo buffer object; and if requested, also creates a
+ * shadow object.
+ * Shadow object is used to backup the original buffer object, and is always
+ * in GTT.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_create(struct amdgpu_device *adev,
+		     struct amdgpu_bo_param *bp,
+		     struct amdgpu_bo **bo_ptr)
+{
+	u64 flags = bp->flags;
+	int r;
+
+	bp->flags = bp->flags & ~AMDGPU_GEM_CREATE_SHADOW;
+	r = amdgpu_bo_do_create(adev, bp, bo_ptr);
+	if (r)
+		return r;
+
+	if ((flags & AMDGPU_GEM_CREATE_SHADOW) && amdgpu_bo_need_backup(adev)) {
+		if (!bp->resv)
+			WARN_ON(reservation_object_lock((*bo_ptr)->tbo.resv,
+							NULL));
+
+		r = amdgpu_bo_create_shadow(adev, bp->size, bp->byte_align, (*bo_ptr));
+
+		if (!bp->resv)
+			reservation_object_unlock((*bo_ptr)->tbo.resv);
+
+		if (r)
+			amdgpu_bo_unref(bo_ptr);
+	}
+
+	return r;
+}
+
+/**
+ * amdgpu_bo_backup_to_shadow - Backs up an &amdgpu_bo buffer object
+ * @adev: amdgpu device object
+ * @ring: amdgpu_ring for the engine handling the buffer operations
+ * @bo: &amdgpu_bo buffer to be backed up
+ * @resv: reservation object with embedded fence
+ * @fence: dma_fence associated with the operation
+ * @direct: whether to submit the job directly
+ *
+ * Copies an &amdgpu_bo buffer object to its shadow object.
+ * Not used for now.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_backup_to_shadow(struct amdgpu_device *adev,
+			       struct amdgpu_ring *ring,
+			       struct amdgpu_bo *bo,
+			       struct reservation_object *resv,
+			       struct dma_fence **fence,
+			       bool direct)
+
+{
+	struct amdgpu_bo *shadow = bo->shadow;
+	uint64_t bo_addr, shadow_addr;
+	int r;
+
+	if (!shadow)
+		return -EINVAL;
+
+	bo_addr = amdgpu_bo_gpu_offset(bo);
+	shadow_addr = amdgpu_bo_gpu_offset(bo->shadow);
+
+	r = reservation_object_reserve_shared(bo->tbo.resv);
+	if (r)
+		goto err;
+
+	r = amdgpu_copy_buffer(ring, bo_addr, shadow_addr,
+			       amdgpu_bo_size(bo), resv, fence,
+			       direct, false);
+	if (!r)
+		amdgpu_bo_fence(bo, *fence, true);
+
+err:
+	return r;
+}
+
+/**
+ * amdgpu_bo_validate - validate an &amdgpu_bo buffer object
+ * @bo: pointer to the buffer object
+ *
+ * Sets placement according to domain; and changes placement and caching
+ * policy of the buffer object according to the placement.
+ * This is used for validating shadow bos.  It calls ttm_bo_validate() to
+ * make sure the buffer is resident where it needs to be.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_validate(struct amdgpu_bo *bo)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+	uint32_t domain;
+	int r;
+
+	if (bo->pin_count)
+		return 0;
+
+	domain = bo->preferred_domains;
+
+retry:
+	amdgpu_bo_placement_from_domain(bo, domain);
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains) {
+		domain = bo->allowed_domains;
+		goto retry;
+	}
+
+	return r;
+}
+
+/**
+ * amdgpu_bo_restore_from_shadow - restore an &amdgpu_bo buffer object
+ * @adev: amdgpu device object
+ * @ring: amdgpu_ring for the engine handling the buffer operations
+ * @bo: &amdgpu_bo buffer to be restored
+ * @resv: reservation object with embedded fence
+ * @fence: dma_fence associated with the operation
+ * @direct: whether to submit the job directly
+ *
+ * Copies a buffer object's shadow content back to the object.
+ * This is used for recovering a buffer from its shadow in case of a gpu
+ * reset where vram context may be lost.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_restore_from_shadow(struct amdgpu_device *adev,
+				  struct amdgpu_ring *ring,
+				  struct amdgpu_bo *bo,
+				  struct reservation_object *resv,
+				  struct dma_fence **fence,
+				  bool direct)
+
+{
+	struct amdgpu_bo *shadow = bo->shadow;
+	uint64_t bo_addr, shadow_addr;
+	int r;
+
+	if (!shadow)
+		return -EINVAL;
+
+	bo_addr = amdgpu_bo_gpu_offset(bo);
+	shadow_addr = amdgpu_bo_gpu_offset(bo->shadow);
+
+	r = reservation_object_reserve_shared(bo->tbo.resv);
+	if (r)
+		goto err;
+
+	r = amdgpu_copy_buffer(ring, shadow_addr, bo_addr,
+			       amdgpu_bo_size(bo), resv, fence,
+			       direct, false);
+	if (!r)
+		amdgpu_bo_fence(bo, *fence, true);
+
+err:
+	return r;
+}
+
+/**
+ * amdgpu_bo_kmap - map an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object to be mapped
+ * @ptr: kernel virtual address to be returned
+ *
+ * Calls ttm_bo_kmap() to set up the kernel virtual mapping; calls
+ * amdgpu_bo_kptr() to get the kernel virtual address.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr)
+{
+	void *kptr;
+	long r;
+
+	if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)
+		return -EPERM;
+
+	kptr = amdgpu_bo_kptr(bo);
+	if (kptr) {
+		if (ptr)
+			*ptr = kptr;
+		return 0;
+	}
+
+	r = reservation_object_wait_timeout_rcu(bo->tbo.resv, false, false,
+						MAX_SCHEDULE_TIMEOUT);
+	if (r < 0)
+		return r;
+
+	r = ttm_bo_kmap(&bo->tbo, 0, bo->tbo.num_pages, &bo->kmap);
+	if (r)
+		return r;
+
+	if (ptr)
+		*ptr = amdgpu_bo_kptr(bo);
+
+	return 0;
+}
+
+/**
+ * amdgpu_bo_kptr - returns a kernel virtual address of the buffer object
+ * @bo: &amdgpu_bo buffer object
+ *
+ * Calls ttm_kmap_obj_virtual() to get the kernel virtual address
+ *
+ * Returns:
+ * the virtual address of a buffer object area.
+ */
+void *amdgpu_bo_kptr(struct amdgpu_bo *bo)
+{
+	bool is_iomem;
+
+	return ttm_kmap_obj_virtual(&bo->kmap, &is_iomem);
+}
+
+/**
+ * amdgpu_bo_kunmap - unmap an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object to be unmapped
+ *
+ * Unmaps a kernel map set up by amdgpu_bo_kmap().
+ */
+void amdgpu_bo_kunmap(struct amdgpu_bo *bo)
+{
+	if (bo->kmap.bo)
+		ttm_bo_kunmap(&bo->kmap);
+}
+
+/**
+ * amdgpu_bo_ref - reference an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object
+ *
+ * References the contained &ttm_buffer_object.
+ *
+ * Returns:
+ * a refcounted pointer to the &amdgpu_bo buffer object.
+ */
+struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo)
+{
+	if (bo == NULL)
+		return NULL;
+
+	ttm_bo_get(&bo->tbo);
+	return bo;
+}
+
+/**
+ * amdgpu_bo_unref - unreference an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object
+ *
+ * Unreferences the contained &ttm_buffer_object and clear the pointer
+ */
+void amdgpu_bo_unref(struct amdgpu_bo **bo)
+{
+	struct ttm_buffer_object *tbo;
+
+	if ((*bo) == NULL)
+		return;
+
+	tbo = &((*bo)->tbo);
+	ttm_bo_put(tbo);
+	*bo = NULL;
+}
+
+/**
+ * amdgpu_bo_pin_restricted - pin an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object to be pinned
+ * @domain: domain to be pinned to
+ * @min_offset: the start of requested address range
+ * @max_offset: the end of requested address range
+ *
+ * Pins the buffer object according to requested domain and address range. If
+ * the memory is unbound gart memory, binds the pages into gart table. Adjusts
+ * pin_count and pin_size accordingly.
+ *
+ * Pinning means to lock pages in memory along with keeping them at a fixed
+ * offset. It is required when a buffer can not be moved, for example, when
+ * a display buffer is being scanned out.
+ *
+ * Compared with amdgpu_bo_pin(), this function gives more flexibility on
+ * where to pin a buffer if there are specific restrictions on where a buffer
+ * must be located.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
+			     u64 min_offset, u64 max_offset)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct ttm_operation_ctx ctx = { false, false };
+	int r, i;
+
+	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm))
+		return -EPERM;
+
+	if (WARN_ON_ONCE(min_offset > max_offset))
+		return -EINVAL;
+
+	/* A shared bo cannot be migrated to VRAM */
+	if (bo->prime_shared_count) {
+		if (domain & AMDGPU_GEM_DOMAIN_GTT)
+			domain = AMDGPU_GEM_DOMAIN_GTT;
+		else
+			return -EINVAL;
+	}
+
+	/* This assumes only APU display buffers are pinned with (VRAM|GTT).
+	 * See function amdgpu_display_supported_domains()
+	 */
+	domain = amdgpu_bo_get_preferred_pin_domain(adev, domain);
+
+	if (bo->pin_count) {
+		uint32_t mem_type = bo->tbo.mem.mem_type;
+
+		if (!(domain & amdgpu_mem_type_to_domain(mem_type)))
+			return -EINVAL;
+
+		bo->pin_count++;
+
+		if (max_offset != 0) {
+			u64 domain_start = bo->tbo.bdev->man[mem_type].gpu_offset;
+			WARN_ON_ONCE(max_offset <
+				     (amdgpu_bo_gpu_offset(bo) - domain_start));
+		}
+
+		return 0;
+	}
+
+	bo->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	/* force to pin into visible video ram */
+	if (!(bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS))
+		bo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+	amdgpu_bo_placement_from_domain(bo, domain);
+	for (i = 0; i < bo->placement.num_placement; i++) {
+		unsigned fpfn, lpfn;
+
+		fpfn = min_offset >> PAGE_SHIFT;
+		lpfn = max_offset >> PAGE_SHIFT;
+
+		if (fpfn > bo->placements[i].fpfn)
+			bo->placements[i].fpfn = fpfn;
+		if (!bo->placements[i].lpfn ||
+		    (lpfn && lpfn < bo->placements[i].lpfn))
+			bo->placements[i].lpfn = lpfn;
+		bo->placements[i].flags |= TTM_PL_FLAG_NO_EVICT;
+	}
+
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (unlikely(r)) {
+		dev_err(adev->dev, "%p pin failed\n", bo);
+		goto error;
+	}
+
+	bo->pin_count = 1;
+
+	domain = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type);
+	if (domain == AMDGPU_GEM_DOMAIN_VRAM) {
+		atomic64_add(amdgpu_bo_size(bo), &adev->vram_pin_size);
+		atomic64_add(amdgpu_vram_mgr_bo_visible_size(bo),
+			     &adev->visible_pin_size);
+	} else if (domain == AMDGPU_GEM_DOMAIN_GTT) {
+		atomic64_add(amdgpu_bo_size(bo), &adev->gart_pin_size);
+	}
+
+error:
+	return r;
+}
+
+/**
+ * amdgpu_bo_pin - pin an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object to be pinned
+ * @domain: domain to be pinned to
+ *
+ * A simple wrapper to amdgpu_bo_pin_restricted().
+ * Provides a simpler API for buffers that do not have any strict restrictions
+ * on where a buffer must be located.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain)
+{
+	return amdgpu_bo_pin_restricted(bo, domain, 0, 0);
+}
+
+/**
+ * amdgpu_bo_unpin - unpin an &amdgpu_bo buffer object
+ * @bo: &amdgpu_bo buffer object to be unpinned
+ *
+ * Decreases the pin_count, and clears the flags if pin_count reaches 0.
+ * Changes placement and pin size accordingly.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_unpin(struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct ttm_operation_ctx ctx = { false, false };
+	int r, i;
+
+	if (!bo->pin_count) {
+		dev_warn(adev->dev, "%p unpin not necessary\n", bo);
+		return 0;
+	}
+	bo->pin_count--;
+	if (bo->pin_count)
+		return 0;
+
+	amdgpu_bo_subtract_pin_size(bo);
+
+	for (i = 0; i < bo->placement.num_placement; i++) {
+		bo->placements[i].lpfn = 0;
+		bo->placements[i].flags &= ~TTM_PL_FLAG_NO_EVICT;
+	}
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (unlikely(r))
+		dev_err(adev->dev, "%p validate failed for unpin\n", bo);
+
+	return r;
+}
+
+/**
+ * amdgpu_bo_evict_vram - evict VRAM buffers
+ * @adev: amdgpu device object
+ *
+ * Evicts all VRAM buffers on the lru list of the memory type.
+ * Mainly used for evicting vram at suspend time.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_evict_vram(struct amdgpu_device *adev)
+{
+	/* late 2.6.33 fix IGP hibernate - we need pm ops to do this correct */
+	if (0 && (adev->flags & AMD_IS_APU)) {
+		/* Useless to evict on IGP chips */
+		return 0;
+	}
+	return ttm_bo_evict_mm(&adev->mman.bdev, TTM_PL_VRAM);
+}
+
+static const char *amdgpu_vram_names[] = {
+	"UNKNOWN",
+	"GDDR1",
+	"DDR2",
+	"GDDR3",
+	"GDDR4",
+	"GDDR5",
+	"HBM",
+	"DDR3",
+	"DDR4",
+};
+
+/**
+ * amdgpu_bo_init - initialize memory manager
+ * @adev: amdgpu device object
+ *
+ * Calls amdgpu_ttm_init() to initialize amdgpu memory manager.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_init(struct amdgpu_device *adev)
+{
+	paddr_t start, end;
+
+#ifdef __linux__
+	/* reserve PAT memory space to WC for VRAM */
+	arch_io_reserve_memtype_wc(adev->gmc.aper_base,
+				   adev->gmc.aper_size);
+
+	/* Add an MTRR for the VRAM */
+	adev->gmc.vram_mtrr = arch_phys_wc_add(adev->gmc.aper_base,
+					      adev->gmc.aper_size);
+#else
+	drm_mtrr_add(adev->gmc.aper_base, adev->gmc.aper_size, DRM_MTRR_WC);
+
+	start = atop(bus_space_mmap(adev->memt, adev->gmc.aper_base, 0, 0, 0));
+	end = start + atop(adev->gmc.aper_size);
+	uvm_page_physload(start, end, start, end, PHYSLOAD_DEVICE);
+#endif
+	DRM_INFO("Detected VRAM RAM=%lluM, BAR=%lluM\n",
+		 adev->gmc.mc_vram_size >> 20,
+		 (unsigned long long)adev->gmc.aper_size >> 20);
+	DRM_INFO("RAM width %dbits %s\n",
+		 adev->gmc.vram_width, amdgpu_vram_names[adev->gmc.vram_type]);
+	return amdgpu_ttm_init(adev);
+}
+
+/**
+ * amdgpu_bo_late_init - late init
+ * @adev: amdgpu device object
+ *
+ * Calls amdgpu_ttm_late_init() to free resources used earlier during
+ * initialization.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_late_init(struct amdgpu_device *adev)
+{
+	amdgpu_ttm_late_init(adev);
+
+	return 0;
+}
+
+/**
+ * amdgpu_bo_fini - tear down memory manager
+ * @adev: amdgpu device object
+ *
+ * Reverses amdgpu_bo_init() to tear down memory manager.
+ */
+void amdgpu_bo_fini(struct amdgpu_device *adev)
+{
+	amdgpu_ttm_fini(adev);
+#ifdef __linux__
+	arch_phys_wc_del(adev->gmc.vram_mtrr);
+	arch_io_free_memtype_wc(adev->gmc.aper_base, adev->gmc.aper_size);
+#else
+	drm_mtrr_del(0, adev->gmc.aper_base, adev->gmc.aper_size, DRM_MTRR_WC);
+#endif
+}
+
+#ifdef notyet
+/**
+ * amdgpu_bo_fbdev_mmap - mmap fbdev memory
+ * @bo: &amdgpu_bo buffer object
+ * @vma: vma as input from the fbdev mmap method
+ *
+ * Calls ttm_fbdev_mmap() to mmap fbdev memory if it is backed by a bo.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_fbdev_mmap(struct amdgpu_bo *bo,
+			     struct vm_area_struct *vma)
+{
+	return ttm_fbdev_mmap(vma, &bo->tbo);
+}
+#endif
+
+/**
+ * amdgpu_bo_set_tiling_flags - set tiling flags
+ * @bo: &amdgpu_bo buffer object
+ * @tiling_flags: new flags
+ *
+ * Sets buffer object's tiling flags with the new one. Used by GEM ioctl or
+ * kernel driver to set the tiling flags on a buffer.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_set_tiling_flags(struct amdgpu_bo *bo, u64 tiling_flags)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+
+	if (adev->family <= AMDGPU_FAMILY_CZ &&
+	    AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT) > 6)
+		return -EINVAL;
+
+	bo->tiling_flags = tiling_flags;
+	return 0;
+}
+
+/**
+ * amdgpu_bo_get_tiling_flags - get tiling flags
+ * @bo: &amdgpu_bo buffer object
+ * @tiling_flags: returned flags
+ *
+ * Gets buffer object's tiling flags. Used by GEM ioctl or kernel driver to
+ * set the tiling flags on a buffer.
+ */
+void amdgpu_bo_get_tiling_flags(struct amdgpu_bo *bo, u64 *tiling_flags)
+{
+	lockdep_assert_held(&bo->tbo.resv->lock.base);
+
+	if (tiling_flags)
+		*tiling_flags = bo->tiling_flags;
+}
+
+/**
+ * amdgpu_bo_set_metadata - set metadata
+ * @bo: &amdgpu_bo buffer object
+ * @metadata: new metadata
+ * @metadata_size: size of the new metadata
+ * @flags: flags of the new metadata
+ *
+ * Sets buffer object's metadata, its size and flags.
+ * Used via GEM ioctl.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_set_metadata (struct amdgpu_bo *bo, void *metadata,
+			    uint32_t metadata_size, uint64_t flags)
+{
+	void *buffer;
+
+	if (!metadata_size) {
+		if (bo->metadata_size) {
+			kfree(bo->metadata);
+			bo->metadata = NULL;
+			bo->metadata_size = 0;
+		}
+		return 0;
+	}
+
+	if (metadata == NULL)
+		return -EINVAL;
+
+	buffer = kmemdup(metadata, metadata_size, GFP_KERNEL);
+	if (buffer == NULL)
+		return -ENOMEM;
+
+	kfree(bo->metadata);
+	bo->metadata_flags = flags;
+	bo->metadata = buffer;
+	bo->metadata_size = metadata_size;
+
+	return 0;
+}
+
+/**
+ * amdgpu_bo_get_metadata - get metadata
+ * @bo: &amdgpu_bo buffer object
+ * @buffer: returned metadata
+ * @buffer_size: size of the buffer
+ * @metadata_size: size of the returned metadata
+ * @flags: flags of the returned metadata
+ *
+ * Gets buffer object's metadata, its size and flags. buffer_size shall not be
+ * less than metadata_size.
+ * Used via GEM ioctl.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_get_metadata(struct amdgpu_bo *bo, void *buffer,
+			   size_t buffer_size, uint32_t *metadata_size,
+			   uint64_t *flags)
+{
+	if (!buffer && !metadata_size)
+		return -EINVAL;
+
+	if (buffer) {
+		if (buffer_size < bo->metadata_size)
+			return -EINVAL;
+
+		if (bo->metadata_size)
+			memcpy(buffer, bo->metadata, bo->metadata_size);
+	}
+
+	if (metadata_size)
+		*metadata_size = bo->metadata_size;
+	if (flags)
+		*flags = bo->metadata_flags;
+
+	return 0;
+}
+
+/**
+ * amdgpu_bo_move_notify - notification about a memory move
+ * @bo: pointer to a buffer object
+ * @evict: if this move is evicting the buffer from the graphics address space
+ * @new_mem: new information of the bufer object
+ *
+ * Marks the corresponding &amdgpu_bo buffer object as invalid, also performs
+ * bookkeeping.
+ * TTM driver callback which is called when ttm moves a buffer.
+ */
+void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
+			   bool evict,
+			   struct ttm_mem_reg *new_mem)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
+	struct amdgpu_bo *abo;
+	struct ttm_mem_reg *old_mem = &bo->mem;
+
+	if (!amdgpu_bo_is_amdgpu_bo(bo))
+		return;
+
+	abo = ttm_to_amdgpu_bo(bo);
+	amdgpu_vm_bo_invalidate(adev, abo, evict);
+
+	amdgpu_bo_kunmap(abo);
+
+	/* remember the eviction */
+	if (evict)
+		atomic64_inc(&adev->num_evictions);
+
+	/* update statistics */
+	if (!new_mem)
+		return;
+
+	/* move_notify is called before move happens */
+	trace_amdgpu_bo_move(abo, new_mem->mem_type, old_mem->mem_type);
+}
+
+/**
+ * amdgpu_bo_fault_reserve_notify - notification about a memory fault
+ * @bo: pointer to a buffer object
+ *
+ * Notifies the driver we are taking a fault on this BO and have reserved it,
+ * also performs bookkeeping.
+ * TTM driver callback for dealing with vm faults.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_bo *abo;
+	unsigned long offset, size;
+	int r;
+
+	if (!amdgpu_bo_is_amdgpu_bo(bo))
+		return 0;
+
+	abo = ttm_to_amdgpu_bo(bo);
+
+	/* Remember that this BO was accessed by the CPU */
+	abo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+
+	if (bo->mem.mem_type != TTM_PL_VRAM)
+		return 0;
+
+	size = bo->mem.num_pages << PAGE_SHIFT;
+	offset = bo->mem.start << PAGE_SHIFT;
+	if ((offset + size) <= adev->gmc.visible_vram_size)
+		return 0;
+
+	/* Can't move a pinned BO to visible VRAM */
+	if (abo->pin_count > 0)
+		return -EINVAL;
+
+	/* hurrah the memory is not visible ! */
+	atomic64_inc(&adev->num_vram_cpu_page_faults);
+	amdgpu_bo_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM |
+					AMDGPU_GEM_DOMAIN_GTT);
+
+	/* Avoid costly evictions; only set GTT as a busy placement */
+	abo->placement.num_busy_placement = 1;
+	abo->placement.busy_placement = &abo->placements[1];
+
+	r = ttm_bo_validate(bo, &abo->placement, &ctx);
+	if (unlikely(r != 0))
+		return r;
+
+	offset = bo->mem.start << PAGE_SHIFT;
+	/* this should never happen */
+	if (bo->mem.mem_type == TTM_PL_VRAM &&
+	    (offset + size) > adev->gmc.visible_vram_size)
+		return -EINVAL;
+
+	return 0;
+}
+
+/**
+ * amdgpu_bo_fence - add fence to buffer object
+ *
+ * @bo: buffer object in question
+ * @fence: fence to add
+ * @shared: true if fence should be added shared
+ *
+ */
+void amdgpu_bo_fence(struct amdgpu_bo *bo, struct dma_fence *fence,
+		     bool shared)
+{
+	struct reservation_object *resv = bo->tbo.resv;
+
+	if (shared)
+		reservation_object_add_shared_fence(resv, fence);
+	else
+		reservation_object_add_excl_fence(resv, fence);
+}
+
+/**
+ * amdgpu_bo_gpu_offset - return GPU offset of bo
+ * @bo:	amdgpu object for which we query the offset
+ *
+ * Note: object should either be pinned or reserved when calling this
+ * function, it might be useful to add check for this for debugging.
+ *
+ * Returns:
+ * current GPU offset of the object.
+ */
+u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo)
+{
+	WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_SYSTEM);
+	WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_TT &&
+		     !amdgpu_gtt_mgr_has_gart_addr(&bo->tbo.mem));
+	WARN_ON_ONCE(!ww_mutex_is_locked(&bo->tbo.resv->lock) &&
+		     !bo->pin_count);
+	WARN_ON_ONCE(bo->tbo.mem.start == AMDGPU_BO_INVALID_OFFSET);
+	WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_VRAM &&
+		     !(bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS));
+
+	return bo->tbo.offset;
+}
+
+/**
+ * amdgpu_bo_get_preferred_pin_domain - get preferred domain for scanout
+ * @adev: amdgpu device object
+ * @domain: allowed :ref:`memory domains <amdgpu_memory_domains>`
+ *
+ * Returns:
+ * Which of the allowed domains is preferred for pinning the BO for scanout.
+ */
+uint32_t amdgpu_bo_get_preferred_pin_domain(struct amdgpu_device *adev,
+					    uint32_t domain)
+{
+	if (domain == (AMDGPU_GEM_DOMAIN_VRAM | AMDGPU_GEM_DOMAIN_GTT)) {
+		domain = AMDGPU_GEM_DOMAIN_VRAM;
+		if (adev->gmc.real_vram_size <= AMDGPU_SG_THRESHOLD)
+			domain = AMDGPU_GEM_DOMAIN_GTT;
+	}
+	return domain;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_object.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_object.h
new file mode 100644
index 00000000000..3a8635a027b
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_object.h
@@ -0,0 +1,334 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#ifndef __AMDGPU_OBJECT_H__
+#define __AMDGPU_OBJECT_H__
+
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+#define AMDGPU_BO_INVALID_OFFSET	LONG_MAX
+#define AMDGPU_BO_MAX_PLACEMENTS	3
+
+struct amdgpu_bo_param {
+	unsigned long			size;
+	int				byte_align;
+	u32				domain;
+	u32				preferred_domain;
+	u64				flags;
+	enum ttm_bo_type		type;
+	struct reservation_object	*resv;
+};
+
+/* bo virtual addresses in a vm */
+struct amdgpu_bo_va_mapping {
+	struct amdgpu_bo_va		*bo_va;
+	struct list_head		list;
+	struct rb_node			rb;
+	uint64_t			start;
+	uint64_t			last;
+	uint64_t			__subtree_last;
+	uint64_t			offset;
+	uint64_t			flags;
+};
+
+/* User space allocated BO in a VM */
+struct amdgpu_bo_va {
+	struct amdgpu_vm_bo_base	base;
+
+	/* protected by bo being reserved */
+	unsigned			ref_count;
+
+	/* all other members protected by the VM PD being reserved */
+	struct dma_fence	        *last_pt_update;
+
+	/* mappings for this bo_va */
+	struct list_head		invalids;
+	struct list_head		valids;
+
+	/* If the mappings are cleared or filled */
+	bool				cleared;
+};
+
+struct amdgpu_bo {
+	/* Protected by tbo.reserved */
+	u32				preferred_domains;
+	u32				allowed_domains;
+	struct ttm_place		placements[AMDGPU_BO_MAX_PLACEMENTS];
+	struct ttm_placement		placement;
+	struct ttm_buffer_object	tbo;
+	struct ttm_bo_kmap_obj		kmap;
+	u64				flags;
+	unsigned			pin_count;
+	u64				tiling_flags;
+	u64				metadata_flags;
+	void				*metadata;
+	u32				metadata_size;
+	unsigned			prime_shared_count;
+	/* list of all virtual address to which this bo is associated to */
+	struct list_head		va;
+	/* Constant after initialization */
+	struct amdgpu_device		*adev;
+	struct drm_gem_object		gem_base;
+	struct amdgpu_bo		*parent;
+	struct amdgpu_bo		*shadow;
+
+	struct ttm_bo_kmap_obj		dma_buf_vmap;
+	struct amdgpu_mn		*mn;
+
+	union {
+		struct list_head	mn_list;
+		struct list_head	shadow_list;
+	};
+
+	struct kgd_mem                  *kfd_bo;
+};
+
+static inline struct amdgpu_bo *ttm_to_amdgpu_bo(struct ttm_buffer_object *tbo)
+{
+	return container_of(tbo, struct amdgpu_bo, tbo);
+}
+
+/**
+ * amdgpu_mem_type_to_domain - return domain corresponding to mem_type
+ * @mem_type:	ttm memory type
+ *
+ * Returns corresponding domain of the ttm mem_type
+ */
+static inline unsigned amdgpu_mem_type_to_domain(u32 mem_type)
+{
+	switch (mem_type) {
+	case TTM_PL_VRAM:
+		return AMDGPU_GEM_DOMAIN_VRAM;
+	case TTM_PL_TT:
+		return AMDGPU_GEM_DOMAIN_GTT;
+	case TTM_PL_SYSTEM:
+		return AMDGPU_GEM_DOMAIN_CPU;
+	case AMDGPU_PL_GDS:
+		return AMDGPU_GEM_DOMAIN_GDS;
+	case AMDGPU_PL_GWS:
+		return AMDGPU_GEM_DOMAIN_GWS;
+	case AMDGPU_PL_OA:
+		return AMDGPU_GEM_DOMAIN_OA;
+	default:
+		break;
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_bo_reserve - reserve bo
+ * @bo:		bo structure
+ * @no_intr:	don't return -ERESTARTSYS on pending signal
+ *
+ * Returns:
+ * -ERESTARTSYS: A wait for the buffer to become unreserved was interrupted by
+ * a signal. Release all buffer reservations and return to user-space.
+ */
+static inline int amdgpu_bo_reserve(struct amdgpu_bo *bo, bool no_intr)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	int r;
+
+	r = ttm_bo_reserve(&bo->tbo, !no_intr, false, NULL);
+	if (unlikely(r != 0)) {
+		if (r != -ERESTARTSYS)
+			dev_err(adev->dev, "%p reserve failed\n", bo);
+		return r;
+	}
+	return 0;
+}
+
+static inline void amdgpu_bo_unreserve(struct amdgpu_bo *bo)
+{
+	ttm_bo_unreserve(&bo->tbo);
+}
+
+static inline unsigned long amdgpu_bo_size(struct amdgpu_bo *bo)
+{
+	return bo->tbo.num_pages << PAGE_SHIFT;
+}
+
+static inline unsigned amdgpu_bo_ngpu_pages(struct amdgpu_bo *bo)
+{
+	return (bo->tbo.num_pages << PAGE_SHIFT) / AMDGPU_GPU_PAGE_SIZE;
+}
+
+static inline unsigned amdgpu_bo_gpu_page_alignment(struct amdgpu_bo *bo)
+{
+	return (bo->tbo.mem.page_alignment << PAGE_SHIFT) / AMDGPU_GPU_PAGE_SIZE;
+}
+
+/**
+ * amdgpu_bo_mmap_offset - return mmap offset of bo
+ * @bo:	amdgpu object for which we query the offset
+ *
+ * Returns mmap offset of the object.
+ */
+static inline u64 amdgpu_bo_mmap_offset(struct amdgpu_bo *bo)
+{
+	return drm_vma_node_offset_addr(&bo->tbo.vma_node);
+}
+
+/**
+ * amdgpu_bo_gpu_accessible - return whether the bo is currently in memory that
+ * is accessible to the GPU.
+ */
+static inline bool amdgpu_bo_gpu_accessible(struct amdgpu_bo *bo)
+{
+	switch (bo->tbo.mem.mem_type) {
+	case TTM_PL_TT: return amdgpu_gtt_mgr_has_gart_addr(&bo->tbo.mem);
+	case TTM_PL_VRAM: return true;
+	default: return false;
+	}
+}
+
+/**
+ * amdgpu_bo_in_cpu_visible_vram - check if BO is (partly) in visible VRAM
+ */
+static inline bool amdgpu_bo_in_cpu_visible_vram(struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	unsigned fpfn = adev->gmc.visible_vram_size >> PAGE_SHIFT;
+	struct drm_mm_node *node = bo->tbo.mem.mm_node;
+	unsigned long pages_left;
+
+	if (bo->tbo.mem.mem_type != TTM_PL_VRAM)
+		return false;
+
+	for (pages_left = bo->tbo.mem.num_pages; pages_left;
+	     pages_left -= node->size, node++)
+		if (node->start < fpfn)
+			return true;
+
+	return false;
+}
+
+/**
+ * amdgpu_bo_explicit_sync - return whether the bo is explicitly synced
+ */
+static inline bool amdgpu_bo_explicit_sync(struct amdgpu_bo *bo)
+{
+	return bo->flags & AMDGPU_GEM_CREATE_EXPLICIT_SYNC;
+}
+
+bool amdgpu_bo_is_amdgpu_bo(struct ttm_buffer_object *bo);
+void amdgpu_bo_placement_from_domain(struct amdgpu_bo *abo, u32 domain);
+
+int amdgpu_bo_create(struct amdgpu_device *adev,
+		     struct amdgpu_bo_param *bp,
+		     struct amdgpu_bo **bo_ptr);
+int amdgpu_bo_create_reserved(struct amdgpu_device *adev,
+			      unsigned long size, int align,
+			      u32 domain, struct amdgpu_bo **bo_ptr,
+			      u64 *gpu_addr, void **cpu_addr);
+int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
+			    unsigned long size, int align,
+			    u32 domain, struct amdgpu_bo **bo_ptr,
+			    u64 *gpu_addr, void **cpu_addr);
+void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr,
+			   void **cpu_addr);
+int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr);
+void *amdgpu_bo_kptr(struct amdgpu_bo *bo);
+void amdgpu_bo_kunmap(struct amdgpu_bo *bo);
+struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo);
+void amdgpu_bo_unref(struct amdgpu_bo **bo);
+int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain);
+int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
+			     u64 min_offset, u64 max_offset);
+int amdgpu_bo_unpin(struct amdgpu_bo *bo);
+int amdgpu_bo_evict_vram(struct amdgpu_device *adev);
+int amdgpu_bo_init(struct amdgpu_device *adev);
+int amdgpu_bo_late_init(struct amdgpu_device *adev);
+void amdgpu_bo_fini(struct amdgpu_device *adev);
+#ifdef notyet
+int amdgpu_bo_fbdev_mmap(struct amdgpu_bo *bo,
+				struct vm_area_struct *vma);
+#endif
+int amdgpu_bo_set_tiling_flags(struct amdgpu_bo *bo, u64 tiling_flags);
+void amdgpu_bo_get_tiling_flags(struct amdgpu_bo *bo, u64 *tiling_flags);
+int amdgpu_bo_set_metadata (struct amdgpu_bo *bo, void *metadata,
+			    uint32_t metadata_size, uint64_t flags);
+int amdgpu_bo_get_metadata(struct amdgpu_bo *bo, void *buffer,
+			   size_t buffer_size, uint32_t *metadata_size,
+			   uint64_t *flags);
+void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
+			   bool evict,
+			   struct ttm_mem_reg *new_mem);
+int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo);
+void amdgpu_bo_fence(struct amdgpu_bo *bo, struct dma_fence *fence,
+		     bool shared);
+u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo);
+int amdgpu_bo_backup_to_shadow(struct amdgpu_device *adev,
+			       struct amdgpu_ring *ring,
+			       struct amdgpu_bo *bo,
+			       struct reservation_object *resv,
+			       struct dma_fence **fence, bool direct);
+int amdgpu_bo_validate(struct amdgpu_bo *bo);
+int amdgpu_bo_restore_from_shadow(struct amdgpu_device *adev,
+				  struct amdgpu_ring *ring,
+				  struct amdgpu_bo *bo,
+				  struct reservation_object *resv,
+				  struct dma_fence **fence,
+				  bool direct);
+uint32_t amdgpu_bo_get_preferred_pin_domain(struct amdgpu_device *adev,
+					    uint32_t domain);
+
+/*
+ * sub allocation
+ */
+
+static inline uint64_t amdgpu_sa_bo_gpu_addr(struct amdgpu_sa_bo *sa_bo)
+{
+	return sa_bo->manager->gpu_addr + sa_bo->soffset;
+}
+
+static inline void * amdgpu_sa_bo_cpu_addr(struct amdgpu_sa_bo *sa_bo)
+{
+	return sa_bo->manager->cpu_ptr + sa_bo->soffset;
+}
+
+int amdgpu_sa_bo_manager_init(struct amdgpu_device *adev,
+				     struct amdgpu_sa_manager *sa_manager,
+				     unsigned size, u32 align, u32 domain);
+void amdgpu_sa_bo_manager_fini(struct amdgpu_device *adev,
+				      struct amdgpu_sa_manager *sa_manager);
+int amdgpu_sa_bo_manager_start(struct amdgpu_device *adev,
+				      struct amdgpu_sa_manager *sa_manager);
+int amdgpu_sa_bo_new(struct amdgpu_sa_manager *sa_manager,
+		     struct amdgpu_sa_bo **sa_bo,
+		     unsigned size, unsigned align);
+void amdgpu_sa_bo_free(struct amdgpu_device *adev,
+			      struct amdgpu_sa_bo **sa_bo,
+			      struct dma_fence *fence);
+#if defined(CONFIG_DEBUG_FS)
+void amdgpu_sa_bo_dump_debug_info(struct amdgpu_sa_manager *sa_manager,
+					 struct seq_file *m);
+#endif
+
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_pll.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_pll.c
new file mode 100644
index 00000000000..8e67c1210d7
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_pll.c
@@ -0,0 +1,351 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "atom.h"
+#include "atombios_encoders.h"
+#include "amdgpu_pll.h"
+#include <asm/div64.h>
+#include <linux/gcd.h>
+
+/**
+ * amdgpu_pll_reduce_ratio - fractional number reduction
+ *
+ * @nom: nominator
+ * @den: denominator
+ * @nom_min: minimum value for nominator
+ * @den_min: minimum value for denominator
+ *
+ * Find the greatest common divisor and apply it on both nominator and
+ * denominator, but make nominator and denominator are at least as large
+ * as their minimum values.
+ */
+static void amdgpu_pll_reduce_ratio(unsigned *nom, unsigned *den,
+				    unsigned nom_min, unsigned den_min)
+{
+	unsigned tmp;
+
+	/* reduce the numbers to a simpler ratio */
+	tmp = gcd(*nom, *den);
+	*nom /= tmp;
+	*den /= tmp;
+
+	/* make sure nominator is large enough */
+	if (*nom < nom_min) {
+		tmp = DIV_ROUND_UP(nom_min, *nom);
+		*nom *= tmp;
+		*den *= tmp;
+	}
+
+	/* make sure the denominator is large enough */
+	if (*den < den_min) {
+		tmp = DIV_ROUND_UP(den_min, *den);
+		*nom *= tmp;
+		*den *= tmp;
+	}
+}
+
+/**
+ * amdgpu_pll_get_fb_ref_div - feedback and ref divider calculation
+ *
+ * @nom: nominator
+ * @den: denominator
+ * @post_div: post divider
+ * @fb_div_max: feedback divider maximum
+ * @ref_div_max: reference divider maximum
+ * @fb_div: resulting feedback divider
+ * @ref_div: resulting reference divider
+ *
+ * Calculate feedback and reference divider for a given post divider. Makes
+ * sure we stay within the limits.
+ */
+static void amdgpu_pll_get_fb_ref_div(unsigned nom, unsigned den, unsigned post_div,
+				      unsigned fb_div_max, unsigned ref_div_max,
+				      unsigned *fb_div, unsigned *ref_div)
+{
+	/* limit reference * post divider to a maximum */
+	ref_div_max = min(128 / post_div, ref_div_max);
+
+	/* get matching reference and feedback divider */
+	*ref_div = min(max(DIV_ROUND_CLOSEST(den, post_div), 1u), ref_div_max);
+	*fb_div = DIV_ROUND_CLOSEST(nom * *ref_div * post_div, den);
+
+	/* limit fb divider to its maximum */
+	if (*fb_div > fb_div_max) {
+		*ref_div = DIV_ROUND_CLOSEST(*ref_div * fb_div_max, *fb_div);
+		*fb_div = fb_div_max;
+	}
+}
+
+/**
+ * amdgpu_pll_compute - compute PLL paramaters
+ *
+ * @pll: information about the PLL
+ * @dot_clock_p: resulting pixel clock
+ * fb_div_p: resulting feedback divider
+ * frac_fb_div_p: fractional part of the feedback divider
+ * ref_div_p: resulting reference divider
+ * post_div_p: resulting reference divider
+ *
+ * Try to calculate the PLL parameters to generate the given frequency:
+ * dot_clock = (ref_freq * feedback_div) / (ref_div * post_div)
+ */
+void amdgpu_pll_compute(struct amdgpu_pll *pll,
+			u32 freq,
+			u32 *dot_clock_p,
+			u32 *fb_div_p,
+			u32 *frac_fb_div_p,
+			u32 *ref_div_p,
+			u32 *post_div_p)
+{
+	unsigned target_clock = pll->flags & AMDGPU_PLL_USE_FRAC_FB_DIV ?
+		freq : freq / 10;
+
+	unsigned fb_div_min, fb_div_max, fb_div;
+	unsigned post_div_min, post_div_max, post_div;
+	unsigned ref_div_min, ref_div_max, ref_div;
+	unsigned post_div_best, diff_best;
+	unsigned nom, den;
+
+	/* determine allowed feedback divider range */
+	fb_div_min = pll->min_feedback_div;
+	fb_div_max = pll->max_feedback_div;
+
+	if (pll->flags & AMDGPU_PLL_USE_FRAC_FB_DIV) {
+		fb_div_min *= 10;
+		fb_div_max *= 10;
+	}
+
+	/* determine allowed ref divider range */
+	if (pll->flags & AMDGPU_PLL_USE_REF_DIV)
+		ref_div_min = pll->reference_div;
+	else
+		ref_div_min = pll->min_ref_div;
+
+	if (pll->flags & AMDGPU_PLL_USE_FRAC_FB_DIV &&
+	    pll->flags & AMDGPU_PLL_USE_REF_DIV)
+		ref_div_max = pll->reference_div;
+	else
+		ref_div_max = pll->max_ref_div;
+
+	/* determine allowed post divider range */
+	if (pll->flags & AMDGPU_PLL_USE_POST_DIV) {
+		post_div_min = pll->post_div;
+		post_div_max = pll->post_div;
+	} else {
+		unsigned vco_min, vco_max;
+
+		if (pll->flags & AMDGPU_PLL_IS_LCD) {
+			vco_min = pll->lcd_pll_out_min;
+			vco_max = pll->lcd_pll_out_max;
+		} else {
+			vco_min = pll->pll_out_min;
+			vco_max = pll->pll_out_max;
+		}
+
+		if (pll->flags & AMDGPU_PLL_USE_FRAC_FB_DIV) {
+			vco_min *= 10;
+			vco_max *= 10;
+		}
+
+		post_div_min = vco_min / target_clock;
+		if ((target_clock * post_div_min) < vco_min)
+			++post_div_min;
+		if (post_div_min < pll->min_post_div)
+			post_div_min = pll->min_post_div;
+
+		post_div_max = vco_max / target_clock;
+		if ((target_clock * post_div_max) > vco_max)
+			--post_div_max;
+		if (post_div_max > pll->max_post_div)
+			post_div_max = pll->max_post_div;
+	}
+
+	/* represent the searched ratio as fractional number */
+	nom = target_clock;
+	den = pll->reference_freq;
+
+	/* reduce the numbers to a simpler ratio */
+	amdgpu_pll_reduce_ratio(&nom, &den, fb_div_min, post_div_min);
+
+	/* now search for a post divider */
+	if (pll->flags & AMDGPU_PLL_PREFER_MINM_OVER_MAXP)
+		post_div_best = post_div_min;
+	else
+		post_div_best = post_div_max;
+	diff_best = ~0;
+
+	for (post_div = post_div_min; post_div <= post_div_max; ++post_div) {
+		unsigned diff;
+		amdgpu_pll_get_fb_ref_div(nom, den, post_div, fb_div_max,
+					  ref_div_max, &fb_div, &ref_div);
+		diff = abs(target_clock - (pll->reference_freq * fb_div) /
+			(ref_div * post_div));
+
+		if (diff < diff_best || (diff == diff_best &&
+		    !(pll->flags & AMDGPU_PLL_PREFER_MINM_OVER_MAXP))) {
+
+			post_div_best = post_div;
+			diff_best = diff;
+		}
+	}
+	post_div = post_div_best;
+
+	/* get the feedback and reference divider for the optimal value */
+	amdgpu_pll_get_fb_ref_div(nom, den, post_div, fb_div_max, ref_div_max,
+				  &fb_div, &ref_div);
+
+	/* reduce the numbers to a simpler ratio once more */
+	/* this also makes sure that the reference divider is large enough */
+	amdgpu_pll_reduce_ratio(&fb_div, &ref_div, fb_div_min, ref_div_min);
+
+	/* avoid high jitter with small fractional dividers */
+	if (pll->flags & AMDGPU_PLL_USE_FRAC_FB_DIV && (fb_div % 10)) {
+		fb_div_min = max(fb_div_min, (9 - (fb_div % 10)) * 20 + 60);
+		if (fb_div < fb_div_min) {
+			unsigned tmp = DIV_ROUND_UP(fb_div_min, fb_div);
+			fb_div *= tmp;
+			ref_div *= tmp;
+		}
+	}
+
+	/* and finally save the result */
+	if (pll->flags & AMDGPU_PLL_USE_FRAC_FB_DIV) {
+		*fb_div_p = fb_div / 10;
+		*frac_fb_div_p = fb_div % 10;
+	} else {
+		*fb_div_p = fb_div;
+		*frac_fb_div_p = 0;
+	}
+
+	*dot_clock_p = ((pll->reference_freq * *fb_div_p * 10) +
+			(pll->reference_freq * *frac_fb_div_p)) /
+		       (ref_div * post_div * 10);
+	*ref_div_p = ref_div;
+	*post_div_p = post_div;
+
+	DRM_DEBUG_KMS("%d - %d, pll dividers - fb: %d.%d ref: %d, post %d\n",
+		      freq, *dot_clock_p * 10, *fb_div_p, *frac_fb_div_p,
+		      ref_div, post_div);
+}
+
+/**
+ * amdgpu_pll_get_use_mask - look up a mask of which pplls are in use
+ *
+ * @crtc: drm crtc
+ *
+ * Returns the mask of which PPLLs (Pixel PLLs) are in use.
+ */
+u32 amdgpu_pll_get_use_mask(struct drm_crtc *crtc)
+{
+	struct drm_device *dev = crtc->dev;
+	struct drm_crtc *test_crtc;
+	struct amdgpu_crtc *test_amdgpu_crtc;
+	u32 pll_in_use = 0;
+
+	list_for_each_entry(test_crtc, &dev->mode_config.crtc_list, head) {
+		if (crtc == test_crtc)
+			continue;
+
+		test_amdgpu_crtc = to_amdgpu_crtc(test_crtc);
+		if (test_amdgpu_crtc->pll_id != ATOM_PPLL_INVALID)
+			pll_in_use |= (1 << test_amdgpu_crtc->pll_id);
+	}
+	return pll_in_use;
+}
+
+/**
+ * amdgpu_pll_get_shared_dp_ppll - return the PPLL used by another crtc for DP
+ *
+ * @crtc: drm crtc
+ *
+ * Returns the PPLL (Pixel PLL) used by another crtc/encoder which is
+ * also in DP mode.  For DP, a single PPLL can be used for all DP
+ * crtcs/encoders.
+ */
+int amdgpu_pll_get_shared_dp_ppll(struct drm_crtc *crtc)
+{
+	struct drm_device *dev = crtc->dev;
+	struct drm_crtc *test_crtc;
+	struct amdgpu_crtc *test_amdgpu_crtc;
+
+	list_for_each_entry(test_crtc, &dev->mode_config.crtc_list, head) {
+		if (crtc == test_crtc)
+			continue;
+		test_amdgpu_crtc = to_amdgpu_crtc(test_crtc);
+		if (test_amdgpu_crtc->encoder &&
+		    ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(test_amdgpu_crtc->encoder))) {
+			/* for DP use the same PLL for all */
+			if (test_amdgpu_crtc->pll_id != ATOM_PPLL_INVALID)
+				return test_amdgpu_crtc->pll_id;
+		}
+	}
+	return ATOM_PPLL_INVALID;
+}
+
+/**
+ * amdgpu_pll_get_shared_nondp_ppll - return the PPLL used by another non-DP crtc
+ *
+ * @crtc: drm crtc
+ * @encoder: drm encoder
+ *
+ * Returns the PPLL (Pixel PLL) used by another non-DP crtc/encoder which can
+ * be shared (i.e., same clock).
+ */
+int amdgpu_pll_get_shared_nondp_ppll(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct drm_crtc *test_crtc;
+	struct amdgpu_crtc *test_amdgpu_crtc;
+	u32 adjusted_clock, test_adjusted_clock;
+
+	adjusted_clock = amdgpu_crtc->adjusted_clock;
+
+	if (adjusted_clock == 0)
+		return ATOM_PPLL_INVALID;
+
+	list_for_each_entry(test_crtc, &dev->mode_config.crtc_list, head) {
+		if (crtc == test_crtc)
+			continue;
+		test_amdgpu_crtc = to_amdgpu_crtc(test_crtc);
+		if (test_amdgpu_crtc->encoder &&
+		    !ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(test_amdgpu_crtc->encoder))) {
+			/* check if we are already driving this connector with another crtc */
+			if (test_amdgpu_crtc->connector == amdgpu_crtc->connector) {
+				/* if we are, return that pll */
+				if (test_amdgpu_crtc->pll_id != ATOM_PPLL_INVALID)
+					return test_amdgpu_crtc->pll_id;
+			}
+			/* for non-DP check the clock */
+			test_adjusted_clock = test_amdgpu_crtc->adjusted_clock;
+			if ((crtc->mode.clock == test_crtc->mode.clock) &&
+			    (adjusted_clock == test_adjusted_clock) &&
+			    (amdgpu_crtc->ss_enabled == test_amdgpu_crtc->ss_enabled) &&
+			    (test_amdgpu_crtc->pll_id != ATOM_PPLL_INVALID))
+				return test_amdgpu_crtc->pll_id;
+		}
+	}
+	return ATOM_PPLL_INVALID;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_pll.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_pll.h
new file mode 100644
index 00000000000..db6136f68b8
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_pll.h
@@ -0,0 +1,38 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_PLL_H__
+#define __AMDGPU_PLL_H__
+
+void amdgpu_pll_compute(struct amdgpu_pll *pll,
+			 u32 freq,
+			 u32 *dot_clock_p,
+			 u32 *fb_div_p,
+			 u32 *frac_fb_div_p,
+			 u32 *ref_div_p,
+			 u32 *post_div_p);
+u32 amdgpu_pll_get_use_mask(struct drm_crtc *crtc);
+int amdgpu_pll_get_shared_dp_ppll(struct drm_crtc *crtc);
+int amdgpu_pll_get_shared_nondp_ppll(struct drm_crtc *crtc);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_pm.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_pm.c
new file mode 100644
index 00000000000..b811ba056b0
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_pm.c
@@ -0,0 +1,2160 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Rafał Miłecki <zajec5@gmail.com>
+ *          Alex Deucher <alexdeucher@gmail.com>
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_drv.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_dpm.h"
+#include "atom.h"
+#include <linux/power_supply.h>
+#include <linux/hwmon.h>
+#include <linux/hwmon-sysfs.h>
+#include <linux/nospec.h>
+
+static int amdgpu_debugfs_pm_init(struct amdgpu_device *adev);
+
+static const struct cg_flag_name clocks[] = {
+	{AMD_CG_SUPPORT_GFX_MGCG, "Graphics Medium Grain Clock Gating"},
+	{AMD_CG_SUPPORT_GFX_MGLS, "Graphics Medium Grain memory Light Sleep"},
+	{AMD_CG_SUPPORT_GFX_CGCG, "Graphics Coarse Grain Clock Gating"},
+	{AMD_CG_SUPPORT_GFX_CGLS, "Graphics Coarse Grain memory Light Sleep"},
+	{AMD_CG_SUPPORT_GFX_CGTS, "Graphics Coarse Grain Tree Shader Clock Gating"},
+	{AMD_CG_SUPPORT_GFX_CGTS_LS, "Graphics Coarse Grain Tree Shader Light Sleep"},
+	{AMD_CG_SUPPORT_GFX_CP_LS, "Graphics Command Processor Light Sleep"},
+	{AMD_CG_SUPPORT_GFX_RLC_LS, "Graphics Run List Controller Light Sleep"},
+	{AMD_CG_SUPPORT_GFX_3D_CGCG, "Graphics 3D Coarse Grain Clock Gating"},
+	{AMD_CG_SUPPORT_GFX_3D_CGLS, "Graphics 3D Coarse Grain memory Light Sleep"},
+	{AMD_CG_SUPPORT_MC_LS, "Memory Controller Light Sleep"},
+	{AMD_CG_SUPPORT_MC_MGCG, "Memory Controller Medium Grain Clock Gating"},
+	{AMD_CG_SUPPORT_SDMA_LS, "System Direct Memory Access Light Sleep"},
+	{AMD_CG_SUPPORT_SDMA_MGCG, "System Direct Memory Access Medium Grain Clock Gating"},
+	{AMD_CG_SUPPORT_BIF_MGCG, "Bus Interface Medium Grain Clock Gating"},
+	{AMD_CG_SUPPORT_BIF_LS, "Bus Interface Light Sleep"},
+	{AMD_CG_SUPPORT_UVD_MGCG, "Unified Video Decoder Medium Grain Clock Gating"},
+	{AMD_CG_SUPPORT_VCE_MGCG, "Video Compression Engine Medium Grain Clock Gating"},
+	{AMD_CG_SUPPORT_HDP_LS, "Host Data Path Light Sleep"},
+	{AMD_CG_SUPPORT_HDP_MGCG, "Host Data Path Medium Grain Clock Gating"},
+	{AMD_CG_SUPPORT_DRM_MGCG, "Digital Right Management Medium Grain Clock Gating"},
+	{AMD_CG_SUPPORT_DRM_LS, "Digital Right Management Light Sleep"},
+	{AMD_CG_SUPPORT_ROM_MGCG, "Rom Medium Grain Clock Gating"},
+	{AMD_CG_SUPPORT_DF_MGCG, "Data Fabric Medium Grain Clock Gating"},
+	{0, NULL},
+};
+
+void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev)
+{
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		if (power_supply_is_system_supplied() > 0)
+			adev->pm.ac_power = true;
+		else
+			adev->pm.ac_power = false;
+		if (adev->powerplay.pp_funcs->enable_bapm)
+			amdgpu_dpm_enable_bapm(adev, adev->pm.ac_power);
+		mutex_unlock(&adev->pm.mutex);
+	}
+}
+
+/**
+ * DOC: power_dpm_state
+ *
+ * The power_dpm_state file is a legacy interface and is only provided for
+ * backwards compatibility. The amdgpu driver provides a sysfs API for adjusting
+ * certain power related parameters.  The file power_dpm_state is used for this.
+ * It accepts the following arguments:
+ *
+ * - battery
+ *
+ * - balanced
+ *
+ * - performance
+ *
+ * battery
+ *
+ * On older GPUs, the vbios provided a special power state for battery
+ * operation.  Selecting battery switched to this state.  This is no
+ * longer provided on newer GPUs so the option does nothing in that case.
+ *
+ * balanced
+ *
+ * On older GPUs, the vbios provided a special power state for balanced
+ * operation.  Selecting balanced switched to this state.  This is no
+ * longer provided on newer GPUs so the option does nothing in that case.
+ *
+ * performance
+ *
+ * On older GPUs, the vbios provided a special power state for performance
+ * operation.  Selecting performance switched to this state.  This is no
+ * longer provided on newer GPUs so the option does nothing in that case.
+ *
+ */
+
+static ssize_t amdgpu_get_dpm_state(struct device *dev,
+				    struct device_attribute *attr,
+				    char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	enum amd_pm_state_type pm;
+
+	if (adev->powerplay.pp_funcs->get_current_power_state)
+		pm = amdgpu_dpm_get_current_power_state(adev);
+	else
+		pm = adev->pm.dpm.user_state;
+
+	return snprintf(buf, PAGE_SIZE, "%s\n",
+			(pm == POWER_STATE_TYPE_BATTERY) ? "battery" :
+			(pm == POWER_STATE_TYPE_BALANCED) ? "balanced" : "performance");
+}
+
+static ssize_t amdgpu_set_dpm_state(struct device *dev,
+				    struct device_attribute *attr,
+				    const char *buf,
+				    size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	enum amd_pm_state_type  state;
+
+	if (strncmp("battery", buf, strlen("battery")) == 0)
+		state = POWER_STATE_TYPE_BATTERY;
+	else if (strncmp("balanced", buf, strlen("balanced")) == 0)
+		state = POWER_STATE_TYPE_BALANCED;
+	else if (strncmp("performance", buf, strlen("performance")) == 0)
+		state = POWER_STATE_TYPE_PERFORMANCE;
+	else {
+		count = -EINVAL;
+		goto fail;
+	}
+
+	if (adev->powerplay.pp_funcs->dispatch_tasks) {
+		amdgpu_dpm_dispatch_task(adev, AMD_PP_TASK_ENABLE_USER_STATE, &state);
+	} else {
+		mutex_lock(&adev->pm.mutex);
+		adev->pm.dpm.user_state = state;
+		mutex_unlock(&adev->pm.mutex);
+
+		/* Can't set dpm state when the card is off */
+#ifdef notyet
+		if (!(adev->flags & AMD_IS_PX) ||
+		    (ddev->switch_power_state == DRM_SWITCH_POWER_ON))
+#endif
+			amdgpu_pm_compute_clocks(adev);
+	}
+fail:
+	return count;
+}
+
+
+/**
+ * DOC: power_dpm_force_performance_level
+ *
+ * The amdgpu driver provides a sysfs API for adjusting certain power
+ * related parameters.  The file power_dpm_force_performance_level is
+ * used for this.  It accepts the following arguments:
+ *
+ * - auto
+ *
+ * - low
+ *
+ * - high
+ *
+ * - manual
+ *
+ * - profile_standard
+ *
+ * - profile_min_sclk
+ *
+ * - profile_min_mclk
+ *
+ * - profile_peak
+ *
+ * auto
+ *
+ * When auto is selected, the driver will attempt to dynamically select
+ * the optimal power profile for current conditions in the driver.
+ *
+ * low
+ *
+ * When low is selected, the clocks are forced to the lowest power state.
+ *
+ * high
+ *
+ * When high is selected, the clocks are forced to the highest power state.
+ *
+ * manual
+ *
+ * When manual is selected, the user can manually adjust which power states
+ * are enabled for each clock domain via the sysfs pp_dpm_mclk, pp_dpm_sclk,
+ * and pp_dpm_pcie files and adjust the power state transition heuristics
+ * via the pp_power_profile_mode sysfs file.
+ *
+ * profile_standard
+ * profile_min_sclk
+ * profile_min_mclk
+ * profile_peak
+ *
+ * When the profiling modes are selected, clock and power gating are
+ * disabled and the clocks are set for different profiling cases. This
+ * mode is recommended for profiling specific work loads where you do
+ * not want clock or power gating for clock fluctuation to interfere
+ * with your results. profile_standard sets the clocks to a fixed clock
+ * level which varies from asic to asic.  profile_min_sclk forces the sclk
+ * to the lowest level.  profile_min_mclk forces the mclk to the lowest level.
+ * profile_peak sets all clocks (mclk, sclk, pcie) to the highest levels.
+ *
+ */
+
+static ssize_t amdgpu_get_dpm_forced_performance_level(struct device *dev,
+						struct device_attribute *attr,
+								char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	enum amd_dpm_forced_level level = 0xff;
+
+#ifdef notyet
+	if  ((adev->flags & AMD_IS_PX) &&
+	     (ddev->switch_power_state != DRM_SWITCH_POWER_ON))
+		return snprintf(buf, PAGE_SIZE, "off\n");
+#endif
+
+	if (adev->powerplay.pp_funcs->get_performance_level)
+		level = amdgpu_dpm_get_performance_level(adev);
+	else
+		level = adev->pm.dpm.forced_level;
+
+	return snprintf(buf, PAGE_SIZE, "%s\n",
+			(level == AMD_DPM_FORCED_LEVEL_AUTO) ? "auto" :
+			(level == AMD_DPM_FORCED_LEVEL_LOW) ? "low" :
+			(level == AMD_DPM_FORCED_LEVEL_HIGH) ? "high" :
+			(level == AMD_DPM_FORCED_LEVEL_MANUAL) ? "manual" :
+			(level == AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD) ? "profile_standard" :
+			(level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK) ? "profile_min_sclk" :
+			(level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) ? "profile_min_mclk" :
+			(level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) ? "profile_peak" :
+			"unknown");
+}
+
+static ssize_t amdgpu_set_dpm_forced_performance_level(struct device *dev,
+						       struct device_attribute *attr,
+						       const char *buf,
+						       size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	enum amd_dpm_forced_level level;
+	enum amd_dpm_forced_level current_level = 0xff;
+	int ret = 0;
+
+#ifdef notyet
+	/* Can't force performance level when the card is off */
+	if  ((adev->flags & AMD_IS_PX) &&
+	     (ddev->switch_power_state != DRM_SWITCH_POWER_ON))
+		return -EINVAL;
+#endif
+
+	if (adev->powerplay.pp_funcs->get_performance_level)
+		current_level = amdgpu_dpm_get_performance_level(adev);
+
+	if (strncmp("low", buf, strlen("low")) == 0) {
+		level = AMD_DPM_FORCED_LEVEL_LOW;
+	} else if (strncmp("high", buf, strlen("high")) == 0) {
+		level = AMD_DPM_FORCED_LEVEL_HIGH;
+	} else if (strncmp("auto", buf, strlen("auto")) == 0) {
+		level = AMD_DPM_FORCED_LEVEL_AUTO;
+	} else if (strncmp("manual", buf, strlen("manual")) == 0) {
+		level = AMD_DPM_FORCED_LEVEL_MANUAL;
+	} else if (strncmp("profile_exit", buf, strlen("profile_exit")) == 0) {
+		level = AMD_DPM_FORCED_LEVEL_PROFILE_EXIT;
+	} else if (strncmp("profile_standard", buf, strlen("profile_standard")) == 0) {
+		level = AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD;
+	} else if (strncmp("profile_min_sclk", buf, strlen("profile_min_sclk")) == 0) {
+		level = AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK;
+	} else if (strncmp("profile_min_mclk", buf, strlen("profile_min_mclk")) == 0) {
+		level = AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK;
+	} else if (strncmp("profile_peak", buf, strlen("profile_peak")) == 0) {
+		level = AMD_DPM_FORCED_LEVEL_PROFILE_PEAK;
+	}  else {
+		count = -EINVAL;
+		goto fail;
+	}
+
+	if (current_level == level)
+		return count;
+
+	if (adev->powerplay.pp_funcs->force_performance_level) {
+		mutex_lock(&adev->pm.mutex);
+		if (adev->pm.dpm.thermal_active) {
+			count = -EINVAL;
+			mutex_unlock(&adev->pm.mutex);
+			goto fail;
+		}
+		ret = amdgpu_dpm_force_performance_level(adev, level);
+		if (ret)
+			count = -EINVAL;
+		else
+			adev->pm.dpm.forced_level = level;
+		mutex_unlock(&adev->pm.mutex);
+	}
+
+fail:
+	return count;
+}
+
+static ssize_t amdgpu_get_pp_num_states(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	struct pp_states_info data;
+	int i, buf_len;
+
+	if (adev->powerplay.pp_funcs->get_pp_num_states)
+		amdgpu_dpm_get_pp_num_states(adev, &data);
+
+	buf_len = snprintf(buf, PAGE_SIZE, "states: %d\n", data.nums);
+	for (i = 0; i < data.nums; i++)
+		buf_len += snprintf(buf + buf_len, PAGE_SIZE, "%d %s\n", i,
+				(data.states[i] == POWER_STATE_TYPE_INTERNAL_BOOT) ? "boot" :
+				(data.states[i] == POWER_STATE_TYPE_BATTERY) ? "battery" :
+				(data.states[i] == POWER_STATE_TYPE_BALANCED) ? "balanced" :
+				(data.states[i] == POWER_STATE_TYPE_PERFORMANCE) ? "performance" : "default");
+
+	return buf_len;
+}
+
+static ssize_t amdgpu_get_pp_cur_state(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	struct pp_states_info data;
+	enum amd_pm_state_type pm = 0;
+	int i = 0;
+
+	if (adev->powerplay.pp_funcs->get_current_power_state
+		 && adev->powerplay.pp_funcs->get_pp_num_states) {
+		pm = amdgpu_dpm_get_current_power_state(adev);
+		amdgpu_dpm_get_pp_num_states(adev, &data);
+
+		for (i = 0; i < data.nums; i++) {
+			if (pm == data.states[i])
+				break;
+		}
+
+		if (i == data.nums)
+			i = -EINVAL;
+	}
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", i);
+}
+
+static ssize_t amdgpu_get_pp_force_state(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+
+	if (adev->pp_force_state_enabled)
+		return amdgpu_get_pp_cur_state(dev, attr, buf);
+	else
+		return snprintf(buf, PAGE_SIZE, "\n");
+}
+
+static ssize_t amdgpu_set_pp_force_state(struct device *dev,
+		struct device_attribute *attr,
+		const char *buf,
+		size_t count)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	enum amd_pm_state_type state = 0;
+	unsigned long idx;
+	int ret;
+
+	if (strlen(buf) == 1)
+		adev->pp_force_state_enabled = false;
+	else if (adev->powerplay.pp_funcs->dispatch_tasks &&
+			adev->powerplay.pp_funcs->get_pp_num_states) {
+		struct pp_states_info data;
+
+		ret = kstrtoul(buf, 0, &idx);
+		if (ret || idx >= ARRAY_SIZE(data.states)) {
+			count = -EINVAL;
+			goto fail;
+		}
+		idx = array_index_nospec(idx, ARRAY_SIZE(data.states));
+
+		amdgpu_dpm_get_pp_num_states(adev, &data);
+		state = data.states[idx];
+		/* only set user selected power states */
+		if (state != POWER_STATE_TYPE_INTERNAL_BOOT &&
+		    state != POWER_STATE_TYPE_DEFAULT) {
+			amdgpu_dpm_dispatch_task(adev,
+					AMD_PP_TASK_ENABLE_USER_STATE, &state);
+			adev->pp_force_state_enabled = true;
+		}
+	}
+fail:
+	return count;
+#endif
+}
+
+/**
+ * DOC: pp_table
+ *
+ * The amdgpu driver provides a sysfs API for uploading new powerplay
+ * tables.  The file pp_table is used for this.  Reading the file
+ * will dump the current power play table.  Writing to the file
+ * will attempt to upload a new powerplay table and re-initialize
+ * powerplay using that new table.
+ *
+ */
+
+static ssize_t amdgpu_get_pp_table(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	char *table = NULL;
+	int size;
+
+	if (adev->powerplay.pp_funcs->get_pp_table)
+		size = amdgpu_dpm_get_pp_table(adev, &table);
+	else
+		return 0;
+
+	if (size >= PAGE_SIZE)
+		size = PAGE_SIZE - 1;
+
+	memcpy(buf, table, size);
+
+	return size;
+}
+
+static ssize_t amdgpu_set_pp_table(struct device *dev,
+		struct device_attribute *attr,
+		const char *buf,
+		size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+
+	if (adev->powerplay.pp_funcs->set_pp_table)
+		amdgpu_dpm_set_pp_table(adev, buf, count);
+
+	return count;
+}
+
+/**
+ * DOC: pp_od_clk_voltage
+ *
+ * The amdgpu driver provides a sysfs API for adjusting the clocks and voltages
+ * in each power level within a power state.  The pp_od_clk_voltage is used for
+ * this.
+ *
+ * Reading the file will display:
+ *
+ * - a list of engine clock levels and voltages labeled OD_SCLK
+ *
+ * - a list of memory clock levels and voltages labeled OD_MCLK
+ *
+ * - a list of valid ranges for sclk, mclk, and voltage labeled OD_RANGE
+ *
+ * To manually adjust these settings, first select manual using
+ * power_dpm_force_performance_level. Enter a new value for each
+ * level by writing a string that contains "s/m level clock voltage" to
+ * the file.  E.g., "s 1 500 820" will update sclk level 1 to be 500 MHz
+ * at 820 mV; "m 0 350 810" will update mclk level 0 to be 350 MHz at
+ * 810 mV.  When you have edited all of the states as needed, write
+ * "c" (commit) to the file to commit your changes.  If you want to reset to the
+ * default power levels, write "r" (reset) to the file to reset them.
+ *
+ */
+
+static ssize_t amdgpu_set_pp_od_clk_voltage(struct device *dev,
+		struct device_attribute *attr,
+		const char *buf,
+		size_t count)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	int ret;
+	uint32_t parameter_size = 0;
+	long parameter[64];
+	char buf_cpy[128];
+	char *tmp_str;
+	char *sub_str;
+	const char delimiter[3] = {' ', '\n', '\0'};
+	uint32_t type;
+
+	if (count > 127)
+		return -EINVAL;
+
+	if (*buf == 's')
+		type = PP_OD_EDIT_SCLK_VDDC_TABLE;
+	else if (*buf == 'm')
+		type = PP_OD_EDIT_MCLK_VDDC_TABLE;
+	else if(*buf == 'r')
+		type = PP_OD_RESTORE_DEFAULT_TABLE;
+	else if (*buf == 'c')
+		type = PP_OD_COMMIT_DPM_TABLE;
+	else
+		return -EINVAL;
+
+	memcpy(buf_cpy, buf, count+1);
+
+	tmp_str = buf_cpy;
+
+	while (isspace(*++tmp_str));
+
+	while (tmp_str[0]) {
+		sub_str = strsep(&tmp_str, delimiter);
+		ret = kstrtol(sub_str, 0, &parameter[parameter_size]);
+		if (ret)
+			return -EINVAL;
+		parameter_size++;
+
+		while (isspace(*tmp_str))
+			tmp_str++;
+	}
+
+	if (adev->powerplay.pp_funcs->odn_edit_dpm_table)
+		ret = amdgpu_dpm_odn_edit_dpm_table(adev, type,
+						parameter, parameter_size);
+
+	if (ret)
+		return -EINVAL;
+
+	if (type == PP_OD_COMMIT_DPM_TABLE) {
+		if (adev->powerplay.pp_funcs->dispatch_tasks) {
+			amdgpu_dpm_dispatch_task(adev, AMD_PP_TASK_READJUST_POWER_STATE, NULL);
+			return count;
+		} else {
+			return -EINVAL;
+		}
+	}
+
+	return count;
+#endif
+}
+
+static ssize_t amdgpu_get_pp_od_clk_voltage(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	uint32_t size = 0;
+
+	if (adev->powerplay.pp_funcs->print_clock_levels) {
+		size = amdgpu_dpm_print_clock_levels(adev, OD_SCLK, buf);
+		size += amdgpu_dpm_print_clock_levels(adev, OD_MCLK, buf+size);
+		size += amdgpu_dpm_print_clock_levels(adev, OD_RANGE, buf+size);
+		return size;
+	} else {
+		return snprintf(buf, PAGE_SIZE, "\n");
+	}
+
+}
+
+/**
+ * DOC: pp_dpm_sclk pp_dpm_mclk pp_dpm_pcie
+ *
+ * The amdgpu driver provides a sysfs API for adjusting what power levels
+ * are enabled for a given power state.  The files pp_dpm_sclk, pp_dpm_mclk,
+ * and pp_dpm_pcie are used for this.
+ *
+ * Reading back the files will show you the available power levels within
+ * the power state and the clock information for those levels.
+ *
+ * To manually adjust these states, first select manual using
+ * power_dpm_force_performance_level.
+ * Secondly,Enter a new value for each level by inputing a string that
+ * contains " echo xx xx xx > pp_dpm_sclk/mclk/pcie"
+ * E.g., echo 4 5 6 to > pp_dpm_sclk will enable sclk levels 4, 5, and 6.
+ */
+
+static ssize_t amdgpu_get_pp_dpm_sclk(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+
+	if (adev->powerplay.pp_funcs->print_clock_levels)
+		return amdgpu_dpm_print_clock_levels(adev, PP_SCLK, buf);
+	else
+		return snprintf(buf, PAGE_SIZE, "\n");
+}
+
+/*
+ * Worst case: 32 bits individually specified, in octal at 12 characters
+ * per line (+1 for \n).
+ */
+#define AMDGPU_MASK_BUF_MAX	(32 * 13)
+
+static ssize_t amdgpu_read_mask(const char *buf, size_t count, uint32_t *mask)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	int ret;
+	long level;
+	char *sub_str = NULL;
+	char *tmp;
+	char buf_cpy[AMDGPU_MASK_BUF_MAX + 1];
+	const char delimiter[3] = {' ', '\n', '\0'};
+	size_t bytes;
+
+	*mask = 0;
+
+	bytes = min(count, sizeof(buf_cpy) - 1);
+	memcpy(buf_cpy, buf, bytes);
+	buf_cpy[bytes] = '\0';
+	tmp = buf_cpy;
+	while (tmp[0]) {
+		sub_str = strsep(&tmp, delimiter);
+		if (strlen(sub_str)) {
+			ret = kstrtol(sub_str, 0, &level);
+			if (ret)
+				return -EINVAL;
+			*mask |= 1 << level;
+		} else
+			break;
+	}
+
+	return 0;
+#endif
+}
+
+static ssize_t amdgpu_set_pp_dpm_sclk(struct device *dev,
+		struct device_attribute *attr,
+		const char *buf,
+		size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	int ret;
+	uint32_t mask = 0;
+
+	ret = amdgpu_read_mask(buf, count, &mask);
+	if (ret)
+		return ret;
+
+	if (adev->powerplay.pp_funcs->force_clock_level)
+		amdgpu_dpm_force_clock_level(adev, PP_SCLK, mask);
+
+	return count;
+}
+
+static ssize_t amdgpu_get_pp_dpm_mclk(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+
+	if (adev->powerplay.pp_funcs->print_clock_levels)
+		return amdgpu_dpm_print_clock_levels(adev, PP_MCLK, buf);
+	else
+		return snprintf(buf, PAGE_SIZE, "\n");
+}
+
+static ssize_t amdgpu_set_pp_dpm_mclk(struct device *dev,
+		struct device_attribute *attr,
+		const char *buf,
+		size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	int ret;
+	uint32_t mask = 0;
+
+	ret = amdgpu_read_mask(buf, count, &mask);
+	if (ret)
+		return ret;
+
+	if (adev->powerplay.pp_funcs->force_clock_level)
+		amdgpu_dpm_force_clock_level(adev, PP_MCLK, mask);
+
+	return count;
+}
+
+static ssize_t amdgpu_get_pp_dpm_pcie(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+
+	if (adev->powerplay.pp_funcs->print_clock_levels)
+		return amdgpu_dpm_print_clock_levels(adev, PP_PCIE, buf);
+	else
+		return snprintf(buf, PAGE_SIZE, "\n");
+}
+
+static ssize_t amdgpu_set_pp_dpm_pcie(struct device *dev,
+		struct device_attribute *attr,
+		const char *buf,
+		size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	int ret;
+	uint32_t mask = 0;
+
+	ret = amdgpu_read_mask(buf, count, &mask);
+	if (ret)
+		return ret;
+
+	if (adev->powerplay.pp_funcs->force_clock_level)
+		amdgpu_dpm_force_clock_level(adev, PP_PCIE, mask);
+
+	return count;
+}
+
+static ssize_t amdgpu_get_pp_sclk_od(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	uint32_t value = 0;
+
+	if (adev->powerplay.pp_funcs->get_sclk_od)
+		value = amdgpu_dpm_get_sclk_od(adev);
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", value);
+}
+
+static ssize_t amdgpu_set_pp_sclk_od(struct device *dev,
+		struct device_attribute *attr,
+		const char *buf,
+		size_t count)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	int ret;
+	long int value;
+
+	ret = kstrtol(buf, 0, &value);
+
+	if (ret) {
+		count = -EINVAL;
+		goto fail;
+	}
+	if (adev->powerplay.pp_funcs->set_sclk_od)
+		amdgpu_dpm_set_sclk_od(adev, (uint32_t)value);
+
+	if (adev->powerplay.pp_funcs->dispatch_tasks) {
+		amdgpu_dpm_dispatch_task(adev, AMD_PP_TASK_READJUST_POWER_STATE, NULL);
+	} else {
+		adev->pm.dpm.current_ps = adev->pm.dpm.boot_ps;
+		amdgpu_pm_compute_clocks(adev);
+	}
+
+fail:
+	return count;
+#endif
+}
+
+static ssize_t amdgpu_get_pp_mclk_od(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	uint32_t value = 0;
+
+	if (adev->powerplay.pp_funcs->get_mclk_od)
+		value = amdgpu_dpm_get_mclk_od(adev);
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", value);
+}
+
+static ssize_t amdgpu_set_pp_mclk_od(struct device *dev,
+		struct device_attribute *attr,
+		const char *buf,
+		size_t count)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	int ret;
+	long int value;
+
+	ret = kstrtol(buf, 0, &value);
+
+	if (ret) {
+		count = -EINVAL;
+		goto fail;
+	}
+	if (adev->powerplay.pp_funcs->set_mclk_od)
+		amdgpu_dpm_set_mclk_od(adev, (uint32_t)value);
+
+	if (adev->powerplay.pp_funcs->dispatch_tasks) {
+		amdgpu_dpm_dispatch_task(adev, AMD_PP_TASK_READJUST_POWER_STATE, NULL);
+	} else {
+		adev->pm.dpm.current_ps = adev->pm.dpm.boot_ps;
+		amdgpu_pm_compute_clocks(adev);
+	}
+
+fail:
+	return count;
+#endif
+}
+
+/**
+ * DOC: pp_power_profile_mode
+ *
+ * The amdgpu driver provides a sysfs API for adjusting the heuristics
+ * related to switching between power levels in a power state.  The file
+ * pp_power_profile_mode is used for this.
+ *
+ * Reading this file outputs a list of all of the predefined power profiles
+ * and the relevant heuristics settings for that profile.
+ *
+ * To select a profile or create a custom profile, first select manual using
+ * power_dpm_force_performance_level.  Writing the number of a predefined
+ * profile to pp_power_profile_mode will enable those heuristics.  To
+ * create a custom set of heuristics, write a string of numbers to the file
+ * starting with the number of the custom profile along with a setting
+ * for each heuristic parameter.  Due to differences across asic families
+ * the heuristic parameters vary from family to family.
+ *
+ */
+
+static ssize_t amdgpu_get_pp_power_profile_mode(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+
+	if (adev->powerplay.pp_funcs->get_power_profile_mode)
+		return amdgpu_dpm_get_power_profile_mode(adev, buf);
+
+	return snprintf(buf, PAGE_SIZE, "\n");
+}
+
+
+static ssize_t amdgpu_set_pp_power_profile_mode(struct device *dev,
+		struct device_attribute *attr,
+		const char *buf,
+		size_t count)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	int ret = 0xff;
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	uint32_t parameter_size = 0;
+	long parameter[64];
+	char *sub_str, buf_cpy[128];
+	char *tmp_str;
+	uint32_t i = 0;
+	char tmp[2];
+	long int profile_mode = 0;
+	const char delimiter[3] = {' ', '\n', '\0'};
+
+	tmp[0] = *(buf);
+	tmp[1] = '\0';
+	ret = kstrtol(tmp, 0, &profile_mode);
+	if (ret)
+		goto fail;
+
+	if (profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) {
+		if (count < 2 || count > 127)
+			return -EINVAL;
+		while (isspace(*++buf))
+			i++;
+		memcpy(buf_cpy, buf, count-i);
+		tmp_str = buf_cpy;
+		while (tmp_str[0]) {
+			sub_str = strsep(&tmp_str, delimiter);
+			ret = kstrtol(sub_str, 0, &parameter[parameter_size]);
+			if (ret) {
+				count = -EINVAL;
+				goto fail;
+			}
+			parameter_size++;
+			while (isspace(*tmp_str))
+				tmp_str++;
+		}
+	}
+	parameter[parameter_size] = profile_mode;
+	if (adev->powerplay.pp_funcs->set_power_profile_mode)
+		ret = amdgpu_dpm_set_power_profile_mode(adev, parameter, parameter_size);
+
+	if (!ret)
+		return count;
+fail:
+	return -EINVAL;
+#endif
+}
+
+/**
+ * DOC: busy_percent
+ *
+ * The amdgpu driver provides a sysfs API for reading how busy the GPU
+ * is as a percentage.  The file gpu_busy_percent is used for this.
+ * The SMU firmware computes a percentage of load based on the
+ * aggregate activity level in the IP cores.
+ */
+static ssize_t amdgpu_get_busy_percent(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	int r, value, size = sizeof(value);
+
+	/* sanity check PP is enabled */
+	if (!(adev->powerplay.pp_funcs &&
+	      adev->powerplay.pp_funcs->read_sensor))
+		return -EINVAL;
+
+	/* read the IP busy sensor */
+	r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_LOAD,
+				   (void *)&value, &size);
+	if (r)
+		return r;
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", value);
+}
+
+static DEVICE_ATTR(power_dpm_state, S_IRUGO | S_IWUSR, amdgpu_get_dpm_state, amdgpu_set_dpm_state);
+static DEVICE_ATTR(power_dpm_force_performance_level, S_IRUGO | S_IWUSR,
+		   amdgpu_get_dpm_forced_performance_level,
+		   amdgpu_set_dpm_forced_performance_level);
+static DEVICE_ATTR(pp_num_states, S_IRUGO, amdgpu_get_pp_num_states, NULL);
+static DEVICE_ATTR(pp_cur_state, S_IRUGO, amdgpu_get_pp_cur_state, NULL);
+static DEVICE_ATTR(pp_force_state, S_IRUGO | S_IWUSR,
+		amdgpu_get_pp_force_state,
+		amdgpu_set_pp_force_state);
+static DEVICE_ATTR(pp_table, S_IRUGO | S_IWUSR,
+		amdgpu_get_pp_table,
+		amdgpu_set_pp_table);
+static DEVICE_ATTR(pp_dpm_sclk, S_IRUGO | S_IWUSR,
+		amdgpu_get_pp_dpm_sclk,
+		amdgpu_set_pp_dpm_sclk);
+static DEVICE_ATTR(pp_dpm_mclk, S_IRUGO | S_IWUSR,
+		amdgpu_get_pp_dpm_mclk,
+		amdgpu_set_pp_dpm_mclk);
+static DEVICE_ATTR(pp_dpm_pcie, S_IRUGO | S_IWUSR,
+		amdgpu_get_pp_dpm_pcie,
+		amdgpu_set_pp_dpm_pcie);
+static DEVICE_ATTR(pp_sclk_od, S_IRUGO | S_IWUSR,
+		amdgpu_get_pp_sclk_od,
+		amdgpu_set_pp_sclk_od);
+static DEVICE_ATTR(pp_mclk_od, S_IRUGO | S_IWUSR,
+		amdgpu_get_pp_mclk_od,
+		amdgpu_set_pp_mclk_od);
+static DEVICE_ATTR(pp_power_profile_mode, S_IRUGO | S_IWUSR,
+		amdgpu_get_pp_power_profile_mode,
+		amdgpu_set_pp_power_profile_mode);
+static DEVICE_ATTR(pp_od_clk_voltage, S_IRUGO | S_IWUSR,
+		amdgpu_get_pp_od_clk_voltage,
+		amdgpu_set_pp_od_clk_voltage);
+static DEVICE_ATTR(gpu_busy_percent, S_IRUGO,
+		amdgpu_get_busy_percent, NULL);
+
+static ssize_t amdgpu_hwmon_show_temp(struct device *dev,
+				      struct device_attribute *attr,
+				      char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	struct drm_device *ddev = adev->ddev;
+	int r, temp, size = sizeof(temp);
+
+#ifdef notyet
+	/* Can't get temperature when the card is off */
+	if  ((adev->flags & AMD_IS_PX) &&
+	     (ddev->switch_power_state != DRM_SWITCH_POWER_ON))
+		return -EINVAL;
+#endif
+
+	/* sanity check PP is enabled */
+	if (!(adev->powerplay.pp_funcs &&
+	      adev->powerplay.pp_funcs->read_sensor))
+		return -EINVAL;
+
+	/* get the temperature */
+	r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_TEMP,
+				   (void *)&temp, &size);
+	if (r)
+		return r;
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", temp);
+}
+
+static ssize_t amdgpu_hwmon_show_temp_thresh(struct device *dev,
+					     struct device_attribute *attr,
+					     char *buf)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int hyst = to_sensor_dev_attr(attr)->index;
+	int temp;
+
+	if (hyst)
+		temp = adev->pm.dpm.thermal.min_temp;
+	else
+		temp = adev->pm.dpm.thermal.max_temp;
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", temp);
+#endif
+}
+
+static ssize_t amdgpu_hwmon_get_pwm1_enable(struct device *dev,
+					    struct device_attribute *attr,
+					    char *buf)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	u32 pwm_mode = 0;
+
+	if (!adev->powerplay.pp_funcs->get_fan_control_mode)
+		return -EINVAL;
+
+	pwm_mode = amdgpu_dpm_get_fan_control_mode(adev);
+
+	return sprintf(buf, "%i\n", pwm_mode);
+#endif
+}
+
+static ssize_t amdgpu_hwmon_set_pwm1_enable(struct device *dev,
+					    struct device_attribute *attr,
+					    const char *buf,
+					    size_t count)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int err;
+	int value;
+
+	/* Can't adjust fan when the card is off */
+	if  ((adev->flags & AMD_IS_PX) &&
+	     (adev->ddev->switch_power_state != DRM_SWITCH_POWER_ON))
+		return -EINVAL;
+
+	if (!adev->powerplay.pp_funcs->set_fan_control_mode)
+		return -EINVAL;
+
+	err = kstrtoint(buf, 10, &value);
+	if (err)
+		return err;
+
+	amdgpu_dpm_set_fan_control_mode(adev, value);
+
+	return count;
+#endif
+}
+
+static ssize_t amdgpu_hwmon_get_pwm1_min(struct device *dev,
+					 struct device_attribute *attr,
+					 char *buf)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	return sprintf(buf, "%i\n", 0);
+#endif
+}
+
+static ssize_t amdgpu_hwmon_get_pwm1_max(struct device *dev,
+					 struct device_attribute *attr,
+					 char *buf)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	return sprintf(buf, "%i\n", 255);
+#endif
+}
+
+static ssize_t amdgpu_hwmon_set_pwm1(struct device *dev,
+				     struct device_attribute *attr,
+				     const char *buf, size_t count)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int err;
+	u32 value;
+
+#ifdef notyet
+	/* Can't adjust fan when the card is off */
+	if  ((adev->flags & AMD_IS_PX) &&
+	     (adev->ddev->switch_power_state != DRM_SWITCH_POWER_ON))
+		return -EINVAL;
+#endif
+
+	err = kstrtou32(buf, 10, &value);
+	if (err)
+		return err;
+
+	value = (value * 100) / 255;
+
+	if (adev->powerplay.pp_funcs->set_fan_speed_percent) {
+		err = amdgpu_dpm_set_fan_speed_percent(adev, value);
+		if (err)
+			return err;
+	}
+
+	return count;
+#endif
+}
+
+static ssize_t amdgpu_hwmon_get_pwm1(struct device *dev,
+				     struct device_attribute *attr,
+				     char *buf)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int err;
+	u32 speed = 0;
+
+	/* Can't adjust fan when the card is off */
+	if  ((adev->flags & AMD_IS_PX) &&
+	     (adev->ddev->switch_power_state != DRM_SWITCH_POWER_ON))
+		return -EINVAL;
+
+	if (adev->powerplay.pp_funcs->get_fan_speed_percent) {
+		err = amdgpu_dpm_get_fan_speed_percent(adev, &speed);
+		if (err)
+			return err;
+	}
+
+	speed = (speed * 255) / 100;
+
+	return sprintf(buf, "%i\n", speed);
+#endif
+}
+
+#ifdef notyet
+
+static ssize_t amdgpu_hwmon_get_fan1_input(struct device *dev,
+					   struct device_attribute *attr,
+					   char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int err;
+	u32 speed = 0;
+
+	/* Can't adjust fan when the card is off */
+	if  ((adev->flags & AMD_IS_PX) &&
+	     (adev->ddev->switch_power_state != DRM_SWITCH_POWER_ON))
+		return -EINVAL;
+
+	if (adev->powerplay.pp_funcs->get_fan_speed_rpm) {
+		err = amdgpu_dpm_get_fan_speed_rpm(adev, &speed);
+		if (err)
+			return err;
+	}
+
+	return sprintf(buf, "%i\n", speed);
+}
+
+static ssize_t amdgpu_hwmon_show_vddgfx(struct device *dev,
+					struct device_attribute *attr,
+					char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	struct drm_device *ddev = adev->ddev;
+	u32 vddgfx;
+	int r, size = sizeof(vddgfx);
+
+	/* Can't get voltage when the card is off */
+	if  ((adev->flags & AMD_IS_PX) &&
+	     (ddev->switch_power_state != DRM_SWITCH_POWER_ON))
+		return -EINVAL;
+
+	/* sanity check PP is enabled */
+	if (!(adev->powerplay.pp_funcs &&
+	      adev->powerplay.pp_funcs->read_sensor))
+	      return -EINVAL;
+
+	/* get the voltage */
+	r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_VDDGFX,
+				   (void *)&vddgfx, &size);
+	if (r)
+		return r;
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", vddgfx);
+}
+
+static ssize_t amdgpu_hwmon_show_vddgfx_label(struct device *dev,
+					      struct device_attribute *attr,
+					      char *buf)
+{
+	return snprintf(buf, PAGE_SIZE, "vddgfx\n");
+}
+
+static ssize_t amdgpu_hwmon_show_vddnb(struct device *dev,
+				       struct device_attribute *attr,
+				       char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	struct drm_device *ddev = adev->ddev;
+	u32 vddnb;
+	int r, size = sizeof(vddnb);
+
+	/* only APUs have vddnb */
+	if  (!(adev->flags & AMD_IS_APU))
+		return -EINVAL;
+
+	/* Can't get voltage when the card is off */
+	if  ((adev->flags & AMD_IS_PX) &&
+	     (ddev->switch_power_state != DRM_SWITCH_POWER_ON))
+		return -EINVAL;
+
+	/* sanity check PP is enabled */
+	if (!(adev->powerplay.pp_funcs &&
+	      adev->powerplay.pp_funcs->read_sensor))
+	      return -EINVAL;
+
+	/* get the voltage */
+	r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_VDDNB,
+				   (void *)&vddnb, &size);
+	if (r)
+		return r;
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", vddnb);
+}
+
+static ssize_t amdgpu_hwmon_show_vddnb_label(struct device *dev,
+					      struct device_attribute *attr,
+					      char *buf)
+{
+	return snprintf(buf, PAGE_SIZE, "vddnb\n");
+}
+
+static ssize_t amdgpu_hwmon_show_power_avg(struct device *dev,
+					   struct device_attribute *attr,
+					   char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	struct drm_device *ddev = adev->ddev;
+	u32 query = 0;
+	int r, size = sizeof(u32);
+	unsigned uw;
+
+	/* Can't get power when the card is off */
+	if  ((adev->flags & AMD_IS_PX) &&
+	     (ddev->switch_power_state != DRM_SWITCH_POWER_ON))
+		return -EINVAL;
+
+	/* sanity check PP is enabled */
+	if (!(adev->powerplay.pp_funcs &&
+	      adev->powerplay.pp_funcs->read_sensor))
+	      return -EINVAL;
+
+	/* get the voltage */
+	r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_POWER,
+				   (void *)&query, &size);
+	if (r)
+		return r;
+
+	/* convert to microwatts */
+	uw = (query >> 8) * 1000000 + (query & 0xff) * 1000;
+
+	return snprintf(buf, PAGE_SIZE, "%u\n", uw);
+}
+
+static ssize_t amdgpu_hwmon_show_power_cap_min(struct device *dev,
+					 struct device_attribute *attr,
+					 char *buf)
+{
+	return sprintf(buf, "%i\n", 0);
+}
+
+static ssize_t amdgpu_hwmon_show_power_cap_max(struct device *dev,
+					 struct device_attribute *attr,
+					 char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	uint32_t limit = 0;
+
+	if (adev->powerplay.pp_funcs && adev->powerplay.pp_funcs->get_power_limit) {
+		adev->powerplay.pp_funcs->get_power_limit(adev->powerplay.pp_handle, &limit, true);
+		return snprintf(buf, PAGE_SIZE, "%u\n", limit * 1000000);
+	} else {
+		return snprintf(buf, PAGE_SIZE, "\n");
+	}
+}
+
+static ssize_t amdgpu_hwmon_show_power_cap(struct device *dev,
+					 struct device_attribute *attr,
+					 char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	uint32_t limit = 0;
+
+	if (adev->powerplay.pp_funcs && adev->powerplay.pp_funcs->get_power_limit) {
+		adev->powerplay.pp_funcs->get_power_limit(adev->powerplay.pp_handle, &limit, false);
+		return snprintf(buf, PAGE_SIZE, "%u\n", limit * 1000000);
+	} else {
+		return snprintf(buf, PAGE_SIZE, "\n");
+	}
+}
+
+
+static ssize_t amdgpu_hwmon_set_power_cap(struct device *dev,
+		struct device_attribute *attr,
+		const char *buf,
+		size_t count)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int err;
+	u32 value;
+
+	err = kstrtou32(buf, 10, &value);
+	if (err)
+		return err;
+
+	value = value / 1000000; /* convert to Watt */
+	if (adev->powerplay.pp_funcs && adev->powerplay.pp_funcs->set_power_limit) {
+		err = adev->powerplay.pp_funcs->set_power_limit(adev->powerplay.pp_handle, value);
+		if (err)
+			return err;
+	} else {
+		return -EINVAL;
+	}
+
+	return count;
+}
+
+
+/**
+ * DOC: hwmon
+ *
+ * The amdgpu driver exposes the following sensor interfaces:
+ *
+ * - GPU temperature (via the on-die sensor)
+ *
+ * - GPU voltage
+ *
+ * - Northbridge voltage (APUs only)
+ *
+ * - GPU power
+ *
+ * - GPU fan
+ *
+ * hwmon interfaces for GPU temperature:
+ *
+ * - temp1_input: the on die GPU temperature in millidegrees Celsius
+ *
+ * - temp1_crit: temperature critical max value in millidegrees Celsius
+ *
+ * - temp1_crit_hyst: temperature hysteresis for critical limit in millidegrees Celsius
+ *
+ * hwmon interfaces for GPU voltage:
+ *
+ * - in0_input: the voltage on the GPU in millivolts
+ *
+ * - in1_input: the voltage on the Northbridge in millivolts
+ *
+ * hwmon interfaces for GPU power:
+ *
+ * - power1_average: average power used by the GPU in microWatts
+ *
+ * - power1_cap_min: minimum cap supported in microWatts
+ *
+ * - power1_cap_max: maximum cap supported in microWatts
+ *
+ * - power1_cap: selected power cap in microWatts
+ *
+ * hwmon interfaces for GPU fan:
+ *
+ * - pwm1: pulse width modulation fan level (0-255)
+ *
+ * - pwm1_enable: pulse width modulation fan control method (0: no fan speed control, 1: manual fan speed control using pwm interface, 2: automatic fan speed control)
+ *
+ * - pwm1_min: pulse width modulation fan control minimum level (0)
+ *
+ * - pwm1_max: pulse width modulation fan control maximum level (255)
+ *
+ * - fan1_input: fan speed in RPM
+ *
+ * You can use hwmon tools like sensors to view this information on your system.
+ *
+ */
+
+static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp1_crit, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp1_crit_hyst, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(pwm1, S_IRUGO | S_IWUSR, amdgpu_hwmon_get_pwm1, amdgpu_hwmon_set_pwm1, 0);
+static SENSOR_DEVICE_ATTR(pwm1_enable, S_IRUGO | S_IWUSR, amdgpu_hwmon_get_pwm1_enable, amdgpu_hwmon_set_pwm1_enable, 0);
+static SENSOR_DEVICE_ATTR(pwm1_min, S_IRUGO, amdgpu_hwmon_get_pwm1_min, NULL, 0);
+static SENSOR_DEVICE_ATTR(pwm1_max, S_IRUGO, amdgpu_hwmon_get_pwm1_max, NULL, 0);
+static SENSOR_DEVICE_ATTR(fan1_input, S_IRUGO, amdgpu_hwmon_get_fan1_input, NULL, 0);
+static SENSOR_DEVICE_ATTR(in0_input, S_IRUGO, amdgpu_hwmon_show_vddgfx, NULL, 0);
+static SENSOR_DEVICE_ATTR(in0_label, S_IRUGO, amdgpu_hwmon_show_vddgfx_label, NULL, 0);
+static SENSOR_DEVICE_ATTR(in1_input, S_IRUGO, amdgpu_hwmon_show_vddnb, NULL, 0);
+static SENSOR_DEVICE_ATTR(in1_label, S_IRUGO, amdgpu_hwmon_show_vddnb_label, NULL, 0);
+static SENSOR_DEVICE_ATTR(power1_average, S_IRUGO, amdgpu_hwmon_show_power_avg, NULL, 0);
+static SENSOR_DEVICE_ATTR(power1_cap_max, S_IRUGO, amdgpu_hwmon_show_power_cap_max, NULL, 0);
+static SENSOR_DEVICE_ATTR(power1_cap_min, S_IRUGO, amdgpu_hwmon_show_power_cap_min, NULL, 0);
+static SENSOR_DEVICE_ATTR(power1_cap, S_IRUGO | S_IWUSR, amdgpu_hwmon_show_power_cap, amdgpu_hwmon_set_power_cap, 0);
+
+static struct attribute *hwmon_attributes[] = {
+	&sensor_dev_attr_temp1_input.dev_attr.attr,
+	&sensor_dev_attr_temp1_crit.dev_attr.attr,
+	&sensor_dev_attr_temp1_crit_hyst.dev_attr.attr,
+	&sensor_dev_attr_pwm1.dev_attr.attr,
+	&sensor_dev_attr_pwm1_enable.dev_attr.attr,
+	&sensor_dev_attr_pwm1_min.dev_attr.attr,
+	&sensor_dev_attr_pwm1_max.dev_attr.attr,
+	&sensor_dev_attr_fan1_input.dev_attr.attr,
+	&sensor_dev_attr_in0_input.dev_attr.attr,
+	&sensor_dev_attr_in0_label.dev_attr.attr,
+	&sensor_dev_attr_in1_input.dev_attr.attr,
+	&sensor_dev_attr_in1_label.dev_attr.attr,
+	&sensor_dev_attr_power1_average.dev_attr.attr,
+	&sensor_dev_attr_power1_cap_max.dev_attr.attr,
+	&sensor_dev_attr_power1_cap_min.dev_attr.attr,
+	&sensor_dev_attr_power1_cap.dev_attr.attr,
+	NULL
+};
+
+static umode_t hwmon_attributes_visible(struct kobject *kobj,
+					struct attribute *attr, int index)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	umode_t effective_mode = attr->mode;
+
+
+	/* Skip fan attributes if fan is not present */
+	if (adev->pm.no_fan && (attr == &sensor_dev_attr_pwm1.dev_attr.attr ||
+	    attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr ||
+	    attr == &sensor_dev_attr_pwm1_max.dev_attr.attr ||
+	    attr == &sensor_dev_attr_pwm1_min.dev_attr.attr ||
+	    attr == &sensor_dev_attr_fan1_input.dev_attr.attr))
+		return 0;
+
+	/* Skip limit attributes if DPM is not enabled */
+	if (!adev->pm.dpm_enabled &&
+	    (attr == &sensor_dev_attr_temp1_crit.dev_attr.attr ||
+	     attr == &sensor_dev_attr_temp1_crit_hyst.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1_max.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1_min.dev_attr.attr))
+		return 0;
+
+	/* mask fan attributes if we have no bindings for this asic to expose */
+	if ((!adev->powerplay.pp_funcs->get_fan_speed_percent &&
+	     attr == &sensor_dev_attr_pwm1.dev_attr.attr) || /* can't query fan */
+	    (!adev->powerplay.pp_funcs->get_fan_control_mode &&
+	     attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr)) /* can't query state */
+		effective_mode &= ~S_IRUGO;
+
+	if ((!adev->powerplay.pp_funcs->set_fan_speed_percent &&
+	     attr == &sensor_dev_attr_pwm1.dev_attr.attr) || /* can't manage fan */
+	    (!adev->powerplay.pp_funcs->set_fan_control_mode &&
+	     attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr)) /* can't manage state */
+		effective_mode &= ~S_IWUSR;
+
+	if ((adev->flags & AMD_IS_APU) &&
+	    (attr == &sensor_dev_attr_power1_average.dev_attr.attr ||
+	     attr == &sensor_dev_attr_power1_cap_max.dev_attr.attr ||
+	     attr == &sensor_dev_attr_power1_cap_min.dev_attr.attr||
+	     attr == &sensor_dev_attr_power1_cap.dev_attr.attr))
+		return 0;
+
+	/* hide max/min values if we can't both query and manage the fan */
+	if ((!adev->powerplay.pp_funcs->set_fan_speed_percent &&
+	     !adev->powerplay.pp_funcs->get_fan_speed_percent) &&
+	    (attr == &sensor_dev_attr_pwm1_max.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1_min.dev_attr.attr))
+		return 0;
+
+	/* only APUs have vddnb */
+	if (!(adev->flags & AMD_IS_APU) &&
+	    (attr == &sensor_dev_attr_in1_input.dev_attr.attr ||
+	     attr == &sensor_dev_attr_in1_label.dev_attr.attr))
+		return 0;
+
+	return effective_mode;
+}
+
+static const struct attribute_group hwmon_attrgroup = {
+	.attrs = hwmon_attributes,
+	.is_visible = hwmon_attributes_visible,
+};
+
+static const struct attribute_group *hwmon_groups[] = {
+	&hwmon_attrgroup,
+	NULL
+};
+#endif /* notyet */
+
+void amdgpu_dpm_thermal_work_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device,
+			     pm.dpm.thermal.work);
+	/* switch to the thermal state */
+	enum amd_pm_state_type dpm_state = POWER_STATE_TYPE_INTERNAL_THERMAL;
+	int temp, size = sizeof(temp);
+
+	if (!adev->pm.dpm_enabled)
+		return;
+
+	if (adev->powerplay.pp_funcs &&
+	    adev->powerplay.pp_funcs->read_sensor &&
+	    !amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_TEMP,
+				    (void *)&temp, &size)) {
+		if (temp < adev->pm.dpm.thermal.min_temp)
+			/* switch back the user state */
+			dpm_state = adev->pm.dpm.user_state;
+	} else {
+		if (adev->pm.dpm.thermal.high_to_low)
+			/* switch back the user state */
+			dpm_state = adev->pm.dpm.user_state;
+	}
+	mutex_lock(&adev->pm.mutex);
+	if (dpm_state == POWER_STATE_TYPE_INTERNAL_THERMAL)
+		adev->pm.dpm.thermal_active = true;
+	else
+		adev->pm.dpm.thermal_active = false;
+	adev->pm.dpm.state = dpm_state;
+	mutex_unlock(&adev->pm.mutex);
+
+	amdgpu_pm_compute_clocks(adev);
+}
+
+static struct amdgpu_ps *amdgpu_dpm_pick_power_state(struct amdgpu_device *adev,
+						     enum amd_pm_state_type dpm_state)
+{
+	int i;
+	struct amdgpu_ps *ps;
+	u32 ui_class;
+	bool single_display = (adev->pm.dpm.new_active_crtc_count < 2) ?
+		true : false;
+
+	/* check if the vblank period is too short to adjust the mclk */
+	if (single_display && adev->powerplay.pp_funcs->vblank_too_short) {
+		if (amdgpu_dpm_vblank_too_short(adev))
+			single_display = false;
+	}
+
+	/* certain older asics have a separare 3D performance state,
+	 * so try that first if the user selected performance
+	 */
+	if (dpm_state == POWER_STATE_TYPE_PERFORMANCE)
+		dpm_state = POWER_STATE_TYPE_INTERNAL_3DPERF;
+	/* balanced states don't exist at the moment */
+	if (dpm_state == POWER_STATE_TYPE_BALANCED)
+		dpm_state = POWER_STATE_TYPE_PERFORMANCE;
+
+restart_search:
+	/* Pick the best power state based on current conditions */
+	for (i = 0; i < adev->pm.dpm.num_ps; i++) {
+		ps = &adev->pm.dpm.ps[i];
+		ui_class = ps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK;
+		switch (dpm_state) {
+		/* user states */
+		case POWER_STATE_TYPE_BATTERY:
+			if (ui_class == ATOM_PPLIB_CLASSIFICATION_UI_BATTERY) {
+				if (ps->caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY) {
+					if (single_display)
+						return ps;
+				} else
+					return ps;
+			}
+			break;
+		case POWER_STATE_TYPE_BALANCED:
+			if (ui_class == ATOM_PPLIB_CLASSIFICATION_UI_BALANCED) {
+				if (ps->caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY) {
+					if (single_display)
+						return ps;
+				} else
+					return ps;
+			}
+			break;
+		case POWER_STATE_TYPE_PERFORMANCE:
+			if (ui_class == ATOM_PPLIB_CLASSIFICATION_UI_PERFORMANCE) {
+				if (ps->caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY) {
+					if (single_display)
+						return ps;
+				} else
+					return ps;
+			}
+			break;
+		/* internal states */
+		case POWER_STATE_TYPE_INTERNAL_UVD:
+			if (adev->pm.dpm.uvd_ps)
+				return adev->pm.dpm.uvd_ps;
+			else
+				break;
+		case POWER_STATE_TYPE_INTERNAL_UVD_SD:
+			if (ps->class & ATOM_PPLIB_CLASSIFICATION_SDSTATE)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_UVD_HD:
+			if (ps->class & ATOM_PPLIB_CLASSIFICATION_HDSTATE)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_UVD_HD2:
+			if (ps->class & ATOM_PPLIB_CLASSIFICATION_HD2STATE)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_UVD_MVC:
+			if (ps->class2 & ATOM_PPLIB_CLASSIFICATION2_MVC)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_BOOT:
+			return adev->pm.dpm.boot_ps;
+		case POWER_STATE_TYPE_INTERNAL_THERMAL:
+			if (ps->class & ATOM_PPLIB_CLASSIFICATION_THERMAL)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_ACPI:
+			if (ps->class & ATOM_PPLIB_CLASSIFICATION_ACPI)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_ULV:
+			if (ps->class2 & ATOM_PPLIB_CLASSIFICATION2_ULV)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_3DPERF:
+			if (ps->class & ATOM_PPLIB_CLASSIFICATION_3DPERFORMANCE)
+				return ps;
+			break;
+		default:
+			break;
+		}
+	}
+	/* use a fallback state if we didn't match */
+	switch (dpm_state) {
+	case POWER_STATE_TYPE_INTERNAL_UVD_SD:
+		dpm_state = POWER_STATE_TYPE_INTERNAL_UVD_HD;
+		goto restart_search;
+	case POWER_STATE_TYPE_INTERNAL_UVD_HD:
+	case POWER_STATE_TYPE_INTERNAL_UVD_HD2:
+	case POWER_STATE_TYPE_INTERNAL_UVD_MVC:
+		if (adev->pm.dpm.uvd_ps) {
+			return adev->pm.dpm.uvd_ps;
+		} else {
+			dpm_state = POWER_STATE_TYPE_PERFORMANCE;
+			goto restart_search;
+		}
+	case POWER_STATE_TYPE_INTERNAL_THERMAL:
+		dpm_state = POWER_STATE_TYPE_INTERNAL_ACPI;
+		goto restart_search;
+	case POWER_STATE_TYPE_INTERNAL_ACPI:
+		dpm_state = POWER_STATE_TYPE_BATTERY;
+		goto restart_search;
+	case POWER_STATE_TYPE_BATTERY:
+	case POWER_STATE_TYPE_BALANCED:
+	case POWER_STATE_TYPE_INTERNAL_3DPERF:
+		dpm_state = POWER_STATE_TYPE_PERFORMANCE;
+		goto restart_search;
+	default:
+		break;
+	}
+
+	return NULL;
+}
+
+static void amdgpu_dpm_change_power_state_locked(struct amdgpu_device *adev)
+{
+	struct amdgpu_ps *ps;
+	enum amd_pm_state_type dpm_state;
+	int ret;
+	bool equal = false;
+
+	/* if dpm init failed */
+	if (!adev->pm.dpm_enabled)
+		return;
+
+	if (adev->pm.dpm.user_state != adev->pm.dpm.state) {
+		/* add other state override checks here */
+		if ((!adev->pm.dpm.thermal_active) &&
+		    (!adev->pm.dpm.uvd_active))
+			adev->pm.dpm.state = adev->pm.dpm.user_state;
+	}
+	dpm_state = adev->pm.dpm.state;
+
+	ps = amdgpu_dpm_pick_power_state(adev, dpm_state);
+	if (ps)
+		adev->pm.dpm.requested_ps = ps;
+	else
+		return;
+
+	if (amdgpu_dpm == 1 && adev->powerplay.pp_funcs->print_power_state) {
+		printk("switching from power state:\n");
+		amdgpu_dpm_print_power_state(adev, adev->pm.dpm.current_ps);
+		printk("switching to power state:\n");
+		amdgpu_dpm_print_power_state(adev, adev->pm.dpm.requested_ps);
+	}
+
+	/* update whether vce is active */
+	ps->vce_active = adev->pm.dpm.vce_active;
+	if (adev->powerplay.pp_funcs->display_configuration_changed)
+		amdgpu_dpm_display_configuration_changed(adev);
+
+	ret = amdgpu_dpm_pre_set_power_state(adev);
+	if (ret)
+		return;
+
+	if (adev->powerplay.pp_funcs->check_state_equal) {
+		if (0 != amdgpu_dpm_check_state_equal(adev, adev->pm.dpm.current_ps, adev->pm.dpm.requested_ps, &equal))
+			equal = false;
+	}
+
+	if (equal)
+		return;
+
+	amdgpu_dpm_set_power_state(adev);
+	amdgpu_dpm_post_set_power_state(adev);
+
+	adev->pm.dpm.current_active_crtcs = adev->pm.dpm.new_active_crtcs;
+	adev->pm.dpm.current_active_crtc_count = adev->pm.dpm.new_active_crtc_count;
+
+	if (adev->powerplay.pp_funcs->force_performance_level) {
+		if (adev->pm.dpm.thermal_active) {
+			enum amd_dpm_forced_level level = adev->pm.dpm.forced_level;
+			/* force low perf level for thermal */
+			amdgpu_dpm_force_performance_level(adev, AMD_DPM_FORCED_LEVEL_LOW);
+			/* save the user's level */
+			adev->pm.dpm.forced_level = level;
+		} else {
+			/* otherwise, user selected level */
+			amdgpu_dpm_force_performance_level(adev, adev->pm.dpm.forced_level);
+		}
+	}
+}
+
+void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable)
+{
+	if (adev->powerplay.pp_funcs->set_powergating_by_smu) {
+		/* enable/disable UVD */
+		mutex_lock(&adev->pm.mutex);
+		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_UVD, !enable);
+		mutex_unlock(&adev->pm.mutex);
+	} else {
+		if (enable) {
+			mutex_lock(&adev->pm.mutex);
+			adev->pm.dpm.uvd_active = true;
+			adev->pm.dpm.state = POWER_STATE_TYPE_INTERNAL_UVD;
+			mutex_unlock(&adev->pm.mutex);
+		} else {
+			mutex_lock(&adev->pm.mutex);
+			adev->pm.dpm.uvd_active = false;
+			mutex_unlock(&adev->pm.mutex);
+		}
+		amdgpu_pm_compute_clocks(adev);
+	}
+}
+
+void amdgpu_dpm_enable_vce(struct amdgpu_device *adev, bool enable)
+{
+	if (adev->powerplay.pp_funcs->set_powergating_by_smu) {
+		/* enable/disable VCE */
+		mutex_lock(&adev->pm.mutex);
+		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_VCE, !enable);
+		mutex_unlock(&adev->pm.mutex);
+	} else {
+		if (enable) {
+			mutex_lock(&adev->pm.mutex);
+			adev->pm.dpm.vce_active = true;
+			/* XXX select vce level based on ring/task */
+			adev->pm.dpm.vce_level = AMD_VCE_LEVEL_AC_ALL;
+			mutex_unlock(&adev->pm.mutex);
+			amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							       AMD_CG_STATE_UNGATE);
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							       AMD_PG_STATE_UNGATE);
+			amdgpu_pm_compute_clocks(adev);
+		} else {
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							       AMD_PG_STATE_GATE);
+			amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							       AMD_CG_STATE_GATE);
+			mutex_lock(&adev->pm.mutex);
+			adev->pm.dpm.vce_active = false;
+			mutex_unlock(&adev->pm.mutex);
+			amdgpu_pm_compute_clocks(adev);
+		}
+
+	}
+}
+
+void amdgpu_pm_print_power_states(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (adev->powerplay.pp_funcs->print_power_state == NULL)
+		return;
+
+	for (i = 0; i < adev->pm.dpm.num_ps; i++)
+		amdgpu_dpm_print_power_state(adev, &adev->pm.dpm.ps[i]);
+
+}
+
+int amdgpu_pm_sysfs_init(struct amdgpu_device *adev)
+{
+	return 0;
+#ifdef __linux__
+	int ret;
+
+	if (adev->pm.sysfs_initialized)
+		return 0;
+
+	if (adev->pm.dpm_enabled == 0)
+		return 0;
+
+	adev->pm.int_hwmon_dev = hwmon_device_register_with_groups(adev->dev,
+								   DRIVER_NAME, adev,
+								   hwmon_groups);
+	if (IS_ERR(adev->pm.int_hwmon_dev)) {
+		ret = PTR_ERR(adev->pm.int_hwmon_dev);
+		dev_err(adev->dev,
+			"Unable to register hwmon device: %d\n", ret);
+		return ret;
+	}
+
+	ret = device_create_file(adev->dev, &dev_attr_power_dpm_state);
+	if (ret) {
+		DRM_ERROR("failed to create device file for dpm state\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_power_dpm_force_performance_level);
+	if (ret) {
+		DRM_ERROR("failed to create device file for dpm state\n");
+		return ret;
+	}
+
+
+	ret = device_create_file(adev->dev, &dev_attr_pp_num_states);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_num_states\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_pp_cur_state);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_cur_state\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_pp_force_state);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_force_state\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_pp_table);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_table\n");
+		return ret;
+	}
+
+	ret = device_create_file(adev->dev, &dev_attr_pp_dpm_sclk);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_dpm_sclk\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_pp_dpm_mclk);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_dpm_mclk\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_pp_dpm_pcie);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_dpm_pcie\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_pp_sclk_od);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_sclk_od\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_pp_mclk_od);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_mclk_od\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev,
+			&dev_attr_pp_power_profile_mode);
+	if (ret) {
+		DRM_ERROR("failed to create device file	"
+				"pp_power_profile_mode\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev,
+			&dev_attr_pp_od_clk_voltage);
+	if (ret) {
+		DRM_ERROR("failed to create device file	"
+				"pp_od_clk_voltage\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev,
+			&dev_attr_gpu_busy_percent);
+	if (ret) {
+		DRM_ERROR("failed to create device file	"
+				"gpu_busy_level\n");
+		return ret;
+	}
+	ret = amdgpu_debugfs_pm_init(adev);
+	if (ret) {
+		DRM_ERROR("Failed to register debugfs file for dpm!\n");
+		return ret;
+	}
+
+	adev->pm.sysfs_initialized = true;
+
+	return 0;
+#endif
+}
+
+void amdgpu_pm_sysfs_fini(struct amdgpu_device *adev)
+{
+#ifdef __linux__
+	if (adev->pm.dpm_enabled == 0)
+		return;
+
+	if (adev->pm.int_hwmon_dev)
+		hwmon_device_unregister(adev->pm.int_hwmon_dev);
+	device_remove_file(adev->dev, &dev_attr_power_dpm_state);
+	device_remove_file(adev->dev, &dev_attr_power_dpm_force_performance_level);
+
+	device_remove_file(adev->dev, &dev_attr_pp_num_states);
+	device_remove_file(adev->dev, &dev_attr_pp_cur_state);
+	device_remove_file(adev->dev, &dev_attr_pp_force_state);
+	device_remove_file(adev->dev, &dev_attr_pp_table);
+
+	device_remove_file(adev->dev, &dev_attr_pp_dpm_sclk);
+	device_remove_file(adev->dev, &dev_attr_pp_dpm_mclk);
+	device_remove_file(adev->dev, &dev_attr_pp_dpm_pcie);
+	device_remove_file(adev->dev, &dev_attr_pp_sclk_od);
+	device_remove_file(adev->dev, &dev_attr_pp_mclk_od);
+	device_remove_file(adev->dev,
+			&dev_attr_pp_power_profile_mode);
+	device_remove_file(adev->dev,
+			&dev_attr_pp_od_clk_voltage);
+	device_remove_file(adev->dev, &dev_attr_gpu_busy_percent);
+#endif
+}
+
+void amdgpu_pm_compute_clocks(struct amdgpu_device *adev)
+{
+	int i = 0;
+
+	if (!adev->pm.dpm_enabled)
+		return;
+
+	if (adev->mode_info.num_crtc)
+		amdgpu_display_bandwidth_update(adev);
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		if (ring && ring->ready)
+			amdgpu_fence_wait_empty(ring);
+	}
+
+	if (adev->powerplay.pp_funcs->dispatch_tasks) {
+		if (!amdgpu_device_has_dc_support(adev)) {
+			mutex_lock(&adev->pm.mutex);
+			amdgpu_dpm_get_active_displays(adev);
+			adev->pm.pm_display_cfg.num_display = adev->pm.dpm.new_active_crtc_count;
+			adev->pm.pm_display_cfg.vrefresh = amdgpu_dpm_get_vrefresh(adev);
+			adev->pm.pm_display_cfg.min_vblank_time = amdgpu_dpm_get_vblank_time(adev);
+			/* we have issues with mclk switching with refresh rates over 120 hz on the non-DC code. */
+			if (adev->pm.pm_display_cfg.vrefresh > 120)
+				adev->pm.pm_display_cfg.min_vblank_time = 0;
+			if (adev->powerplay.pp_funcs->display_configuration_change)
+				adev->powerplay.pp_funcs->display_configuration_change(
+								adev->powerplay.pp_handle,
+								&adev->pm.pm_display_cfg);
+			mutex_unlock(&adev->pm.mutex);
+		}
+		amdgpu_dpm_dispatch_task(adev, AMD_PP_TASK_DISPLAY_CONFIG_CHANGE, NULL);
+	} else {
+		mutex_lock(&adev->pm.mutex);
+		amdgpu_dpm_get_active_displays(adev);
+		amdgpu_dpm_change_power_state_locked(adev);
+		mutex_unlock(&adev->pm.mutex);
+	}
+}
+
+/*
+ * Debugfs info
+ */
+#if defined(CONFIG_DEBUG_FS)
+
+static int amdgpu_debugfs_pm_info_pp(struct seq_file *m, struct amdgpu_device *adev)
+{
+	uint32_t value;
+	uint32_t query = 0;
+	int size;
+
+	/* sanity check PP is enabled */
+	if (!(adev->powerplay.pp_funcs &&
+	      adev->powerplay.pp_funcs->read_sensor))
+	      return -EINVAL;
+
+	/* GPU Clocks */
+	size = sizeof(value);
+	seq_printf(m, "GFX Clocks and Power:\n");
+	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GFX_MCLK, (void *)&value, &size))
+		seq_printf(m, "\t%u MHz (MCLK)\n", value/100);
+	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GFX_SCLK, (void *)&value, &size))
+		seq_printf(m, "\t%u MHz (SCLK)\n", value/100);
+	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_STABLE_PSTATE_SCLK, (void *)&value, &size))
+		seq_printf(m, "\t%u MHz (PSTATE_SCLK)\n", value/100);
+	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_STABLE_PSTATE_MCLK, (void *)&value, &size))
+		seq_printf(m, "\t%u MHz (PSTATE_MCLK)\n", value/100);
+	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_VDDGFX, (void *)&value, &size))
+		seq_printf(m, "\t%u mV (VDDGFX)\n", value);
+	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_VDDNB, (void *)&value, &size))
+		seq_printf(m, "\t%u mV (VDDNB)\n", value);
+	size = sizeof(uint32_t);
+	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_POWER, (void *)&query, &size))
+		seq_printf(m, "\t%u.%u W (average GPU)\n", query >> 8, query & 0xff);
+	size = sizeof(value);
+	seq_printf(m, "\n");
+
+	/* GPU Temp */
+	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_TEMP, (void *)&value, &size))
+		seq_printf(m, "GPU Temperature: %u C\n", value/1000);
+
+	/* GPU Load */
+	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_LOAD, (void *)&value, &size))
+		seq_printf(m, "GPU Load: %u %%\n", value);
+	seq_printf(m, "\n");
+
+	/* UVD clocks */
+	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_UVD_POWER, (void *)&value, &size)) {
+		if (!value) {
+			seq_printf(m, "UVD: Disabled\n");
+		} else {
+			seq_printf(m, "UVD: Enabled\n");
+			if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_UVD_DCLK, (void *)&value, &size))
+				seq_printf(m, "\t%u MHz (DCLK)\n", value/100);
+			if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_UVD_VCLK, (void *)&value, &size))
+				seq_printf(m, "\t%u MHz (VCLK)\n", value/100);
+		}
+	}
+	seq_printf(m, "\n");
+
+	/* VCE clocks */
+	if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_VCE_POWER, (void *)&value, &size)) {
+		if (!value) {
+			seq_printf(m, "VCE: Disabled\n");
+		} else {
+			seq_printf(m, "VCE: Enabled\n");
+			if (!amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_VCE_ECCLK, (void *)&value, &size))
+				seq_printf(m, "\t%u MHz (ECCLK)\n", value/100);
+		}
+	}
+
+	return 0;
+}
+
+static void amdgpu_parse_cg_state(struct seq_file *m, u32 flags)
+{
+	int i;
+
+	for (i = 0; clocks[i].flag; i++)
+		seq_printf(m, "\t%s: %s\n", clocks[i].name,
+			   (flags & clocks[i].flag) ? "On" : "Off");
+}
+
+static int amdgpu_debugfs_pm_info(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_device *ddev = adev->ddev;
+	u32 flags = 0;
+
+	amdgpu_device_ip_get_clockgating_state(adev, &flags);
+	seq_printf(m, "Clock Gating Flags Mask: 0x%x\n", flags);
+	amdgpu_parse_cg_state(m, flags);
+	seq_printf(m, "\n");
+
+	if (!adev->pm.dpm_enabled) {
+		seq_printf(m, "dpm not enabled\n");
+		return 0;
+	}
+	if  ((adev->flags & AMD_IS_PX) &&
+	     (ddev->switch_power_state != DRM_SWITCH_POWER_ON)) {
+		seq_printf(m, "PX asic powered off\n");
+	} else if (adev->powerplay.pp_funcs->debugfs_print_current_performance_level) {
+		mutex_lock(&adev->pm.mutex);
+		if (adev->powerplay.pp_funcs->debugfs_print_current_performance_level)
+			adev->powerplay.pp_funcs->debugfs_print_current_performance_level(adev, m);
+		else
+			seq_printf(m, "Debugfs support not implemented for this asic\n");
+		mutex_unlock(&adev->pm.mutex);
+	} else {
+		return amdgpu_debugfs_pm_info_pp(m, adev);
+	}
+
+	return 0;
+}
+
+static const struct drm_info_list amdgpu_pm_info_list[] = {
+	{"amdgpu_pm_info", amdgpu_debugfs_pm_info, 0, NULL},
+};
+#endif
+
+static int amdgpu_debugfs_pm_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	return amdgpu_debugfs_add_files(adev, amdgpu_pm_info_list, ARRAY_SIZE(amdgpu_pm_info_list));
+#else
+	return 0;
+#endif
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_pm.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_pm.h
new file mode 100644
index 00000000000..f21a7716b90
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_pm.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_PM_H__
+#define __AMDGPU_PM_H__
+
+struct cg_flag_name
+{
+	u32 flag;
+	const char *name;
+};
+
+void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev);
+int amdgpu_pm_sysfs_init(struct amdgpu_device *adev);
+void amdgpu_pm_sysfs_fini(struct amdgpu_device *adev);
+void amdgpu_pm_print_power_states(struct amdgpu_device *adev);
+void amdgpu_pm_compute_clocks(struct amdgpu_device *adev);
+void amdgpu_dpm_thermal_work_handler(struct work_struct *work);
+void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable);
+void amdgpu_dpm_enable_vce(struct amdgpu_device *adev, bool enable);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_prime.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_prime.c
new file mode 100644
index 00000000000..89b3300d40e
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_prime.c
@@ -0,0 +1,466 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * based on nouveau_prime.c
+ *
+ * Authors: Alex Deucher
+ */
+
+/**
+ * DOC: PRIME Buffer Sharing
+ *
+ * The following callback implementations are used for :ref:`sharing GEM buffer
+ * objects between different devices via PRIME <prime_buffer_sharing>`.
+ */
+
+#include <drm/drmP.h>
+
+#include "amdgpu.h"
+#include "amdgpu_display.h"
+#include <drm/amdgpu_drm.h>
+#include <linux/dma-buf.h>
+#include <linux/dma-fence-array.h>
+
+static const struct dma_buf_ops amdgpu_dmabuf_ops;
+
+#ifdef notyet
+
+/**
+ * amdgpu_gem_prime_get_sg_table - &drm_driver.gem_prime_get_sg_table
+ * implementation
+ * @obj: GEM buffer object
+ *
+ * Returns:
+ * A scatter/gather table for the pinned pages of the buffer object's memory.
+ */
+struct sg_table *amdgpu_gem_prime_get_sg_table(struct drm_gem_object *obj)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	int npages = bo->tbo.num_pages;
+
+	return drm_prime_pages_to_sg(bo->tbo.ttm->pages, npages);
+}
+
+/**
+ * amdgpu_gem_prime_vmap - &dma_buf_ops.vmap implementation
+ * @obj: GEM buffer object
+ *
+ * Sets up an in-kernel virtual mapping of the buffer object's memory.
+ *
+ * Returns:
+ * The virtual address of the mapping or an error pointer.
+ */
+void *amdgpu_gem_prime_vmap(struct drm_gem_object *obj)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	int ret;
+
+	ret = ttm_bo_kmap(&bo->tbo, 0, bo->tbo.num_pages,
+			  &bo->dma_buf_vmap);
+	if (ret)
+		return ERR_PTR(ret);
+
+	return bo->dma_buf_vmap.virtual;
+}
+
+/**
+ * amdgpu_gem_prime_vunmap - &dma_buf_ops.vunmap implementation
+ * @obj: GEM buffer object
+ * @vaddr: virtual address (unused)
+ *
+ * Tears down the in-kernel virtual mapping of the buffer object's memory.
+ */
+void amdgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+	ttm_bo_kunmap(&bo->dma_buf_vmap);
+}
+
+/**
+ * amdgpu_gem_prime_mmap - &drm_driver.gem_prime_mmap implementation
+ * @obj: GEM buffer object
+ * @vma: virtual memory area
+ *
+ * Sets up a userspace mapping of the buffer object's memory in the given
+ * virtual memory area.
+ *
+ * Returns:
+ * 0 on success or negative error code.
+ */
+int amdgpu_gem_prime_mmap(struct drm_gem_object *obj, struct vm_area_struct *vma)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	unsigned asize = amdgpu_bo_size(bo);
+	int ret;
+
+	if (!vma->vm_file)
+		return -ENODEV;
+
+	if (adev == NULL)
+		return -ENODEV;
+
+	/* Check for valid size. */
+	if (asize < vma->vm_end - vma->vm_start)
+		return -EINVAL;
+
+	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm) ||
+	    (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)) {
+		return -EPERM;
+	}
+	vma->vm_pgoff += amdgpu_bo_mmap_offset(bo) >> PAGE_SHIFT;
+
+	/* prime mmap does not need to check access, so allow here */
+	ret = drm_vma_node_allow(&obj->vma_node, vma->vm_file->private_data);
+	if (ret)
+		return ret;
+
+	ret = ttm_bo_mmap(vma->vm_file, vma, &adev->mman.bdev);
+	drm_vma_node_revoke(&obj->vma_node, vma->vm_file->private_data);
+
+	return ret;
+}
+
+/**
+ * amdgpu_gem_prime_import_sg_table - &drm_driver.gem_prime_import_sg_table
+ * implementation
+ * @dev: DRM device
+ * @attach: DMA-buf attachment
+ * @sg: Scatter/gather table
+ *
+ * Import shared DMA buffer memory exported by another device.
+ *
+ * Returns:
+ * A new GEM buffer object of the given DRM device, representing the memory
+ * described by the given DMA-buf attachment and scatter/gather table.
+ */
+struct drm_gem_object *
+amdgpu_gem_prime_import_sg_table(struct drm_device *dev,
+				 struct dma_buf_attachment *attach,
+				 struct sg_table *sg)
+{
+	struct reservation_object *resv = attach->dmabuf->resv;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_bo *bo;
+	struct amdgpu_bo_param bp;
+	int ret;
+
+	memset(&bp, 0, sizeof(bp));
+	bp.size = attach->dmabuf->size;
+	bp.byte_align = PAGE_SIZE;
+	bp.domain = AMDGPU_GEM_DOMAIN_CPU;
+	bp.flags = 0;
+	bp.type = ttm_bo_type_sg;
+	bp.resv = resv;
+	ww_mutex_lock(&resv->lock, NULL);
+	ret = amdgpu_bo_create(adev, &bp, &bo);
+	if (ret)
+		goto error;
+
+	bo->tbo.sg = sg;
+	bo->tbo.ttm->sg = sg;
+	bo->allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
+	bo->preferred_domains = AMDGPU_GEM_DOMAIN_GTT;
+	if (attach->dmabuf->ops != &amdgpu_dmabuf_ops)
+		bo->prime_shared_count = 1;
+
+	ww_mutex_unlock(&resv->lock);
+	return &bo->gem_base;
+
+error:
+	ww_mutex_unlock(&resv->lock);
+	return ERR_PTR(ret);
+}
+
+static int
+__reservation_object_make_exclusive(struct reservation_object *obj)
+{
+	struct dma_fence **fences;
+	unsigned int count;
+	int r;
+
+	if (!reservation_object_get_list(obj)) /* no shared fences to convert */
+		return 0;
+
+	r = reservation_object_get_fences_rcu(obj, NULL, &count, &fences);
+	if (r)
+		return r;
+
+	if (count == 0) {
+		/* Now that was unexpected. */
+	} else if (count == 1) {
+		reservation_object_add_excl_fence(obj, fences[0]);
+		dma_fence_put(fences[0]);
+		kfree(fences);
+	} else {
+		struct dma_fence_array *array;
+
+		array = dma_fence_array_create(count, fences,
+					       dma_fence_context_alloc(1), 0,
+					       false);
+		if (!array)
+			goto err_fences_put;
+
+		reservation_object_add_excl_fence(obj, &array->base);
+		dma_fence_put(&array->base);
+	}
+
+	return 0;
+
+err_fences_put:
+	while (count--)
+		dma_fence_put(fences[count]);
+	kfree(fences);
+	return -ENOMEM;
+}
+
+/**
+ * amdgpu_gem_map_attach - &dma_buf_ops.attach implementation
+ * @dma_buf: shared DMA buffer
+ * @attach: DMA-buf attachment
+ *
+ * Makes sure that the shared DMA buffer can be accessed by the target device.
+ * For now, simply pins it to the GTT domain, where it should be accessible by
+ * all DMA devices.
+ *
+ * Returns:
+ * 0 on success or negative error code.
+ */
+static int amdgpu_gem_map_attach(struct dma_buf *dma_buf,
+				 struct dma_buf_attachment *attach)
+{
+	struct drm_gem_object *obj = dma_buf->priv;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	long r;
+
+	r = drm_gem_map_attach(dma_buf, attach);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_reserve(bo, false);
+	if (unlikely(r != 0))
+		goto error_detach;
+
+
+	if (attach->dev->driver != adev->dev->driver) {
+		/*
+		 * We only create shared fences for internal use, but importers
+		 * of the dmabuf rely on exclusive fences for implicitly
+		 * tracking write hazards. As any of the current fences may
+		 * correspond to a write, we need to convert all existing
+		 * fences on the reservation object into a single exclusive
+		 * fence.
+		 */
+		r = __reservation_object_make_exclusive(bo->tbo.resv);
+		if (r)
+			goto error_unreserve;
+	}
+
+	/* pin buffer into GTT */
+	r = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT);
+	if (r)
+		goto error_unreserve;
+
+	if (attach->dev->driver != adev->dev->driver)
+		bo->prime_shared_count++;
+
+error_unreserve:
+	amdgpu_bo_unreserve(bo);
+
+error_detach:
+	if (r)
+		drm_gem_map_detach(dma_buf, attach);
+	return r;
+}
+
+/**
+ * amdgpu_gem_map_detach - &dma_buf_ops.detach implementation
+ * @dma_buf: shared DMA buffer
+ * @attach: DMA-buf attachment
+ *
+ * This is called when a shared DMA buffer no longer needs to be accessible by
+ * the other device. For now, simply unpins the buffer from GTT.
+ */
+static void amdgpu_gem_map_detach(struct dma_buf *dma_buf,
+				  struct dma_buf_attachment *attach)
+{
+	struct drm_gem_object *obj = dma_buf->priv;
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	int ret = 0;
+
+	ret = amdgpu_bo_reserve(bo, true);
+	if (unlikely(ret != 0))
+		goto error;
+
+	amdgpu_bo_unpin(bo);
+	if (attach->dev->driver != adev->dev->driver && bo->prime_shared_count)
+		bo->prime_shared_count--;
+	amdgpu_bo_unreserve(bo);
+
+error:
+	drm_gem_map_detach(dma_buf, attach);
+}
+
+#endif /* notyet */
+
+/**
+ * amdgpu_gem_prime_res_obj - &drm_driver.gem_prime_res_obj implementation
+ * @obj: GEM buffer object
+ *
+ * Returns:
+ * The buffer object's reservation object.
+ */
+struct reservation_object *amdgpu_gem_prime_res_obj(struct drm_gem_object *obj)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+	return bo->tbo.resv;
+}
+
+#ifdef notyet
+
+/**
+ * amdgpu_gem_begin_cpu_access - &dma_buf_ops.begin_cpu_access implementation
+ * @dma_buf: shared DMA buffer
+ * @direction: direction of DMA transfer
+ *
+ * This is called before CPU access to the shared DMA buffer's memory. If it's
+ * a read access, the buffer is moved to the GTT domain if possible, for optimal
+ * CPU read performance.
+ *
+ * Returns:
+ * 0 on success or negative error code.
+ */
+static int amdgpu_gem_begin_cpu_access(struct dma_buf *dma_buf,
+				       enum dma_data_direction direction)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(dma_buf->priv);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct ttm_operation_ctx ctx = { true, false };
+	u32 domain = amdgpu_display_supported_domains(adev);
+	int ret;
+	bool reads = (direction == DMA_BIDIRECTIONAL ||
+		      direction == DMA_FROM_DEVICE);
+
+	if (!reads || !(domain & AMDGPU_GEM_DOMAIN_GTT))
+		return 0;
+
+	/* move to gtt */
+	ret = amdgpu_bo_reserve(bo, false);
+	if (unlikely(ret != 0))
+		return ret;
+
+	if (!bo->pin_count && (bo->allowed_domains & AMDGPU_GEM_DOMAIN_GTT)) {
+		amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
+		ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	}
+
+	amdgpu_bo_unreserve(bo);
+	return ret;
+}
+
+#endif /* notyet */
+
+static const struct dma_buf_ops amdgpu_dmabuf_ops = {
+#ifdef notyet
+	.attach = amdgpu_gem_map_attach,
+	.detach = amdgpu_gem_map_detach,
+	.map_dma_buf = drm_gem_map_dma_buf,
+	.unmap_dma_buf = drm_gem_unmap_dma_buf,
+#endif
+	.release = drm_gem_dmabuf_release,
+#ifdef notyet
+	.begin_cpu_access = amdgpu_gem_begin_cpu_access,
+	.map = drm_gem_dmabuf_kmap,
+	.unmap = drm_gem_dmabuf_kunmap,
+	.mmap = drm_gem_dmabuf_mmap,
+	.vmap = drm_gem_dmabuf_vmap,
+	.vunmap = drm_gem_dmabuf_vunmap,
+#endif
+};
+
+/**
+ * amdgpu_gem_prime_export - &drm_driver.gem_prime_export implementation
+ * @dev: DRM device
+ * @gobj: GEM buffer object
+ * @flags: flags like DRM_CLOEXEC and DRM_RDWR
+ *
+ * The main work is done by the &drm_gem_prime_export helper, which in turn
+ * uses &amdgpu_gem_prime_res_obj.
+ *
+ * Returns:
+ * Shared DMA buffer representing the GEM buffer object from the given device.
+ */
+struct dma_buf *amdgpu_gem_prime_export(struct drm_device *dev,
+					struct drm_gem_object *gobj,
+					int flags)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(gobj);
+	struct dma_buf *buf;
+
+	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm) ||
+	    bo->flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID)
+		return ERR_PTR(-EPERM);
+
+	buf = drm_gem_prime_export(dev, gobj, flags);
+	if (!IS_ERR(buf)) {
+#ifdef __linux__
+		buf->file->f_mapping = dev->anon_inode->i_mapping;
+#endif
+		buf->ops = &amdgpu_dmabuf_ops;
+	}
+
+	return buf;
+}
+
+/**
+ * amdgpu_gem_prime_import - &drm_driver.gem_prime_import implementation
+ * @dev: DRM device
+ * @dma_buf: Shared DMA buffer
+ *
+ * The main work is done by the &drm_gem_prime_import helper, which in turn
+ * uses &amdgpu_gem_prime_import_sg_table.
+ *
+ * Returns:
+ * GEM buffer object representing the shared DMA buffer for the given device.
+ */
+struct drm_gem_object *amdgpu_gem_prime_import(struct drm_device *dev,
+					    struct dma_buf *dma_buf)
+{
+	struct drm_gem_object *obj;
+
+	if (dma_buf->ops == &amdgpu_dmabuf_ops) {
+		obj = dma_buf->priv;
+		if (obj->dev == dev) {
+			/*
+			 * Importing dmabuf exported from out own gem increases
+			 * refcount on gem itself instead of f_count of dmabuf.
+			 */
+			drm_gem_object_get(obj);
+			return obj;
+		}
+	}
+
+	return drm_gem_prime_import(dev, dma_buf);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_psp.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_psp.c
new file mode 100644
index 00000000000..2921a7bd37d
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_psp.c
@@ -0,0 +1,596 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: Huang Rui
+ *
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_ucode.h"
+#include "soc15_common.h"
+#include "psp_v3_1.h"
+#include "psp_v10_0.h"
+
+static void psp_set_funcs(struct amdgpu_device *adev);
+
+static int psp_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	psp_set_funcs(adev);
+
+	return 0;
+}
+
+static int psp_sw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct psp_context *psp = &adev->psp;
+	int ret;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+		psp_v3_1_set_psp_funcs(psp);
+		break;
+	case CHIP_RAVEN:
+		psp_v10_0_set_psp_funcs(psp);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	psp->adev = adev;
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		return 0;
+
+	ret = psp_init_microcode(psp);
+	if (ret) {
+		DRM_ERROR("Failed to load psp firmware!\n");
+		return ret;
+	}
+
+	return 0;
+}
+
+static int psp_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		return 0;
+
+	release_firmware(adev->psp.sos_fw);
+	adev->psp.sos_fw = NULL;
+	release_firmware(adev->psp.asd_fw);
+	adev->psp.asd_fw = NULL;
+	return 0;
+}
+
+int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
+		 uint32_t reg_val, uint32_t mask, bool check_changed)
+{
+	uint32_t val;
+	int i;
+	struct amdgpu_device *adev = psp->adev;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32(reg_index);
+		if (check_changed) {
+			if (val != reg_val)
+				return 0;
+		} else {
+			if ((val & mask) == reg_val)
+				return 0;
+		}
+		udelay(1);
+	}
+
+	return -ETIME;
+}
+
+static int
+psp_cmd_submit_buf(struct psp_context *psp,
+		   struct amdgpu_firmware_info *ucode,
+		   struct psp_gfx_cmd_resp *cmd, uint64_t fence_mc_addr,
+		   int index)
+{
+	int ret;
+
+	memset(psp->cmd_buf_mem, 0, PSP_CMD_BUFFER_SIZE);
+
+	memcpy(psp->cmd_buf_mem, cmd, sizeof(struct psp_gfx_cmd_resp));
+
+	ret = psp_cmd_submit(psp, ucode, psp->cmd_buf_mc_addr,
+			     fence_mc_addr, index);
+
+	while (*((unsigned int *)psp->fence_buf) != index) {
+		drm_msleep(1);
+	}
+
+	if (ucode) {
+		ucode->tmr_mc_addr_lo = psp->cmd_buf_mem->resp.fw_addr_lo;
+		ucode->tmr_mc_addr_hi = psp->cmd_buf_mem->resp.fw_addr_hi;
+	}
+
+	return ret;
+}
+
+static void psp_prep_tmr_cmd_buf(struct psp_gfx_cmd_resp *cmd,
+				 uint64_t tmr_mc, uint32_t size)
+{
+	cmd->cmd_id = GFX_CMD_ID_SETUP_TMR;
+	cmd->cmd.cmd_setup_tmr.buf_phy_addr_lo = lower_32_bits(tmr_mc);
+	cmd->cmd.cmd_setup_tmr.buf_phy_addr_hi = upper_32_bits(tmr_mc);
+	cmd->cmd.cmd_setup_tmr.buf_size = size;
+}
+
+/* Set up Trusted Memory Region */
+static int psp_tmr_init(struct psp_context *psp)
+{
+	int ret;
+
+	/*
+	 * Allocate 3M memory aligned to 1M from Frame Buffer (local
+	 * physical).
+	 *
+	 * Note: this memory need be reserved till the driver
+	 * uninitializes.
+	 */
+	ret = amdgpu_bo_create_kernel(psp->adev, 0x300000, 0x100000,
+				      AMDGPU_GEM_DOMAIN_VRAM,
+				      &psp->tmr_bo, &psp->tmr_mc_addr, &psp->tmr_buf);
+
+	return ret;
+}
+
+static int psp_tmr_load(struct psp_context *psp)
+{
+	int ret;
+	struct psp_gfx_cmd_resp *cmd;
+
+	cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
+	if (!cmd)
+		return -ENOMEM;
+
+	psp_prep_tmr_cmd_buf(cmd, psp->tmr_mc_addr, 0x300000);
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd,
+				 psp->fence_buf_mc_addr, 1);
+	if (ret)
+		goto failed;
+
+	kfree(cmd);
+
+	return 0;
+
+failed:
+	kfree(cmd);
+	return ret;
+}
+
+static void psp_prep_asd_cmd_buf(struct psp_gfx_cmd_resp *cmd,
+				 uint64_t asd_mc, uint64_t asd_mc_shared,
+				 uint32_t size, uint32_t shared_size)
+{
+	cmd->cmd_id = GFX_CMD_ID_LOAD_ASD;
+	cmd->cmd.cmd_load_ta.app_phy_addr_lo = lower_32_bits(asd_mc);
+	cmd->cmd.cmd_load_ta.app_phy_addr_hi = upper_32_bits(asd_mc);
+	cmd->cmd.cmd_load_ta.app_len = size;
+
+	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_lo = lower_32_bits(asd_mc_shared);
+	cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_hi = upper_32_bits(asd_mc_shared);
+	cmd->cmd.cmd_load_ta.cmd_buf_len = shared_size;
+}
+
+static int psp_asd_init(struct psp_context *psp)
+{
+	int ret;
+
+	/*
+	 * Allocate 16k memory aligned to 4k from Frame Buffer (local
+	 * physical) for shared ASD <-> Driver
+	 */
+	ret = amdgpu_bo_create_kernel(psp->adev, PSP_ASD_SHARED_MEM_SIZE,
+				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
+				      &psp->asd_shared_bo,
+				      &psp->asd_shared_mc_addr,
+				      &psp->asd_shared_buf);
+
+	return ret;
+}
+
+static int psp_asd_load(struct psp_context *psp)
+{
+	int ret;
+	struct psp_gfx_cmd_resp *cmd;
+
+	/* If PSP version doesn't match ASD version, asd loading will be failed.
+	 * add workaround to bypass it for sriov now.
+	 * TODO: add version check to make it common
+	 */
+	if (amdgpu_sriov_vf(psp->adev))
+		return 0;
+
+	cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
+	if (!cmd)
+		return -ENOMEM;
+
+	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+	memcpy(psp->fw_pri_buf, psp->asd_start_addr, psp->asd_ucode_size);
+
+	psp_prep_asd_cmd_buf(cmd, psp->fw_pri_mc_addr, psp->asd_shared_mc_addr,
+			     psp->asd_ucode_size, PSP_ASD_SHARED_MEM_SIZE);
+
+	ret = psp_cmd_submit_buf(psp, NULL, cmd,
+				 psp->fence_buf_mc_addr, 2);
+
+	kfree(cmd);
+
+	return ret;
+}
+
+static int psp_hw_start(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	int ret;
+
+	if (!amdgpu_sriov_vf(adev) || !adev->in_gpu_reset) {
+		ret = psp_bootloader_load_sysdrv(psp);
+		if (ret)
+			return ret;
+
+		ret = psp_bootloader_load_sos(psp);
+		if (ret)
+			return ret;
+	}
+
+	ret = psp_ring_create(psp, PSP_RING_TYPE__KM);
+	if (ret)
+		return ret;
+
+	ret = psp_tmr_load(psp);
+	if (ret)
+		return ret;
+
+	ret = psp_asd_load(psp);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int psp_np_fw_load(struct psp_context *psp)
+{
+	int i, ret;
+	struct amdgpu_firmware_info *ucode;
+	struct amdgpu_device* adev = psp->adev;
+
+	for (i = 0; i < adev->firmware.max_ucodes; i++) {
+		ucode = &adev->firmware.ucode[i];
+		if (!ucode->fw)
+			continue;
+
+		if (ucode->ucode_id == AMDGPU_UCODE_ID_SMC &&
+		    psp_smu_reload_quirk(psp))
+			continue;
+		if (amdgpu_sriov_vf(adev) &&
+		   (ucode->ucode_id == AMDGPU_UCODE_ID_SDMA0
+		    || ucode->ucode_id == AMDGPU_UCODE_ID_SDMA1
+		    || ucode->ucode_id == AMDGPU_UCODE_ID_RLC_G))
+			/*skip ucode loading in SRIOV VF */
+			continue;
+
+		ret = psp_prep_cmd_buf(ucode, psp->cmd);
+		if (ret)
+			return ret;
+
+		ret = psp_cmd_submit_buf(psp, ucode, psp->cmd,
+					 psp->fence_buf_mc_addr, i + 3);
+		if (ret)
+			return ret;
+
+#if 0
+		/* check if firmware loaded sucessfully */
+		if (!amdgpu_psp_check_fw_loading_status(adev, i))
+			return -EINVAL;
+#endif
+	}
+
+	return 0;
+}
+
+static int psp_load_fw(struct amdgpu_device *adev)
+{
+	int ret;
+	struct psp_context *psp = &adev->psp;
+
+	if (amdgpu_sriov_vf(adev) && adev->in_gpu_reset != 0)
+		goto skip_memalloc;
+
+	psp->cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
+	if (!psp->cmd)
+		return -ENOMEM;
+
+	ret = amdgpu_bo_create_kernel(adev, PSP_1_MEG, PSP_1_MEG,
+					AMDGPU_GEM_DOMAIN_GTT,
+					&psp->fw_pri_bo,
+					&psp->fw_pri_mc_addr,
+					&psp->fw_pri_buf);
+	if (ret)
+		goto failed;
+
+	ret = amdgpu_bo_create_kernel(adev, PSP_FENCE_BUFFER_SIZE, PAGE_SIZE,
+					AMDGPU_GEM_DOMAIN_VRAM,
+					&psp->fence_buf_bo,
+					&psp->fence_buf_mc_addr,
+					&psp->fence_buf);
+	if (ret)
+		goto failed_mem2;
+
+	ret = amdgpu_bo_create_kernel(adev, PSP_CMD_BUFFER_SIZE, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM,
+				      &psp->cmd_buf_bo, &psp->cmd_buf_mc_addr,
+				      (void **)&psp->cmd_buf_mem);
+	if (ret)
+		goto failed_mem1;
+
+	memset(psp->fence_buf, 0, PSP_FENCE_BUFFER_SIZE);
+
+	ret = psp_ring_init(psp, PSP_RING_TYPE__KM);
+	if (ret)
+		goto failed_mem;
+
+	ret = psp_tmr_init(psp);
+	if (ret)
+		goto failed_mem;
+
+	ret = psp_asd_init(psp);
+	if (ret)
+		goto failed_mem;
+
+skip_memalloc:
+	ret = psp_hw_start(psp);
+	if (ret)
+		goto failed_mem;
+
+	ret = psp_np_fw_load(psp);
+	if (ret)
+		goto failed_mem;
+
+	return 0;
+
+failed_mem:
+	amdgpu_bo_free_kernel(&psp->cmd_buf_bo,
+			      &psp->cmd_buf_mc_addr,
+			      (void **)&psp->cmd_buf_mem);
+failed_mem1:
+	amdgpu_bo_free_kernel(&psp->fence_buf_bo,
+			      &psp->fence_buf_mc_addr, &psp->fence_buf);
+failed_mem2:
+	amdgpu_bo_free_kernel(&psp->fw_pri_bo,
+			      &psp->fw_pri_mc_addr, &psp->fw_pri_buf);
+failed:
+	kfree(psp->cmd);
+	psp->cmd = NULL;
+	return ret;
+}
+
+static int psp_hw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		return 0;
+
+	mutex_lock(&adev->firmware.mutex);
+	/*
+	 * This sequence is just used on hw_init only once, no need on
+	 * resume.
+	 */
+	ret = amdgpu_ucode_init_bo(adev);
+	if (ret)
+		goto failed;
+
+	ret = psp_load_fw(adev);
+	if (ret) {
+		DRM_ERROR("PSP firmware loading failed\n");
+		goto failed;
+	}
+
+	mutex_unlock(&adev->firmware.mutex);
+	return 0;
+
+failed:
+	adev->firmware.load_type = AMDGPU_FW_LOAD_DIRECT;
+	mutex_unlock(&adev->firmware.mutex);
+	return -EINVAL;
+}
+
+static int psp_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct psp_context *psp = &adev->psp;
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		return 0;
+
+	amdgpu_ucode_fini_bo(adev);
+
+	psp_ring_destroy(psp, PSP_RING_TYPE__KM);
+
+	amdgpu_bo_free_kernel(&psp->tmr_bo, &psp->tmr_mc_addr, &psp->tmr_buf);
+	amdgpu_bo_free_kernel(&psp->fw_pri_bo,
+			      &psp->fw_pri_mc_addr, &psp->fw_pri_buf);
+	amdgpu_bo_free_kernel(&psp->fence_buf_bo,
+			      &psp->fence_buf_mc_addr, &psp->fence_buf);
+	amdgpu_bo_free_kernel(&psp->asd_shared_bo, &psp->asd_shared_mc_addr,
+			      &psp->asd_shared_buf);
+	amdgpu_bo_free_kernel(&psp->cmd_buf_bo, &psp->cmd_buf_mc_addr,
+			      (void **)&psp->cmd_buf_mem);
+
+	kfree(psp->cmd);
+	psp->cmd = NULL;
+
+	return 0;
+}
+
+static int psp_suspend(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct psp_context *psp = &adev->psp;
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		return 0;
+
+	ret = psp_ring_stop(psp, PSP_RING_TYPE__KM);
+	if (ret) {
+		DRM_ERROR("PSP ring stop failed\n");
+		return ret;
+	}
+
+	return 0;
+}
+
+static int psp_resume(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct psp_context *psp = &adev->psp;
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		return 0;
+
+	DRM_INFO("PSP is resuming...\n");
+
+	mutex_lock(&adev->firmware.mutex);
+
+	ret = psp_hw_start(psp);
+	if (ret)
+		goto failed;
+
+	ret = psp_np_fw_load(psp);
+	if (ret)
+		goto failed;
+
+	mutex_unlock(&adev->firmware.mutex);
+
+	return 0;
+
+failed:
+	DRM_ERROR("PSP resume failed\n");
+	mutex_unlock(&adev->firmware.mutex);
+	return ret;
+}
+
+int psp_gpu_reset(struct amdgpu_device *adev)
+{
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		return 0;
+
+	return psp_mode1_reset(&adev->psp);
+}
+
+static bool psp_check_fw_loading_status(struct amdgpu_device *adev,
+					enum AMDGPU_UCODE_ID ucode_type)
+{
+	struct amdgpu_firmware_info *ucode = NULL;
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+		DRM_INFO("firmware is not loaded by PSP\n");
+		return true;
+	}
+
+	if (!adev->firmware.fw_size)
+		return false;
+
+	ucode = &adev->firmware.ucode[ucode_type];
+	if (!ucode->fw || !ucode->ucode_size)
+		return false;
+
+	return psp_compare_sram_data(&adev->psp, ucode, ucode_type);
+}
+
+static int psp_set_clockgating_state(void *handle,
+				     enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int psp_set_powergating_state(void *handle,
+				     enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs psp_ip_funcs = {
+	.name = "psp",
+	.early_init = psp_early_init,
+	.late_init = NULL,
+	.sw_init = psp_sw_init,
+	.sw_fini = psp_sw_fini,
+	.hw_init = psp_hw_init,
+	.hw_fini = psp_hw_fini,
+	.suspend = psp_suspend,
+	.resume = psp_resume,
+	.is_idle = NULL,
+	.check_soft_reset = NULL,
+	.wait_for_idle = NULL,
+	.soft_reset = NULL,
+	.set_clockgating_state = psp_set_clockgating_state,
+	.set_powergating_state = psp_set_powergating_state,
+};
+
+static const struct amdgpu_psp_funcs psp_funcs = {
+	.check_fw_loading_status = psp_check_fw_loading_status,
+};
+
+static void psp_set_funcs(struct amdgpu_device *adev)
+{
+	if (NULL == adev->firmware.funcs)
+		adev->firmware.funcs = &psp_funcs;
+}
+
+const struct amdgpu_ip_block_version psp_v3_1_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_PSP,
+	.major = 3,
+	.minor = 1,
+	.rev = 0,
+	.funcs = &psp_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version psp_v10_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_PSP,
+	.major = 10,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &psp_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_psp.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_psp.h
new file mode 100644
index 00000000000..12920968684
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_psp.h
@@ -0,0 +1,163 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: Huang Rui
+ *
+ */
+#ifndef __AMDGPU_PSP_H__
+#define __AMDGPU_PSP_H__
+
+#include "amdgpu.h"
+#include "psp_gfx_if.h"
+
+#define PSP_FENCE_BUFFER_SIZE	0x1000
+#define PSP_CMD_BUFFER_SIZE	0x1000
+#define PSP_ASD_SHARED_MEM_SIZE	0x4000
+#define PSP_1_MEG		0x100000
+
+struct psp_context;
+
+enum psp_ring_type
+{
+	PSP_RING_TYPE__INVALID = 0,
+	/*
+	 * These values map to the way the PSP kernel identifies the
+	 * rings.
+	 */
+	PSP_RING_TYPE__UM = 1, /* User mode ring (formerly called RBI) */
+	PSP_RING_TYPE__KM = 2  /* Kernel mode ring (formerly called GPCOM) */
+};
+
+struct psp_ring
+{
+	enum psp_ring_type		ring_type;
+	struct psp_gfx_rb_frame		*ring_mem;
+	uint64_t			ring_mem_mc_addr;
+	void				*ring_mem_handle;
+	uint32_t			ring_size;
+};
+
+struct psp_funcs
+{
+	int (*init_microcode)(struct psp_context *psp);
+	int (*bootloader_load_sysdrv)(struct psp_context *psp);
+	int (*bootloader_load_sos)(struct psp_context *psp);
+	int (*prep_cmd_buf)(struct amdgpu_firmware_info *ucode,
+			    struct psp_gfx_cmd_resp *cmd);
+	int (*ring_init)(struct psp_context *psp, enum psp_ring_type ring_type);
+	int (*ring_create)(struct psp_context *psp, enum psp_ring_type ring_type);
+	int (*ring_stop)(struct psp_context *psp,
+			    enum psp_ring_type ring_type);
+	int (*ring_destroy)(struct psp_context *psp,
+			    enum psp_ring_type ring_type);
+	int (*cmd_submit)(struct psp_context *psp, struct amdgpu_firmware_info *ucode,
+			  uint64_t cmd_buf_mc_addr, uint64_t fence_mc_addr, int index);
+	bool (*compare_sram_data)(struct psp_context *psp,
+				  struct amdgpu_firmware_info *ucode,
+				  enum AMDGPU_UCODE_ID ucode_type);
+	bool (*smu_reload_quirk)(struct psp_context *psp);
+	int (*mode1_reset)(struct psp_context *psp);
+};
+
+struct psp_context
+{
+	struct amdgpu_device            *adev;
+	struct psp_ring                 km_ring;
+	struct psp_gfx_cmd_resp		*cmd;
+
+	const struct psp_funcs 		*funcs;
+
+	/* fence buffer */
+	struct amdgpu_bo 		*fw_pri_bo;
+	uint64_t 			fw_pri_mc_addr;
+	void				*fw_pri_buf;
+
+	/* sos firmware */
+	const struct firmware		*sos_fw;
+	uint32_t			sos_fw_version;
+	uint32_t			sos_feature_version;
+	uint32_t			sys_bin_size;
+	uint32_t			sos_bin_size;
+	uint8_t				*sys_start_addr;
+	uint8_t				*sos_start_addr;
+
+	/* tmr buffer */
+	struct amdgpu_bo 		*tmr_bo;
+	uint64_t 			tmr_mc_addr;
+	void				*tmr_buf;
+
+	/* asd firmware and buffer */
+	const struct firmware		*asd_fw;
+	uint32_t			asd_fw_version;
+	uint32_t			asd_feature_version;
+	uint32_t			asd_ucode_size;
+	uint8_t				*asd_start_addr;
+	struct amdgpu_bo 		*asd_shared_bo;
+	uint64_t 			asd_shared_mc_addr;
+	void				*asd_shared_buf;
+
+	/* fence buffer */
+	struct amdgpu_bo 		*fence_buf_bo;
+	uint64_t 			fence_buf_mc_addr;
+	void				*fence_buf;
+
+	/* cmd buffer */
+	struct amdgpu_bo		*cmd_buf_bo;
+	uint64_t			cmd_buf_mc_addr;
+	struct psp_gfx_cmd_resp		*cmd_buf_mem;
+};
+
+struct amdgpu_psp_funcs {
+	bool (*check_fw_loading_status)(struct amdgpu_device *adev,
+					enum AMDGPU_UCODE_ID);
+};
+
+#define psp_prep_cmd_buf(ucode, type) (psp)->funcs->prep_cmd_buf((ucode), (type))
+#define psp_ring_init(psp, type) (psp)->funcs->ring_init((psp), (type))
+#define psp_ring_create(psp, type) (psp)->funcs->ring_create((psp), (type))
+#define psp_ring_stop(psp, type) (psp)->funcs->ring_stop((psp), (type))
+#define psp_ring_destroy(psp, type) ((psp)->funcs->ring_destroy((psp), (type)))
+#define psp_cmd_submit(psp, ucode, cmd_mc, fence_mc, index) \
+		(psp)->funcs->cmd_submit((psp), (ucode), (cmd_mc), (fence_mc), (index))
+#define psp_compare_sram_data(psp, ucode, type) \
+		(psp)->funcs->compare_sram_data((psp), (ucode), (type))
+#define psp_init_microcode(psp) \
+		((psp)->funcs->init_microcode ? (psp)->funcs->init_microcode((psp)) : 0)
+#define psp_bootloader_load_sysdrv(psp) \
+		((psp)->funcs->bootloader_load_sysdrv ? (psp)->funcs->bootloader_load_sysdrv((psp)) : 0)
+#define psp_bootloader_load_sos(psp) \
+		((psp)->funcs->bootloader_load_sos ? (psp)->funcs->bootloader_load_sos((psp)) : 0)
+#define psp_smu_reload_quirk(psp) \
+		((psp)->funcs->smu_reload_quirk ? (psp)->funcs->smu_reload_quirk((psp)) : false)
+#define psp_mode1_reset(psp) \
+		((psp)->funcs->mode1_reset ? (psp)->funcs->mode1_reset((psp)) : false)
+
+extern const struct amd_ip_funcs psp_ip_funcs;
+
+extern const struct amdgpu_ip_block_version psp_v3_1_ip_block;
+extern int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
+			uint32_t field_val, uint32_t mask, bool check_changed);
+
+extern const struct amdgpu_ip_block_version psp_v10_0_ip_block;
+
+int psp_gpu_reset(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_queue_mgr.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_queue_mgr.c
new file mode 100644
index 00000000000..967c762b1dd
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_queue_mgr.c
@@ -0,0 +1,316 @@
+/*
+ * Copyright 2017 Valve Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Andres Rodriguez
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_ring.h"
+
+static int amdgpu_queue_mapper_init(struct amdgpu_queue_mapper *mapper,
+				    int hw_ip)
+{
+	if (!mapper)
+		return -EINVAL;
+
+	if (hw_ip > AMDGPU_MAX_IP_NUM)
+		return -EINVAL;
+
+	mapper->hw_ip = hw_ip;
+	rw_init(&mapper->lock, "agqmap");
+
+	memset(mapper->queue_map, 0, sizeof(mapper->queue_map));
+
+	return 0;
+}
+
+static struct amdgpu_ring *amdgpu_get_cached_map(struct amdgpu_queue_mapper *mapper,
+					  int ring)
+{
+	return mapper->queue_map[ring];
+}
+
+static int amdgpu_update_cached_map(struct amdgpu_queue_mapper *mapper,
+			     int ring, struct amdgpu_ring *pring)
+{
+	if (WARN_ON(mapper->queue_map[ring])) {
+		DRM_ERROR("Un-expected ring re-map\n");
+		return -EINVAL;
+	}
+
+	mapper->queue_map[ring] = pring;
+
+	return 0;
+}
+
+static int amdgpu_identity_map(struct amdgpu_device *adev,
+			       struct amdgpu_queue_mapper *mapper,
+			       u32 ring,
+			       struct amdgpu_ring **out_ring)
+{
+	switch (mapper->hw_ip) {
+	case AMDGPU_HW_IP_GFX:
+		*out_ring = &adev->gfx.gfx_ring[ring];
+		break;
+	case AMDGPU_HW_IP_COMPUTE:
+		*out_ring = &adev->gfx.compute_ring[ring];
+		break;
+	case AMDGPU_HW_IP_DMA:
+		*out_ring = &adev->sdma.instance[ring].ring;
+		break;
+	case AMDGPU_HW_IP_UVD:
+		*out_ring = &adev->uvd.inst[0].ring;
+		break;
+	case AMDGPU_HW_IP_VCE:
+		*out_ring = &adev->vce.ring[ring];
+		break;
+	case AMDGPU_HW_IP_UVD_ENC:
+		*out_ring = &adev->uvd.inst[0].ring_enc[ring];
+		break;
+	case AMDGPU_HW_IP_VCN_DEC:
+		*out_ring = &adev->vcn.ring_dec;
+		break;
+	case AMDGPU_HW_IP_VCN_ENC:
+		*out_ring = &adev->vcn.ring_enc[ring];
+		break;
+	case AMDGPU_HW_IP_VCN_JPEG:
+		*out_ring = &adev->vcn.ring_jpeg;
+		break;
+	default:
+		*out_ring = NULL;
+		DRM_ERROR("unknown HW IP type: %d\n", mapper->hw_ip);
+		return -EINVAL;
+	}
+
+	return amdgpu_update_cached_map(mapper, ring, *out_ring);
+}
+
+static enum amdgpu_ring_type amdgpu_hw_ip_to_ring_type(int hw_ip)
+{
+	switch (hw_ip) {
+	case AMDGPU_HW_IP_GFX:
+		return AMDGPU_RING_TYPE_GFX;
+	case AMDGPU_HW_IP_COMPUTE:
+		return AMDGPU_RING_TYPE_COMPUTE;
+	case AMDGPU_HW_IP_DMA:
+		return AMDGPU_RING_TYPE_SDMA;
+	case AMDGPU_HW_IP_UVD:
+		return AMDGPU_RING_TYPE_UVD;
+	case AMDGPU_HW_IP_VCE:
+		return AMDGPU_RING_TYPE_VCE;
+	default:
+		DRM_ERROR("Invalid HW IP specified %d\n", hw_ip);
+		return -1;
+	}
+}
+
+static int amdgpu_lru_map(struct amdgpu_device *adev,
+			  struct amdgpu_queue_mapper *mapper,
+			  u32 user_ring, bool lru_pipe_order,
+			  struct amdgpu_ring **out_ring)
+{
+	int r, i, j;
+	int ring_type = amdgpu_hw_ip_to_ring_type(mapper->hw_ip);
+	int ring_blacklist[AMDGPU_MAX_RINGS];
+	struct amdgpu_ring *ring;
+
+	/* 0 is a valid ring index, so initialize to -1 */
+	memset(ring_blacklist, 0xff, sizeof(ring_blacklist));
+
+	for (i = 0, j = 0; i < AMDGPU_MAX_RINGS; i++) {
+		ring = mapper->queue_map[i];
+		if (ring)
+			ring_blacklist[j++] = ring->idx;
+	}
+
+	r = amdgpu_ring_lru_get(adev, ring_type, ring_blacklist,
+				j, lru_pipe_order, out_ring);
+	if (r)
+		return r;
+
+	return amdgpu_update_cached_map(mapper, user_ring, *out_ring);
+}
+
+/**
+ * amdgpu_queue_mgr_init - init an amdgpu_queue_mgr struct
+ *
+ * @adev: amdgpu_device pointer
+ * @mgr: amdgpu_queue_mgr structure holding queue information
+ *
+ * Initialize the the selected @mgr (all asics).
+ *
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_queue_mgr_init(struct amdgpu_device *adev,
+			  struct amdgpu_queue_mgr *mgr)
+{
+	int i, r;
+
+	if (!adev || !mgr)
+		return -EINVAL;
+
+	memset(mgr, 0, sizeof(*mgr));
+
+	for (i = 0; i < AMDGPU_MAX_IP_NUM; ++i) {
+		r = amdgpu_queue_mapper_init(&mgr->mapper[i], i);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_queue_mgr_fini - de-initialize an amdgpu_queue_mgr struct
+ *
+ * @adev: amdgpu_device pointer
+ * @mgr: amdgpu_queue_mgr structure holding queue information
+ *
+ * De-initialize the the selected @mgr (all asics).
+ *
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_queue_mgr_fini(struct amdgpu_device *adev,
+			  struct amdgpu_queue_mgr *mgr)
+{
+	return 0;
+}
+
+/**
+ * amdgpu_queue_mgr_map - Map a userspace ring id to an amdgpu_ring
+ *
+ * @adev: amdgpu_device pointer
+ * @mgr: amdgpu_queue_mgr structure holding queue information
+ * @hw_ip: HW IP enum
+ * @instance: HW instance
+ * @ring: user ring id
+ * @our_ring: pointer to mapped amdgpu_ring
+ *
+ * Map a userspace ring id to an appropriate kernel ring. Different
+ * policies are configurable at a HW IP level.
+ *
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_queue_mgr_map(struct amdgpu_device *adev,
+			 struct amdgpu_queue_mgr *mgr,
+			 u32 hw_ip, u32 instance, u32 ring,
+			 struct amdgpu_ring **out_ring)
+{
+	int i, r, ip_num_rings = 0;
+	struct amdgpu_queue_mapper *mapper = &mgr->mapper[hw_ip];
+
+	if (!adev || !mgr || !out_ring)
+		return -EINVAL;
+
+	if (hw_ip >= AMDGPU_MAX_IP_NUM)
+		return -EINVAL;
+
+	if (ring >= AMDGPU_MAX_RINGS)
+		return -EINVAL;
+
+	/* Right now all IPs have only one instance - multiple rings. */
+	if (instance != 0) {
+		DRM_DEBUG("invalid ip instance: %d\n", instance);
+		return -EINVAL;
+	}
+
+	switch (hw_ip) {
+	case AMDGPU_HW_IP_GFX:
+		ip_num_rings = adev->gfx.num_gfx_rings;
+		break;
+	case AMDGPU_HW_IP_COMPUTE:
+		ip_num_rings = adev->gfx.num_compute_rings;
+		break;
+	case AMDGPU_HW_IP_DMA:
+		ip_num_rings = adev->sdma.num_instances;
+		break;
+	case AMDGPU_HW_IP_UVD:
+		for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+			if (!(adev->uvd.harvest_config & (1 << i)))
+				ip_num_rings++;
+		}
+		break;
+	case AMDGPU_HW_IP_VCE:
+		ip_num_rings = adev->vce.num_rings;
+		break;
+	case AMDGPU_HW_IP_UVD_ENC:
+		for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+			if (!(adev->uvd.harvest_config & (1 << i)))
+				ip_num_rings++;
+		}
+		ip_num_rings =
+			adev->uvd.num_enc_rings * ip_num_rings;
+		break;
+	case AMDGPU_HW_IP_VCN_DEC:
+		ip_num_rings = 1;
+		break;
+	case AMDGPU_HW_IP_VCN_ENC:
+		ip_num_rings = adev->vcn.num_enc_rings;
+		break;
+	case AMDGPU_HW_IP_VCN_JPEG:
+		ip_num_rings = 1;
+		break;
+	default:
+		DRM_DEBUG("unknown ip type: %d\n", hw_ip);
+		return -EINVAL;
+	}
+
+	if (ring >= ip_num_rings) {
+		DRM_DEBUG("Ring index:%d exceeds maximum:%d for ip:%d\n",
+			  ring, ip_num_rings, hw_ip);
+		return -EINVAL;
+	}
+
+	mutex_lock(&mapper->lock);
+
+	*out_ring = amdgpu_get_cached_map(mapper, ring);
+	if (*out_ring) {
+		/* cache hit */
+		r = 0;
+		goto out_unlock;
+	}
+
+	switch (mapper->hw_ip) {
+	case AMDGPU_HW_IP_GFX:
+	case AMDGPU_HW_IP_UVD:
+	case AMDGPU_HW_IP_VCE:
+	case AMDGPU_HW_IP_UVD_ENC:
+	case AMDGPU_HW_IP_VCN_DEC:
+	case AMDGPU_HW_IP_VCN_ENC:
+	case AMDGPU_HW_IP_VCN_JPEG:
+		r = amdgpu_identity_map(adev, mapper, ring, out_ring);
+		break;
+	case AMDGPU_HW_IP_DMA:
+		r = amdgpu_lru_map(adev, mapper, ring, false, out_ring);
+		break;
+	case AMDGPU_HW_IP_COMPUTE:
+		r = amdgpu_lru_map(adev, mapper, ring, true, out_ring);
+		break;
+	default:
+		*out_ring = NULL;
+		r = -EINVAL;
+		DRM_DEBUG("unknown HW IP type: %d\n", mapper->hw_ip);
+	}
+
+out_unlock:
+	mutex_unlock(&mapper->lock);
+	return r;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ring.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ring.c
new file mode 100644
index 00000000000..ee18ba3ce97
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ring.c
@@ -0,0 +1,575 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ *          Christian König
+ */
+#include <linux/seq_file.h>
+#include <linux/slab.h>
+#include <linux/debugfs.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "atom.h"
+
+/*
+ * Rings
+ * Most engines on the GPU are fed via ring buffers.  Ring
+ * buffers are areas of GPU accessible memory that the host
+ * writes commands into and the GPU reads commands out of.
+ * There is a rptr (read pointer) that determines where the
+ * GPU is currently reading, and a wptr (write pointer)
+ * which determines where the host has written.  When the
+ * pointers are equal, the ring is idle.  When the host
+ * writes commands to the ring buffer, it increments the
+ * wptr.  The GPU then starts fetching commands and executes
+ * them until the pointers are equal again.
+ */
+static int amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
+				    struct amdgpu_ring *ring);
+static void amdgpu_debugfs_ring_fini(struct amdgpu_ring *ring);
+
+/**
+ * amdgpu_ring_alloc - allocate space on the ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring structure holding ring information
+ * @ndw: number of dwords to allocate in the ring buffer
+ *
+ * Allocate @ndw dwords in the ring buffer (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw)
+{
+	/* Align requested size with padding so unlock_commit can
+	 * pad safely */
+	ndw = (ndw + ring->funcs->align_mask) & ~ring->funcs->align_mask;
+
+	/* Make sure we aren't trying to allocate more space
+	 * than the maximum for one submission
+	 */
+	if (WARN_ON_ONCE(ndw > ring->max_dw))
+		return -ENOMEM;
+
+	ring->count_dw = ndw;
+	ring->wptr_old = ring->wptr;
+
+	if (ring->funcs->begin_use)
+		ring->funcs->begin_use(ring);
+
+	return 0;
+}
+
+/** amdgpu_ring_insert_nop - insert NOP packets
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @count: the number of NOP packets to insert
+ *
+ * This is the generic insert_nop function for rings except SDMA
+ */
+void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	for (i = 0; i < count; i++)
+		amdgpu_ring_write(ring, ring->funcs->nop);
+}
+
+/** amdgpu_ring_generic_pad_ib - pad IB with NOP packets
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @ib: IB to add NOP packets to
+ *
+ * This is the generic pad_ib function for rings except SDMA
+ */
+void amdgpu_ring_generic_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
+{
+	while (ib->length_dw & ring->funcs->align_mask)
+		ib->ptr[ib->length_dw++] = ring->funcs->nop;
+}
+
+/**
+ * amdgpu_ring_commit - tell the GPU to execute the new
+ * commands on the ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Update the wptr (write pointer) to tell the GPU to
+ * execute new commands on the ring buffer (all asics).
+ */
+void amdgpu_ring_commit(struct amdgpu_ring *ring)
+{
+	uint32_t count;
+
+	/* We pad to match fetch size */
+	count = ring->funcs->align_mask + 1 -
+		(ring->wptr & ring->funcs->align_mask);
+	count %= ring->funcs->align_mask + 1;
+	ring->funcs->insert_nop(ring, count);
+
+	mb();
+	amdgpu_ring_set_wptr(ring);
+
+	if (ring->funcs->end_use)
+		ring->funcs->end_use(ring);
+
+	if (ring->funcs->type != AMDGPU_RING_TYPE_KIQ)
+		amdgpu_ring_lru_touch(ring->adev, ring);
+}
+
+/**
+ * amdgpu_ring_undo - reset the wptr
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Reset the driver's copy of the wptr (all asics).
+ */
+void amdgpu_ring_undo(struct amdgpu_ring *ring)
+{
+	ring->wptr = ring->wptr_old;
+
+	if (ring->funcs->end_use)
+		ring->funcs->end_use(ring);
+}
+
+/**
+ * amdgpu_ring_priority_put - restore a ring's priority
+ *
+ * @ring: amdgpu_ring structure holding the information
+ * @priority: target priority
+ *
+ * Release a request for executing at @priority
+ */
+void amdgpu_ring_priority_put(struct amdgpu_ring *ring,
+			      enum drm_sched_priority priority)
+{
+	int i;
+
+	if (!ring->funcs->set_priority)
+		return;
+
+	if (atomic_dec_return(&ring->num_jobs[priority]) > 0)
+		return;
+
+	/* no need to restore if the job is already at the lowest priority */
+	if (priority == DRM_SCHED_PRIORITY_NORMAL)
+		return;
+
+	mutex_lock(&ring->priority_mutex);
+	/* something higher prio is executing, no need to decay */
+	if (ring->priority > priority)
+		goto out_unlock;
+
+	/* decay priority to the next level with a job available */
+	for (i = priority; i >= DRM_SCHED_PRIORITY_MIN; i--) {
+		if (i == DRM_SCHED_PRIORITY_NORMAL
+				|| atomic_read(&ring->num_jobs[i])) {
+			ring->priority = i;
+			ring->funcs->set_priority(ring, i);
+			break;
+		}
+	}
+
+out_unlock:
+	mutex_unlock(&ring->priority_mutex);
+}
+
+/**
+ * amdgpu_ring_priority_get - change the ring's priority
+ *
+ * @ring: amdgpu_ring structure holding the information
+ * @priority: target priority
+ *
+ * Request a ring's priority to be raised to @priority (refcounted).
+ */
+void amdgpu_ring_priority_get(struct amdgpu_ring *ring,
+			      enum drm_sched_priority priority)
+{
+	if (!ring->funcs->set_priority)
+		return;
+
+	if (atomic_inc_return(&ring->num_jobs[priority]) <= 0)
+		return;
+
+	mutex_lock(&ring->priority_mutex);
+	if (priority <= ring->priority)
+		goto out_unlock;
+
+	ring->priority = priority;
+	ring->funcs->set_priority(ring, priority);
+
+out_unlock:
+	mutex_unlock(&ring->priority_mutex);
+}
+
+/**
+ * amdgpu_ring_init - init driver ring struct.
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring structure holding ring information
+ * @max_ndw: maximum number of dw for ring alloc
+ * @nop: nop packet for this ring
+ *
+ * Initialize the driver information for the selected ring (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
+		     unsigned max_dw, struct amdgpu_irq_src *irq_src,
+		     unsigned irq_type)
+{
+	int r, i;
+	int sched_hw_submission = amdgpu_sched_hw_submission;
+
+	/* Set the hw submission limit higher for KIQ because
+	 * it's used for a number of gfx/compute tasks by both
+	 * KFD and KGD which may have outstanding fences and
+	 * it doesn't really use the gpu scheduler anyway;
+	 * KIQ tasks get submitted directly to the ring.
+	 */
+	if (ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
+		sched_hw_submission = max(sched_hw_submission, 256);
+
+	if (ring->adev == NULL) {
+		if (adev->num_rings >= AMDGPU_MAX_RINGS)
+			return -EINVAL;
+
+		ring->adev = adev;
+		ring->idx = adev->num_rings++;
+		adev->rings[ring->idx] = ring;
+		r = amdgpu_fence_driver_init_ring(ring, sched_hw_submission);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_device_wb_get(adev, &ring->rptr_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring rptr_offs wb alloc failed\n", r);
+		return r;
+	}
+
+	r = amdgpu_device_wb_get(adev, &ring->wptr_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring wptr_offs wb alloc failed\n", r);
+		return r;
+	}
+
+	r = amdgpu_device_wb_get(adev, &ring->fence_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring fence_offs wb alloc failed\n", r);
+		return r;
+	}
+
+	r = amdgpu_device_wb_get(adev, &ring->cond_exe_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring cond_exec_polling wb alloc failed\n", r);
+		return r;
+	}
+	ring->cond_exe_gpu_addr = adev->wb.gpu_addr + (ring->cond_exe_offs * 4);
+	ring->cond_exe_cpu_addr = &adev->wb.wb[ring->cond_exe_offs];
+	/* always set cond_exec_polling to CONTINUE */
+	*ring->cond_exe_cpu_addr = 1;
+
+	r = amdgpu_fence_driver_start_ring(ring, irq_src, irq_type);
+	if (r) {
+		dev_err(adev->dev, "failed initializing fences (%d).\n", r);
+		return r;
+	}
+
+	ring->ring_size = roundup_pow_of_two(max_dw * 4 * sched_hw_submission);
+
+	ring->buf_mask = (ring->ring_size / 4) - 1;
+	ring->ptr_mask = ring->funcs->support_64bit_ptrs ?
+		0xffffffffffffffff : ring->buf_mask;
+	/* Allocate ring buffer */
+	if (ring->ring_obj == NULL) {
+		r = amdgpu_bo_create_kernel(adev, ring->ring_size + ring->funcs->extra_dw, PAGE_SIZE,
+					    AMDGPU_GEM_DOMAIN_GTT,
+					    &ring->ring_obj,
+					    &ring->gpu_addr,
+					    (void **)&ring->ring);
+		if (r) {
+			dev_err(adev->dev, "(%d) ring create failed\n", r);
+			return r;
+		}
+		amdgpu_ring_clear_ring(ring);
+	}
+
+	ring->max_dw = max_dw;
+	ring->priority = DRM_SCHED_PRIORITY_NORMAL;
+	rw_init(&ring->priority_mutex, "ringpri");
+	INIT_LIST_HEAD(&ring->lru_list);
+	amdgpu_ring_lru_touch(adev, ring);
+
+	for (i = 0; i < DRM_SCHED_PRIORITY_MAX; ++i)
+		atomic_set(&ring->num_jobs[i], 0);
+
+	if (amdgpu_debugfs_ring_init(adev, ring)) {
+		DRM_ERROR("Failed to register debugfs file for rings !\n");
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_ring_fini - tear down the driver ring struct.
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Tear down the driver information for the selected ring (all asics).
+ */
+void amdgpu_ring_fini(struct amdgpu_ring *ring)
+{
+	ring->ready = false;
+
+	/* Not to finish a ring which is not initialized */
+	if (!(ring->adev) || !(ring->adev->rings[ring->idx]))
+		return;
+
+	amdgpu_device_wb_free(ring->adev, ring->rptr_offs);
+	amdgpu_device_wb_free(ring->adev, ring->wptr_offs);
+
+	amdgpu_device_wb_free(ring->adev, ring->cond_exe_offs);
+	amdgpu_device_wb_free(ring->adev, ring->fence_offs);
+
+	amdgpu_bo_free_kernel(&ring->ring_obj,
+			      &ring->gpu_addr,
+			      (void **)&ring->ring);
+
+	amdgpu_debugfs_ring_fini(ring);
+
+	dma_fence_put(ring->vmid_wait);
+	ring->vmid_wait = NULL;
+	ring->me = 0;
+
+	ring->adev->rings[ring->idx] = NULL;
+}
+
+static void amdgpu_ring_lru_touch_locked(struct amdgpu_device *adev,
+					 struct amdgpu_ring *ring)
+{
+	/* list_move_tail handles the case where ring isn't part of the list */
+	list_move_tail(&ring->lru_list, &adev->ring_lru_list);
+}
+
+static bool amdgpu_ring_is_blacklisted(struct amdgpu_ring *ring,
+				       int *blacklist, int num_blacklist)
+{
+	int i;
+
+	for (i = 0; i < num_blacklist; i++) {
+		if (ring->idx == blacklist[i])
+			return true;
+	}
+
+	return false;
+}
+
+/**
+ * amdgpu_ring_lru_get - get the least recently used ring for a HW IP block
+ *
+ * @adev: amdgpu_device pointer
+ * @type: amdgpu_ring_type enum
+ * @blacklist: blacklisted ring ids array
+ * @num_blacklist: number of entries in @blacklist
+ * @lru_pipe_order: find a ring from the least recently used pipe
+ * @ring: output ring
+ *
+ * Retrieve the amdgpu_ring structure for the least recently used ring of
+ * a specific IP block (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_ring_lru_get(struct amdgpu_device *adev, int type,
+			int *blacklist,	int num_blacklist,
+			bool lru_pipe_order, struct amdgpu_ring **ring)
+{
+	struct amdgpu_ring *entry;
+
+	/* List is sorted in LRU order, find first entry corresponding
+	 * to the desired HW IP */
+	*ring = NULL;
+	spin_lock(&adev->ring_lru_list_lock);
+	list_for_each_entry(entry, &adev->ring_lru_list, lru_list) {
+		if (entry->funcs->type != type)
+			continue;
+
+		if (amdgpu_ring_is_blacklisted(entry, blacklist, num_blacklist))
+			continue;
+
+		if (!*ring) {
+			*ring = entry;
+
+			/* We are done for ring LRU */
+			if (!lru_pipe_order)
+				break;
+		}
+
+		/* Move all rings on the same pipe to the end of the list */
+		if (entry->pipe == (*ring)->pipe)
+			amdgpu_ring_lru_touch_locked(adev, entry);
+	}
+
+	/* Move the ring we found to the end of the list */
+	if (*ring)
+		amdgpu_ring_lru_touch_locked(adev, *ring);
+
+	spin_unlock(&adev->ring_lru_list_lock);
+
+	if (!*ring) {
+		DRM_ERROR("Ring LRU contains no entries for ring type:%d\n", type);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_ring_lru_touch - mark a ring as recently being used
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: ring to touch
+ *
+ * Move @ring to the tail of the lru list
+ */
+void amdgpu_ring_lru_touch(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	spin_lock(&adev->ring_lru_list_lock);
+	amdgpu_ring_lru_touch_locked(adev, ring);
+	spin_unlock(&adev->ring_lru_list_lock);
+}
+
+/**
+ * amdgpu_ring_emit_reg_write_reg_wait_helper - ring helper
+ *
+ * @adev: amdgpu_device pointer
+ * @reg0: register to write
+ * @reg1: register to wait on
+ * @ref: reference value to write/wait on
+ * @mask: mask to wait on
+ *
+ * Helper for rings that don't support write and wait in a
+ * single oneshot packet.
+ */
+void amdgpu_ring_emit_reg_write_reg_wait_helper(struct amdgpu_ring *ring,
+						uint32_t reg0, uint32_t reg1,
+						uint32_t ref, uint32_t mask)
+{
+	amdgpu_ring_emit_wreg(ring, reg0, ref);
+	amdgpu_ring_emit_reg_wait(ring, reg1, mask, mask);
+}
+
+/*
+ * Debugfs info
+ */
+#if defined(CONFIG_DEBUG_FS)
+
+/* Layout of file is 12 bytes consisting of
+ * - rptr
+ * - wptr
+ * - driver's copy of wptr
+ *
+ * followed by n-words of ring data
+ */
+static ssize_t amdgpu_debugfs_ring_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_ring *ring = file_inode(f)->i_private;
+	int r, i;
+	uint32_t value, result, early[3];
+
+	if (*pos & 3 || size & 3)
+		return -EINVAL;
+
+	result = 0;
+
+	if (*pos < 12) {
+		early[0] = amdgpu_ring_get_rptr(ring) & ring->buf_mask;
+		early[1] = amdgpu_ring_get_wptr(ring) & ring->buf_mask;
+		early[2] = ring->wptr & ring->buf_mask;
+		for (i = *pos / 4; i < 3 && size; i++) {
+			r = put_user(early[i], (uint32_t *)buf);
+			if (r)
+				return r;
+			buf += 4;
+			result += 4;
+			size -= 4;
+			*pos += 4;
+		}
+	}
+
+	while (size) {
+		if (*pos >= (ring->ring_size + 12))
+			return result;
+
+		value = ring->ring[(*pos - 12)/4];
+		r = put_user(value, (uint32_t*)buf);
+		if (r)
+			return r;
+		buf += 4;
+		result += 4;
+		size -= 4;
+		*pos += 4;
+	}
+
+	return result;
+}
+
+static const struct file_operations amdgpu_debugfs_ring_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_ring_read,
+	.llseek = default_llseek
+};
+
+#endif
+
+static int amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
+				    struct amdgpu_ring *ring)
+{
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev->ddev->primary;
+	struct dentry *ent, *root = minor->debugfs_root;
+	char name[32];
+
+	snprintf(name, sizeof(name), "amdgpu_ring_%s", ring->name);
+
+	ent = debugfs_create_file(name,
+				  S_IFREG | S_IRUGO, root,
+				  ring, &amdgpu_debugfs_ring_fops);
+	if (!ent)
+		return -ENOMEM;
+
+	i_size_write(ent->d_inode, ring->ring_size + 12);
+	ring->ent = ent;
+#endif
+	return 0;
+}
+
+static void amdgpu_debugfs_ring_fini(struct amdgpu_ring *ring)
+{
+#if defined(CONFIG_DEBUG_FS)
+	debugfs_remove(ring->ent);
+#endif
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ring.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ring.h
new file mode 100644
index 00000000000..5be66e89532
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ring.h
@@ -0,0 +1,293 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König
+ */
+#ifndef __AMDGPU_RING_H__
+#define __AMDGPU_RING_H__
+
+#include <drm/amdgpu_drm.h>
+#include <drm/gpu_scheduler.h>
+#include <drm/drm_print.h>
+
+/* max number of rings */
+#define AMDGPU_MAX_RINGS		21
+#define AMDGPU_MAX_GFX_RINGS		1
+#define AMDGPU_MAX_COMPUTE_RINGS	8
+#define AMDGPU_MAX_VCE_RINGS		3
+#define AMDGPU_MAX_UVD_ENC_RINGS	2
+
+/* some special values for the owner field */
+#define AMDGPU_FENCE_OWNER_UNDEFINED	((void *)0ul)
+#define AMDGPU_FENCE_OWNER_VM		((void *)1ul)
+#define AMDGPU_FENCE_OWNER_KFD		((void *)2ul)
+
+#define AMDGPU_FENCE_FLAG_64BIT         (1 << 0)
+#define AMDGPU_FENCE_FLAG_INT           (1 << 1)
+#define AMDGPU_FENCE_FLAG_TC_WB_ONLY    (1 << 2)
+
+#define to_amdgpu_ring(s) container_of((s), struct amdgpu_ring, sched)
+
+enum amdgpu_ring_type {
+	AMDGPU_RING_TYPE_GFX,
+	AMDGPU_RING_TYPE_COMPUTE,
+	AMDGPU_RING_TYPE_SDMA,
+	AMDGPU_RING_TYPE_UVD,
+	AMDGPU_RING_TYPE_VCE,
+	AMDGPU_RING_TYPE_KIQ,
+	AMDGPU_RING_TYPE_UVD_ENC,
+	AMDGPU_RING_TYPE_VCN_DEC,
+	AMDGPU_RING_TYPE_VCN_ENC,
+	AMDGPU_RING_TYPE_VCN_JPEG
+};
+
+struct amdgpu_device;
+struct amdgpu_ring;
+struct amdgpu_ib;
+struct amdgpu_cs_parser;
+struct amdgpu_job;
+
+/*
+ * Fences.
+ */
+struct amdgpu_fence_driver {
+	uint64_t			gpu_addr;
+	volatile uint32_t		*cpu_addr;
+	/* sync_seq is protected by ring emission lock */
+	uint32_t			sync_seq;
+	atomic_t			last_seq;
+	bool				initialized;
+	struct amdgpu_irq_src		*irq_src;
+	unsigned			irq_type;
+	struct timeout			fallback_timer;
+	unsigned			num_fences_mask;
+	spinlock_t			lock;
+	struct dma_fence		**fences;
+};
+
+int amdgpu_fence_driver_init(struct amdgpu_device *adev);
+void amdgpu_fence_driver_fini(struct amdgpu_device *adev);
+void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring);
+
+int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring,
+				  unsigned num_hw_submission);
+int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring,
+				   struct amdgpu_irq_src *irq_src,
+				   unsigned irq_type);
+void amdgpu_fence_driver_suspend(struct amdgpu_device *adev);
+void amdgpu_fence_driver_resume(struct amdgpu_device *adev);
+int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **fence,
+		      unsigned flags);
+int amdgpu_fence_emit_polling(struct amdgpu_ring *ring, uint32_t *s);
+void amdgpu_fence_process(struct amdgpu_ring *ring);
+int amdgpu_fence_wait_empty(struct amdgpu_ring *ring);
+signed long amdgpu_fence_wait_polling(struct amdgpu_ring *ring,
+				      uint32_t wait_seq,
+				      signed long timeout);
+unsigned amdgpu_fence_count_emitted(struct amdgpu_ring *ring);
+
+/*
+ * Rings.
+ */
+
+/* provided by hw blocks that expose a ring buffer for commands */
+struct amdgpu_ring_funcs {
+	enum amdgpu_ring_type	type;
+	uint32_t		align_mask;
+	u32			nop;
+	bool			support_64bit_ptrs;
+	unsigned		vmhub;
+	unsigned		extra_dw;
+
+	/* ring read/write ptr handling */
+	u64 (*get_rptr)(struct amdgpu_ring *ring);
+	u64 (*get_wptr)(struct amdgpu_ring *ring);
+	void (*set_wptr)(struct amdgpu_ring *ring);
+	/* validating and patching of IBs */
+	int (*parse_cs)(struct amdgpu_cs_parser *p, uint32_t ib_idx);
+	int (*patch_cs_in_place)(struct amdgpu_cs_parser *p, uint32_t ib_idx);
+	/* constants to calculate how many DW are needed for an emit */
+	unsigned emit_frame_size;
+	unsigned emit_ib_size;
+	/* command emit functions */
+	void (*emit_ib)(struct amdgpu_ring *ring,
+			struct amdgpu_ib *ib,
+			unsigned vmid, bool ctx_switch);
+	void (*emit_fence)(struct amdgpu_ring *ring, uint64_t addr,
+			   uint64_t seq, unsigned flags);
+	void (*emit_pipeline_sync)(struct amdgpu_ring *ring);
+	void (*emit_vm_flush)(struct amdgpu_ring *ring, unsigned vmid,
+			      uint64_t pd_addr);
+	void (*emit_hdp_flush)(struct amdgpu_ring *ring);
+	void (*emit_gds_switch)(struct amdgpu_ring *ring, uint32_t vmid,
+				uint32_t gds_base, uint32_t gds_size,
+				uint32_t gws_base, uint32_t gws_size,
+				uint32_t oa_base, uint32_t oa_size);
+	/* testing functions */
+	int (*test_ring)(struct amdgpu_ring *ring);
+	int (*test_ib)(struct amdgpu_ring *ring, long timeout);
+	/* insert NOP packets */
+	void (*insert_nop)(struct amdgpu_ring *ring, uint32_t count);
+	void (*insert_start)(struct amdgpu_ring *ring);
+	void (*insert_end)(struct amdgpu_ring *ring);
+	/* pad the indirect buffer to the necessary number of dw */
+	void (*pad_ib)(struct amdgpu_ring *ring, struct amdgpu_ib *ib);
+	unsigned (*init_cond_exec)(struct amdgpu_ring *ring);
+	void (*patch_cond_exec)(struct amdgpu_ring *ring, unsigned offset);
+	/* note usage for clock and power gating */
+	void (*begin_use)(struct amdgpu_ring *ring);
+	void (*end_use)(struct amdgpu_ring *ring);
+	void (*emit_switch_buffer) (struct amdgpu_ring *ring);
+	void (*emit_cntxcntl) (struct amdgpu_ring *ring, uint32_t flags);
+	void (*emit_rreg)(struct amdgpu_ring *ring, uint32_t reg);
+	void (*emit_wreg)(struct amdgpu_ring *ring, uint32_t reg, uint32_t val);
+	void (*emit_reg_wait)(struct amdgpu_ring *ring, uint32_t reg,
+			      uint32_t val, uint32_t mask);
+	void (*emit_reg_write_reg_wait)(struct amdgpu_ring *ring,
+					uint32_t reg0, uint32_t reg1,
+					uint32_t ref, uint32_t mask);
+	void (*emit_tmz)(struct amdgpu_ring *ring, bool start);
+	/* priority functions */
+	void (*set_priority) (struct amdgpu_ring *ring,
+			      enum drm_sched_priority priority);
+};
+
+struct amdgpu_ring {
+	struct amdgpu_device		*adev;
+	const struct amdgpu_ring_funcs	*funcs;
+	struct amdgpu_fence_driver	fence_drv;
+	struct drm_gpu_scheduler	sched;
+	struct list_head		lru_list;
+
+	struct amdgpu_bo	*ring_obj;
+	volatile uint32_t	*ring;
+	unsigned		rptr_offs;
+	u64			wptr;
+	u64			wptr_old;
+	unsigned		ring_size;
+	unsigned		max_dw;
+	int			count_dw;
+	uint64_t		gpu_addr;
+	uint64_t		ptr_mask;
+	uint32_t		buf_mask;
+	bool			ready;
+	u32			idx;
+	u32			me;
+	u32			pipe;
+	u32			queue;
+	struct amdgpu_bo	*mqd_obj;
+	uint64_t                mqd_gpu_addr;
+	void                    *mqd_ptr;
+	uint64_t                eop_gpu_addr;
+	u32			doorbell_index;
+	bool			use_doorbell;
+	bool			use_pollmem;
+	unsigned		wptr_offs;
+	unsigned		fence_offs;
+	uint64_t		current_ctx;
+	char			name[16];
+	unsigned		cond_exe_offs;
+	u64			cond_exe_gpu_addr;
+	volatile u32		*cond_exe_cpu_addr;
+	unsigned		vm_inv_eng;
+	struct dma_fence	*vmid_wait;
+	bool			has_compute_vm_bug;
+
+	atomic_t		num_jobs[DRM_SCHED_PRIORITY_MAX];
+	struct rwlock		priority_mutex;
+	/* protected by priority_mutex */
+	int			priority;
+
+#if defined(CONFIG_DEBUG_FS)
+	struct dentry *ent;
+#endif
+};
+
+int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw);
+void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count);
+void amdgpu_ring_generic_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib);
+void amdgpu_ring_commit(struct amdgpu_ring *ring);
+void amdgpu_ring_undo(struct amdgpu_ring *ring);
+void amdgpu_ring_priority_get(struct amdgpu_ring *ring,
+			      enum drm_sched_priority priority);
+void amdgpu_ring_priority_put(struct amdgpu_ring *ring,
+			      enum drm_sched_priority priority);
+int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
+		     unsigned ring_size, struct amdgpu_irq_src *irq_src,
+		     unsigned irq_type);
+void amdgpu_ring_fini(struct amdgpu_ring *ring);
+int amdgpu_ring_lru_get(struct amdgpu_device *adev, int type,
+			int *blacklist, int num_blacklist,
+			bool lru_pipe_order, struct amdgpu_ring **ring);
+void amdgpu_ring_lru_touch(struct amdgpu_device *adev, struct amdgpu_ring *ring);
+void amdgpu_ring_emit_reg_write_reg_wait_helper(struct amdgpu_ring *ring,
+						uint32_t reg0, uint32_t val0,
+						uint32_t reg1, uint32_t val1);
+
+static inline void amdgpu_ring_clear_ring(struct amdgpu_ring *ring)
+{
+	int i = 0;
+	while (i <= ring->buf_mask)
+		ring->ring[i++] = ring->funcs->nop;
+
+}
+
+static inline void amdgpu_ring_write(struct amdgpu_ring *ring, uint32_t v)
+{
+	if (ring->count_dw <= 0)
+		DRM_ERROR("amdgpu: writing more dwords to the ring than expected!\n");
+	ring->ring[ring->wptr++ & ring->buf_mask] = v;
+	ring->wptr &= ring->ptr_mask;
+	ring->count_dw--;
+}
+
+static inline void amdgpu_ring_write_multiple(struct amdgpu_ring *ring,
+					      void *src, int count_dw)
+{
+	unsigned occupied, chunk1, chunk2;
+	void *dst;
+
+	if (unlikely(ring->count_dw < count_dw))
+		DRM_ERROR("amdgpu: writing more dwords to the ring than expected!\n");
+
+	occupied = ring->wptr & ring->buf_mask;
+	dst = (void *)&ring->ring[occupied];
+	chunk1 = ring->buf_mask + 1 - occupied;
+	chunk1 = (chunk1 >= count_dw) ? count_dw: chunk1;
+	chunk2 = count_dw - chunk1;
+	chunk1 <<= 2;
+	chunk2 <<= 2;
+
+	if (chunk1)
+		memcpy(dst, src, chunk1);
+
+	if (chunk2) {
+		src += chunk1;
+		dst = (void *)ring->ring;
+		memcpy(dst, src, chunk2);
+	}
+
+	ring->wptr += count_dw;
+	ring->wptr &= ring->ptr_mask;
+	ring->count_dw -= count_dw;
+}
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_sa.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_sa.c
new file mode 100644
index 00000000000..614cba6767f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_sa.c
@@ -0,0 +1,403 @@
+/*
+ * Copyright 2011 Red Hat Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Jerome Glisse <glisse@freedesktop.org>
+ */
+/* Algorithm:
+ *
+ * We store the last allocated bo in "hole", we always try to allocate
+ * after the last allocated bo. Principle is that in a linear GPU ring
+ * progression was is after last is the oldest bo we allocated and thus
+ * the first one that should no longer be in use by the GPU.
+ *
+ * If it's not the case we skip over the bo after last to the closest
+ * done bo if such one exist. If none exist and we are not asked to
+ * block we report failure to allocate.
+ *
+ * If we are asked to block we wait on all the oldest fence of all
+ * rings. We just wait for any of those fence to complete.
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+
+static void amdgpu_sa_bo_remove_locked(struct amdgpu_sa_bo *sa_bo);
+static void amdgpu_sa_bo_try_free(struct amdgpu_sa_manager *sa_manager);
+
+int amdgpu_sa_bo_manager_init(struct amdgpu_device *adev,
+			      struct amdgpu_sa_manager *sa_manager,
+			      unsigned size, u32 align, u32 domain)
+{
+	int i, r;
+
+	init_waitqueue_head(&sa_manager->wq);
+	sa_manager->bo = NULL;
+	sa_manager->size = size;
+	sa_manager->domain = domain;
+	sa_manager->align = align;
+	sa_manager->hole = &sa_manager->olist;
+	INIT_LIST_HEAD(&sa_manager->olist);
+	for (i = 0; i < AMDGPU_SA_NUM_FENCE_LISTS; ++i)
+		INIT_LIST_HEAD(&sa_manager->flist[i]);
+
+	r = amdgpu_bo_create_kernel(adev, size, align, domain, &sa_manager->bo,
+				&sa_manager->gpu_addr, &sa_manager->cpu_ptr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate bo for manager\n", r);
+		return r;
+	}
+
+	memset(sa_manager->cpu_ptr, 0, sa_manager->size);
+	return r;
+}
+
+void amdgpu_sa_bo_manager_fini(struct amdgpu_device *adev,
+                              struct amdgpu_sa_manager *sa_manager)
+{
+	struct amdgpu_sa_bo *sa_bo, *tmp;
+
+	if (sa_manager->bo == NULL) {
+		dev_err(adev->dev, "no bo for sa manager\n");
+		return;
+	}
+
+	if (!list_empty(&sa_manager->olist)) {
+		sa_manager->hole = &sa_manager->olist,
+		amdgpu_sa_bo_try_free(sa_manager);
+		if (!list_empty(&sa_manager->olist)) {
+			dev_err(adev->dev, "sa_manager is not empty, clearing anyway\n");
+		}
+	}
+	list_for_each_entry_safe(sa_bo, tmp, &sa_manager->olist, olist) {
+		amdgpu_sa_bo_remove_locked(sa_bo);
+	}
+
+	amdgpu_bo_free_kernel(&sa_manager->bo, &sa_manager->gpu_addr, &sa_manager->cpu_ptr);
+	sa_manager->size = 0;
+}
+
+static void amdgpu_sa_bo_remove_locked(struct amdgpu_sa_bo *sa_bo)
+{
+	struct amdgpu_sa_manager *sa_manager = sa_bo->manager;
+	if (sa_manager->hole == &sa_bo->olist) {
+		sa_manager->hole = sa_bo->olist.prev;
+	}
+	list_del_init(&sa_bo->olist);
+	list_del_init(&sa_bo->flist);
+	dma_fence_put(sa_bo->fence);
+	kfree(sa_bo);
+}
+
+static void amdgpu_sa_bo_try_free(struct amdgpu_sa_manager *sa_manager)
+{
+	struct amdgpu_sa_bo *sa_bo, *tmp;
+
+	if (sa_manager->hole->next == &sa_manager->olist)
+		return;
+
+	sa_bo = list_entry(sa_manager->hole->next, struct amdgpu_sa_bo, olist);
+	list_for_each_entry_safe_from(sa_bo, tmp, &sa_manager->olist, olist) {
+		if (sa_bo->fence == NULL ||
+		    !dma_fence_is_signaled(sa_bo->fence)) {
+			return;
+		}
+		amdgpu_sa_bo_remove_locked(sa_bo);
+	}
+}
+
+static inline unsigned amdgpu_sa_bo_hole_soffset(struct amdgpu_sa_manager *sa_manager)
+{
+	struct list_head *hole = sa_manager->hole;
+
+	if (hole != &sa_manager->olist) {
+		return list_entry(hole, struct amdgpu_sa_bo, olist)->eoffset;
+	}
+	return 0;
+}
+
+static inline unsigned amdgpu_sa_bo_hole_eoffset(struct amdgpu_sa_manager *sa_manager)
+{
+	struct list_head *hole = sa_manager->hole;
+
+	if (hole->next != &sa_manager->olist) {
+		return list_entry(hole->next, struct amdgpu_sa_bo, olist)->soffset;
+	}
+	return sa_manager->size;
+}
+
+static bool amdgpu_sa_bo_try_alloc(struct amdgpu_sa_manager *sa_manager,
+				   struct amdgpu_sa_bo *sa_bo,
+				   unsigned size, unsigned align)
+{
+	unsigned soffset, eoffset, wasted;
+
+	soffset = amdgpu_sa_bo_hole_soffset(sa_manager);
+	eoffset = amdgpu_sa_bo_hole_eoffset(sa_manager);
+	wasted = (align - (soffset % align)) % align;
+
+	if ((eoffset - soffset) >= (size + wasted)) {
+		soffset += wasted;
+
+		sa_bo->manager = sa_manager;
+		sa_bo->soffset = soffset;
+		sa_bo->eoffset = soffset + size;
+		list_add(&sa_bo->olist, sa_manager->hole);
+		INIT_LIST_HEAD(&sa_bo->flist);
+		sa_manager->hole = &sa_bo->olist;
+		return true;
+	}
+	return false;
+}
+
+/**
+ * amdgpu_sa_event - Check if we can stop waiting
+ *
+ * @sa_manager: pointer to the sa_manager
+ * @size: number of bytes we want to allocate
+ * @align: alignment we need to match
+ *
+ * Check if either there is a fence we can wait for or
+ * enough free memory to satisfy the allocation directly
+ */
+static bool amdgpu_sa_event(struct amdgpu_sa_manager *sa_manager,
+			    unsigned size, unsigned align)
+{
+	unsigned soffset, eoffset, wasted;
+	int i;
+
+	for (i = 0; i < AMDGPU_SA_NUM_FENCE_LISTS; ++i)
+		if (!list_empty(&sa_manager->flist[i]))
+			return true;
+
+	soffset = amdgpu_sa_bo_hole_soffset(sa_manager);
+	eoffset = amdgpu_sa_bo_hole_eoffset(sa_manager);
+	wasted = (align - (soffset % align)) % align;
+
+	if ((eoffset - soffset) >= (size + wasted)) {
+		return true;
+	}
+
+	return false;
+}
+
+static bool amdgpu_sa_bo_next_hole(struct amdgpu_sa_manager *sa_manager,
+				   struct dma_fence **fences,
+				   unsigned *tries)
+{
+	struct amdgpu_sa_bo *best_bo = NULL;
+	unsigned i, soffset, best, tmp;
+
+	/* if hole points to the end of the buffer */
+	if (sa_manager->hole->next == &sa_manager->olist) {
+		/* try again with its beginning */
+		sa_manager->hole = &sa_manager->olist;
+		return true;
+	}
+
+	soffset = amdgpu_sa_bo_hole_soffset(sa_manager);
+	/* to handle wrap around we add sa_manager->size */
+	best = sa_manager->size * 2;
+	/* go over all fence list and try to find the closest sa_bo
+	 * of the current last
+	 */
+	for (i = 0; i < AMDGPU_SA_NUM_FENCE_LISTS; ++i) {
+		struct amdgpu_sa_bo *sa_bo;
+
+		if (list_empty(&sa_manager->flist[i]))
+			continue;
+
+		sa_bo = list_first_entry(&sa_manager->flist[i],
+					 struct amdgpu_sa_bo, flist);
+
+		if (!dma_fence_is_signaled(sa_bo->fence)) {
+			fences[i] = sa_bo->fence;
+			continue;
+		}
+
+		/* limit the number of tries each ring gets */
+		if (tries[i] > 2) {
+			continue;
+		}
+
+		tmp = sa_bo->soffset;
+		if (tmp < soffset) {
+			/* wrap around, pretend it's after */
+			tmp += sa_manager->size;
+		}
+		tmp -= soffset;
+		if (tmp < best) {
+			/* this sa bo is the closest one */
+			best = tmp;
+			best_bo = sa_bo;
+		}
+	}
+
+	if (best_bo) {
+		uint32_t idx = best_bo->fence->context;
+
+		idx %= AMDGPU_SA_NUM_FENCE_LISTS;
+		++tries[idx];
+		sa_manager->hole = best_bo->olist.prev;
+
+		/* we knew that this one is signaled,
+		   so it's save to remote it */
+		amdgpu_sa_bo_remove_locked(best_bo);
+		return true;
+	}
+	return false;
+}
+
+int amdgpu_sa_bo_new(struct amdgpu_sa_manager *sa_manager,
+		     struct amdgpu_sa_bo **sa_bo,
+		     unsigned size, unsigned align)
+{
+	struct dma_fence *fences[AMDGPU_SA_NUM_FENCE_LISTS];
+	unsigned tries[AMDGPU_SA_NUM_FENCE_LISTS];
+	unsigned count;
+	int i, r;
+	signed long t;
+
+	if (WARN_ON_ONCE(align > sa_manager->align))
+		return -EINVAL;
+
+	if (WARN_ON_ONCE(size > sa_manager->size))
+		return -EINVAL;
+
+	*sa_bo = kmalloc(sizeof(struct amdgpu_sa_bo), GFP_KERNEL);
+	if (!(*sa_bo))
+		return -ENOMEM;
+	(*sa_bo)->manager = sa_manager;
+	(*sa_bo)->fence = NULL;
+	INIT_LIST_HEAD(&(*sa_bo)->olist);
+	INIT_LIST_HEAD(&(*sa_bo)->flist);
+
+	spin_lock(&sa_manager->wq.lock);
+	do {
+		for (i = 0; i < AMDGPU_SA_NUM_FENCE_LISTS; ++i) {
+			fences[i] = NULL;
+			tries[i] = 0;
+		}
+
+		do {
+			amdgpu_sa_bo_try_free(sa_manager);
+
+			if (amdgpu_sa_bo_try_alloc(sa_manager, *sa_bo,
+						   size, align)) {
+				spin_unlock(&sa_manager->wq.lock);
+				return 0;
+			}
+
+			/* see if we can skip over some allocations */
+		} while (amdgpu_sa_bo_next_hole(sa_manager, fences, tries));
+
+		for (i = 0, count = 0; i < AMDGPU_SA_NUM_FENCE_LISTS; ++i)
+			if (fences[i])
+				fences[count++] = dma_fence_get(fences[i]);
+
+		if (count) {
+			spin_unlock(&sa_manager->wq.lock);
+#ifdef notyet
+			t = dma_fence_wait_any_timeout(fences, count, false,
+						       MAX_SCHEDULE_TIMEOUT,
+						       NULL);
+#else
+			STUB();
+			t = -EINVAL;
+#endif
+			for (i = 0; i < count; ++i)
+				dma_fence_put(fences[i]);
+
+			r = (t > 0) ? 0 : t;
+			spin_lock(&sa_manager->wq.lock);
+		} else {
+			/* if we have nothing to wait for block */
+			r = wait_event_interruptible_locked(
+				sa_manager->wq,
+				amdgpu_sa_event(sa_manager, size, align)
+			);
+		}
+
+	} while (!r);
+
+	spin_unlock(&sa_manager->wq.lock);
+	kfree(*sa_bo);
+	*sa_bo = NULL;
+	return r;
+}
+
+void amdgpu_sa_bo_free(struct amdgpu_device *adev, struct amdgpu_sa_bo **sa_bo,
+		       struct dma_fence *fence)
+{
+	struct amdgpu_sa_manager *sa_manager;
+
+	if (sa_bo == NULL || *sa_bo == NULL) {
+		return;
+	}
+
+	sa_manager = (*sa_bo)->manager;
+	spin_lock(&sa_manager->wq.lock);
+	if (fence && !dma_fence_is_signaled(fence)) {
+		uint32_t idx;
+
+		(*sa_bo)->fence = dma_fence_get(fence);
+		idx = fence->context % AMDGPU_SA_NUM_FENCE_LISTS;
+		list_add_tail(&(*sa_bo)->flist, &sa_manager->flist[idx]);
+	} else {
+		amdgpu_sa_bo_remove_locked(*sa_bo);
+	}
+	wake_up_all_locked(&sa_manager->wq);
+	spin_unlock(&sa_manager->wq.lock);
+	*sa_bo = NULL;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+void amdgpu_sa_bo_dump_debug_info(struct amdgpu_sa_manager *sa_manager,
+				  struct seq_file *m)
+{
+	struct amdgpu_sa_bo *i;
+
+	spin_lock(&sa_manager->wq.lock);
+	list_for_each_entry(i, &sa_manager->olist, olist) {
+		uint64_t soffset = i->soffset + sa_manager->gpu_addr;
+		uint64_t eoffset = i->eoffset + sa_manager->gpu_addr;
+		if (&i->olist == sa_manager->hole) {
+			seq_printf(m, ">");
+		} else {
+			seq_printf(m, " ");
+		}
+		seq_printf(m, "[0x%010llx 0x%010llx] size %8lld",
+			   soffset, eoffset, eoffset - soffset);
+
+		if (i->fence)
+			seq_printf(m, " protected by 0x%08x on context %llu",
+				   i->fence->seqno, i->fence->context);
+
+		seq_printf(m, "\n");
+	}
+	spin_unlock(&sa_manager->wq.lock);
+}
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_sched.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_sched.c
new file mode 100644
index 00000000000..1be20214f8b
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_sched.c
@@ -0,0 +1,104 @@
+/*
+ * Copyright 2017 Valve Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Andres Rodriguez <andresx7@gmail.com>
+ */
+
+#include <linux/fdtable.h>
+#include <linux/pid.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+#include "amdgpu_vm.h"
+
+enum drm_sched_priority amdgpu_to_sched_priority(int amdgpu_priority)
+{
+	switch (amdgpu_priority) {
+	case AMDGPU_CTX_PRIORITY_VERY_HIGH:
+		return DRM_SCHED_PRIORITY_HIGH_HW;
+	case AMDGPU_CTX_PRIORITY_HIGH:
+		return DRM_SCHED_PRIORITY_HIGH_SW;
+	case AMDGPU_CTX_PRIORITY_NORMAL:
+		return DRM_SCHED_PRIORITY_NORMAL;
+	case AMDGPU_CTX_PRIORITY_LOW:
+	case AMDGPU_CTX_PRIORITY_VERY_LOW:
+		return DRM_SCHED_PRIORITY_LOW;
+	case AMDGPU_CTX_PRIORITY_UNSET:
+		return DRM_SCHED_PRIORITY_UNSET;
+	default:
+		WARN(1, "Invalid context priority %d\n", amdgpu_priority);
+		return DRM_SCHED_PRIORITY_INVALID;
+	}
+}
+
+static int amdgpu_sched_process_priority_override(struct amdgpu_device *adev,
+						  int fd,
+						  enum drm_sched_priority priority)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct file *filp = fget(fd);
+	struct drm_file *file;
+	struct amdgpu_fpriv *fpriv;
+	struct amdgpu_ctx *ctx;
+	uint32_t id;
+
+	if (!filp)
+		return -EINVAL;
+
+	file = filp->private_data;
+	fpriv = file->driver_priv;
+	idr_for_each_entry(&fpriv->ctx_mgr.ctx_handles, ctx, id)
+		amdgpu_ctx_priority_override(ctx, priority);
+
+	fput(filp);
+
+	return 0;
+#endif
+}
+
+int amdgpu_sched_ioctl(struct drm_device *dev, void *data,
+		       struct drm_file *filp)
+{
+	union drm_amdgpu_sched *args = data;
+	struct amdgpu_device *adev = dev->dev_private;
+	enum drm_sched_priority priority;
+	int r;
+
+	priority = amdgpu_to_sched_priority(args->in.priority);
+	if (args->in.flags || priority == DRM_SCHED_PRIORITY_INVALID)
+		return -EINVAL;
+
+	switch (args->in.op) {
+	case AMDGPU_SCHED_OP_PROCESS_PRIORITY_OVERRIDE:
+		r = amdgpu_sched_process_priority_override(adev,
+							   args->in.fd,
+							   priority);
+		break;
+	default:
+		DRM_ERROR("Invalid sched op specified: %d\n", args->in.op);
+		r = -EINVAL;
+		break;
+	}
+
+	return r;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_sched.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_sched.h
new file mode 100644
index 00000000000..2a1a0c734bd
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_sched.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2017 Valve Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Andres Rodriguez <andresx7@gmail.com>
+ */
+
+#ifndef __AMDGPU_SCHED_H__
+#define __AMDGPU_SCHED_H__
+
+#include <drm/drmP.h>
+
+enum drm_sched_priority amdgpu_to_sched_priority(int amdgpu_priority);
+int amdgpu_sched_ioctl(struct drm_device *dev, void *data,
+		       struct drm_file *filp);
+
+#endif // __AMDGPU_SCHED_H__
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_si.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_si.c
new file mode 100644
index 00000000000..85387adc330
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_si.c
@@ -0,0 +1,2120 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+#include "atom.h"
+#include "amd_pcie.h"
+#include "si_dpm.h"
+#include "sid.h"
+#include "si_ih.h"
+#include "gfx_v6_0.h"
+#include "gmc_v6_0.h"
+#include "si_dma.h"
+#include "dce_v6_0.h"
+#include "si.h"
+#include "dce_virtual.h"
+#include "gca/gfx_6_0_d.h"
+#include "oss/oss_1_0_d.h"
+#include "gmc/gmc_6_0_d.h"
+#include "dce/dce_6_0_d.h"
+#include "uvd/uvd_4_0_d.h"
+#include "bif/bif_3_0_d.h"
+
+static const u32 tahiti_golden_registers[] =
+{
+	mmAZALIA_SCLK_CONTROL, 0x00000030, 0x00000011,
+	mmCB_HW_CONTROL, 0x00010000, 0x00018208,
+	mmDB_DEBUG, 0xffffffff, 0x00000000,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmDB_DEBUG3, 0x0002021c, 0x00020200,
+	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
+	0x340c, 0x000000c0, 0x00800040,
+	0x360c, 0x000000c0, 0x00800040,
+	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
+	mmFBC_MISC, 0x00200000, 0x50100000,
+	mmDIG0_HDMI_CONTROL, 0x31000311, 0x00000011,
+	mmMC_ARB_WTM_CNTL_RD, 0x00000003, 0x000007ff,
+	mmMC_XPB_P2P_BAR_CFG, 0x000007ff, 0x00000000,
+	mmPA_CL_ENHANCE, 0xf000001f, 0x00000007,
+	mmPA_SC_FORCE_EOV_MAX_CNTS, 0xffffffff, 0x00ffffff,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmPA_SC_MODE_CNTL_1, 0x07ffffff, 0x4e000000,
+	mmPA_SC_RASTER_CONFIG, 0x3f3f3fff, 0x2a00126a,
+	0x000c, 0xffffffff, 0x0040,
+	0x000d, 0x00000040, 0x00004040,
+	mmSPI_CONFIG_CNTL, 0x07ffffff, 0x03000000,
+	mmSQ_DED_CNT, 0x01ff1f3f, 0x00000000,
+	mmSQ_SEC_CNT, 0x01ff1f3f, 0x00000000,
+	mmSX_DEBUG_1, 0x0000007f, 0x00000020,
+	mmTA_CNTL_AUX, 0x00010000, 0x00010000,
+	mmTCP_ADDR_CONFIG, 0x00000200, 0x000002fb,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x0000543b,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0xa9210876,
+	mmVGT_FIFO_DEPTHS, 0xffffffff, 0x000fff40,
+	mmVGT_GS_VERTEX_REUSE, 0x0000001f, 0x00000010,
+	mmVM_CONTEXT0_CNTL, 0x20000000, 0x20fffed8,
+	mmVM_L2_CG, 0x000c0fc0, 0x000c0400,
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0xffffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+};
+
+static const u32 tahiti_golden_registers2[] =
+{
+	mmMCIF_MEM_CONTROL, 0x00000001, 0x00000001,
+};
+
+static const u32 tahiti_golden_rlc_registers[] =
+{
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x12011003,
+	mmRLC_LB_PARAMS, 0xffffffff, 0x00601005,
+	0x311f, 0xffffffff, 0x10104040,
+	0x3122, 0xffffffff, 0x0100000a,
+	mmRLC_LB_CNTR_MAX, 0xffffffff, 0x00000800,
+	mmRLC_LB_CNTL, 0xffffffff, 0x800000f4,
+	mmUVD_CGC_GATE, 0x00000008, 0x00000000,
+};
+
+static const u32 pitcairn_golden_registers[] =
+{
+	mmAZALIA_SCLK_CONTROL, 0x00000030, 0x00000011,
+	mmCB_HW_CONTROL, 0x00010000, 0x00018208,
+	mmDB_DEBUG, 0xffffffff, 0x00000000,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmDB_DEBUG3, 0x0002021c, 0x00020200,
+	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
+	0x340c, 0x000300c0, 0x00800040,
+	0x360c, 0x000300c0, 0x00800040,
+	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
+	mmFBC_MISC, 0x00200000, 0x50100000,
+	mmDIG0_HDMI_CONTROL, 0x31000311, 0x00000011,
+	mmMC_SEQ_PMG_PG_HWCNTL, 0x00073ffe, 0x000022a2,
+	mmMC_XPB_P2P_BAR_CFG, 0x000007ff, 0x00000000,
+	mmPA_CL_ENHANCE, 0xf000001f, 0x00000007,
+	mmPA_SC_FORCE_EOV_MAX_CNTS, 0xffffffff, 0x00ffffff,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmPA_SC_MODE_CNTL_1, 0x07ffffff, 0x4e000000,
+	mmPA_SC_RASTER_CONFIG, 0x3f3f3fff, 0x2a00126a,
+	0x000c, 0xffffffff, 0x0040,
+	0x000d, 0x00000040, 0x00004040,
+	mmSPI_CONFIG_CNTL, 0x07ffffff, 0x03000000,
+	mmSX_DEBUG_1, 0x0000007f, 0x00000020,
+	mmTA_CNTL_AUX, 0x00010000, 0x00010000,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000f7,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x32761054,
+	mmVGT_GS_VERTEX_REUSE, 0x0000001f, 0x00000010,
+	mmVM_L2_CG, 0x000c0fc0, 0x000c0400,
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0xffffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+};
+
+static const u32 pitcairn_golden_rlc_registers[] =
+{
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x12011003,
+	mmRLC_LB_PARAMS, 0xffffffff, 0x00601004,
+	0x311f, 0xffffffff, 0x10102020,
+	0x3122, 0xffffffff, 0x01000020,
+	mmRLC_LB_CNTR_MAX, 0xffffffff, 0x00000800,
+	mmRLC_LB_CNTL, 0xffffffff, 0x800000a4,
+};
+
+static const u32 verde_pg_init[] =
+{
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x40000,
+	mmGMCON_PGFSM_CONFIG, 0xffffffff, 0x200010ff,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x7007,
+	mmGMCON_PGFSM_CONFIG, 0xffffffff, 0x300010ff,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x400000,
+	mmGMCON_PGFSM_CONFIG, 0xffffffff, 0x100010ff,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x120200,
+	mmGMCON_PGFSM_CONFIG, 0xffffffff, 0x500010ff,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x1e1e16,
+	mmGMCON_PGFSM_CONFIG, 0xffffffff, 0x600010ff,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x171f1e,
+	mmGMCON_PGFSM_CONFIG, 0xffffffff, 0x700010ff,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_WRITE, 0xffffffff, 0x0,
+	mmGMCON_PGFSM_CONFIG, 0xffffffff, 0x9ff,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x0,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x10000800,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0xf,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0xf,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x4,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x1000051e,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0xffff,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0xffff,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x8,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x80500,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x12,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x9050c,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x1d,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0xb052c,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x2a,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x1053e,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x2d,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x10546,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x30,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0xa054e,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x3c,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x1055f,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x3f,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x10567,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x42,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x1056f,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x45,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x10572,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x48,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x20575,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x4c,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x190801,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x67,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x1082a,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x6a,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x1b082d,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x87,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x310851,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0xba,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x891,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0xbc,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x893,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0xbe,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x20895,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0xc2,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x20899,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0xc6,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x2089d,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0xca,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x8a1,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0xcc,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x8a3,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0xce,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x308a5,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0xd3,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x6d08cd,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x142,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x2000095a,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x1,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x144,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x301f095b,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x165,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0xc094d,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x173,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0xf096d,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x184,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x15097f,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x19b,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0xc0998,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x1a9,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x409a7,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x1af,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0xcdc,
+	mmGMCON_RENG_RAM_INDEX, 0xffffffff, 0x1b1,
+	mmGMCON_RENG_RAM_DATA, 0xffffffff, 0x800,
+	mmGMCON_RENG_EXECUTE, 0xffffffff, 0x6c9b2000,
+	mmGMCON_MISC2, 0xfc00, 0x2000,
+	mmGMCON_MISC3, 0xffffffff, 0xfc0,
+	mmMC_PMG_AUTO_CFG, 0x00000100, 0x100,
+};
+
+static const u32 verde_golden_rlc_registers[] =
+{
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x02010002,
+	mmRLC_LB_PARAMS, 0xffffffff, 0x033f1005,
+	0x311f, 0xffffffff, 0x10808020,
+	0x3122, 0xffffffff, 0x00800008,
+	mmRLC_LB_CNTR_MAX, 0xffffffff, 0x00001000,
+	mmRLC_LB_CNTL, 0xffffffff, 0x80010014,
+};
+
+static const u32 verde_golden_registers[] =
+{
+	mmAZALIA_SCLK_CONTROL, 0x00000030, 0x00000011,
+	mmCB_HW_CONTROL, 0x00010000, 0x00018208,
+	mmDB_DEBUG, 0xffffffff, 0x00000000,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmDB_DEBUG3, 0x0002021c, 0x00020200,
+	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
+	0x340c, 0x000300c0, 0x00800040,
+	0x360c, 0x000300c0, 0x00800040,
+	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
+	mmFBC_MISC, 0x00200000, 0x50100000,
+	mmDIG0_HDMI_CONTROL, 0x31000311, 0x00000011,
+	mmMC_SEQ_PMG_PG_HWCNTL, 0x00073ffe, 0x000022a2,
+	mmMC_XPB_P2P_BAR_CFG, 0x000007ff, 0x00000000,
+	mmPA_CL_ENHANCE, 0xf000001f, 0x00000007,
+	mmPA_SC_FORCE_EOV_MAX_CNTS, 0xffffffff, 0x00ffffff,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmPA_SC_MODE_CNTL_1, 0x07ffffff, 0x4e000000,
+	mmPA_SC_RASTER_CONFIG, 0x3f3f3fff, 0x0000124a,
+	0x000c, 0xffffffff, 0x0040,
+	0x000d, 0x00000040, 0x00004040,
+	mmSPI_CONFIG_CNTL, 0x07ffffff, 0x03000000,
+	mmSQ_DED_CNT, 0x01ff1f3f, 0x00000000,
+	mmSQ_SEC_CNT, 0x01ff1f3f, 0x00000000,
+	mmSX_DEBUG_1, 0x0000007f, 0x00000020,
+	mmTA_CNTL_AUX, 0x00010000, 0x00010000,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x00000003,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00001032,
+	mmVGT_GS_VERTEX_REUSE, 0x0000001f, 0x00000010,
+	mmVM_L2_CG, 0x000c0fc0, 0x000c0400,
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0xffffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+};
+
+static const u32 oland_golden_registers[] =
+{
+	mmAZALIA_SCLK_CONTROL, 0x00000030, 0x00000011,
+	mmCB_HW_CONTROL, 0x00010000, 0x00018208,
+	mmDB_DEBUG, 0xffffffff, 0x00000000,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmDB_DEBUG3, 0x0002021c, 0x00020200,
+	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
+	0x340c, 0x000300c0, 0x00800040,
+	0x360c, 0x000300c0, 0x00800040,
+	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
+	mmFBC_MISC, 0x00200000, 0x50100000,
+	mmDIG0_HDMI_CONTROL, 0x31000311, 0x00000011,
+	mmMC_SEQ_PMG_PG_HWCNTL, 0x00073ffe, 0x000022a2,
+	mmMC_XPB_P2P_BAR_CFG, 0x000007ff, 0x00000000,
+	mmPA_CL_ENHANCE, 0xf000001f, 0x00000007,
+	mmPA_SC_FORCE_EOV_MAX_CNTS, 0xffffffff, 0x00ffffff,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmPA_SC_MODE_CNTL_1, 0x07ffffff, 0x4e000000,
+	mmPA_SC_RASTER_CONFIG, 0x3f3f3fff, 0x00000082,
+	0x000c, 0xffffffff, 0x0040,
+	0x000d, 0x00000040, 0x00004040,
+	mmSPI_CONFIG_CNTL, 0x07ffffff, 0x03000000,
+	mmSX_DEBUG_1, 0x0000007f, 0x00000020,
+	mmTA_CNTL_AUX, 0x00010000, 0x00010000,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000f3,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00003210,
+	mmVGT_GS_VERTEX_REUSE, 0x0000001f, 0x00000010,
+	mmVM_L2_CG, 0x000c0fc0, 0x000c0400,
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0xffffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+
+};
+
+static const u32 oland_golden_rlc_registers[] =
+{
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x02010002,
+	mmRLC_LB_PARAMS, 0xffffffff, 0x00601005,
+	0x311f, 0xffffffff, 0x10104040,
+	0x3122, 0xffffffff, 0x0100000a,
+	mmRLC_LB_CNTR_MAX, 0xffffffff, 0x00000800,
+	mmRLC_LB_CNTL, 0xffffffff, 0x800000f4,
+};
+
+static const u32 hainan_golden_registers[] =
+{
+	0x17bc, 0x00000030, 0x00000011,
+	mmCB_HW_CONTROL, 0x00010000, 0x00018208,
+	mmDB_DEBUG, 0xffffffff, 0x00000000,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmDB_DEBUG3, 0x0002021c, 0x00020200,
+	0x031e, 0x00000080, 0x00000000,
+	0x3430, 0xff000fff, 0x00000100,
+	0x340c, 0x000300c0, 0x00800040,
+	0x3630, 0xff000fff, 0x00000100,
+	0x360c, 0x000300c0, 0x00800040,
+	0x16ec, 0x000000f0, 0x00000070,
+	0x16f0, 0x00200000, 0x50100000,
+	0x1c0c, 0x31000311, 0x00000011,
+	mmMC_SEQ_PMG_PG_HWCNTL, 0x00073ffe, 0x000022a2,
+	mmMC_XPB_P2P_BAR_CFG, 0x000007ff, 0x00000000,
+	mmPA_CL_ENHANCE, 0xf000001f, 0x00000007,
+	mmPA_SC_FORCE_EOV_MAX_CNTS, 0xffffffff, 0x00ffffff,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmPA_SC_MODE_CNTL_1, 0x07ffffff, 0x4e000000,
+	mmPA_SC_RASTER_CONFIG, 0x3f3f3fff, 0x00000000,
+	0x000c, 0xffffffff, 0x0040,
+	0x000d, 0x00000040, 0x00004040,
+	mmSPI_CONFIG_CNTL, 0x03e00000, 0x03600000,
+	mmSX_DEBUG_1, 0x0000007f, 0x00000020,
+	mmTA_CNTL_AUX, 0x00010000, 0x00010000,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000f1,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00003210,
+	mmVGT_GS_VERTEX_REUSE, 0x0000001f, 0x00000010,
+	mmVM_L2_CG, 0x000c0fc0, 0x000c0400,
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0xffffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+};
+
+static const u32 hainan_golden_registers2[] =
+{
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x2011003,
+};
+
+static const u32 tahiti_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xfffffffc,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	0x2458, 0xffffffff, 0x00010000,
+	0x2459, 0xffffffff, 0x00030002,
+	0x245a, 0xffffffff, 0x00040007,
+	0x245b, 0xffffffff, 0x00060005,
+	0x245c, 0xffffffff, 0x00090008,
+	0x245d, 0xffffffff, 0x00020001,
+	0x245e, 0xffffffff, 0x00040003,
+	0x245f, 0xffffffff, 0x00000007,
+	0x2460, 0xffffffff, 0x00060005,
+	0x2461, 0xffffffff, 0x00090008,
+	0x2462, 0xffffffff, 0x00030002,
+	0x2463, 0xffffffff, 0x00050004,
+	0x2464, 0xffffffff, 0x00000008,
+	0x2465, 0xffffffff, 0x00070006,
+	0x2466, 0xffffffff, 0x000a0009,
+	0x2467, 0xffffffff, 0x00040003,
+	0x2468, 0xffffffff, 0x00060005,
+	0x2469, 0xffffffff, 0x00000009,
+	0x246a, 0xffffffff, 0x00080007,
+	0x246b, 0xffffffff, 0x000b000a,
+	0x246c, 0xffffffff, 0x00050004,
+	0x246d, 0xffffffff, 0x00070006,
+	0x246e, 0xffffffff, 0x0008000b,
+	0x246f, 0xffffffff, 0x000a0009,
+	0x2470, 0xffffffff, 0x000d000c,
+	0x2471, 0xffffffff, 0x00060005,
+	0x2472, 0xffffffff, 0x00080007,
+	0x2473, 0xffffffff, 0x0000000b,
+	0x2474, 0xffffffff, 0x000a0009,
+	0x2475, 0xffffffff, 0x000d000c,
+	0x2476, 0xffffffff, 0x00070006,
+	0x2477, 0xffffffff, 0x00090008,
+	0x2478, 0xffffffff, 0x0000000c,
+	0x2479, 0xffffffff, 0x000b000a,
+	0x247a, 0xffffffff, 0x000e000d,
+	0x247b, 0xffffffff, 0x00080007,
+	0x247c, 0xffffffff, 0x000a0009,
+	0x247d, 0xffffffff, 0x0000000d,
+	0x247e, 0xffffffff, 0x000c000b,
+	0x247f, 0xffffffff, 0x000f000e,
+	0x2480, 0xffffffff, 0x00090008,
+	0x2481, 0xffffffff, 0x000b000a,
+	0x2482, 0xffffffff, 0x000c000f,
+	0x2483, 0xffffffff, 0x000e000d,
+	0x2484, 0xffffffff, 0x00110010,
+	0x2485, 0xffffffff, 0x000a0009,
+	0x2486, 0xffffffff, 0x000c000b,
+	0x2487, 0xffffffff, 0x0000000f,
+	0x2488, 0xffffffff, 0x000e000d,
+	0x2489, 0xffffffff, 0x00110010,
+	0x248a, 0xffffffff, 0x000b000a,
+	0x248b, 0xffffffff, 0x000d000c,
+	0x248c, 0xffffffff, 0x00000010,
+	0x248d, 0xffffffff, 0x000f000e,
+	0x248e, 0xffffffff, 0x00120011,
+	0x248f, 0xffffffff, 0x000c000b,
+	0x2490, 0xffffffff, 0x000e000d,
+	0x2491, 0xffffffff, 0x00000011,
+	0x2492, 0xffffffff, 0x0010000f,
+	0x2493, 0xffffffff, 0x00130012,
+	0x2494, 0xffffffff, 0x000d000c,
+	0x2495, 0xffffffff, 0x000f000e,
+	0x2496, 0xffffffff, 0x00100013,
+	0x2497, 0xffffffff, 0x00120011,
+	0x2498, 0xffffffff, 0x00150014,
+	0x2499, 0xffffffff, 0x000e000d,
+	0x249a, 0xffffffff, 0x0010000f,
+	0x249b, 0xffffffff, 0x00000013,
+	0x249c, 0xffffffff, 0x00120011,
+	0x249d, 0xffffffff, 0x00150014,
+	0x249e, 0xffffffff, 0x000f000e,
+	0x249f, 0xffffffff, 0x00110010,
+	0x24a0, 0xffffffff, 0x00000014,
+	0x24a1, 0xffffffff, 0x00130012,
+	0x24a2, 0xffffffff, 0x00160015,
+	0x24a3, 0xffffffff, 0x0010000f,
+	0x24a4, 0xffffffff, 0x00120011,
+	0x24a5, 0xffffffff, 0x00000015,
+	0x24a6, 0xffffffff, 0x00140013,
+	0x24a7, 0xffffffff, 0x00170016,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96940200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_GCPM_GENERAL_3, 0xffffffff, 0x00000080,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003f,
+	0x000c, 0xffffffff, 0x0000001c,
+	0x000d, 0x000f0000, 0x000f0000,
+	0x0583, 0xffffffff, 0x00000100,
+	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
+	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104,
+	mmMC_CITF_MISC_WR_CG, 0x000c0000, 0x000c0000,
+	mmMC_CITF_MISC_RD_CG, 0x000c0000, 0x000c0000,
+	mmCGTT_DRM_CLK_CTRL0, 0xff000fff, 0x00000100,
+	0x157a, 0x00000001, 0x00000001,
+	mmHDP_MEM_POWER_LS, 0x00000001, 0x00000001,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
+	0x3430, 0xfffffff0, 0x00000100,
+	0x3630, 0xfffffff0, 0x00000100,
+};
+static const u32 pitcairn_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xfffffffc,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	0x2458, 0xffffffff, 0x00010000,
+	0x2459, 0xffffffff, 0x00030002,
+	0x245a, 0xffffffff, 0x00040007,
+	0x245b, 0xffffffff, 0x00060005,
+	0x245c, 0xffffffff, 0x00090008,
+	0x245d, 0xffffffff, 0x00020001,
+	0x245e, 0xffffffff, 0x00040003,
+	0x245f, 0xffffffff, 0x00000007,
+	0x2460, 0xffffffff, 0x00060005,
+	0x2461, 0xffffffff, 0x00090008,
+	0x2462, 0xffffffff, 0x00030002,
+	0x2463, 0xffffffff, 0x00050004,
+	0x2464, 0xffffffff, 0x00000008,
+	0x2465, 0xffffffff, 0x00070006,
+	0x2466, 0xffffffff, 0x000a0009,
+	0x2467, 0xffffffff, 0x00040003,
+	0x2468, 0xffffffff, 0x00060005,
+	0x2469, 0xffffffff, 0x00000009,
+	0x246a, 0xffffffff, 0x00080007,
+	0x246b, 0xffffffff, 0x000b000a,
+	0x246c, 0xffffffff, 0x00050004,
+	0x246d, 0xffffffff, 0x00070006,
+	0x246e, 0xffffffff, 0x0008000b,
+	0x246f, 0xffffffff, 0x000a0009,
+	0x2470, 0xffffffff, 0x000d000c,
+	0x2480, 0xffffffff, 0x00090008,
+	0x2481, 0xffffffff, 0x000b000a,
+	0x2482, 0xffffffff, 0x000c000f,
+	0x2483, 0xffffffff, 0x000e000d,
+	0x2484, 0xffffffff, 0x00110010,
+	0x2485, 0xffffffff, 0x000a0009,
+	0x2486, 0xffffffff, 0x000c000b,
+	0x2487, 0xffffffff, 0x0000000f,
+	0x2488, 0xffffffff, 0x000e000d,
+	0x2489, 0xffffffff, 0x00110010,
+	0x248a, 0xffffffff, 0x000b000a,
+	0x248b, 0xffffffff, 0x000d000c,
+	0x248c, 0xffffffff, 0x00000010,
+	0x248d, 0xffffffff, 0x000f000e,
+	0x248e, 0xffffffff, 0x00120011,
+	0x248f, 0xffffffff, 0x000c000b,
+	0x2490, 0xffffffff, 0x000e000d,
+	0x2491, 0xffffffff, 0x00000011,
+	0x2492, 0xffffffff, 0x0010000f,
+	0x2493, 0xffffffff, 0x00130012,
+	0x2494, 0xffffffff, 0x000d000c,
+	0x2495, 0xffffffff, 0x000f000e,
+	0x2496, 0xffffffff, 0x00100013,
+	0x2497, 0xffffffff, 0x00120011,
+	0x2498, 0xffffffff, 0x00150014,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96940200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_GCPM_GENERAL_3, 0xffffffff, 0x00000080,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003f,
+	0x000c, 0xffffffff, 0x0000001c,
+	0x000d, 0x000f0000, 0x000f0000,
+	0x0583, 0xffffffff, 0x00000100,
+	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
+	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104,
+	mmCGTT_DRM_CLK_CTRL0, 0xff000fff, 0x00000100,
+	0x157a, 0x00000001, 0x00000001,
+	mmHDP_MEM_POWER_LS, 0x00000001, 0x00000001,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
+	0x3430, 0xfffffff0, 0x00000100,
+	0x3630, 0xfffffff0, 0x00000100,
+};
+
+static const u32 verde_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xfffffffc,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	0x2458, 0xffffffff, 0x00010000,
+	0x2459, 0xffffffff, 0x00030002,
+	0x245a, 0xffffffff, 0x00040007,
+	0x245b, 0xffffffff, 0x00060005,
+	0x245c, 0xffffffff, 0x00090008,
+	0x245d, 0xffffffff, 0x00020001,
+	0x245e, 0xffffffff, 0x00040003,
+	0x245f, 0xffffffff, 0x00000007,
+	0x2460, 0xffffffff, 0x00060005,
+	0x2461, 0xffffffff, 0x00090008,
+	0x2462, 0xffffffff, 0x00030002,
+	0x2463, 0xffffffff, 0x00050004,
+	0x2464, 0xffffffff, 0x00000008,
+	0x2465, 0xffffffff, 0x00070006,
+	0x2466, 0xffffffff, 0x000a0009,
+	0x2467, 0xffffffff, 0x00040003,
+	0x2468, 0xffffffff, 0x00060005,
+	0x2469, 0xffffffff, 0x00000009,
+	0x246a, 0xffffffff, 0x00080007,
+	0x246b, 0xffffffff, 0x000b000a,
+	0x246c, 0xffffffff, 0x00050004,
+	0x246d, 0xffffffff, 0x00070006,
+	0x246e, 0xffffffff, 0x0008000b,
+	0x246f, 0xffffffff, 0x000a0009,
+	0x2470, 0xffffffff, 0x000d000c,
+	0x2480, 0xffffffff, 0x00090008,
+	0x2481, 0xffffffff, 0x000b000a,
+	0x2482, 0xffffffff, 0x000c000f,
+	0x2483, 0xffffffff, 0x000e000d,
+	0x2484, 0xffffffff, 0x00110010,
+	0x2485, 0xffffffff, 0x000a0009,
+	0x2486, 0xffffffff, 0x000c000b,
+	0x2487, 0xffffffff, 0x0000000f,
+	0x2488, 0xffffffff, 0x000e000d,
+	0x2489, 0xffffffff, 0x00110010,
+	0x248a, 0xffffffff, 0x000b000a,
+	0x248b, 0xffffffff, 0x000d000c,
+	0x248c, 0xffffffff, 0x00000010,
+	0x248d, 0xffffffff, 0x000f000e,
+	0x248e, 0xffffffff, 0x00120011,
+	0x248f, 0xffffffff, 0x000c000b,
+	0x2490, 0xffffffff, 0x000e000d,
+	0x2491, 0xffffffff, 0x00000011,
+	0x2492, 0xffffffff, 0x0010000f,
+	0x2493, 0xffffffff, 0x00130012,
+	0x2494, 0xffffffff, 0x000d000c,
+	0x2495, 0xffffffff, 0x000f000e,
+	0x2496, 0xffffffff, 0x00100013,
+	0x2497, 0xffffffff, 0x00120011,
+	0x2498, 0xffffffff, 0x00150014,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96940200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_GCPM_GENERAL_3, 0xffffffff, 0x00000080,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003f,
+	0x000c, 0xffffffff, 0x0000001c,
+	0x000d, 0x000f0000, 0x000f0000,
+	0x0583, 0xffffffff, 0x00000100,
+	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
+	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104,
+	mmMC_CITF_MISC_WR_CG, 0x000c0000, 0x000c0000,
+	mmMC_CITF_MISC_RD_CG, 0x000c0000, 0x000c0000,
+	mmCGTT_DRM_CLK_CTRL0, 0xff000fff, 0x00000100,
+	0x157a, 0x00000001, 0x00000001,
+	mmHDP_MEM_POWER_LS, 0x00000001, 0x00000001,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
+	0x3430, 0xfffffff0, 0x00000100,
+	0x3630, 0xfffffff0, 0x00000100,
+};
+
+static const u32 oland_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xfffffffc,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	0x2458, 0xffffffff, 0x00010000,
+	0x2459, 0xffffffff, 0x00030002,
+	0x245a, 0xffffffff, 0x00040007,
+	0x245b, 0xffffffff, 0x00060005,
+	0x245c, 0xffffffff, 0x00090008,
+	0x245d, 0xffffffff, 0x00020001,
+	0x245e, 0xffffffff, 0x00040003,
+	0x245f, 0xffffffff, 0x00000007,
+	0x2460, 0xffffffff, 0x00060005,
+	0x2461, 0xffffffff, 0x00090008,
+	0x2462, 0xffffffff, 0x00030002,
+	0x2463, 0xffffffff, 0x00050004,
+	0x2464, 0xffffffff, 0x00000008,
+	0x2465, 0xffffffff, 0x00070006,
+	0x2466, 0xffffffff, 0x000a0009,
+	0x2467, 0xffffffff, 0x00040003,
+	0x2468, 0xffffffff, 0x00060005,
+	0x2469, 0xffffffff, 0x00000009,
+	0x246a, 0xffffffff, 0x00080007,
+	0x246b, 0xffffffff, 0x000b000a,
+	0x246c, 0xffffffff, 0x00050004,
+	0x246d, 0xffffffff, 0x00070006,
+	0x246e, 0xffffffff, 0x0008000b,
+	0x246f, 0xffffffff, 0x000a0009,
+	0x2470, 0xffffffff, 0x000d000c,
+	0x2471, 0xffffffff, 0x00060005,
+	0x2472, 0xffffffff, 0x00080007,
+	0x2473, 0xffffffff, 0x0000000b,
+	0x2474, 0xffffffff, 0x000a0009,
+	0x2475, 0xffffffff, 0x000d000c,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96940200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_GCPM_GENERAL_3, 0xffffffff, 0x00000080,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003f,
+	0x000c, 0xffffffff, 0x0000001c,
+	0x000d, 0x000f0000, 0x000f0000,
+	0x0583, 0xffffffff, 0x00000100,
+	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
+	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104,
+	mmMC_CITF_MISC_WR_CG, 0x000c0000, 0x000c0000,
+	mmMC_CITF_MISC_RD_CG, 0x000c0000, 0x000c0000,
+	mmCGTT_DRM_CLK_CTRL0, 0xff000fff, 0x00000100,
+	0x157a, 0x00000001, 0x00000001,
+	mmHDP_MEM_POWER_LS, 0x00000001, 0x00000001,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
+	0x3430, 0xfffffff0, 0x00000100,
+	0x3630, 0xfffffff0, 0x00000100,
+};
+
+static const u32 hainan_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xfffffffc,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	0x2458, 0xffffffff, 0x00010000,
+	0x2459, 0xffffffff, 0x00030002,
+	0x245a, 0xffffffff, 0x00040007,
+	0x245b, 0xffffffff, 0x00060005,
+	0x245c, 0xffffffff, 0x00090008,
+	0x245d, 0xffffffff, 0x00020001,
+	0x245e, 0xffffffff, 0x00040003,
+	0x245f, 0xffffffff, 0x00000007,
+	0x2460, 0xffffffff, 0x00060005,
+	0x2461, 0xffffffff, 0x00090008,
+	0x2462, 0xffffffff, 0x00030002,
+	0x2463, 0xffffffff, 0x00050004,
+	0x2464, 0xffffffff, 0x00000008,
+	0x2465, 0xffffffff, 0x00070006,
+	0x2466, 0xffffffff, 0x000a0009,
+	0x2467, 0xffffffff, 0x00040003,
+	0x2468, 0xffffffff, 0x00060005,
+	0x2469, 0xffffffff, 0x00000009,
+	0x246a, 0xffffffff, 0x00080007,
+	0x246b, 0xffffffff, 0x000b000a,
+	0x246c, 0xffffffff, 0x00050004,
+	0x246d, 0xffffffff, 0x00070006,
+	0x246e, 0xffffffff, 0x0008000b,
+	0x246f, 0xffffffff, 0x000a0009,
+	0x2470, 0xffffffff, 0x000d000c,
+	0x2471, 0xffffffff, 0x00060005,
+	0x2472, 0xffffffff, 0x00080007,
+	0x2473, 0xffffffff, 0x0000000b,
+	0x2474, 0xffffffff, 0x000a0009,
+	0x2475, 0xffffffff, 0x000d000c,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96940200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_GCPM_GENERAL_3, 0xffffffff, 0x00000080,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003f,
+	0x000c, 0xffffffff, 0x0000001c,
+	0x000d, 0x000f0000, 0x000f0000,
+	0x0583, 0xffffffff, 0x00000100,
+	0x0409, 0xffffffff, 0x00000100,
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104,
+	mmMC_CITF_MISC_WR_CG, 0x000c0000, 0x000c0000,
+	mmMC_CITF_MISC_RD_CG, 0x000c0000, 0x000c0000,
+	mmHDP_MEM_POWER_LS, 0x00000001, 0x00000001,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
+	0x3430, 0xfffffff0, 0x00000100,
+	0x3630, 0xfffffff0, 0x00000100,
+};
+
+static u32 si_pcie_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(AMDGPU_PCIE_INDEX, reg);
+	(void)RREG32(AMDGPU_PCIE_INDEX);
+	r = RREG32(AMDGPU_PCIE_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+	return r;
+}
+
+static void si_pcie_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(AMDGPU_PCIE_INDEX, reg);
+	(void)RREG32(AMDGPU_PCIE_INDEX);
+	WREG32(AMDGPU_PCIE_DATA, v);
+	(void)RREG32(AMDGPU_PCIE_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+static u32 si_pciep_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(PCIE_PORT_INDEX, ((reg) & 0xff));
+	(void)RREG32(PCIE_PORT_INDEX);
+	r = RREG32(PCIE_PORT_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+	return r;
+}
+
+static void si_pciep_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(PCIE_PORT_INDEX, ((reg) & 0xff));
+	(void)RREG32(PCIE_PORT_INDEX);
+	WREG32(PCIE_PORT_DATA, (v));
+	(void)RREG32(PCIE_PORT_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+static u32 si_smc_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(SMC_IND_INDEX_0, (reg));
+	r = RREG32(SMC_IND_DATA_0);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return r;
+}
+
+static void si_smc_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(SMC_IND_INDEX_0, (reg));
+	WREG32(SMC_IND_DATA_0, (v));
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+}
+
+static struct amdgpu_allowed_register_entry si_allowed_read_registers[] = {
+	{GRBM_STATUS},
+	{GB_ADDR_CONFIG},
+	{MC_ARB_RAMCFG},
+	{GB_TILE_MODE0},
+	{GB_TILE_MODE1},
+	{GB_TILE_MODE2},
+	{GB_TILE_MODE3},
+	{GB_TILE_MODE4},
+	{GB_TILE_MODE5},
+	{GB_TILE_MODE6},
+	{GB_TILE_MODE7},
+	{GB_TILE_MODE8},
+	{GB_TILE_MODE9},
+	{GB_TILE_MODE10},
+	{GB_TILE_MODE11},
+	{GB_TILE_MODE12},
+	{GB_TILE_MODE13},
+	{GB_TILE_MODE14},
+	{GB_TILE_MODE15},
+	{GB_TILE_MODE16},
+	{GB_TILE_MODE17},
+	{GB_TILE_MODE18},
+	{GB_TILE_MODE19},
+	{GB_TILE_MODE20},
+	{GB_TILE_MODE21},
+	{GB_TILE_MODE22},
+	{GB_TILE_MODE23},
+	{GB_TILE_MODE24},
+	{GB_TILE_MODE25},
+	{GB_TILE_MODE26},
+	{GB_TILE_MODE27},
+	{GB_TILE_MODE28},
+	{GB_TILE_MODE29},
+	{GB_TILE_MODE30},
+	{GB_TILE_MODE31},
+	{CC_RB_BACKEND_DISABLE, true},
+	{GC_USER_RB_BACKEND_DISABLE, true},
+	{PA_SC_RASTER_CONFIG, true},
+};
+
+static uint32_t si_get_register_value(struct amdgpu_device *adev,
+				      bool indexed, u32 se_num,
+				      u32 sh_num, u32 reg_offset)
+{
+	if (indexed) {
+		uint32_t val;
+		unsigned se_idx = (se_num == 0xffffffff) ? 0 : se_num;
+		unsigned sh_idx = (sh_num == 0xffffffff) ? 0 : sh_num;
+
+		switch (reg_offset) {
+		case mmCC_RB_BACKEND_DISABLE:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].rb_backend_disable;
+		case mmGC_USER_RB_BACKEND_DISABLE:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].user_rb_backend_disable;
+		case mmPA_SC_RASTER_CONFIG:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].raster_config;
+		}
+
+		mutex_lock(&adev->grbm_idx_mutex);
+		if (se_num != 0xffffffff || sh_num != 0xffffffff)
+			amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+
+		val = RREG32(reg_offset);
+
+		if (se_num != 0xffffffff || sh_num != 0xffffffff)
+			amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		mutex_unlock(&adev->grbm_idx_mutex);
+		return val;
+	} else {
+		unsigned idx;
+
+		switch (reg_offset) {
+		case mmGB_ADDR_CONFIG:
+			return adev->gfx.config.gb_addr_config;
+		case mmMC_ARB_RAMCFG:
+			return adev->gfx.config.mc_arb_ramcfg;
+		case mmGB_TILE_MODE0:
+		case mmGB_TILE_MODE1:
+		case mmGB_TILE_MODE2:
+		case mmGB_TILE_MODE3:
+		case mmGB_TILE_MODE4:
+		case mmGB_TILE_MODE5:
+		case mmGB_TILE_MODE6:
+		case mmGB_TILE_MODE7:
+		case mmGB_TILE_MODE8:
+		case mmGB_TILE_MODE9:
+		case mmGB_TILE_MODE10:
+		case mmGB_TILE_MODE11:
+		case mmGB_TILE_MODE12:
+		case mmGB_TILE_MODE13:
+		case mmGB_TILE_MODE14:
+		case mmGB_TILE_MODE15:
+		case mmGB_TILE_MODE16:
+		case mmGB_TILE_MODE17:
+		case mmGB_TILE_MODE18:
+		case mmGB_TILE_MODE19:
+		case mmGB_TILE_MODE20:
+		case mmGB_TILE_MODE21:
+		case mmGB_TILE_MODE22:
+		case mmGB_TILE_MODE23:
+		case mmGB_TILE_MODE24:
+		case mmGB_TILE_MODE25:
+		case mmGB_TILE_MODE26:
+		case mmGB_TILE_MODE27:
+		case mmGB_TILE_MODE28:
+		case mmGB_TILE_MODE29:
+		case mmGB_TILE_MODE30:
+		case mmGB_TILE_MODE31:
+			idx = (reg_offset - mmGB_TILE_MODE0);
+			return adev->gfx.config.tile_mode_array[idx];
+		default:
+			return RREG32(reg_offset);
+		}
+	}
+}
+static int si_read_register(struct amdgpu_device *adev, u32 se_num,
+			     u32 sh_num, u32 reg_offset, u32 *value)
+{
+	uint32_t i;
+
+	*value = 0;
+	for (i = 0; i < ARRAY_SIZE(si_allowed_read_registers); i++) {
+		bool indexed = si_allowed_read_registers[i].grbm_indexed;
+
+		if (reg_offset != si_allowed_read_registers[i].reg_offset)
+			continue;
+
+		*value = si_get_register_value(adev, indexed, se_num, sh_num,
+					       reg_offset);
+		return 0;
+	}
+	return -EINVAL;
+}
+
+static bool si_read_disabled_bios(struct amdgpu_device *adev)
+{
+	u32 bus_cntl;
+	u32 d1vga_control = 0;
+	u32 d2vga_control = 0;
+	u32 vga_render_control = 0;
+	u32 rom_cntl;
+	bool r;
+
+	bus_cntl = RREG32(R600_BUS_CNTL);
+	if (adev->mode_info.num_crtc) {
+		d1vga_control = RREG32(AVIVO_D1VGA_CONTROL);
+		d2vga_control = RREG32(AVIVO_D2VGA_CONTROL);
+		vga_render_control = RREG32(VGA_RENDER_CONTROL);
+	}
+	rom_cntl = RREG32(R600_ROM_CNTL);
+
+	/* enable the rom */
+	WREG32(R600_BUS_CNTL, (bus_cntl & ~R600_BIOS_ROM_DIS));
+	if (adev->mode_info.num_crtc) {
+		/* Disable VGA mode */
+		WREG32(AVIVO_D1VGA_CONTROL,
+		       (d1vga_control & ~(AVIVO_DVGA_CONTROL_MODE_ENABLE |
+					  AVIVO_DVGA_CONTROL_TIMING_SELECT)));
+		WREG32(AVIVO_D2VGA_CONTROL,
+		       (d2vga_control & ~(AVIVO_DVGA_CONTROL_MODE_ENABLE |
+					  AVIVO_DVGA_CONTROL_TIMING_SELECT)));
+		WREG32(VGA_RENDER_CONTROL,
+		       (vga_render_control & C_000300_VGA_VSTATUS_CNTL));
+	}
+	WREG32(R600_ROM_CNTL, rom_cntl | R600_SCK_OVERWRITE);
+
+	r = amdgpu_read_bios(adev);
+
+	/* restore regs */
+	WREG32(R600_BUS_CNTL, bus_cntl);
+	if (adev->mode_info.num_crtc) {
+		WREG32(AVIVO_D1VGA_CONTROL, d1vga_control);
+		WREG32(AVIVO_D2VGA_CONTROL, d2vga_control);
+		WREG32(VGA_RENDER_CONTROL, vga_render_control);
+	}
+	WREG32(R600_ROM_CNTL, rom_cntl);
+	return r;
+}
+
+#define mmROM_INDEX 0x2A
+#define mmROM_DATA  0x2B
+
+static bool si_read_bios_from_rom(struct amdgpu_device *adev,
+				  u8 *bios, u32 length_bytes)
+{
+	u32 *dw_ptr;
+	u32 i, length_dw;
+
+	if (bios == NULL)
+		return false;
+	if (length_bytes == 0)
+		return false;
+	/* APU vbios image is part of sbios image */
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	dw_ptr = (u32 *)bios;
+	length_dw = roundup2(length_bytes, 4) / 4;
+	/* set rom index to 0 */
+	WREG32(mmROM_INDEX, 0);
+	for (i = 0; i < length_dw; i++)
+		dw_ptr[i] = RREG32(mmROM_DATA);
+
+	return true;
+}
+
+//xxx: not implemented
+static int si_asic_reset(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
+static u32 si_get_config_memsize(struct amdgpu_device *adev)
+{
+	return RREG32(mmCONFIG_MEMSIZE);
+}
+
+static void si_vga_set_state(struct amdgpu_device *adev, bool state)
+{
+	uint32_t temp;
+
+	temp = RREG32(CONFIG_CNTL);
+	if (state == false) {
+		temp &= ~(1<<0);
+		temp |= (1<<1);
+	} else {
+		temp &= ~(1<<1);
+	}
+	WREG32(CONFIG_CNTL, temp);
+}
+
+static u32 si_get_xclk(struct amdgpu_device *adev)
+{
+        u32 reference_clock = adev->clock.spll.reference_freq;
+	u32 tmp;
+
+	tmp = RREG32(CG_CLKPIN_CNTL_2);
+	if (tmp & MUX_TCLK_TO_XCLK)
+		return TCLK;
+
+	tmp = RREG32(CG_CLKPIN_CNTL);
+	if (tmp & XTALIN_DIVIDE)
+		return reference_clock / 4;
+
+	return reference_clock;
+}
+
+//xxx:not implemented
+static int si_set_uvd_clocks(struct amdgpu_device *adev, u32 vclk, u32 dclk)
+{
+	return 0;
+}
+
+static void si_detect_hw_virtualization(struct amdgpu_device *adev)
+{
+	if (is_virtual_machine()) /* passthrough mode */
+		adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
+}
+
+static void si_flush_hdp(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 1);
+		RREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL);
+	} else {
+		amdgpu_ring_emit_wreg(ring, mmHDP_MEM_COHERENCY_FLUSH_CNTL, 1);
+	}
+}
+
+static void si_invalidate_hdp(struct amdgpu_device *adev,
+			      struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32(mmHDP_DEBUG0, 1);
+		RREG32(mmHDP_DEBUG0);
+	} else {
+		amdgpu_ring_emit_wreg(ring, mmHDP_DEBUG0, 1);
+	}
+}
+
+static bool si_need_full_reset(struct amdgpu_device *adev)
+{
+	/* change this when we support soft reset */
+	return true;
+}
+
+static int si_get_pcie_lanes(struct amdgpu_device *adev)
+{
+	u32 link_width_cntl;
+
+	if (adev->flags & AMD_IS_APU)
+		return 0;
+
+	link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
+
+	switch ((link_width_cntl & LC_LINK_WIDTH_RD_MASK) >> LC_LINK_WIDTH_RD_SHIFT) {
+	case LC_LINK_WIDTH_X1:
+		return 1;
+	case LC_LINK_WIDTH_X2:
+		return 2;
+	case LC_LINK_WIDTH_X4:
+		return 4;
+	case LC_LINK_WIDTH_X8:
+		return 8;
+	case LC_LINK_WIDTH_X0:
+	case LC_LINK_WIDTH_X16:
+	default:
+		return 16;
+	}
+}
+
+static void si_set_pcie_lanes(struct amdgpu_device *adev, int lanes)
+{
+	u32 link_width_cntl, mask;
+
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	switch (lanes) {
+	case 0:
+		mask = LC_LINK_WIDTH_X0;
+		break;
+	case 1:
+		mask = LC_LINK_WIDTH_X1;
+		break;
+	case 2:
+		mask = LC_LINK_WIDTH_X2;
+		break;
+	case 4:
+		mask = LC_LINK_WIDTH_X4;
+		break;
+	case 8:
+		mask = LC_LINK_WIDTH_X8;
+		break;
+	case 16:
+		mask = LC_LINK_WIDTH_X16;
+		break;
+	default:
+		DRM_ERROR("invalid pcie lane request: %d\n", lanes);
+		return;
+	}
+
+	link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
+	link_width_cntl &= ~LC_LINK_WIDTH_MASK;
+	link_width_cntl |= mask << LC_LINK_WIDTH_SHIFT;
+	link_width_cntl |= (LC_RECONFIG_NOW |
+			    LC_RECONFIG_ARC_MISSING_ESCAPE);
+
+	WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
+}
+
+static const struct amdgpu_asic_funcs si_asic_funcs =
+{
+	.read_disabled_bios = &si_read_disabled_bios,
+	.read_bios_from_rom = &si_read_bios_from_rom,
+	.read_register = &si_read_register,
+	.reset = &si_asic_reset,
+	.set_vga_state = &si_vga_set_state,
+	.get_xclk = &si_get_xclk,
+	.set_uvd_clocks = &si_set_uvd_clocks,
+	.set_vce_clocks = NULL,
+	.get_pcie_lanes = &si_get_pcie_lanes,
+	.set_pcie_lanes = &si_set_pcie_lanes,
+	.get_config_memsize = &si_get_config_memsize,
+	.flush_hdp = &si_flush_hdp,
+	.invalidate_hdp = &si_invalidate_hdp,
+	.need_full_reset = &si_need_full_reset,
+};
+
+static uint32_t si_get_rev_id(struct amdgpu_device *adev)
+{
+	return (RREG32(CC_DRM_ID_STRAPS) & CC_DRM_ID_STRAPS__ATI_REV_ID_MASK)
+		>> CC_DRM_ID_STRAPS__ATI_REV_ID__SHIFT;
+}
+
+static int si_common_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->smc_rreg = &si_smc_rreg;
+	adev->smc_wreg = &si_smc_wreg;
+	adev->pcie_rreg = &si_pcie_rreg;
+	adev->pcie_wreg = &si_pcie_wreg;
+	adev->pciep_rreg = &si_pciep_rreg;
+	adev->pciep_wreg = &si_pciep_wreg;
+	adev->uvd_ctx_rreg = NULL;
+	adev->uvd_ctx_wreg = NULL;
+	adev->didt_rreg = NULL;
+	adev->didt_wreg = NULL;
+
+	adev->asic_funcs = &si_asic_funcs;
+
+	adev->rev_id = si_get_rev_id(adev);
+	adev->external_rev_id = 0xFF;
+	switch (adev->asic_type) {
+	case CHIP_TAHITI:
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			/*AMD_CG_SUPPORT_GFX_CGCG |*/
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_VCE_MGCG |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_MGCG;
+			adev->pg_flags = 0;
+		adev->external_rev_id = (adev->rev_id == 0) ? 1 :
+					(adev->rev_id == 1) ? 5 : 6;
+		break;
+	case CHIP_PITCAIRN:
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			/*AMD_CG_SUPPORT_GFX_CGCG |*/
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_VCE_MGCG |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 20;
+		break;
+
+	case CHIP_VERDE:
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_CGTS_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_VCE_MGCG |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_MGCG;
+		adev->pg_flags = 0;
+		//???
+		adev->external_rev_id = adev->rev_id + 40;
+		break;
+	case CHIP_OLAND:
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			/*AMD_CG_SUPPORT_GFX_CGCG |*/
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = 60;
+		break;
+	case CHIP_HAINAN:
+		adev->cg_flags =
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			/*AMD_CG_SUPPORT_GFX_CGCG |*/
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_HDP_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = 70;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int si_common_sw_init(void *handle)
+{
+	return 0;
+}
+
+static int si_common_sw_fini(void *handle)
+{
+	return 0;
+}
+
+
+static void si_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_TAHITI:
+		amdgpu_device_program_register_sequence(adev,
+							tahiti_golden_registers,
+							ARRAY_SIZE(tahiti_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							tahiti_golden_rlc_registers,
+							ARRAY_SIZE(tahiti_golden_rlc_registers));
+		amdgpu_device_program_register_sequence(adev,
+							tahiti_mgcg_cgcg_init,
+							ARRAY_SIZE(tahiti_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							tahiti_golden_registers2,
+							ARRAY_SIZE(tahiti_golden_registers2));
+		break;
+	case CHIP_PITCAIRN:
+		amdgpu_device_program_register_sequence(adev,
+							pitcairn_golden_registers,
+							ARRAY_SIZE(pitcairn_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							pitcairn_golden_rlc_registers,
+							ARRAY_SIZE(pitcairn_golden_rlc_registers));
+		amdgpu_device_program_register_sequence(adev,
+							pitcairn_mgcg_cgcg_init,
+							ARRAY_SIZE(pitcairn_mgcg_cgcg_init));
+		break;
+	case CHIP_VERDE:
+		amdgpu_device_program_register_sequence(adev,
+							verde_golden_registers,
+							ARRAY_SIZE(verde_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							verde_golden_rlc_registers,
+							ARRAY_SIZE(verde_golden_rlc_registers));
+		amdgpu_device_program_register_sequence(adev,
+							verde_mgcg_cgcg_init,
+							ARRAY_SIZE(verde_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							verde_pg_init,
+							ARRAY_SIZE(verde_pg_init));
+		break;
+	case CHIP_OLAND:
+		amdgpu_device_program_register_sequence(adev,
+							oland_golden_registers,
+							ARRAY_SIZE(oland_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							oland_golden_rlc_registers,
+							ARRAY_SIZE(oland_golden_rlc_registers));
+		amdgpu_device_program_register_sequence(adev,
+							oland_mgcg_cgcg_init,
+							ARRAY_SIZE(oland_mgcg_cgcg_init));
+		break;
+	case CHIP_HAINAN:
+		amdgpu_device_program_register_sequence(adev,
+							hainan_golden_registers,
+							ARRAY_SIZE(hainan_golden_registers));
+		amdgpu_device_program_register_sequence(adev,
+							hainan_golden_registers2,
+							ARRAY_SIZE(hainan_golden_registers2));
+		amdgpu_device_program_register_sequence(adev,
+							hainan_mgcg_cgcg_init,
+							ARRAY_SIZE(hainan_mgcg_cgcg_init));
+		break;
+
+
+	default:
+		BUG();
+	}
+}
+
+static void si_pcie_gen3_enable(struct amdgpu_device *adev)
+{
+	struct pci_dev *root = adev->pdev->bus->self;
+	int bridge_pos, gpu_pos;
+	u32 speed_cntl, current_data_rate;
+	int i;
+	u16 tmp16;
+
+	if (pci_is_root_bus(adev->pdev->bus))
+		return;
+
+	if (amdgpu_pcie_gen2 == 0)
+		return;
+
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	if (!(adev->pm.pcie_gen_mask & (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+					CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)))
+		return;
+
+	speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
+	current_data_rate = (speed_cntl & LC_CURRENT_DATA_RATE_MASK) >>
+		LC_CURRENT_DATA_RATE_SHIFT;
+	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3) {
+		if (current_data_rate == 2) {
+			DRM_INFO("PCIE gen 3 link speeds already enabled\n");
+			return;
+		}
+		DRM_INFO("enabling PCIE gen 3 link speeds, disable with amdgpu.pcie_gen2=0\n");
+	} else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2) {
+		if (current_data_rate == 1) {
+			DRM_INFO("PCIE gen 2 link speeds already enabled\n");
+			return;
+		}
+		DRM_INFO("enabling PCIE gen 2 link speeds, disable with amdgpu.pcie_gen2=0\n");
+	}
+
+	bridge_pos = pci_pcie_cap(root);
+	if (!bridge_pos)
+		return;
+
+	gpu_pos = pci_pcie_cap(adev->pdev);
+	if (!gpu_pos)
+		return;
+
+	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3) {
+		if (current_data_rate != 2) {
+			u16 bridge_cfg, gpu_cfg;
+			u16 bridge_cfg2, gpu_cfg2;
+			u32 max_lw, current_lw, tmp;
+
+			pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &bridge_cfg);
+			pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &gpu_cfg);
+
+			tmp16 = bridge_cfg | PCI_EXP_LNKCTL_HAWD;
+			pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL, tmp16);
+
+			tmp16 = gpu_cfg | PCI_EXP_LNKCTL_HAWD;
+			pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, tmp16);
+
+			tmp = RREG32_PCIE(PCIE_LC_STATUS1);
+			max_lw = (tmp & LC_DETECTED_LINK_WIDTH_MASK) >> LC_DETECTED_LINK_WIDTH_SHIFT;
+			current_lw = (tmp & LC_OPERATING_LINK_WIDTH_MASK) >> LC_OPERATING_LINK_WIDTH_SHIFT;
+
+			if (current_lw < max_lw) {
+				tmp = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
+				if (tmp & LC_RENEGOTIATION_SUPPORT) {
+					tmp &= ~(LC_LINK_WIDTH_MASK | LC_UPCONFIGURE_DIS);
+					tmp |= (max_lw << LC_LINK_WIDTH_SHIFT);
+					tmp |= LC_UPCONFIGURE_SUPPORT | LC_RENEGOTIATE_EN | LC_RECONFIG_NOW;
+					WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, tmp);
+				}
+			}
+
+			for (i = 0; i < 10; i++) {
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_DEVSTA, &tmp16);
+				if (tmp16 & PCI_EXP_DEVSTA_TRPND)
+					break;
+
+				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &bridge_cfg);
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &gpu_cfg);
+
+				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, &bridge_cfg2);
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &gpu_cfg2);
+
+				tmp = RREG32_PCIE_PORT(PCIE_LC_CNTL4);
+				tmp |= LC_SET_QUIESCE;
+				WREG32_PCIE_PORT(PCIE_LC_CNTL4, tmp);
+
+				tmp = RREG32_PCIE_PORT(PCIE_LC_CNTL4);
+				tmp |= LC_REDO_EQ;
+				WREG32_PCIE_PORT(PCIE_LC_CNTL4, tmp);
+
+				mdelay(100);
+
+				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &tmp16);
+				tmp16 &= ~PCI_EXP_LNKCTL_HAWD;
+				tmp16 |= (bridge_cfg & PCI_EXP_LNKCTL_HAWD);
+				pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL, tmp16);
+
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &tmp16);
+				tmp16 &= ~PCI_EXP_LNKCTL_HAWD;
+				tmp16 |= (gpu_cfg & PCI_EXP_LNKCTL_HAWD);
+				pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, tmp16);
+
+				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, &tmp16);
+				tmp16 &= ~((1 << 4) | (7 << 9));
+				tmp16 |= (bridge_cfg2 & ((1 << 4) | (7 << 9)));
+				pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, tmp16);
+
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &tmp16);
+				tmp16 &= ~((1 << 4) | (7 << 9));
+				tmp16 |= (gpu_cfg2 & ((1 << 4) | (7 << 9)));
+				pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, tmp16);
+
+				tmp = RREG32_PCIE_PORT(PCIE_LC_CNTL4);
+				tmp &= ~LC_SET_QUIESCE;
+				WREG32_PCIE_PORT(PCIE_LC_CNTL4, tmp);
+			}
+		}
+	}
+
+	speed_cntl |= LC_FORCE_EN_SW_SPEED_CHANGE | LC_FORCE_DIS_HW_SPEED_CHANGE;
+	speed_cntl &= ~LC_FORCE_DIS_SW_SPEED_CHANGE;
+	WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
+
+	pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &tmp16);
+	tmp16 &= ~0xf;
+	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)
+		tmp16 |= 3;
+	else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2)
+		tmp16 |= 2;
+	else
+		tmp16 |= 1;
+	pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, tmp16);
+
+	speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
+	speed_cntl |= LC_INITIATE_LINK_SPEED_CHANGE;
+	WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
+		if ((speed_cntl & LC_INITIATE_LINK_SPEED_CHANGE) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+static inline u32 si_pif_phy0_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(EVERGREEN_PIF_PHY0_INDEX, ((reg) & 0xffff));
+	r = RREG32(EVERGREEN_PIF_PHY0_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+	return r;
+}
+
+static inline void si_pif_phy0_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(EVERGREEN_PIF_PHY0_INDEX, ((reg) & 0xffff));
+	WREG32(EVERGREEN_PIF_PHY0_DATA, (v));
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+static inline u32 si_pif_phy1_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(EVERGREEN_PIF_PHY1_INDEX, ((reg) & 0xffff));
+	r = RREG32(EVERGREEN_PIF_PHY1_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+	return r;
+}
+
+static inline void si_pif_phy1_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(EVERGREEN_PIF_PHY1_INDEX, ((reg) & 0xffff));
+	WREG32(EVERGREEN_PIF_PHY1_DATA, (v));
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+static void si_program_aspm(struct amdgpu_device *adev)
+{
+	u32 data, orig;
+	bool disable_l0s = false, disable_l1 = false, disable_plloff_in_l1 = false;
+	bool disable_clkreq = false;
+
+	if (amdgpu_aspm == 0)
+		return;
+
+	if (adev->flags & AMD_IS_APU)
+		return;
+	orig = data = RREG32_PCIE_PORT(PCIE_LC_N_FTS_CNTL);
+	data &= ~LC_XMIT_N_FTS_MASK;
+	data |= LC_XMIT_N_FTS(0x24) | LC_XMIT_N_FTS_OVERRIDE_EN;
+	if (orig != data)
+		WREG32_PCIE_PORT(PCIE_LC_N_FTS_CNTL, data);
+
+	orig = data = RREG32_PCIE_PORT(PCIE_LC_CNTL3);
+	data |= LC_GO_TO_RECOVERY;
+	if (orig != data)
+		WREG32_PCIE_PORT(PCIE_LC_CNTL3, data);
+
+	orig = data = RREG32_PCIE(PCIE_P_CNTL);
+	data |= P_IGNORE_EDB_ERR;
+	if (orig != data)
+		WREG32_PCIE(PCIE_P_CNTL, data);
+
+	orig = data = RREG32_PCIE_PORT(PCIE_LC_CNTL);
+	data &= ~(LC_L0S_INACTIVITY_MASK | LC_L1_INACTIVITY_MASK);
+	data |= LC_PMI_TO_L1_DIS;
+	if (!disable_l0s)
+		data |= LC_L0S_INACTIVITY(7);
+
+	if (!disable_l1) {
+		data |= LC_L1_INACTIVITY(7);
+		data &= ~LC_PMI_TO_L1_DIS;
+		if (orig != data)
+			WREG32_PCIE_PORT(PCIE_LC_CNTL, data);
+
+		if (!disable_plloff_in_l1) {
+			bool clk_req_support;
+
+			orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_0);
+			data &= ~(PLL_POWER_STATE_IN_OFF_0_MASK | PLL_POWER_STATE_IN_TXS2_0_MASK);
+			data |= PLL_POWER_STATE_IN_OFF_0(7) | PLL_POWER_STATE_IN_TXS2_0(7);
+			if (orig != data)
+				si_pif_phy0_wreg(adev,PB0_PIF_PWRDOWN_0, data);
+
+			orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_1);
+			data &= ~(PLL_POWER_STATE_IN_OFF_1_MASK | PLL_POWER_STATE_IN_TXS2_1_MASK);
+			data |= PLL_POWER_STATE_IN_OFF_1(7) | PLL_POWER_STATE_IN_TXS2_1(7);
+			if (orig != data)
+				si_pif_phy0_wreg(adev,PB0_PIF_PWRDOWN_1, data);
+
+			orig = data = si_pif_phy1_rreg(adev,PB1_PIF_PWRDOWN_0);
+			data &= ~(PLL_POWER_STATE_IN_OFF_0_MASK | PLL_POWER_STATE_IN_TXS2_0_MASK);
+			data |= PLL_POWER_STATE_IN_OFF_0(7) | PLL_POWER_STATE_IN_TXS2_0(7);
+			if (orig != data)
+				si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_0, data);
+
+			orig = data = si_pif_phy1_rreg(adev,PB1_PIF_PWRDOWN_1);
+			data &= ~(PLL_POWER_STATE_IN_OFF_1_MASK | PLL_POWER_STATE_IN_TXS2_1_MASK);
+			data |= PLL_POWER_STATE_IN_OFF_1(7) | PLL_POWER_STATE_IN_TXS2_1(7);
+			if (orig != data)
+				si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_1, data);
+
+			if ((adev->family != CHIP_OLAND) && (adev->family != CHIP_HAINAN)) {
+				orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_0);
+				data &= ~PLL_RAMP_UP_TIME_0_MASK;
+				if (orig != data)
+					si_pif_phy0_wreg(adev,PB0_PIF_PWRDOWN_0, data);
+
+				orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_1);
+				data &= ~PLL_RAMP_UP_TIME_1_MASK;
+				if (orig != data)
+					si_pif_phy0_wreg(adev,PB0_PIF_PWRDOWN_1, data);
+
+				orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_2);
+				data &= ~PLL_RAMP_UP_TIME_2_MASK;
+				if (orig != data)
+					si_pif_phy0_wreg(adev,PB0_PIF_PWRDOWN_2, data);
+
+				orig = data = si_pif_phy0_rreg(adev,PB0_PIF_PWRDOWN_3);
+				data &= ~PLL_RAMP_UP_TIME_3_MASK;
+				if (orig != data)
+					si_pif_phy0_wreg(adev,PB0_PIF_PWRDOWN_3, data);
+
+				orig = data = si_pif_phy1_rreg(adev,PB1_PIF_PWRDOWN_0);
+				data &= ~PLL_RAMP_UP_TIME_0_MASK;
+				if (orig != data)
+					si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_0, data);
+
+				orig = data = si_pif_phy1_rreg(adev,PB1_PIF_PWRDOWN_1);
+				data &= ~PLL_RAMP_UP_TIME_1_MASK;
+				if (orig != data)
+					si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_1, data);
+
+				orig = data = si_pif_phy1_rreg(adev,PB1_PIF_PWRDOWN_2);
+				data &= ~PLL_RAMP_UP_TIME_2_MASK;
+				if (orig != data)
+					si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_2, data);
+
+				orig = data = si_pif_phy1_rreg(adev,PB1_PIF_PWRDOWN_3);
+				data &= ~PLL_RAMP_UP_TIME_3_MASK;
+				if (orig != data)
+					si_pif_phy1_wreg(adev,PB1_PIF_PWRDOWN_3, data);
+			}
+			orig = data = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
+			data &= ~LC_DYN_LANES_PWR_STATE_MASK;
+			data |= LC_DYN_LANES_PWR_STATE(3);
+			if (orig != data)
+				WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, data);
+
+			orig = data = si_pif_phy0_rreg(adev,PB0_PIF_CNTL);
+			data &= ~LS2_EXIT_TIME_MASK;
+			if ((adev->family == CHIP_OLAND) || (adev->family == CHIP_HAINAN))
+				data |= LS2_EXIT_TIME(5);
+			if (orig != data)
+				si_pif_phy0_wreg(adev,PB0_PIF_CNTL, data);
+
+			orig = data = si_pif_phy1_rreg(adev,PB1_PIF_CNTL);
+			data &= ~LS2_EXIT_TIME_MASK;
+			if ((adev->family == CHIP_OLAND) || (adev->family == CHIP_HAINAN))
+				data |= LS2_EXIT_TIME(5);
+			if (orig != data)
+				si_pif_phy1_wreg(adev,PB1_PIF_CNTL, data);
+
+			if (!disable_clkreq &&
+			    !pci_is_root_bus(adev->pdev->bus)) {
+				struct pci_dev *root = adev->pdev->bus->self;
+				u32 lnkcap;
+
+				clk_req_support = false;
+				pcie_capability_read_dword(root, PCI_EXP_LNKCAP, &lnkcap);
+				if (lnkcap & PCI_EXP_LNKCAP_CLKPM)
+					clk_req_support = true;
+			} else {
+				clk_req_support = false;
+			}
+
+			if (clk_req_support) {
+				orig = data = RREG32_PCIE_PORT(PCIE_LC_CNTL2);
+				data |= LC_ALLOW_PDWN_IN_L1 | LC_ALLOW_PDWN_IN_L23;
+				if (orig != data)
+					WREG32_PCIE_PORT(PCIE_LC_CNTL2, data);
+
+				orig = data = RREG32(THM_CLK_CNTL);
+				data &= ~(CMON_CLK_SEL_MASK | TMON_CLK_SEL_MASK);
+				data |= CMON_CLK_SEL(1) | TMON_CLK_SEL(1);
+				if (orig != data)
+					WREG32(THM_CLK_CNTL, data);
+
+				orig = data = RREG32(MISC_CLK_CNTL);
+				data &= ~(DEEP_SLEEP_CLK_SEL_MASK | ZCLK_SEL_MASK);
+				data |= DEEP_SLEEP_CLK_SEL(1) | ZCLK_SEL(1);
+				if (orig != data)
+					WREG32(MISC_CLK_CNTL, data);
+
+				orig = data = RREG32(CG_CLKPIN_CNTL);
+				data &= ~BCLK_AS_XCLK;
+				if (orig != data)
+					WREG32(CG_CLKPIN_CNTL, data);
+
+				orig = data = RREG32(CG_CLKPIN_CNTL_2);
+				data &= ~FORCE_BIF_REFCLK_EN;
+				if (orig != data)
+					WREG32(CG_CLKPIN_CNTL_2, data);
+
+				orig = data = RREG32(MPLL_BYPASSCLK_SEL);
+				data &= ~MPLL_CLKOUT_SEL_MASK;
+				data |= MPLL_CLKOUT_SEL(4);
+				if (orig != data)
+					WREG32(MPLL_BYPASSCLK_SEL, data);
+
+				orig = data = RREG32(SPLL_CNTL_MODE);
+				data &= ~SPLL_REFCLK_SEL_MASK;
+				if (orig != data)
+					WREG32(SPLL_CNTL_MODE, data);
+			}
+		}
+	} else {
+		if (orig != data)
+			WREG32_PCIE_PORT(PCIE_LC_CNTL, data);
+	}
+
+	orig = data = RREG32_PCIE(PCIE_CNTL2);
+	data |= SLV_MEM_LS_EN | MST_MEM_LS_EN | REPLAY_MEM_LS_EN;
+	if (orig != data)
+		WREG32_PCIE(PCIE_CNTL2, data);
+
+	if (!disable_l0s) {
+		data = RREG32_PCIE_PORT(PCIE_LC_N_FTS_CNTL);
+		if((data & LC_N_FTS_MASK) == LC_N_FTS_MASK) {
+			data = RREG32_PCIE(PCIE_LC_STATUS1);
+			if ((data & LC_REVERSE_XMIT) && (data & LC_REVERSE_RCVR)) {
+				orig = data = RREG32_PCIE_PORT(PCIE_LC_CNTL);
+				data &= ~LC_L0S_INACTIVITY_MASK;
+				if (orig != data)
+					WREG32_PCIE_PORT(PCIE_LC_CNTL, data);
+			}
+		}
+	}
+}
+
+#ifdef __linux__
+static void si_fix_pci_max_read_req_size(struct amdgpu_device *adev)
+{
+	int readrq;
+	u16 v;
+
+	readrq = pcie_get_readrq(adev->pdev);
+	v = ffs(readrq) - 8;
+	if ((v == 0) || (v == 6) || (v == 7))
+		pcie_set_readrq(adev->pdev, 512);
+}
+#else
+static void si_fix_pci_max_read_req_size(struct amdgpu_device *adev)
+{
+	pcireg_t ctl, v;
+	int off;
+
+	if (pci_get_capability(adev->pc, adev->pa_tag, PCI_CAP_PCIEXPRESS,
+			       &off, &ctl) == 0)
+		return;
+
+	ctl = pci_conf_read(adev->pc, adev->pa_tag, off + PCI_PCIE_DCSR);
+
+	v = (ctl & PCI_PCIE_DCSR_MPS) >> 12;
+	if ((v == 0) || (v == 6) || (v == 7)) {
+		ctl &= ~PCI_PCIE_DCSR_MPS;
+		ctl |= (2 << 12);
+		pci_conf_write(adev->pc, adev->pa_tag, off + PCI_PCIE_DCSR, ctl);
+	}
+}
+#endif
+
+static int si_common_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	si_fix_pci_max_read_req_size(adev);
+	si_init_golden_registers(adev);
+	si_pcie_gen3_enable(adev);
+	si_program_aspm(adev);
+
+	return 0;
+}
+
+static int si_common_hw_fini(void *handle)
+{
+	return 0;
+}
+
+static int si_common_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return si_common_hw_fini(adev);
+}
+
+static int si_common_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return si_common_hw_init(adev);
+}
+
+static bool si_common_is_idle(void *handle)
+{
+	return true;
+}
+
+static int si_common_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static int si_common_soft_reset(void *handle)
+{
+	return 0;
+}
+
+static int si_common_set_clockgating_state(void *handle,
+					    enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int si_common_set_powergating_state(void *handle,
+					    enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs si_common_ip_funcs = {
+	.name = "si_common",
+	.early_init = si_common_early_init,
+	.late_init = NULL,
+	.sw_init = si_common_sw_init,
+	.sw_fini = si_common_sw_fini,
+	.hw_init = si_common_hw_init,
+	.hw_fini = si_common_hw_fini,
+	.suspend = si_common_suspend,
+	.resume = si_common_resume,
+	.is_idle = si_common_is_idle,
+	.wait_for_idle = si_common_wait_for_idle,
+	.soft_reset = si_common_soft_reset,
+	.set_clockgating_state = si_common_set_clockgating_state,
+	.set_powergating_state = si_common_set_powergating_state,
+};
+
+static const struct amdgpu_ip_block_version si_common_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_COMMON,
+	.major = 1,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &si_common_ip_funcs,
+};
+
+int si_set_ip_blocks(struct amdgpu_device *adev)
+{
+	si_detect_hw_virtualization(adev);
+
+	switch (adev->asic_type) {
+	case CHIP_VERDE:
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+		amdgpu_device_ip_block_add(adev, &si_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+		else
+			amdgpu_device_ip_block_add(adev, &dce_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_dma_ip_block);
+		/* amdgpu_device_ip_block_add(adev, &uvd_v3_1_ip_block); */
+		/* amdgpu_device_ip_block_add(adev, &vce_v1_0_ip_block); */
+		break;
+	case CHIP_OLAND:
+		amdgpu_device_ip_block_add(adev, &si_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+		else
+			amdgpu_device_ip_block_add(adev, &dce_v6_4_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_dma_ip_block);
+		/* amdgpu_device_ip_block_add(adev, &uvd_v3_1_ip_block); */
+		/* amdgpu_device_ip_block_add(adev, &vce_v1_0_ip_block); */
+		break;
+	case CHIP_HAINAN:
+		amdgpu_device_ip_block_add(adev, &si_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &si_dma_ip_block);
+		break;
+	default:
+		BUG();
+	}
+	return 0;
+}
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_si_dma.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_si_dma.c
new file mode 100644
index 00000000000..d5dc4fdf4b9
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_si_dma.c
@@ -0,0 +1,901 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+#include "si.h"
+#include "sid.h"
+
+const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
+{
+	DMA0_REGISTER_OFFSET,
+	DMA1_REGISTER_OFFSET
+};
+
+static void si_dma_set_ring_funcs(struct amdgpu_device *adev);
+static void si_dma_set_buffer_funcs(struct amdgpu_device *adev);
+static void si_dma_set_vm_pte_funcs(struct amdgpu_device *adev);
+static void si_dma_set_irq_funcs(struct amdgpu_device *adev);
+
+static uint64_t si_dma_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	return ring->adev->wb.wb[ring->rptr_offs>>2];
+}
+
+static uint64_t si_dma_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 me = (ring == &adev->sdma.instance[0].ring) ? 0 : 1;
+
+	return (RREG32(DMA_RB_WPTR + sdma_offsets[me]) & 0x3fffc) >> 2;
+}
+
+static void si_dma_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 me = (ring == &adev->sdma.instance[0].ring) ? 0 : 1;
+
+	WREG32(DMA_RB_WPTR + sdma_offsets[me],
+	       (lower_32_bits(ring->wptr) << 2) & 0x3fffc);
+}
+
+static void si_dma_ring_emit_ib(struct amdgpu_ring *ring,
+				struct amdgpu_ib *ib,
+				unsigned vmid, bool ctx_switch)
+{
+	/* The indirect buffer packet must end on an 8 DW boundary in the DMA ring.
+	 * Pad as necessary with NOPs.
+	 */
+	while ((lower_32_bits(ring->wptr) & 7) != 5)
+		amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0, 0));
+	amdgpu_ring_write(ring, DMA_IB_PACKET(DMA_PACKET_INDIRECT_BUFFER, vmid, 0));
+	amdgpu_ring_write(ring, (ib->gpu_addr & 0xFFFFFFE0));
+	amdgpu_ring_write(ring, (ib->length_dw << 12) | (upper_32_bits(ib->gpu_addr) & 0xFF));
+
+}
+
+/**
+ * si_dma_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @fence: amdgpu fence object
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed (VI).
+ */
+static void si_dma_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				      unsigned flags)
+{
+
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_FENCE, 0, 0, 0, 0));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xff));
+	amdgpu_ring_write(ring, seq);
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_FENCE, 0, 0, 0, 0));
+		amdgpu_ring_write(ring, addr & 0xfffffffc);
+		amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xff));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+	/* generate an interrupt */
+	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_TRAP, 0, 0, 0, 0));
+}
+
+static void si_dma_stop(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl;
+	unsigned i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		/* dma0 */
+		rb_cntl = RREG32(DMA_RB_CNTL + sdma_offsets[i]);
+		rb_cntl &= ~DMA_RB_ENABLE;
+		WREG32(DMA_RB_CNTL + sdma_offsets[i], rb_cntl);
+
+		if (adev->mman.buffer_funcs_ring == ring)
+			amdgpu_ttm_set_buffer_funcs_status(adev, false);
+		ring->ready = false;
+	}
+}
+
+static int si_dma_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl, dma_cntl, ib_cntl, rb_bufsz;
+	int i, r;
+	uint64_t rptr_addr;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+
+		WREG32(DMA_SEM_INCOMPLETE_TIMER_CNTL + sdma_offsets[i], 0);
+		WREG32(DMA_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i], 0);
+
+		/* Set ring buffer size in dwords */
+		rb_bufsz = order_base_2(ring->ring_size / 4);
+		rb_cntl = rb_bufsz << 1;
+#ifdef __BIG_ENDIAN
+		rb_cntl |= DMA_RB_SWAP_ENABLE | DMA_RPTR_WRITEBACK_SWAP_ENABLE;
+#endif
+		WREG32(DMA_RB_CNTL + sdma_offsets[i], rb_cntl);
+
+		/* Initialize the ring buffer's read and write pointers */
+		WREG32(DMA_RB_RPTR + sdma_offsets[i], 0);
+		WREG32(DMA_RB_WPTR + sdma_offsets[i], 0);
+
+		rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+
+		WREG32(DMA_RB_RPTR_ADDR_LO + sdma_offsets[i], lower_32_bits(rptr_addr));
+		WREG32(DMA_RB_RPTR_ADDR_HI + sdma_offsets[i], upper_32_bits(rptr_addr) & 0xFF);
+
+		rb_cntl |= DMA_RPTR_WRITEBACK_ENABLE;
+
+		WREG32(DMA_RB_BASE + sdma_offsets[i], ring->gpu_addr >> 8);
+
+		/* enable DMA IBs */
+		ib_cntl = DMA_IB_ENABLE | CMD_VMID_FORCE;
+#ifdef __BIG_ENDIAN
+		ib_cntl |= DMA_IB_SWAP_ENABLE;
+#endif
+		WREG32(DMA_IB_CNTL + sdma_offsets[i], ib_cntl);
+
+		dma_cntl = RREG32(DMA_CNTL + sdma_offsets[i]);
+		dma_cntl &= ~CTXEMPTY_INT_ENABLE;
+		WREG32(DMA_CNTL + sdma_offsets[i], dma_cntl);
+
+		ring->wptr = 0;
+		WREG32(DMA_RB_WPTR + sdma_offsets[i], lower_32_bits(ring->wptr) << 2);
+		WREG32(DMA_RB_CNTL + sdma_offsets[i], rb_cntl | DMA_RB_ENABLE);
+
+		ring->ready = true;
+
+		r = amdgpu_ring_test_ring(ring);
+		if (r) {
+			ring->ready = false;
+			return r;
+		}
+
+		if (adev->mman.buffer_funcs_ring == ring)
+			amdgpu_ttm_set_buffer_funcs_status(adev, true);
+	}
+
+	return 0;
+}
+
+/**
+ * si_dma_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory. (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int si_dma_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_alloc(ring, 4);
+	if (r) {
+		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_device_wb_free(adev, index);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 0, 1));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr) & 0xff);
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_device_wb_free(adev, index);
+
+	return r;
+}
+
+/**
+ * si_dma_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test a simple IB in the DMA ring (VI).
+ * Returns 0 on success, error on failure.
+ */
+static int si_dma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned index;
+	u32 tmp = 0;
+	u64 gpu_addr;
+	long r;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, 256, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err0;
+	}
+
+	ib.ptr[0] = DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 0, 1);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr) & 0xff;
+	ib.ptr[3] = 0xDEADBEEF;
+	ib.length_dw = 4;
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err1;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
+		goto err1;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto err1;
+	}
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+	if (tmp == 0xDEADBEEF) {
+		DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
+	} else {
+		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+		r = -EINVAL;
+	}
+
+err1:
+	amdgpu_ib_free(adev, &ib, NULL);
+	dma_fence_put(f);
+err0:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+/**
+ * cik_dma_vm_copy_pte - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using DMA (SI).
+ */
+static void si_dma_vm_copy_pte(struct amdgpu_ib *ib,
+			       uint64_t pe, uint64_t src,
+			       unsigned count)
+{
+	unsigned bytes = count * 8;
+
+	ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_COPY,
+					      1, 0, 0, bytes);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = lower_32_bits(src);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe) & 0xff;
+	ib->ptr[ib->length_dw++] = upper_32_bits(src) & 0xff;
+}
+
+/**
+ * si_dma_vm_write_pte - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @value: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ *
+ * Update PTEs by writing them manually using DMA (SI).
+ */
+static void si_dma_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
+				uint64_t value, unsigned count,
+				uint32_t incr)
+{
+	unsigned ndw = count * 2;
+
+	ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 0, ndw);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	for (; ndw > 0; ndw -= 2) {
+		ib->ptr[ib->length_dw++] = lower_32_bits(value);
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		value += incr;
+	}
+}
+
+/**
+ * si_dma_vm_set_pte_pde - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA (CIK).
+ */
+static void si_dma_vm_set_pte_pde(struct amdgpu_ib *ib,
+				     uint64_t pe,
+				     uint64_t addr, unsigned count,
+				     uint32_t incr, uint64_t flags)
+{
+	uint64_t value;
+	unsigned ndw;
+
+	while (count) {
+		ndw = count * 2;
+		if (ndw > 0xFFFFE)
+			ndw = 0xFFFFE;
+
+		if (flags & AMDGPU_PTE_VALID)
+			value = addr;
+		else
+			value = 0;
+
+		/* for physically contiguous pages (vram) */
+		ib->ptr[ib->length_dw++] = DMA_PTE_PDE_PACKET(ndw);
+		ib->ptr[ib->length_dw++] = pe; /* dst addr */
+		ib->ptr[ib->length_dw++] = upper_32_bits(pe) & 0xff;
+		ib->ptr[ib->length_dw++] = lower_32_bits(flags); /* mask */
+		ib->ptr[ib->length_dw++] = upper_32_bits(flags);
+		ib->ptr[ib->length_dw++] = value; /* value */
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		ib->ptr[ib->length_dw++] = incr; /* increment size */
+		ib->ptr[ib->length_dw++] = 0;
+		pe += ndw * 4;
+		addr += (ndw / 2) * incr;
+		count -= ndw / 2;
+	}
+}
+
+/**
+ * si_dma_pad_ib - pad the IB to the required number of dw
+ *
+ * @ib: indirect buffer to fill with padding
+ *
+ */
+static void si_dma_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
+{
+	while (ib->length_dw & 0x7)
+		ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0, 0);
+}
+
+/**
+ * cik_sdma_ring_emit_pipeline_sync - sync the pipeline
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Make sure all previous operations are completed (CIK).
+ */
+static void si_dma_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	/* wait for idle */
+	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_POLL_REG_MEM, 0, 0, 0, 0) |
+			  (1 << 27)); /* Poll memory */
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, (0xff << 16) | upper_32_bits(addr)); /* retry, addr_hi */
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
+	amdgpu_ring_write(ring, seq); /* value */
+	amdgpu_ring_write(ring, (3 << 28) | 0x20); /* func(equal) | poll interval */
+}
+
+/**
+ * si_dma_ring_emit_vm_flush - cik vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vm: amdgpu_vm pointer
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA (VI).
+ */
+static void si_dma_ring_emit_vm_flush(struct amdgpu_ring *ring,
+				      unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for invalidate to complete */
+	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_POLL_REG_MEM, 0, 0, 0, 0));
+	amdgpu_ring_write(ring, VM_INVALIDATE_REQUEST);
+	amdgpu_ring_write(ring, 0xff << 16); /* retry */
+	amdgpu_ring_write(ring, 1 << vmid); /* mask */
+	amdgpu_ring_write(ring, 0); /* value */
+	amdgpu_ring_write(ring, (0 << 28) | 0x20); /* func(always) | poll interval */
+}
+
+static void si_dma_ring_emit_wreg(struct amdgpu_ring *ring,
+				  uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0, 0));
+	amdgpu_ring_write(ring, (0xf << 16) | reg);
+	amdgpu_ring_write(ring, val);
+}
+
+static int si_dma_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->sdma.num_instances = 2;
+
+	si_dma_set_ring_funcs(adev);
+	si_dma_set_buffer_funcs(adev);
+	si_dma_set_vm_pte_funcs(adev);
+	si_dma_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int si_dma_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* DMA0 trap event */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 224, &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	/* DMA1 trap event */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 244, &adev->sdma.trap_irq_1);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		ring->use_doorbell = false;
+		snprintf(ring->name, sizeof(ring->name), "sdma%d", i);
+		r = amdgpu_ring_init(adev, ring, 1024,
+				     &adev->sdma.trap_irq,
+				     (i == 0) ?
+				     AMDGPU_SDMA_IRQ_TRAP0 :
+				     AMDGPU_SDMA_IRQ_TRAP1);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+static int si_dma_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+
+	return 0;
+}
+
+static int si_dma_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return si_dma_start(adev);
+}
+
+static int si_dma_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	si_dma_stop(adev);
+
+	return 0;
+}
+
+static int si_dma_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return si_dma_hw_fini(adev);
+}
+
+static int si_dma_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return si_dma_hw_init(adev);
+}
+
+static bool si_dma_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(SRBM_STATUS2);
+
+	if (tmp & (DMA_BUSY_MASK | DMA1_BUSY_MASK))
+	    return false;
+
+	return true;
+}
+
+static int si_dma_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (si_dma_is_idle(handle))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int si_dma_soft_reset(void *handle)
+{
+	DRM_INFO("si_dma_soft_reset --- not implemented !!!!!!!\n");
+	return 0;
+}
+
+static int si_dma_set_trap_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *src,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	switch (type) {
+	case AMDGPU_SDMA_IRQ_TRAP0:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(DMA_CNTL + DMA0_REGISTER_OFFSET);
+			sdma_cntl &= ~TRAP_ENABLE;
+			WREG32(DMA_CNTL + DMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(DMA_CNTL + DMA0_REGISTER_OFFSET);
+			sdma_cntl |= TRAP_ENABLE;
+			WREG32(DMA_CNTL + DMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	case AMDGPU_SDMA_IRQ_TRAP1:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(DMA_CNTL + DMA1_REGISTER_OFFSET);
+			sdma_cntl &= ~TRAP_ENABLE;
+			WREG32(DMA_CNTL + DMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(DMA_CNTL + DMA1_REGISTER_OFFSET);
+			sdma_cntl |= TRAP_ENABLE;
+			WREG32(DMA_CNTL + DMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int si_dma_process_trap_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	amdgpu_fence_process(&adev->sdma.instance[0].ring);
+
+	return 0;
+}
+
+static int si_dma_process_trap_irq_1(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	amdgpu_fence_process(&adev->sdma.instance[1].ring);
+
+	return 0;
+}
+
+static int si_dma_process_illegal_inst_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in SDMA command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int si_dma_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	u32 orig, data, offset;
+	int i;
+	bool enable;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	enable = (state == AMD_CG_STATE_GATE) ? true : false;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_MGCG)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			if (i == 0)
+				offset = DMA0_REGISTER_OFFSET;
+			else
+				offset = DMA1_REGISTER_OFFSET;
+			orig = data = RREG32(DMA_POWER_CNTL + offset);
+			data &= ~MEM_POWER_OVERRIDE;
+			if (data != orig)
+				WREG32(DMA_POWER_CNTL + offset, data);
+			WREG32(DMA_CLK_CTRL + offset, 0x00000100);
+		}
+	} else {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			if (i == 0)
+				offset = DMA0_REGISTER_OFFSET;
+			else
+				offset = DMA1_REGISTER_OFFSET;
+			orig = data = RREG32(DMA_POWER_CNTL + offset);
+			data |= MEM_POWER_OVERRIDE;
+			if (data != orig)
+				WREG32(DMA_POWER_CNTL + offset, data);
+
+			orig = data = RREG32(DMA_CLK_CTRL + offset);
+			data = 0xff000000;
+			if (data != orig)
+				WREG32(DMA_CLK_CTRL + offset, data);
+		}
+	}
+
+	return 0;
+}
+
+static int si_dma_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	u32 tmp;
+
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	WREG32(DMA_PGFSM_WRITE,  0x00002000);
+	WREG32(DMA_PGFSM_CONFIG, 0x100010ff);
+
+	for (tmp = 0; tmp < 5; tmp++)
+		WREG32(DMA_PGFSM_WRITE, 0);
+
+	return 0;
+}
+
+static const struct amd_ip_funcs si_dma_ip_funcs = {
+	.name = "si_dma",
+	.early_init = si_dma_early_init,
+	.late_init = NULL,
+	.sw_init = si_dma_sw_init,
+	.sw_fini = si_dma_sw_fini,
+	.hw_init = si_dma_hw_init,
+	.hw_fini = si_dma_hw_fini,
+	.suspend = si_dma_suspend,
+	.resume = si_dma_resume,
+	.is_idle = si_dma_is_idle,
+	.wait_for_idle = si_dma_wait_for_idle,
+	.soft_reset = si_dma_soft_reset,
+	.set_clockgating_state = si_dma_set_clockgating_state,
+	.set_powergating_state = si_dma_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs si_dma_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xf,
+	.nop = DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0, 0),
+	.support_64bit_ptrs = false,
+	.get_rptr = si_dma_ring_get_rptr,
+	.get_wptr = si_dma_ring_get_wptr,
+	.set_wptr = si_dma_ring_set_wptr,
+	.emit_frame_size =
+		3 + 3 + /* hdp flush / invalidate */
+		6 + /* si_dma_ring_emit_pipeline_sync */
+		SI_FLUSH_GPU_TLB_NUM_WREG * 3 + 6 + /* si_dma_ring_emit_vm_flush */
+		9 + 9 + 9, /* si_dma_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 7 + 3, /* si_dma_ring_emit_ib */
+	.emit_ib = si_dma_ring_emit_ib,
+	.emit_fence = si_dma_ring_emit_fence,
+	.emit_pipeline_sync = si_dma_ring_emit_pipeline_sync,
+	.emit_vm_flush = si_dma_ring_emit_vm_flush,
+	.test_ring = si_dma_ring_test_ring,
+	.test_ib = si_dma_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = si_dma_ring_pad_ib,
+	.emit_wreg = si_dma_ring_emit_wreg,
+};
+
+static void si_dma_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		adev->sdma.instance[i].ring.funcs = &si_dma_ring_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs si_dma_trap_irq_funcs = {
+	.set = si_dma_set_trap_irq_state,
+	.process = si_dma_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs si_dma_trap_irq_funcs_1 = {
+	.set = si_dma_set_trap_irq_state,
+	.process = si_dma_process_trap_irq_1,
+};
+
+static const struct amdgpu_irq_src_funcs si_dma_illegal_inst_irq_funcs = {
+	.process = si_dma_process_illegal_inst_irq,
+};
+
+static void si_dma_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_LAST;
+	adev->sdma.trap_irq.funcs = &si_dma_trap_irq_funcs;
+	adev->sdma.trap_irq_1.funcs = &si_dma_trap_irq_funcs_1;
+	adev->sdma.illegal_inst_irq.funcs = &si_dma_illegal_inst_irq_funcs;
+}
+
+/**
+ * si_dma_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Copy GPU buffers using the DMA engine (VI).
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void si_dma_emit_copy_buffer(struct amdgpu_ib *ib,
+				       uint64_t src_offset,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_COPY,
+					      1, 0, 0, byte_count);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset) & 0xff;
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset) & 0xff;
+}
+
+/**
+ * si_dma_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine (VI).
+ */
+static void si_dma_emit_fill_buffer(struct amdgpu_ib *ib,
+				       uint32_t src_data,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_CONSTANT_FILL,
+					      0, 0, 0, byte_count / 4);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset) << 16;
+}
+
+
+static const struct amdgpu_buffer_funcs si_dma_buffer_funcs = {
+	.copy_max_bytes = 0xffff8,
+	.copy_num_dw = 5,
+	.emit_copy_buffer = si_dma_emit_copy_buffer,
+
+	.fill_max_bytes = 0xffff8,
+	.fill_num_dw = 4,
+	.emit_fill_buffer = si_dma_emit_fill_buffer,
+};
+
+static void si_dma_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mman.buffer_funcs == NULL) {
+		adev->mman.buffer_funcs = &si_dma_buffer_funcs;
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+	}
+}
+
+static const struct amdgpu_vm_pte_funcs si_dma_vm_pte_funcs = {
+	.copy_pte_num_dw = 5,
+	.copy_pte = si_dma_vm_copy_pte,
+
+	.write_pte = si_dma_vm_write_pte,
+	.set_pte_pde = si_dma_vm_set_pte_pde,
+};
+
+static void si_dma_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	if (adev->vm_manager.vm_pte_funcs == NULL) {
+		adev->vm_manager.vm_pte_funcs = &si_dma_vm_pte_funcs;
+		for (i = 0; i < adev->sdma.num_instances; i++)
+			adev->vm_manager.vm_pte_rings[i] =
+				&adev->sdma.instance[i].ring;
+
+		adev->vm_manager.vm_pte_num_rings = adev->sdma.num_instances;
+	}
+}
+
+const struct amdgpu_ip_block_version si_dma_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_SDMA,
+	.major = 1,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &si_dma_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_si_dpm.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_si_dpm.c
new file mode 100644
index 00000000000..1de96995e69
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_si_dpm.c
@@ -0,0 +1,8078 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_dpm.h"
+#include "amdgpu_atombios.h"
+#include "amd_pcie.h"
+#include "sid.h"
+#include "r600_dpm.h"
+#include "si_dpm.h"
+#include "atom.h"
+#include "../include/pptable.h"
+#include <linux/math64.h>
+#include <linux/seq_file.h>
+#include <linux/firmware.h>
+
+#define MC_CG_ARB_FREQ_F0           0x0a
+#define MC_CG_ARB_FREQ_F1           0x0b
+#define MC_CG_ARB_FREQ_F2           0x0c
+#define MC_CG_ARB_FREQ_F3           0x0d
+
+#define SMC_RAM_END                 0x20000
+
+#define SCLK_MIN_DEEPSLEEP_FREQ     1350
+
+
+/* sizeof(ATOM_PPLIB_EXTENDEDHEADER) */
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V2 12
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V3 14
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V4 16
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V5 18
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V6 20
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V7 22
+
+#define BIOS_SCRATCH_4                                    0x5cd
+
+MODULE_FIRMWARE("amdgpu/tahiti_smc.bin");
+MODULE_FIRMWARE("amdgpu/pitcairn_smc.bin");
+MODULE_FIRMWARE("amdgpu/pitcairn_k_smc.bin");
+MODULE_FIRMWARE("amdgpu/verde_smc.bin");
+MODULE_FIRMWARE("amdgpu/verde_k_smc.bin");
+MODULE_FIRMWARE("amdgpu/oland_smc.bin");
+MODULE_FIRMWARE("amdgpu/oland_k_smc.bin");
+MODULE_FIRMWARE("amdgpu/hainan_smc.bin");
+MODULE_FIRMWARE("amdgpu/hainan_k_smc.bin");
+MODULE_FIRMWARE("amdgpu/banks_k_2_smc.bin");
+
+static const struct amd_pm_funcs si_dpm_funcs;
+
+union power_info {
+	struct _ATOM_POWERPLAY_INFO info;
+	struct _ATOM_POWERPLAY_INFO_V2 info_2;
+	struct _ATOM_POWERPLAY_INFO_V3 info_3;
+	struct _ATOM_PPLIB_POWERPLAYTABLE pplib;
+	struct _ATOM_PPLIB_POWERPLAYTABLE2 pplib2;
+	struct _ATOM_PPLIB_POWERPLAYTABLE3 pplib3;
+	struct _ATOM_PPLIB_POWERPLAYTABLE4 pplib4;
+	struct _ATOM_PPLIB_POWERPLAYTABLE5 pplib5;
+};
+
+union fan_info {
+	struct _ATOM_PPLIB_FANTABLE fan;
+	struct _ATOM_PPLIB_FANTABLE2 fan2;
+	struct _ATOM_PPLIB_FANTABLE3 fan3;
+};
+
+union pplib_clock_info {
+	struct _ATOM_PPLIB_R600_CLOCK_INFO r600;
+	struct _ATOM_PPLIB_RS780_CLOCK_INFO rs780;
+	struct _ATOM_PPLIB_EVERGREEN_CLOCK_INFO evergreen;
+	struct _ATOM_PPLIB_SUMO_CLOCK_INFO sumo;
+	struct _ATOM_PPLIB_SI_CLOCK_INFO si;
+};
+
+static const u32 r600_utc[R600_PM_NUMBER_OF_TC] =
+{
+	R600_UTC_DFLT_00,
+	R600_UTC_DFLT_01,
+	R600_UTC_DFLT_02,
+	R600_UTC_DFLT_03,
+	R600_UTC_DFLT_04,
+	R600_UTC_DFLT_05,
+	R600_UTC_DFLT_06,
+	R600_UTC_DFLT_07,
+	R600_UTC_DFLT_08,
+	R600_UTC_DFLT_09,
+	R600_UTC_DFLT_10,
+	R600_UTC_DFLT_11,
+	R600_UTC_DFLT_12,
+	R600_UTC_DFLT_13,
+	R600_UTC_DFLT_14,
+};
+
+static const u32 r600_dtc[R600_PM_NUMBER_OF_TC] =
+{
+	R600_DTC_DFLT_00,
+	R600_DTC_DFLT_01,
+	R600_DTC_DFLT_02,
+	R600_DTC_DFLT_03,
+	R600_DTC_DFLT_04,
+	R600_DTC_DFLT_05,
+	R600_DTC_DFLT_06,
+	R600_DTC_DFLT_07,
+	R600_DTC_DFLT_08,
+	R600_DTC_DFLT_09,
+	R600_DTC_DFLT_10,
+	R600_DTC_DFLT_11,
+	R600_DTC_DFLT_12,
+	R600_DTC_DFLT_13,
+	R600_DTC_DFLT_14,
+};
+
+static const struct si_cac_config_reg cac_weights_tahiti[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0xc, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0x101, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0xc, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0x8fc, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0x95, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x34e, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x1a1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0xda, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0x46, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x208, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0xe7, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0x948, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x167, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x35, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x31, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0x18e, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg lcac_tahiti[] =
+{
+	{ 0x143, 0x0001fffe, 1, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x143, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x146, 0x0001fffe, 1, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x146, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x149, 0x0001fffe, 1, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x149, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14c, 0x0001fffe, 1, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x98, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x98, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9b, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9b, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9e, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9e, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x101, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x101, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x104, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x104, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x107, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x107, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10a, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10a, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10d, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10d, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8c, 0x0001fffe, 1, 0x8, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8f, 0x0001fffe, 1, 0x8, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x92, 0x0001fffe, 1, 0x8, SISLANDS_CACCONFIG_CGIND },
+	{ 0x92, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x95, 0x0001fffe, 1, 0x8, SISLANDS_CACCONFIG_CGIND },
+	{ 0x95, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14f, 0x0001fffe, 1, 0x8, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x152, 0x0001fffe, 1, 0x8, SISLANDS_CACCONFIG_CGIND },
+	{ 0x152, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x155, 0x0001fffe, 1, 0x8, SISLANDS_CACCONFIG_CGIND },
+	{ 0x155, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x158, 0x0001fffe, 1, 0x8, SISLANDS_CACCONFIG_CGIND },
+	{ 0x158, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x110, 0x0001fffe, 1, 0x8, SISLANDS_CACCONFIG_CGIND },
+	{ 0x110, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x113, 0x0001fffe, 1, 0x8, SISLANDS_CACCONFIG_CGIND },
+	{ 0x113, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x116, 0x0001fffe, 1, 0x8, SISLANDS_CACCONFIG_CGIND },
+	{ 0x116, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x119, 0x0001fffe, 1, 0x8, SISLANDS_CACCONFIG_CGIND },
+	{ 0x119, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11c, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11f, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x122, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x122, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x125, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x125, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x128, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x128, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12b, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12b, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15b, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15b, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15e, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15e, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x161, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x161, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x164, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x164, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x167, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x167, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16a, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16a, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16d, 0x0001fffe, 1, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16d, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x170, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x170, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x173, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x173, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x176, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x176, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x179, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x179, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17c, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17f, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+
+};
+
+static const struct si_cac_config_reg cac_override_tahiti[] =
+{
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_powertune_data powertune_data_tahiti =
+{
+	((1 << 16) | 27027),
+	6,
+	0,
+	4,
+	95,
+	{
+		0UL,
+		0UL,
+		4521550UL,
+		309631529UL,
+		-1270850L,
+		4513710L,
+		40
+	},
+	595000000UL,
+	12,
+	{
+		0,
+		0,
+		0,
+		0,
+		0,
+		0,
+		0,
+		0
+	},
+	true
+};
+
+static const struct si_dte_data dte_data_tahiti =
+{
+	{ 1159409, 0, 0, 0, 0 },
+	{ 777, 0, 0, 0, 0 },
+	2,
+	54000,
+	127000,
+	25,
+	2,
+	10,
+	13,
+	{ 27, 31, 35, 39, 43, 47, 54, 61, 67, 74, 81, 88, 95, 0, 0, 0 },
+	{ 240888759, 221057860, 235370597, 162287531, 158510299, 131423027, 116673180, 103067515, 87941937, 76209048, 68209175, 64090048, 58301890, 0, 0, 0 },
+	{ 12024, 11189, 11451, 8411, 7939, 6666, 5681, 4905, 4241, 3720, 3354, 3122, 2890, 0, 0, 0 },
+	85,
+	false
+};
+
+#if 0
+static const struct si_dte_data dte_data_tahiti_le =
+{
+	{ 0x1E8480, 0x7A1200, 0x2160EC0, 0x3938700, 0 },
+	{ 0x7D, 0x7D, 0x4E4, 0xB00, 0 },
+	0x5,
+	0xAFC8,
+	0x64,
+	0x32,
+	1,
+	0,
+	0x10,
+	{ 0x78, 0x7C, 0x82, 0x88, 0x8E, 0x94, 0x9A, 0xA0, 0xA6, 0xAC, 0xB0, 0xB4, 0xB8, 0xBC, 0xC0, 0xC4 },
+	{ 0x3938700, 0x3938700, 0x3938700, 0x3938700, 0x3938700, 0x3938700, 0x3938700, 0x3938700, 0x3938700, 0x3938700, 0x3938700, 0x3938700, 0x3938700, 0x3938700, 0x3938700, 0x3938700 },
+	{ 0x2AF8, 0x2AF8, 0x29BB, 0x27F9, 0x2637, 0x2475, 0x22B3, 0x20F1, 0x1F2F, 0x1D6D, 0x1734, 0x1414, 0x10F4, 0xDD4, 0xAB4, 0x794 },
+	85,
+	true
+};
+#endif
+
+static const struct si_dte_data dte_data_tahiti_pro =
+{
+	{ 0x1E8480, 0x3D0900, 0x989680, 0x2625A00, 0x0 },
+	{ 0x0, 0x0, 0x0, 0x0, 0x0 },
+	5,
+	45000,
+	100,
+	0xA,
+	1,
+	0,
+	0x10,
+	{ 0x96, 0xB4, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF },
+	{ 0x895440, 0x3D0900, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680 },
+	{ 0x7D0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	90,
+	true
+};
+
+static const struct si_dte_data dte_data_new_zealand =
+{
+	{ 0x1E8480, 0x3D0900, 0x989680, 0x2625A00, 0 },
+	{ 0x29B, 0x3E9, 0x537, 0x7D2, 0 },
+	0x5,
+	0xAFC8,
+	0x69,
+	0x32,
+	1,
+	0,
+	0x10,
+	{ 0x82, 0xA0, 0xB4, 0xFE, 0xFE, 0xFE, 0xFE, 0xFE, 0xFE, 0xFE, 0xFE, 0xFE, 0xFE, 0xFE, 0xFE, 0xFE },
+	{ 0x895440, 0x3D0900, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680 },
+	{ 0xDAC, 0x1388, 0x685, 0x685, 0x685, 0x685, 0x685, 0x685, 0x685, 0x685, 0x685, 0x685, 0x685, 0x685, 0x685, 0x685 },
+	85,
+	true
+};
+
+static const struct si_dte_data dte_data_aruba_pro =
+{
+	{ 0x1E8480, 0x3D0900, 0x989680, 0x2625A00, 0x0 },
+	{ 0x0, 0x0, 0x0, 0x0, 0x0 },
+	5,
+	45000,
+	100,
+	0xA,
+	1,
+	0,
+	0x10,
+	{ 0x96, 0xB4, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF },
+	{ 0x895440, 0x3D0900, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680 },
+	{ 0x1000, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	90,
+	true
+};
+
+static const struct si_dte_data dte_data_malta =
+{
+	{ 0x1E8480, 0x3D0900, 0x989680, 0x2625A00, 0x0 },
+	{ 0x0, 0x0, 0x0, 0x0, 0x0 },
+	5,
+	45000,
+	100,
+	0xA,
+	1,
+	0,
+	0x10,
+	{ 0x96, 0xB4, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF },
+	{ 0x895440, 0x3D0900, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680 },
+	{ 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	90,
+	true
+};
+
+static const struct si_cac_config_reg cac_weights_pitcairn[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0x8a, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0x24d, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x19, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0x118, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0x76, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0xc11, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0x7f3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0x403, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x367, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x4c9, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0x45d, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0x36d, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x534, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0x5da, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0x880, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0x201, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0x9f, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x1f, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x5de, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x7b, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x13, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0xf9, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x66, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x13, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0x186, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg lcac_pitcairn[] =
+{
+	{ 0x98, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x98, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x104, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x104, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x110, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x110, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14f, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8c, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x143, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x143, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9b, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9b, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x107, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x107, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x113, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x113, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x152, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x152, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8f, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x146, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x146, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9e, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9e, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10a, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10a, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x116, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x116, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x155, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x155, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x92, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x92, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x149, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x149, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x101, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x101, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10d, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10d, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x119, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x119, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x158, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x158, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x95, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x95, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14c, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11c, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11f, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x122, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x122, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x125, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x125, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x128, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x128, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12b, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12b, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x164, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x164, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x167, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x167, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16a, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16a, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15e, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15e, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x161, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x161, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15b, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15b, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16d, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16d, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x170, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x170, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x173, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x173, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x176, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x176, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x179, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x179, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17c, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17f, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg cac_override_pitcairn[] =
+{
+    { 0xFFFFFFFF }
+};
+
+static const struct si_powertune_data powertune_data_pitcairn =
+{
+	((1 << 16) | 27027),
+	5,
+	0,
+	6,
+	100,
+	{
+		51600000UL,
+		1800000UL,
+		7194395UL,
+		309631529UL,
+		-1270850L,
+		4513710L,
+		100
+	},
+	117830498UL,
+	12,
+	{
+		0,
+		0,
+		0,
+		0,
+		0,
+		0,
+		0,
+		0
+	},
+	true
+};
+
+static const struct si_dte_data dte_data_pitcairn =
+{
+	{ 0, 0, 0, 0, 0 },
+	{ 0, 0, 0, 0, 0 },
+	0,
+	0,
+	0,
+	0,
+	0,
+	0,
+	0,
+	{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
+	{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
+	{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
+	0,
+	false
+};
+
+static const struct si_dte_data dte_data_curacao_xt =
+{
+	{ 0x1E8480, 0x3D0900, 0x989680, 0x2625A00, 0x0 },
+	{ 0x0, 0x0, 0x0, 0x0, 0x0 },
+	5,
+	45000,
+	100,
+	0xA,
+	1,
+	0,
+	0x10,
+	{ 0x96, 0xB4, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF },
+	{ 0x895440, 0x3D0900, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680 },
+	{ 0x1D17, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	90,
+	true
+};
+
+static const struct si_dte_data dte_data_curacao_pro =
+{
+	{ 0x1E8480, 0x3D0900, 0x989680, 0x2625A00, 0x0 },
+	{ 0x0, 0x0, 0x0, 0x0, 0x0 },
+	5,
+	45000,
+	100,
+	0xA,
+	1,
+	0,
+	0x10,
+	{ 0x96, 0xB4, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF },
+	{ 0x895440, 0x3D0900, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680 },
+	{ 0x1D17, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	90,
+	true
+};
+
+static const struct si_dte_data dte_data_neptune_xt =
+{
+	{ 0x1E8480, 0x3D0900, 0x989680, 0x2625A00, 0x0 },
+	{ 0x0, 0x0, 0x0, 0x0, 0x0 },
+	5,
+	45000,
+	100,
+	0xA,
+	1,
+	0,
+	0x10,
+	{ 0x96, 0xB4, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF },
+	{ 0x895440, 0x3D0900, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680 },
+	{ 0x3A2F, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	90,
+	true
+};
+
+static const struct si_cac_config_reg cac_weights_chelsea_pro[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0x82, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x4F, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0x153, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0x52, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0x135, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x4F, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0x135, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0xAC, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0x118, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0xBE, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0x110, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x4CD, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x30, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0x37, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0x27, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0xC3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0x35, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0x28, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x26C, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0x3B2, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0x99D, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0xA3F, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0xA, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0xA, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x15, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x34, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0x2BD, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x30, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x7A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0x100, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg cac_weights_chelsea_xt[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0x82, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x4F, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0x153, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0x52, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0x135, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x4F, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0x135, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0xAC, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0x118, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0xBE, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0x110, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x4CD, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x30, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0x37, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0x27, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0xC3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0x35, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0x28, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x26C, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0x3B2, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0x99D, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0xA3F, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0xA, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0xA, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x15, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x34, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0x30A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x30, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x7A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0x100, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg cac_weights_heathrow[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0x82, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x4F, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0x153, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0x52, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0x135, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x4F, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0x135, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0xAC, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0x118, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0xBE, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0x110, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x4CD, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x30, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0x37, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0x27, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0xC3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0x35, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0x28, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x26C, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0x3B2, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0x99D, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0xA3F, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0xA, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0xA, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x15, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x34, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0x362, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x30, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x7A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0x100, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg cac_weights_cape_verde_pro[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0x82, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x4F, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0x153, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0x52, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0x135, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x4F, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0x135, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0xAC, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0x118, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0xBE, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0x110, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x4CD, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x30, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0x37, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0x27, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0xC3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0x35, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0x28, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x26C, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0x3B2, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0x99D, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0xA3F, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0xA, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0xA, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x15, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x34, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0x315, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x30, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x7A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0x100, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg cac_weights_cape_verde[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0x82, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x4F, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0x153, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0x52, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0x135, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x4F, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0x135, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0xAC, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0x118, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0xBE, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0x110, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x4CD, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x30, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0x37, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0x27, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0xC3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0x35, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0x28, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x26C, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0x3B2, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0x99D, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0xA3F, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0xA, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0xA, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x15, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x34, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0x3BA, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x30, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x7A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0x100, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg lcac_cape_verde[] =
+{
+	{ 0x98, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x98, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x104, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x104, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x110, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x110, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14f, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8c, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x143, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x143, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9b, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9b, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x107, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x107, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x113, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x113, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x152, 0x0001fffe, 1, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x152, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8f, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x146, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x146, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11c, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11f, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x164, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x164, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x167, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x167, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16a, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16a, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15e, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15e, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x161, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x161, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15b, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15b, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16d, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16d, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x170, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x170, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x173, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x173, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x176, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x176, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x179, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x179, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17c, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17f, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg cac_override_cape_verde[] =
+{
+    { 0xFFFFFFFF }
+};
+
+static const struct si_powertune_data powertune_data_cape_verde =
+{
+	((1 << 16) | 0x6993),
+	5,
+	0,
+	7,
+	105,
+	{
+		0UL,
+		0UL,
+		7194395UL,
+		309631529UL,
+		-1270850L,
+		4513710L,
+		100
+	},
+	117830498UL,
+	12,
+	{
+		0,
+		0,
+		0,
+		0,
+		0,
+		0,
+		0,
+		0
+	},
+	true
+};
+
+static const struct si_dte_data dte_data_cape_verde =
+{
+	{ 0, 0, 0, 0, 0 },
+	{ 0, 0, 0, 0, 0 },
+	0,
+	0,
+	0,
+	0,
+	0,
+	0,
+	0,
+	{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
+	{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
+	{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
+	0,
+	false
+};
+
+static const struct si_dte_data dte_data_venus_xtx =
+{
+	{ 0x1E8480, 0x3D0900, 0x989680, 0x2625A00, 0x0 },
+	{ 0x71C, 0xAAB, 0xE39, 0x11C7, 0x0 },
+	5,
+	55000,
+	0x69,
+	0xA,
+	1,
+	0,
+	0x3,
+	{ 0x96, 0xB4, 0xFF, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	{ 0x895440, 0x3D0900, 0x989680, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	{ 0xD6D8, 0x88B8, 0x1555, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	90,
+	true
+};
+
+static const struct si_dte_data dte_data_venus_xt =
+{
+	{ 0x1E8480, 0x3D0900, 0x989680, 0x2625A00, 0x0 },
+	{ 0xBDA, 0x11C7, 0x17B4, 0x1DA1, 0x0 },
+	5,
+	55000,
+	0x69,
+	0xA,
+	1,
+	0,
+	0x3,
+	{ 0x96, 0xB4, 0xFF, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	{ 0x895440, 0x3D0900, 0x989680, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	{ 0xAFC8, 0x88B8, 0x238E, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	90,
+	true
+};
+
+static const struct si_dte_data dte_data_venus_pro =
+{
+	{  0x1E8480, 0x3D0900, 0x989680, 0x2625A00, 0x0 },
+	{ 0x11C7, 0x1AAB, 0x238E, 0x2C72, 0x0 },
+	5,
+	55000,
+	0x69,
+	0xA,
+	1,
+	0,
+	0x3,
+	{ 0x96, 0xB4, 0xFF, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	{ 0x895440, 0x3D0900, 0x989680, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	{ 0x88B8, 0x88B8, 0x3555, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	90,
+	true
+};
+
+static const struct si_cac_config_reg cac_weights_oland[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0x82, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x4F, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0x153, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0x52, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0x135, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x4F, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0x135, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0xAC, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0x118, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0xBE, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0x110, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x4CD, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x30, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0x37, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0x27, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0xC3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0x35, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0x28, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x26C, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0x3B2, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0x99D, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0xA3F, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0xA, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0xA, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x15, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x34, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0x3BA, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x30, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x7A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0x100, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg cac_weights_mars_pro[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0x43, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x29, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0xAF, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0x2A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0xA0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x29, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0xA0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0x59, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0x1A5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0x1D6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0x2A3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x8FD, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x76, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0x8A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0xA3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0x71, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0x36, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0xA6, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x81, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0x3D2, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0x27C, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0xA96, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0xB, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x15, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x7, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x36, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x10, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x10, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x32, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x7E, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0x280, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0x7, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0x3C, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0x203, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0xB4, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg cac_weights_mars_xt[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0x43, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x29, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0xAF, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0x2A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0xA0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x29, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0xA0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0x59, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0x1A5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0x1D6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0x2A3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x8FD, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x76, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0x8A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0xA3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0x71, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0x36, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0xA6, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x81, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0x3D2, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0x27C, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0xA96, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0xB, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x15, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x7, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x36, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x10, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x10, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0x60, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x32, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x7E, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0x280, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0x7, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0x3C, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0x203, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0xB4, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg cac_weights_oland_pro[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0x43, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x29, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0xAF, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0x2A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0xA0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x29, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0xA0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0x59, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0x1A5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0x1D6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0x2A3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x8FD, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x76, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0x8A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0xA3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0x71, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0x36, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0xA6, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x81, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0x3D2, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0x27C, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0xA96, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0xB, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x15, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x7, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x36, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x10, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x10, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0x90, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x32, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x7E, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0x280, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0x7, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0x3C, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0x203, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0xB4, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg cac_weights_oland_xt[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0x43, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x29, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0xAF, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0x2A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0xA0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x29, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0xA0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0x59, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0x1A5, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0x1D6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0x2A3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x8FD, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x76, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0x8A, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0xA3, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0x71, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0x36, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0xA6, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x81, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0x3D2, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0x27C, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0xA96, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0x5, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0xB, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x3, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x15, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x7, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x36, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x10, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x10, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0x120, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x32, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x7E, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0x280, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0x7, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0x3C, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0x203, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0xB4, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg lcac_oland[] =
+{
+	{ 0x98, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x98, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x104, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x104, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x110, 0x0001fffe, 1, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x110, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14f, 0x0001fffe, 1, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8c, 0x0001fffe, 1, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x143, 0x0001fffe, 1, 0x4, SISLANDS_CACCONFIG_CGIND },
+	{ 0x143, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11c, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11f, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x164, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x164, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x167, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x167, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16a, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16a, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15e, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15e, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x161, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x161, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15b, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15b, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16d, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16d, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x170, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x170, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x173, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x173, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x176, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x176, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x179, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x179, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17c, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17f, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg lcac_mars_pro[] =
+{
+	{ 0x98, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x98, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x104, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x104, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x110, 0x0001fffe, 1, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x110, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14f, 0x0001fffe, 1, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8c, 0x0001fffe, 1, 0x6, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x143, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x143, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11c, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11f, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x164, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x164, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x167, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x167, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16a, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16a, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15e, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15e, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x161, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x161, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15b, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15b, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16d, 0x0001fffe, 1, 0x2, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16d, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x170, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x170, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x173, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x173, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x176, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x176, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x179, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x179, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17c, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17c, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17f, 0x0001fffe, 1, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17f, 0x00000001, 0, 0x1, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_cac_config_reg cac_override_oland[] =
+{
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_powertune_data powertune_data_oland =
+{
+	((1 << 16) | 0x6993),
+	5,
+	0,
+	7,
+	105,
+	{
+		0UL,
+		0UL,
+		7194395UL,
+		309631529UL,
+		-1270850L,
+		4513710L,
+		100
+	},
+	117830498UL,
+	12,
+	{
+		0,
+		0,
+		0,
+		0,
+		0,
+		0,
+		0,
+		0
+	},
+	true
+};
+
+static const struct si_powertune_data powertune_data_mars_pro =
+{
+	((1 << 16) | 0x6993),
+	5,
+	0,
+	7,
+	105,
+	{
+		0UL,
+		0UL,
+		7194395UL,
+		309631529UL,
+		-1270850L,
+		4513710L,
+		100
+	},
+	117830498UL,
+	12,
+	{
+		0,
+		0,
+		0,
+		0,
+		0,
+		0,
+		0,
+		0
+	},
+	true
+};
+
+static const struct si_dte_data dte_data_oland =
+{
+	{ 0, 0, 0, 0, 0 },
+	{ 0, 0, 0, 0, 0 },
+	0,
+	0,
+	0,
+	0,
+	0,
+	0,
+	0,
+	{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
+	{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
+	{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
+	0,
+	false
+};
+
+static const struct si_dte_data dte_data_mars_pro =
+{
+	{ 0x1E8480, 0x3D0900, 0x989680, 0x2625A00, 0x0 },
+	{ 0x0, 0x0, 0x0, 0x0, 0x0 },
+	5,
+	55000,
+	105,
+	0xA,
+	1,
+	0,
+	0x10,
+	{ 0x96, 0xB4, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF },
+	{ 0x895440, 0x3D0900, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680 },
+	{ 0xF627, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	90,
+	true
+};
+
+static const struct si_dte_data dte_data_sun_xt =
+{
+	{ 0x1E8480, 0x3D0900, 0x989680, 0x2625A00, 0x0 },
+	{ 0x0, 0x0, 0x0, 0x0, 0x0 },
+	5,
+	55000,
+	105,
+	0xA,
+	1,
+	0,
+	0x10,
+	{ 0x96, 0xB4, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF },
+	{ 0x895440, 0x3D0900, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680, 0x989680 },
+	{ 0xD555, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 },
+	90,
+	true
+};
+
+
+static const struct si_cac_config_reg cac_weights_hainan[] =
+{
+	{ 0x0, 0x0000ffff, 0, 0x2d9, SISLANDS_CACCONFIG_CGIND },
+	{ 0x0, 0xffff0000, 16, 0x22b, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0x0000ffff, 0, 0x21c, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1, 0xffff0000, 16, 0x1dc, SISLANDS_CACCONFIG_CGIND },
+	{ 0x2, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0x0000ffff, 0, 0x24e, SISLANDS_CACCONFIG_CGIND },
+	{ 0x3, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0x0000ffff, 0, 0x35e, SISLANDS_CACCONFIG_CGIND },
+	{ 0x5, 0xffff0000, 16, 0x1143, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0x0000ffff, 0, 0xe17, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6, 0xffff0000, 16, 0x441, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18f, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0x0000ffff, 0, 0x28b, SISLANDS_CACCONFIG_CGIND },
+	{ 0x7, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x8, 0xffff0000, 16, 0xabe, SISLANDS_CACCONFIG_CGIND },
+	{ 0x9, 0x0000ffff, 0, 0xf11, SISLANDS_CACCONFIG_CGIND },
+	{ 0xa, 0x0000ffff, 0, 0x907, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0x0000ffff, 0, 0xb45, SISLANDS_CACCONFIG_CGIND },
+	{ 0xb, 0xffff0000, 16, 0xd1e, SISLANDS_CACCONFIG_CGIND },
+	{ 0xc, 0x0000ffff, 0, 0xa2c, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0x0000ffff, 0, 0x62, SISLANDS_CACCONFIG_CGIND },
+	{ 0xd, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0xe, 0x0000ffff, 0, 0x1f3, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0x0000ffff, 0, 0x42, SISLANDS_CACCONFIG_CGIND },
+	{ 0xf, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x10, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0x0000ffff, 0, 0x709, SISLANDS_CACCONFIG_CGIND },
+	{ 0x11, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x12, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x13, 0xffff0000, 16, 0x3a, SISLANDS_CACCONFIG_CGIND },
+	{ 0x14, 0x0000ffff, 0, 0x357, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0x0000ffff, 0, 0x9f, SISLANDS_CACCONFIG_CGIND },
+	{ 0x15, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x4e, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0x0000ffff, 0, 0x314, SISLANDS_CACCONFIG_CGIND },
+	{ 0x16, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x17, 0x0000ffff, 0, 0x6d, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x18, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0x0000ffff, 0, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x19, 0xffff0000, 16, 0x0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1a, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1b, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1c, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1d, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1e, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x1f, 0xffff0000, 16, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x20, 0x0000ffff, 0, 0, SISLANDS_CACCONFIG_CGIND },
+	{ 0x6d, 0x0000ffff, 0, 0x1b9, SISLANDS_CACCONFIG_CGIND },
+	{ 0xFFFFFFFF }
+};
+
+static const struct si_powertune_data powertune_data_hainan =
+{
+	((1 << 16) | 0x6993),
+	5,
+	0,
+	9,
+	105,
+	{
+		0UL,
+		0UL,
+		7194395UL,
+		309631529UL,
+		-1270850L,
+		4513710L,
+		100
+	},
+	117830498UL,
+	12,
+	{
+		0,
+		0,
+		0,
+		0,
+		0,
+		0,
+		0,
+		0
+	},
+	true
+};
+
+static struct rv7xx_power_info *rv770_get_pi(struct amdgpu_device *adev);
+static struct evergreen_power_info *evergreen_get_pi(struct amdgpu_device *adev);
+static struct ni_power_info *ni_get_pi(struct amdgpu_device *adev);
+static struct  si_ps *si_get_ps(struct amdgpu_ps *rps);
+
+static int si_populate_voltage_value(struct amdgpu_device *adev,
+				     const struct atom_voltage_table *table,
+				     u16 value, SISLANDS_SMC_VOLTAGE_VALUE *voltage);
+static int si_get_std_voltage_value(struct amdgpu_device *adev,
+				    SISLANDS_SMC_VOLTAGE_VALUE *voltage,
+				    u16 *std_voltage);
+static int si_write_smc_soft_register(struct amdgpu_device *adev,
+				      u16 reg_offset, u32 value);
+static int si_convert_power_level_to_smc(struct amdgpu_device *adev,
+					 struct rv7xx_pl *pl,
+					 SISLANDS_SMC_HW_PERFORMANCE_LEVEL *level);
+static int si_calculate_sclk_params(struct amdgpu_device *adev,
+				    u32 engine_clock,
+				    SISLANDS_SMC_SCLK_VALUE *sclk);
+
+static void si_thermal_start_smc_fan_control(struct amdgpu_device *adev);
+static void si_fan_ctrl_set_default_mode(struct amdgpu_device *adev);
+static void si_dpm_set_irq_funcs(struct amdgpu_device *adev);
+
+static struct si_power_info *si_get_pi(struct amdgpu_device *adev)
+{
+	struct si_power_info *pi = adev->pm.dpm.priv;
+	return pi;
+}
+
+static void si_calculate_leakage_for_v_and_t_formula(const struct ni_leakage_coeffients *coeff,
+						     u16 v, s32 t, u32 ileakage, u32 *leakage)
+{
+	s64 kt, kv, leakage_w, i_leakage, vddc;
+	s64 temperature, t_slope, t_intercept, av, bv, t_ref;
+	s64 tmp;
+
+	i_leakage = div64_s64(drm_int2fixp(ileakage), 100);
+	vddc = div64_s64(drm_int2fixp(v), 1000);
+	temperature = div64_s64(drm_int2fixp(t), 1000);
+
+	t_slope = div64_s64(drm_int2fixp(coeff->t_slope), 100000000);
+	t_intercept = div64_s64(drm_int2fixp(coeff->t_intercept), 100000000);
+	av = div64_s64(drm_int2fixp(coeff->av), 100000000);
+	bv = div64_s64(drm_int2fixp(coeff->bv), 100000000);
+	t_ref = drm_int2fixp(coeff->t_ref);
+
+	tmp = drm_fixp_mul(t_slope, vddc) + t_intercept;
+	kt = drm_fixp_exp(drm_fixp_mul(tmp, temperature));
+	kt = drm_fixp_div(kt, drm_fixp_exp(drm_fixp_mul(tmp, t_ref)));
+	kv = drm_fixp_mul(av, drm_fixp_exp(drm_fixp_mul(bv, vddc)));
+
+	leakage_w = drm_fixp_mul(drm_fixp_mul(drm_fixp_mul(i_leakage, kt), kv), vddc);
+
+	*leakage = drm_fixp2int(leakage_w * 1000);
+}
+
+static void si_calculate_leakage_for_v_and_t(struct amdgpu_device *adev,
+					     const struct ni_leakage_coeffients *coeff,
+					     u16 v,
+					     s32 t,
+					     u32 i_leakage,
+					     u32 *leakage)
+{
+	si_calculate_leakage_for_v_and_t_formula(coeff, v, t, i_leakage, leakage);
+}
+
+static void si_calculate_leakage_for_v_formula(const struct ni_leakage_coeffients *coeff,
+					       const u32 fixed_kt, u16 v,
+					       u32 ileakage, u32 *leakage)
+{
+	s64 kt, kv, leakage_w, i_leakage, vddc;
+
+	i_leakage = div64_s64(drm_int2fixp(ileakage), 100);
+	vddc = div64_s64(drm_int2fixp(v), 1000);
+
+	kt = div64_s64(drm_int2fixp(fixed_kt), 100000000);
+	kv = drm_fixp_mul(div64_s64(drm_int2fixp(coeff->av), 100000000),
+			  drm_fixp_exp(drm_fixp_mul(div64_s64(drm_int2fixp(coeff->bv), 100000000), vddc)));
+
+	leakage_w = drm_fixp_mul(drm_fixp_mul(drm_fixp_mul(i_leakage, kt), kv), vddc);
+
+	*leakage = drm_fixp2int(leakage_w * 1000);
+}
+
+static void si_calculate_leakage_for_v(struct amdgpu_device *adev,
+				       const struct ni_leakage_coeffients *coeff,
+				       const u32 fixed_kt,
+				       u16 v,
+				       u32 i_leakage,
+				       u32 *leakage)
+{
+	si_calculate_leakage_for_v_formula(coeff, fixed_kt, v, i_leakage, leakage);
+}
+
+
+static void si_update_dte_from_pl2(struct amdgpu_device *adev,
+				   struct si_dte_data *dte_data)
+{
+	u32 p_limit1 = adev->pm.dpm.tdp_limit;
+	u32 p_limit2 = adev->pm.dpm.near_tdp_limit;
+	u32 k = dte_data->k;
+	u32 t_max = dte_data->max_t;
+	u32 t_split[5] = { 10, 15, 20, 25, 30 };
+	u32 t_0 = dte_data->t0;
+	u32 i;
+
+	if (p_limit2 != 0 && p_limit2 <= p_limit1) {
+		dte_data->tdep_count = 3;
+
+		for (i = 0; i < k; i++) {
+			dte_data->r[i] =
+				(t_split[i] * (t_max - t_0/(u32)1000) * (1 << 14)) /
+				(p_limit2  * (u32)100);
+		}
+
+		dte_data->tdep_r[1] = dte_data->r[4] * 2;
+
+		for (i = 2; i < SMC_SISLANDS_DTE_MAX_TEMPERATURE_DEPENDENT_ARRAY_SIZE; i++) {
+			dte_data->tdep_r[i] = dte_data->r[4];
+		}
+	} else {
+		DRM_ERROR("Invalid PL2! DTE will not be updated.\n");
+	}
+}
+
+static struct rv7xx_power_info *rv770_get_pi(struct amdgpu_device *adev)
+{
+	struct rv7xx_power_info *pi = adev->pm.dpm.priv;
+
+	return pi;
+}
+
+static struct ni_power_info *ni_get_pi(struct amdgpu_device *adev)
+{
+	struct ni_power_info *pi = adev->pm.dpm.priv;
+
+	return pi;
+}
+
+static struct si_ps *si_get_ps(struct amdgpu_ps *aps)
+{
+	struct  si_ps *ps = aps->ps_priv;
+
+	return ps;
+}
+
+static void si_initialize_powertune_defaults(struct amdgpu_device *adev)
+{
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	bool update_dte_from_pl2 = false;
+
+	if (adev->asic_type == CHIP_TAHITI) {
+		si_pi->cac_weights = cac_weights_tahiti;
+		si_pi->lcac_config = lcac_tahiti;
+		si_pi->cac_override = cac_override_tahiti;
+		si_pi->powertune_data = &powertune_data_tahiti;
+		si_pi->dte_data = dte_data_tahiti;
+
+		switch (adev->pdev->device) {
+		case 0x6798:
+			si_pi->dte_data.enable_dte_by_default = true;
+			break;
+		case 0x6799:
+			si_pi->dte_data = dte_data_new_zealand;
+			break;
+		case 0x6790:
+		case 0x6791:
+		case 0x6792:
+		case 0x679E:
+			si_pi->dte_data = dte_data_aruba_pro;
+			update_dte_from_pl2 = true;
+			break;
+		case 0x679B:
+			si_pi->dte_data = dte_data_malta;
+			update_dte_from_pl2 = true;
+			break;
+		case 0x679A:
+			si_pi->dte_data = dte_data_tahiti_pro;
+			update_dte_from_pl2 = true;
+			break;
+		default:
+			if (si_pi->dte_data.enable_dte_by_default == true)
+				DRM_ERROR("DTE is not enabled!\n");
+			break;
+		}
+	} else if (adev->asic_type == CHIP_PITCAIRN) {
+		si_pi->cac_weights = cac_weights_pitcairn;
+		si_pi->lcac_config = lcac_pitcairn;
+		si_pi->cac_override = cac_override_pitcairn;
+		si_pi->powertune_data = &powertune_data_pitcairn;
+
+		switch (adev->pdev->device) {
+		case 0x6810:
+		case 0x6818:
+			si_pi->dte_data = dte_data_curacao_xt;
+			update_dte_from_pl2 = true;
+			break;
+		case 0x6819:
+		case 0x6811:
+			si_pi->dte_data = dte_data_curacao_pro;
+			update_dte_from_pl2 = true;
+			break;
+		case 0x6800:
+		case 0x6806:
+			si_pi->dte_data = dte_data_neptune_xt;
+			update_dte_from_pl2 = true;
+			break;
+		default:
+			si_pi->dte_data = dte_data_pitcairn;
+			break;
+		}
+	} else if (adev->asic_type == CHIP_VERDE) {
+		si_pi->lcac_config = lcac_cape_verde;
+		si_pi->cac_override = cac_override_cape_verde;
+		si_pi->powertune_data = &powertune_data_cape_verde;
+
+		switch (adev->pdev->device) {
+		case 0x683B:
+		case 0x683F:
+		case 0x6829:
+		case 0x6835:
+			si_pi->cac_weights = cac_weights_cape_verde_pro;
+			si_pi->dte_data = dte_data_cape_verde;
+			break;
+		case 0x682C:
+			si_pi->cac_weights = cac_weights_cape_verde_pro;
+			si_pi->dte_data = dte_data_sun_xt;
+			update_dte_from_pl2 = true;
+			break;
+		case 0x6825:
+		case 0x6827:
+			si_pi->cac_weights = cac_weights_heathrow;
+			si_pi->dte_data = dte_data_cape_verde;
+			break;
+		case 0x6824:
+		case 0x682D:
+			si_pi->cac_weights = cac_weights_chelsea_xt;
+			si_pi->dte_data = dte_data_cape_verde;
+			break;
+		case 0x682F:
+			si_pi->cac_weights = cac_weights_chelsea_pro;
+			si_pi->dte_data = dte_data_cape_verde;
+			break;
+		case 0x6820:
+			si_pi->cac_weights = cac_weights_heathrow;
+			si_pi->dte_data = dte_data_venus_xtx;
+			break;
+		case 0x6821:
+			si_pi->cac_weights = cac_weights_heathrow;
+			si_pi->dte_data = dte_data_venus_xt;
+			break;
+		case 0x6823:
+		case 0x682B:
+		case 0x6822:
+		case 0x682A:
+			si_pi->cac_weights = cac_weights_chelsea_pro;
+			si_pi->dte_data = dte_data_venus_pro;
+			break;
+		default:
+			si_pi->cac_weights = cac_weights_cape_verde;
+			si_pi->dte_data = dte_data_cape_verde;
+			break;
+		}
+	} else if (adev->asic_type == CHIP_OLAND) {
+		si_pi->lcac_config = lcac_mars_pro;
+		si_pi->cac_override = cac_override_oland;
+		si_pi->powertune_data = &powertune_data_mars_pro;
+		si_pi->dte_data = dte_data_mars_pro;
+
+		switch (adev->pdev->device) {
+		case 0x6601:
+		case 0x6621:
+		case 0x6603:
+		case 0x6605:
+			si_pi->cac_weights = cac_weights_mars_pro;
+			update_dte_from_pl2 = true;
+			break;
+		case 0x6600:
+		case 0x6606:
+		case 0x6620:
+		case 0x6604:
+			si_pi->cac_weights = cac_weights_mars_xt;
+			update_dte_from_pl2 = true;
+			break;
+		case 0x6611:
+		case 0x6613:
+		case 0x6608:
+			si_pi->cac_weights = cac_weights_oland_pro;
+			update_dte_from_pl2 = true;
+			break;
+		case 0x6610:
+			si_pi->cac_weights = cac_weights_oland_xt;
+			update_dte_from_pl2 = true;
+			break;
+		default:
+			si_pi->cac_weights = cac_weights_oland;
+			si_pi->lcac_config = lcac_oland;
+			si_pi->cac_override = cac_override_oland;
+			si_pi->powertune_data = &powertune_data_oland;
+			si_pi->dte_data = dte_data_oland;
+			break;
+		}
+	} else if (adev->asic_type == CHIP_HAINAN) {
+		si_pi->cac_weights = cac_weights_hainan;
+		si_pi->lcac_config = lcac_oland;
+		si_pi->cac_override = cac_override_oland;
+		si_pi->powertune_data = &powertune_data_hainan;
+		si_pi->dte_data = dte_data_sun_xt;
+		update_dte_from_pl2 = true;
+	} else {
+		DRM_ERROR("Unknown SI asic revision, failed to initialize PowerTune!\n");
+		return;
+	}
+
+	ni_pi->enable_power_containment = false;
+	ni_pi->enable_cac = false;
+	ni_pi->enable_sq_ramping = false;
+	si_pi->enable_dte = false;
+
+	if (si_pi->powertune_data->enable_powertune_by_default) {
+		ni_pi->enable_power_containment = true;
+		ni_pi->enable_cac = true;
+		if (si_pi->dte_data.enable_dte_by_default) {
+			si_pi->enable_dte = true;
+			if (update_dte_from_pl2)
+				si_update_dte_from_pl2(adev, &si_pi->dte_data);
+
+		}
+		ni_pi->enable_sq_ramping = true;
+	}
+
+	ni_pi->driver_calculate_cac_leakage = true;
+	ni_pi->cac_configuration_required = true;
+
+	if (ni_pi->cac_configuration_required) {
+		ni_pi->support_cac_long_term_average = true;
+		si_pi->dyn_powertune_data.l2_lta_window_size =
+			si_pi->powertune_data->l2_lta_window_size_default;
+		si_pi->dyn_powertune_data.lts_truncate =
+			si_pi->powertune_data->lts_truncate_default;
+	} else {
+		ni_pi->support_cac_long_term_average = false;
+		si_pi->dyn_powertune_data.l2_lta_window_size = 0;
+		si_pi->dyn_powertune_data.lts_truncate = 0;
+	}
+
+	si_pi->dyn_powertune_data.disable_uvd_powertune = false;
+}
+
+static u32 si_get_smc_power_scaling_factor(struct amdgpu_device *adev)
+{
+	return 1;
+}
+
+static u32 si_calculate_cac_wintime(struct amdgpu_device *adev)
+{
+	u32 xclk;
+	u32 wintime;
+	u32 cac_window;
+	u32 cac_window_size;
+
+	xclk = amdgpu_asic_get_xclk(adev);
+
+	if (xclk == 0)
+		return 0;
+
+	cac_window = RREG32(CG_CAC_CTRL) & CAC_WINDOW_MASK;
+	cac_window_size = ((cac_window & 0xFFFF0000) >> 16) * (cac_window & 0x0000FFFF);
+
+	wintime = (cac_window_size * 100) / xclk;
+
+	return wintime;
+}
+
+static u32 si_scale_power_for_smc(u32 power_in_watts, u32 scaling_factor)
+{
+	return power_in_watts;
+}
+
+static int si_calculate_adjusted_tdp_limits(struct amdgpu_device *adev,
+					    bool adjust_polarity,
+					    u32 tdp_adjustment,
+					    u32 *tdp_limit,
+					    u32 *near_tdp_limit)
+{
+	u32 adjustment_delta, max_tdp_limit;
+
+	if (tdp_adjustment > (u32)adev->pm.dpm.tdp_od_limit)
+		return -EINVAL;
+
+	max_tdp_limit = ((100 + 100) * adev->pm.dpm.tdp_limit) / 100;
+
+	if (adjust_polarity) {
+		*tdp_limit = ((100 + tdp_adjustment) * adev->pm.dpm.tdp_limit) / 100;
+		*near_tdp_limit = adev->pm.dpm.near_tdp_limit_adjusted + (*tdp_limit - adev->pm.dpm.tdp_limit);
+	} else {
+		*tdp_limit = ((100 - tdp_adjustment) * adev->pm.dpm.tdp_limit) / 100;
+		adjustment_delta  = adev->pm.dpm.tdp_limit - *tdp_limit;
+		if (adjustment_delta < adev->pm.dpm.near_tdp_limit_adjusted)
+			*near_tdp_limit = adev->pm.dpm.near_tdp_limit_adjusted - adjustment_delta;
+		else
+			*near_tdp_limit = 0;
+	}
+
+	if ((*tdp_limit <= 0) || (*tdp_limit > max_tdp_limit))
+		return -EINVAL;
+	if ((*near_tdp_limit <= 0) || (*near_tdp_limit > *tdp_limit))
+		return -EINVAL;
+
+	return 0;
+}
+
+static int si_populate_smc_tdp_limits(struct amdgpu_device *adev,
+				      struct amdgpu_ps *amdgpu_state)
+{
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+
+	if (ni_pi->enable_power_containment) {
+		SISLANDS_SMC_STATETABLE *smc_table = &si_pi->smc_statetable;
+		PP_SIslands_PAPMParameters *papm_parm;
+		struct amdgpu_ppm_table *ppm = adev->pm.dpm.dyn_state.ppm_table;
+		u32 scaling_factor = si_get_smc_power_scaling_factor(adev);
+		u32 tdp_limit;
+		u32 near_tdp_limit;
+		int ret;
+
+		if (scaling_factor == 0)
+			return -EINVAL;
+
+		memset(smc_table, 0, sizeof(SISLANDS_SMC_STATETABLE));
+
+		ret = si_calculate_adjusted_tdp_limits(adev,
+						       false, /* ??? */
+						       adev->pm.dpm.tdp_adjustment,
+						       &tdp_limit,
+						       &near_tdp_limit);
+		if (ret)
+			return ret;
+
+		smc_table->dpm2Params.TDPLimit =
+			cpu_to_be32(si_scale_power_for_smc(tdp_limit, scaling_factor) * 1000);
+		smc_table->dpm2Params.NearTDPLimit =
+			cpu_to_be32(si_scale_power_for_smc(near_tdp_limit, scaling_factor) * 1000);
+		smc_table->dpm2Params.SafePowerLimit =
+			cpu_to_be32(si_scale_power_for_smc((near_tdp_limit * SISLANDS_DPM2_TDP_SAFE_LIMIT_PERCENT) / 100, scaling_factor) * 1000);
+
+		ret = amdgpu_si_copy_bytes_to_smc(adev,
+						  (si_pi->state_table_start + offsetof(SISLANDS_SMC_STATETABLE, dpm2Params) +
+						   offsetof(PP_SIslands_DPM2Parameters, TDPLimit)),
+						  (u8 *)(&(smc_table->dpm2Params.TDPLimit)),
+						  sizeof(u32) * 3,
+						  si_pi->sram_end);
+		if (ret)
+			return ret;
+
+		if (si_pi->enable_ppm) {
+			papm_parm = &si_pi->papm_parm;
+			memset(papm_parm, 0, sizeof(PP_SIslands_PAPMParameters));
+			papm_parm->NearTDPLimitTherm = cpu_to_be32(ppm->dgpu_tdp);
+			papm_parm->dGPU_T_Limit = cpu_to_be32(ppm->tj_max);
+			papm_parm->dGPU_T_Warning = cpu_to_be32(95);
+			papm_parm->dGPU_T_Hysteresis = cpu_to_be32(5);
+			papm_parm->PlatformPowerLimit = 0xffffffff;
+			papm_parm->NearTDPLimitPAPM = 0xffffffff;
+
+			ret = amdgpu_si_copy_bytes_to_smc(adev, si_pi->papm_cfg_table_start,
+							  (u8 *)papm_parm,
+							  sizeof(PP_SIslands_PAPMParameters),
+							  si_pi->sram_end);
+			if (ret)
+				return ret;
+		}
+	}
+	return 0;
+}
+
+static int si_populate_smc_tdp_limits_2(struct amdgpu_device *adev,
+					struct amdgpu_ps *amdgpu_state)
+{
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+
+	if (ni_pi->enable_power_containment) {
+		SISLANDS_SMC_STATETABLE *smc_table = &si_pi->smc_statetable;
+		u32 scaling_factor = si_get_smc_power_scaling_factor(adev);
+		int ret;
+
+		memset(smc_table, 0, sizeof(SISLANDS_SMC_STATETABLE));
+
+		smc_table->dpm2Params.NearTDPLimit =
+			cpu_to_be32(si_scale_power_for_smc(adev->pm.dpm.near_tdp_limit_adjusted, scaling_factor) * 1000);
+		smc_table->dpm2Params.SafePowerLimit =
+			cpu_to_be32(si_scale_power_for_smc((adev->pm.dpm.near_tdp_limit_adjusted * SISLANDS_DPM2_TDP_SAFE_LIMIT_PERCENT) / 100, scaling_factor) * 1000);
+
+		ret = amdgpu_si_copy_bytes_to_smc(adev,
+						  (si_pi->state_table_start +
+						   offsetof(SISLANDS_SMC_STATETABLE, dpm2Params) +
+						   offsetof(PP_SIslands_DPM2Parameters, NearTDPLimit)),
+						  (u8 *)(&(smc_table->dpm2Params.NearTDPLimit)),
+						  sizeof(u32) * 2,
+						  si_pi->sram_end);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static u16 si_calculate_power_efficiency_ratio(struct amdgpu_device *adev,
+					       const u16 prev_std_vddc,
+					       const u16 curr_std_vddc)
+{
+	u64 margin = (u64)SISLANDS_DPM2_PWREFFICIENCYRATIO_MARGIN;
+	u64 prev_vddc = (u64)prev_std_vddc;
+	u64 curr_vddc = (u64)curr_std_vddc;
+	u64 pwr_efficiency_ratio, n, d;
+
+	if ((prev_vddc == 0) || (curr_vddc == 0))
+		return 0;
+
+	n = div64_u64((u64)1024 * curr_vddc * curr_vddc * ((u64)1000 + margin), (u64)1000);
+	d = prev_vddc * prev_vddc;
+	pwr_efficiency_ratio = div64_u64(n, d);
+
+	if (pwr_efficiency_ratio > (u64)0xFFFF)
+		return 0;
+
+	return (u16)pwr_efficiency_ratio;
+}
+
+static bool si_should_disable_uvd_powertune(struct amdgpu_device *adev,
+					    struct amdgpu_ps *amdgpu_state)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+
+	if (si_pi->dyn_powertune_data.disable_uvd_powertune &&
+	    amdgpu_state->vclk && amdgpu_state->dclk)
+		return true;
+
+	return false;
+}
+
+struct evergreen_power_info *evergreen_get_pi(struct amdgpu_device *adev)
+{
+	struct evergreen_power_info *pi = adev->pm.dpm.priv;
+
+	return pi;
+}
+
+static int si_populate_power_containment_values(struct amdgpu_device *adev,
+						struct amdgpu_ps *amdgpu_state,
+						SISLANDS_SMC_SWSTATE *smc_state)
+{
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+	struct  si_ps *state = si_get_ps(amdgpu_state);
+	SISLANDS_SMC_VOLTAGE_VALUE vddc;
+	u32 prev_sclk;
+	u32 max_sclk;
+	u32 min_sclk;
+	u16 prev_std_vddc;
+	u16 curr_std_vddc;
+	int i;
+	u16 pwr_efficiency_ratio;
+	u8 max_ps_percent;
+	bool disable_uvd_power_tune;
+	int ret;
+
+	if (ni_pi->enable_power_containment == false)
+		return 0;
+
+	if (state->performance_level_count == 0)
+		return -EINVAL;
+
+	if (smc_state->levelCount != state->performance_level_count)
+		return -EINVAL;
+
+	disable_uvd_power_tune = si_should_disable_uvd_powertune(adev, amdgpu_state);
+
+	smc_state->levels[0].dpm2.MaxPS = 0;
+	smc_state->levels[0].dpm2.NearTDPDec = 0;
+	smc_state->levels[0].dpm2.AboveSafeInc = 0;
+	smc_state->levels[0].dpm2.BelowSafeInc = 0;
+	smc_state->levels[0].dpm2.PwrEfficiencyRatio = 0;
+
+	for (i = 1; i < state->performance_level_count; i++) {
+		prev_sclk = state->performance_levels[i-1].sclk;
+		max_sclk  = state->performance_levels[i].sclk;
+		if (i == 1)
+			max_ps_percent = SISLANDS_DPM2_MAXPS_PERCENT_M;
+		else
+			max_ps_percent = SISLANDS_DPM2_MAXPS_PERCENT_H;
+
+		if (prev_sclk > max_sclk)
+			return -EINVAL;
+
+		if ((max_ps_percent == 0) ||
+		    (prev_sclk == max_sclk) ||
+		    disable_uvd_power_tune)
+			min_sclk = max_sclk;
+		else if (i == 1)
+			min_sclk = prev_sclk;
+		else
+			min_sclk = (prev_sclk * (u32)max_ps_percent) / 100;
+
+		if (min_sclk < state->performance_levels[0].sclk)
+			min_sclk = state->performance_levels[0].sclk;
+
+		if (min_sclk == 0)
+			return -EINVAL;
+
+		ret = si_populate_voltage_value(adev, &eg_pi->vddc_voltage_table,
+						state->performance_levels[i-1].vddc, &vddc);
+		if (ret)
+			return ret;
+
+		ret = si_get_std_voltage_value(adev, &vddc, &prev_std_vddc);
+		if (ret)
+			return ret;
+
+		ret = si_populate_voltage_value(adev, &eg_pi->vddc_voltage_table,
+						state->performance_levels[i].vddc, &vddc);
+		if (ret)
+			return ret;
+
+		ret = si_get_std_voltage_value(adev, &vddc, &curr_std_vddc);
+		if (ret)
+			return ret;
+
+		pwr_efficiency_ratio = si_calculate_power_efficiency_ratio(adev,
+									   prev_std_vddc, curr_std_vddc);
+
+		smc_state->levels[i].dpm2.MaxPS = (u8)((SISLANDS_DPM2_MAX_PULSE_SKIP * (max_sclk - min_sclk)) / max_sclk);
+		smc_state->levels[i].dpm2.NearTDPDec = SISLANDS_DPM2_NEAR_TDP_DEC;
+		smc_state->levels[i].dpm2.AboveSafeInc = SISLANDS_DPM2_ABOVE_SAFE_INC;
+		smc_state->levels[i].dpm2.BelowSafeInc = SISLANDS_DPM2_BELOW_SAFE_INC;
+		smc_state->levels[i].dpm2.PwrEfficiencyRatio = cpu_to_be16(pwr_efficiency_ratio);
+	}
+
+	return 0;
+}
+
+static int si_populate_sq_ramping_values(struct amdgpu_device *adev,
+					 struct amdgpu_ps *amdgpu_state,
+					 SISLANDS_SMC_SWSTATE *smc_state)
+{
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+	struct  si_ps *state = si_get_ps(amdgpu_state);
+	u32 sq_power_throttle, sq_power_throttle2;
+	bool enable_sq_ramping = ni_pi->enable_sq_ramping;
+	int i;
+
+	if (state->performance_level_count == 0)
+		return -EINVAL;
+
+	if (smc_state->levelCount != state->performance_level_count)
+		return -EINVAL;
+
+	if (adev->pm.dpm.sq_ramping_threshold == 0)
+		return -EINVAL;
+
+	if (SISLANDS_DPM2_SQ_RAMP_MAX_POWER > (MAX_POWER_MASK >> MAX_POWER_SHIFT))
+		enable_sq_ramping = false;
+
+	if (SISLANDS_DPM2_SQ_RAMP_MIN_POWER > (MIN_POWER_MASK >> MIN_POWER_SHIFT))
+		enable_sq_ramping = false;
+
+	if (SISLANDS_DPM2_SQ_RAMP_MAX_POWER_DELTA > (MAX_POWER_DELTA_MASK >> MAX_POWER_DELTA_SHIFT))
+		enable_sq_ramping = false;
+
+	if (SISLANDS_DPM2_SQ_RAMP_STI_SIZE > (STI_SIZE_MASK >> STI_SIZE_SHIFT))
+		enable_sq_ramping = false;
+
+	if (SISLANDS_DPM2_SQ_RAMP_LTI_RATIO > (LTI_RATIO_MASK >> LTI_RATIO_SHIFT))
+		enable_sq_ramping = false;
+
+	for (i = 0; i < state->performance_level_count; i++) {
+		sq_power_throttle = 0;
+		sq_power_throttle2 = 0;
+
+		if ((state->performance_levels[i].sclk >= adev->pm.dpm.sq_ramping_threshold) &&
+		    enable_sq_ramping) {
+			sq_power_throttle |= MAX_POWER(SISLANDS_DPM2_SQ_RAMP_MAX_POWER);
+			sq_power_throttle |= MIN_POWER(SISLANDS_DPM2_SQ_RAMP_MIN_POWER);
+			sq_power_throttle2 |= MAX_POWER_DELTA(SISLANDS_DPM2_SQ_RAMP_MAX_POWER_DELTA);
+			sq_power_throttle2 |= STI_SIZE(SISLANDS_DPM2_SQ_RAMP_STI_SIZE);
+			sq_power_throttle2 |= LTI_RATIO(SISLANDS_DPM2_SQ_RAMP_LTI_RATIO);
+		} else {
+			sq_power_throttle |= MAX_POWER_MASK | MIN_POWER_MASK;
+			sq_power_throttle2 |= MAX_POWER_DELTA_MASK | STI_SIZE_MASK | LTI_RATIO_MASK;
+		}
+
+		smc_state->levels[i].SQPowerThrottle = cpu_to_be32(sq_power_throttle);
+		smc_state->levels[i].SQPowerThrottle_2 = cpu_to_be32(sq_power_throttle2);
+	}
+
+	return 0;
+}
+
+static int si_enable_power_containment(struct amdgpu_device *adev,
+				       struct amdgpu_ps *amdgpu_new_state,
+				       bool enable)
+{
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+	PPSMC_Result smc_result;
+	int ret = 0;
+
+	if (ni_pi->enable_power_containment) {
+		if (enable) {
+			if (!si_should_disable_uvd_powertune(adev, amdgpu_new_state)) {
+				smc_result = amdgpu_si_send_msg_to_smc(adev, PPSMC_TDPClampingActive);
+				if (smc_result != PPSMC_Result_OK) {
+					ret = -EINVAL;
+					ni_pi->pc_enabled = false;
+				} else {
+					ni_pi->pc_enabled = true;
+				}
+			}
+		} else {
+			smc_result = amdgpu_si_send_msg_to_smc(adev, PPSMC_TDPClampingInactive);
+			if (smc_result != PPSMC_Result_OK)
+				ret = -EINVAL;
+			ni_pi->pc_enabled = false;
+		}
+	}
+
+	return ret;
+}
+
+static int si_initialize_smc_dte_tables(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	int ret = 0;
+	struct si_dte_data *dte_data = &si_pi->dte_data;
+	Smc_SIslands_DTE_Configuration *dte_tables = NULL;
+	u32 table_size;
+	u8 tdep_count;
+	u32 i;
+
+	if (dte_data == NULL)
+		si_pi->enable_dte = false;
+
+	if (si_pi->enable_dte == false)
+		return 0;
+
+	if (dte_data->k <= 0)
+		return -EINVAL;
+
+	dte_tables = kzalloc(sizeof(Smc_SIslands_DTE_Configuration), GFP_KERNEL);
+	if (dte_tables == NULL) {
+		si_pi->enable_dte = false;
+		return -ENOMEM;
+	}
+
+	table_size = dte_data->k;
+
+	if (table_size > SMC_SISLANDS_DTE_MAX_FILTER_STAGES)
+		table_size = SMC_SISLANDS_DTE_MAX_FILTER_STAGES;
+
+	tdep_count = dte_data->tdep_count;
+	if (tdep_count > SMC_SISLANDS_DTE_MAX_TEMPERATURE_DEPENDENT_ARRAY_SIZE)
+		tdep_count = SMC_SISLANDS_DTE_MAX_TEMPERATURE_DEPENDENT_ARRAY_SIZE;
+
+	dte_tables->K = cpu_to_be32(table_size);
+	dte_tables->T0 = cpu_to_be32(dte_data->t0);
+	dte_tables->MaxT = cpu_to_be32(dte_data->max_t);
+	dte_tables->WindowSize = dte_data->window_size;
+	dte_tables->temp_select = dte_data->temp_select;
+	dte_tables->DTE_mode = dte_data->dte_mode;
+	dte_tables->Tthreshold = cpu_to_be32(dte_data->t_threshold);
+
+	if (tdep_count > 0)
+		table_size--;
+
+	for (i = 0; i < table_size; i++) {
+		dte_tables->tau[i] = cpu_to_be32(dte_data->tau[i]);
+		dte_tables->R[i]   = cpu_to_be32(dte_data->r[i]);
+	}
+
+	dte_tables->Tdep_count = tdep_count;
+
+	for (i = 0; i < (u32)tdep_count; i++) {
+		dte_tables->T_limits[i] = dte_data->t_limits[i];
+		dte_tables->Tdep_tau[i] = cpu_to_be32(dte_data->tdep_tau[i]);
+		dte_tables->Tdep_R[i] = cpu_to_be32(dte_data->tdep_r[i]);
+	}
+
+	ret = amdgpu_si_copy_bytes_to_smc(adev, si_pi->dte_table_start,
+					  (u8 *)dte_tables,
+					  sizeof(Smc_SIslands_DTE_Configuration),
+					  si_pi->sram_end);
+	kfree(dte_tables);
+
+	return ret;
+}
+
+static int si_get_cac_std_voltage_max_min(struct amdgpu_device *adev,
+					  u16 *max, u16 *min)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct amdgpu_cac_leakage_table *table =
+		&adev->pm.dpm.dyn_state.cac_leakage_table;
+	u32 i;
+	u32 v0_loadline;
+
+	if (table == NULL)
+		return -EINVAL;
+
+	*max = 0;
+	*min = 0xFFFF;
+
+	for (i = 0; i < table->count; i++) {
+		if (table->entries[i].vddc > *max)
+			*max = table->entries[i].vddc;
+		if (table->entries[i].vddc < *min)
+			*min = table->entries[i].vddc;
+	}
+
+	if (si_pi->powertune_data->lkge_lut_v0_percent > 100)
+		return -EINVAL;
+
+	v0_loadline = (*min) * (100 - si_pi->powertune_data->lkge_lut_v0_percent) / 100;
+
+	if (v0_loadline > 0xFFFFUL)
+		return -EINVAL;
+
+	*min = (u16)v0_loadline;
+
+	if ((*min > *max) || (*max == 0) || (*min == 0))
+		return -EINVAL;
+
+	return 0;
+}
+
+static u16 si_get_cac_std_voltage_step(u16 max, u16 min)
+{
+	return ((max - min) + (SMC_SISLANDS_LKGE_LUT_NUM_OF_VOLT_ENTRIES - 1)) /
+		SMC_SISLANDS_LKGE_LUT_NUM_OF_VOLT_ENTRIES;
+}
+
+static int si_init_dte_leakage_table(struct amdgpu_device *adev,
+				     PP_SIslands_CacConfig *cac_tables,
+				     u16 vddc_max, u16 vddc_min, u16 vddc_step,
+				     u16 t0, u16 t_step)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 leakage;
+	unsigned int i, j;
+	s32 t;
+	u32 smc_leakage;
+	u32 scaling_factor;
+	u16 voltage;
+
+	scaling_factor = si_get_smc_power_scaling_factor(adev);
+
+	for (i = 0; i < SMC_SISLANDS_LKGE_LUT_NUM_OF_TEMP_ENTRIES ; i++) {
+		t = (1000 * (i * t_step + t0));
+
+		for (j = 0; j < SMC_SISLANDS_LKGE_LUT_NUM_OF_VOLT_ENTRIES; j++) {
+			voltage = vddc_max - (vddc_step * j);
+
+			si_calculate_leakage_for_v_and_t(adev,
+							 &si_pi->powertune_data->leakage_coefficients,
+							 voltage,
+							 t,
+							 si_pi->dyn_powertune_data.cac_leakage,
+							 &leakage);
+
+			smc_leakage = si_scale_power_for_smc(leakage, scaling_factor) / 4;
+
+			if (smc_leakage > 0xFFFF)
+				smc_leakage = 0xFFFF;
+
+			cac_tables->cac_lkge_lut[i][SMC_SISLANDS_LKGE_LUT_NUM_OF_VOLT_ENTRIES-1-j] =
+				cpu_to_be16((u16)smc_leakage);
+		}
+	}
+	return 0;
+}
+
+static int si_init_simplified_leakage_table(struct amdgpu_device *adev,
+					    PP_SIslands_CacConfig *cac_tables,
+					    u16 vddc_max, u16 vddc_min, u16 vddc_step)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 leakage;
+	unsigned int i, j;
+	u32 smc_leakage;
+	u32 scaling_factor;
+	u16 voltage;
+
+	scaling_factor = si_get_smc_power_scaling_factor(adev);
+
+	for (j = 0; j < SMC_SISLANDS_LKGE_LUT_NUM_OF_VOLT_ENTRIES; j++) {
+		voltage = vddc_max - (vddc_step * j);
+
+		si_calculate_leakage_for_v(adev,
+					   &si_pi->powertune_data->leakage_coefficients,
+					   si_pi->powertune_data->fixed_kt,
+					   voltage,
+					   si_pi->dyn_powertune_data.cac_leakage,
+					   &leakage);
+
+		smc_leakage = si_scale_power_for_smc(leakage, scaling_factor) / 4;
+
+		if (smc_leakage > 0xFFFF)
+			smc_leakage = 0xFFFF;
+
+		for (i = 0; i < SMC_SISLANDS_LKGE_LUT_NUM_OF_TEMP_ENTRIES ; i++)
+			cac_tables->cac_lkge_lut[i][SMC_SISLANDS_LKGE_LUT_NUM_OF_VOLT_ENTRIES-1-j] =
+				cpu_to_be16((u16)smc_leakage);
+	}
+	return 0;
+}
+
+static int si_initialize_smc_cac_tables(struct amdgpu_device *adev)
+{
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	PP_SIslands_CacConfig *cac_tables = NULL;
+	u16 vddc_max, vddc_min, vddc_step;
+	u16 t0, t_step;
+	u32 load_line_slope, reg;
+	int ret = 0;
+	u32 ticks_per_us = amdgpu_asic_get_xclk(adev) / 100;
+
+	if (ni_pi->enable_cac == false)
+		return 0;
+
+	cac_tables = kzalloc(sizeof(PP_SIslands_CacConfig), GFP_KERNEL);
+	if (!cac_tables)
+		return -ENOMEM;
+
+	reg = RREG32(CG_CAC_CTRL) & ~CAC_WINDOW_MASK;
+	reg |= CAC_WINDOW(si_pi->powertune_data->cac_window);
+	WREG32(CG_CAC_CTRL, reg);
+
+	si_pi->dyn_powertune_data.cac_leakage = adev->pm.dpm.cac_leakage;
+	si_pi->dyn_powertune_data.dc_pwr_value =
+		si_pi->powertune_data->dc_cac[NISLANDS_DCCAC_LEVEL_0];
+	si_pi->dyn_powertune_data.wintime = si_calculate_cac_wintime(adev);
+	si_pi->dyn_powertune_data.shift_n = si_pi->powertune_data->shift_n_default;
+
+	si_pi->dyn_powertune_data.leakage_minimum_temperature = 80 * 1000;
+
+	ret = si_get_cac_std_voltage_max_min(adev, &vddc_max, &vddc_min);
+	if (ret)
+		goto done_free;
+
+	vddc_step = si_get_cac_std_voltage_step(vddc_max, vddc_min);
+	vddc_min = vddc_max - (vddc_step * (SMC_SISLANDS_LKGE_LUT_NUM_OF_VOLT_ENTRIES - 1));
+	t_step = 4;
+	t0 = 60;
+
+	if (si_pi->enable_dte || ni_pi->driver_calculate_cac_leakage)
+		ret = si_init_dte_leakage_table(adev, cac_tables,
+						vddc_max, vddc_min, vddc_step,
+						t0, t_step);
+	else
+		ret = si_init_simplified_leakage_table(adev, cac_tables,
+						       vddc_max, vddc_min, vddc_step);
+	if (ret)
+		goto done_free;
+
+	load_line_slope = ((u32)adev->pm.dpm.load_line_slope << SMC_SISLANDS_SCALE_R) / 100;
+
+	cac_tables->l2numWin_TDP = cpu_to_be32(si_pi->dyn_powertune_data.l2_lta_window_size);
+	cac_tables->lts_truncate_n = si_pi->dyn_powertune_data.lts_truncate;
+	cac_tables->SHIFT_N = si_pi->dyn_powertune_data.shift_n;
+	cac_tables->lkge_lut_V0 = cpu_to_be32((u32)vddc_min);
+	cac_tables->lkge_lut_Vstep = cpu_to_be32((u32)vddc_step);
+	cac_tables->R_LL = cpu_to_be32(load_line_slope);
+	cac_tables->WinTime = cpu_to_be32(si_pi->dyn_powertune_data.wintime);
+	cac_tables->calculation_repeats = cpu_to_be32(2);
+	cac_tables->dc_cac = cpu_to_be32(0);
+	cac_tables->log2_PG_LKG_SCALE = 12;
+	cac_tables->cac_temp = si_pi->powertune_data->operating_temp;
+	cac_tables->lkge_lut_T0 = cpu_to_be32((u32)t0);
+	cac_tables->lkge_lut_Tstep = cpu_to_be32((u32)t_step);
+
+	ret = amdgpu_si_copy_bytes_to_smc(adev, si_pi->cac_table_start,
+					  (u8 *)cac_tables,
+					  sizeof(PP_SIslands_CacConfig),
+					  si_pi->sram_end);
+
+	if (ret)
+		goto done_free;
+
+	ret = si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_ticks_per_us, ticks_per_us);
+
+done_free:
+	if (ret) {
+		ni_pi->enable_cac = false;
+		ni_pi->enable_power_containment = false;
+	}
+
+	kfree(cac_tables);
+
+	return ret;
+}
+
+static int si_program_cac_config_registers(struct amdgpu_device *adev,
+					   const struct si_cac_config_reg *cac_config_regs)
+{
+	const struct si_cac_config_reg *config_regs = cac_config_regs;
+	u32 data = 0, offset;
+
+	if (!config_regs)
+		return -EINVAL;
+
+	while (config_regs->offset != 0xFFFFFFFF) {
+		switch (config_regs->type) {
+		case SISLANDS_CACCONFIG_CGIND:
+			offset = SMC_CG_IND_START + config_regs->offset;
+			if (offset < SMC_CG_IND_END)
+				data = RREG32_SMC(offset);
+			break;
+		default:
+			data = RREG32(config_regs->offset);
+			break;
+		}
+
+		data &= ~config_regs->mask;
+		data |= ((config_regs->value << config_regs->shift) & config_regs->mask);
+
+		switch (config_regs->type) {
+		case SISLANDS_CACCONFIG_CGIND:
+			offset = SMC_CG_IND_START + config_regs->offset;
+			if (offset < SMC_CG_IND_END)
+				WREG32_SMC(offset, data);
+			break;
+		default:
+			WREG32(config_regs->offset, data);
+			break;
+		}
+		config_regs++;
+	}
+	return 0;
+}
+
+static int si_initialize_hardware_cac_manager(struct amdgpu_device *adev)
+{
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	int ret;
+
+	if ((ni_pi->enable_cac == false) ||
+	    (ni_pi->cac_configuration_required == false))
+		return 0;
+
+	ret = si_program_cac_config_registers(adev, si_pi->lcac_config);
+	if (ret)
+		return ret;
+	ret = si_program_cac_config_registers(adev, si_pi->cac_override);
+	if (ret)
+		return ret;
+	ret = si_program_cac_config_registers(adev, si_pi->cac_weights);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int si_enable_smc_cac(struct amdgpu_device *adev,
+			     struct amdgpu_ps *amdgpu_new_state,
+			     bool enable)
+{
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	PPSMC_Result smc_result;
+	int ret = 0;
+
+	if (ni_pi->enable_cac) {
+		if (enable) {
+			if (!si_should_disable_uvd_powertune(adev, amdgpu_new_state)) {
+				if (ni_pi->support_cac_long_term_average) {
+					smc_result = amdgpu_si_send_msg_to_smc(adev, PPSMC_CACLongTermAvgEnable);
+					if (smc_result != PPSMC_Result_OK)
+						ni_pi->support_cac_long_term_average = false;
+				}
+
+				smc_result = amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_EnableCac);
+				if (smc_result != PPSMC_Result_OK) {
+					ret = -EINVAL;
+					ni_pi->cac_enabled = false;
+				} else {
+					ni_pi->cac_enabled = true;
+				}
+
+				if (si_pi->enable_dte) {
+					smc_result = amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_EnableDTE);
+					if (smc_result != PPSMC_Result_OK)
+						ret = -EINVAL;
+				}
+			}
+		} else if (ni_pi->cac_enabled) {
+			if (si_pi->enable_dte)
+				smc_result = amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_DisableDTE);
+
+			smc_result = amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_DisableCac);
+
+			ni_pi->cac_enabled = false;
+
+			if (ni_pi->support_cac_long_term_average)
+				smc_result = amdgpu_si_send_msg_to_smc(adev, PPSMC_CACLongTermAvgDisable);
+		}
+	}
+	return ret;
+}
+
+static int si_init_smc_spll_table(struct amdgpu_device *adev)
+{
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	SMC_SISLANDS_SPLL_DIV_TABLE *spll_table;
+	SISLANDS_SMC_SCLK_VALUE sclk_params;
+	u32 fb_div, p_div;
+	u32 clk_s, clk_v;
+	u32 sclk = 0;
+	int ret = 0;
+	u32 tmp;
+	int i;
+
+	if (si_pi->spll_table_start == 0)
+		return -EINVAL;
+
+	spll_table = kzalloc(sizeof(SMC_SISLANDS_SPLL_DIV_TABLE), GFP_KERNEL);
+	if (spll_table == NULL)
+		return -ENOMEM;
+
+	for (i = 0; i < 256; i++) {
+		ret = si_calculate_sclk_params(adev, sclk, &sclk_params);
+		if (ret)
+			break;
+		p_div = (sclk_params.vCG_SPLL_FUNC_CNTL & SPLL_PDIV_A_MASK) >> SPLL_PDIV_A_SHIFT;
+		fb_div = (sclk_params.vCG_SPLL_FUNC_CNTL_3 & SPLL_FB_DIV_MASK) >> SPLL_FB_DIV_SHIFT;
+		clk_s = (sclk_params.vCG_SPLL_SPREAD_SPECTRUM & CLK_S_MASK) >> CLK_S_SHIFT;
+		clk_v = (sclk_params.vCG_SPLL_SPREAD_SPECTRUM_2 & CLK_V_MASK) >> CLK_V_SHIFT;
+
+		fb_div &= ~0x00001FFF;
+		fb_div >>= 1;
+		clk_v >>= 6;
+
+		if (p_div & ~(SMC_SISLANDS_SPLL_DIV_TABLE_PDIV_MASK >> SMC_SISLANDS_SPLL_DIV_TABLE_PDIV_SHIFT))
+			ret = -EINVAL;
+		if (fb_div & ~(SMC_SISLANDS_SPLL_DIV_TABLE_FBDIV_MASK >> SMC_SISLANDS_SPLL_DIV_TABLE_FBDIV_SHIFT))
+			ret = -EINVAL;
+		if (clk_s & ~(SMC_SISLANDS_SPLL_DIV_TABLE_CLKS_MASK >> SMC_SISLANDS_SPLL_DIV_TABLE_CLKS_SHIFT))
+			ret = -EINVAL;
+		if (clk_v & ~(SMC_SISLANDS_SPLL_DIV_TABLE_CLKV_MASK >> SMC_SISLANDS_SPLL_DIV_TABLE_CLKV_SHIFT))
+			ret = -EINVAL;
+
+		if (ret)
+			break;
+
+		tmp = ((fb_div << SMC_SISLANDS_SPLL_DIV_TABLE_FBDIV_SHIFT) & SMC_SISLANDS_SPLL_DIV_TABLE_FBDIV_MASK) |
+			((p_div << SMC_SISLANDS_SPLL_DIV_TABLE_PDIV_SHIFT) & SMC_SISLANDS_SPLL_DIV_TABLE_PDIV_MASK);
+		spll_table->freq[i] = cpu_to_be32(tmp);
+
+		tmp = ((clk_v << SMC_SISLANDS_SPLL_DIV_TABLE_CLKV_SHIFT) & SMC_SISLANDS_SPLL_DIV_TABLE_CLKV_MASK) |
+			((clk_s << SMC_SISLANDS_SPLL_DIV_TABLE_CLKS_SHIFT) & SMC_SISLANDS_SPLL_DIV_TABLE_CLKS_MASK);
+		spll_table->ss[i] = cpu_to_be32(tmp);
+
+		sclk += 512;
+	}
+
+
+	if (!ret)
+		ret = amdgpu_si_copy_bytes_to_smc(adev, si_pi->spll_table_start,
+						  (u8 *)spll_table,
+						  sizeof(SMC_SISLANDS_SPLL_DIV_TABLE),
+						  si_pi->sram_end);
+
+	if (ret)
+		ni_pi->enable_power_containment = false;
+
+	kfree(spll_table);
+
+	return ret;
+}
+
+static u16 si_get_lower_of_leakage_and_vce_voltage(struct amdgpu_device *adev,
+						   u16 vce_voltage)
+{
+	u16 highest_leakage = 0;
+	struct si_power_info *si_pi = si_get_pi(adev);
+	int i;
+
+	for (i = 0; i < si_pi->leakage_voltage.count; i++){
+		if (highest_leakage < si_pi->leakage_voltage.entries[i].voltage)
+			highest_leakage = si_pi->leakage_voltage.entries[i].voltage;
+	}
+
+	if (si_pi->leakage_voltage.count && (highest_leakage < vce_voltage))
+		return highest_leakage;
+
+	return vce_voltage;
+}
+
+static int si_get_vce_clock_voltage(struct amdgpu_device *adev,
+				    u32 evclk, u32 ecclk, u16 *voltage)
+{
+	u32 i;
+	int ret = -EINVAL;
+	struct amdgpu_vce_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+
+	if (((evclk == 0) && (ecclk == 0)) ||
+	    (table && (table->count == 0))) {
+		*voltage = 0;
+		return 0;
+	}
+
+	for (i = 0; i < table->count; i++) {
+		if ((evclk <= table->entries[i].evclk) &&
+		    (ecclk <= table->entries[i].ecclk)) {
+			*voltage = table->entries[i].v;
+			ret = 0;
+			break;
+		}
+	}
+
+	/* if no match return the highest voltage */
+	if (ret)
+		*voltage = table->entries[table->count - 1].v;
+
+	*voltage = si_get_lower_of_leakage_and_vce_voltage(adev, *voltage);
+
+	return ret;
+}
+
+static bool si_dpm_vblank_too_short(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 vblank_time = amdgpu_dpm_get_vblank_time(adev);
+	/* we never hit the non-gddr5 limit so disable it */
+	u32 switch_limit = adev->gmc.vram_type == AMDGPU_VRAM_TYPE_GDDR5 ? 450 : 0;
+
+	if (vblank_time < switch_limit)
+		return true;
+	else
+		return false;
+
+}
+
+static int ni_copy_and_switch_arb_sets(struct amdgpu_device *adev,
+				u32 arb_freq_src, u32 arb_freq_dest)
+{
+	u32 mc_arb_dram_timing;
+	u32 mc_arb_dram_timing2;
+	u32 burst_time;
+	u32 mc_cg_config;
+
+	switch (arb_freq_src) {
+	case MC_CG_ARB_FREQ_F0:
+		mc_arb_dram_timing  = RREG32(MC_ARB_DRAM_TIMING);
+		mc_arb_dram_timing2 = RREG32(MC_ARB_DRAM_TIMING2);
+		burst_time = (RREG32(MC_ARB_BURST_TIME) & STATE0_MASK) >> STATE0_SHIFT;
+		break;
+	case MC_CG_ARB_FREQ_F1:
+		mc_arb_dram_timing  = RREG32(MC_ARB_DRAM_TIMING_1);
+		mc_arb_dram_timing2 = RREG32(MC_ARB_DRAM_TIMING2_1);
+		burst_time = (RREG32(MC_ARB_BURST_TIME) & STATE1_MASK) >> STATE1_SHIFT;
+		break;
+	case MC_CG_ARB_FREQ_F2:
+		mc_arb_dram_timing  = RREG32(MC_ARB_DRAM_TIMING_2);
+		mc_arb_dram_timing2 = RREG32(MC_ARB_DRAM_TIMING2_2);
+		burst_time = (RREG32(MC_ARB_BURST_TIME) & STATE2_MASK) >> STATE2_SHIFT;
+		break;
+	case MC_CG_ARB_FREQ_F3:
+		mc_arb_dram_timing  = RREG32(MC_ARB_DRAM_TIMING_3);
+		mc_arb_dram_timing2 = RREG32(MC_ARB_DRAM_TIMING2_3);
+		burst_time = (RREG32(MC_ARB_BURST_TIME) & STATE3_MASK) >> STATE3_SHIFT;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	switch (arb_freq_dest) {
+	case MC_CG_ARB_FREQ_F0:
+		WREG32(MC_ARB_DRAM_TIMING, mc_arb_dram_timing);
+		WREG32(MC_ARB_DRAM_TIMING2, mc_arb_dram_timing2);
+		WREG32_P(MC_ARB_BURST_TIME, STATE0(burst_time), ~STATE0_MASK);
+		break;
+	case MC_CG_ARB_FREQ_F1:
+		WREG32(MC_ARB_DRAM_TIMING_1, mc_arb_dram_timing);
+		WREG32(MC_ARB_DRAM_TIMING2_1, mc_arb_dram_timing2);
+		WREG32_P(MC_ARB_BURST_TIME, STATE1(burst_time), ~STATE1_MASK);
+		break;
+	case MC_CG_ARB_FREQ_F2:
+		WREG32(MC_ARB_DRAM_TIMING_2, mc_arb_dram_timing);
+		WREG32(MC_ARB_DRAM_TIMING2_2, mc_arb_dram_timing2);
+		WREG32_P(MC_ARB_BURST_TIME, STATE2(burst_time), ~STATE2_MASK);
+		break;
+	case MC_CG_ARB_FREQ_F3:
+		WREG32(MC_ARB_DRAM_TIMING_3, mc_arb_dram_timing);
+		WREG32(MC_ARB_DRAM_TIMING2_3, mc_arb_dram_timing2);
+		WREG32_P(MC_ARB_BURST_TIME, STATE3(burst_time), ~STATE3_MASK);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	mc_cg_config = RREG32(MC_CG_CONFIG) | 0x0000000F;
+	WREG32(MC_CG_CONFIG, mc_cg_config);
+	WREG32_P(MC_ARB_CG, CG_ARB_REQ(arb_freq_dest), ~CG_ARB_REQ_MASK);
+
+	return 0;
+}
+
+static void ni_update_current_ps(struct amdgpu_device *adev,
+			  struct amdgpu_ps *rps)
+{
+	struct si_ps *new_ps = si_get_ps(rps);
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+
+	eg_pi->current_rps = *rps;
+	ni_pi->current_ps = *new_ps;
+	eg_pi->current_rps.ps_priv = &ni_pi->current_ps;
+	adev->pm.dpm.current_ps = &eg_pi->current_rps;
+}
+
+static void ni_update_requested_ps(struct amdgpu_device *adev,
+			    struct amdgpu_ps *rps)
+{
+	struct si_ps *new_ps = si_get_ps(rps);
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+
+	eg_pi->requested_rps = *rps;
+	ni_pi->requested_ps = *new_ps;
+	eg_pi->requested_rps.ps_priv = &ni_pi->requested_ps;
+	adev->pm.dpm.requested_ps = &eg_pi->requested_rps;
+}
+
+static void ni_set_uvd_clock_before_set_eng_clock(struct amdgpu_device *adev,
+					   struct amdgpu_ps *new_ps,
+					   struct amdgpu_ps *old_ps)
+{
+	struct si_ps *new_state = si_get_ps(new_ps);
+	struct si_ps *current_state = si_get_ps(old_ps);
+
+	if ((new_ps->vclk == old_ps->vclk) &&
+	    (new_ps->dclk == old_ps->dclk))
+		return;
+
+	if (new_state->performance_levels[new_state->performance_level_count - 1].sclk >=
+	    current_state->performance_levels[current_state->performance_level_count - 1].sclk)
+		return;
+
+	amdgpu_asic_set_uvd_clocks(adev, new_ps->vclk, new_ps->dclk);
+}
+
+static void ni_set_uvd_clock_after_set_eng_clock(struct amdgpu_device *adev,
+					  struct amdgpu_ps *new_ps,
+					  struct amdgpu_ps *old_ps)
+{
+	struct si_ps *new_state = si_get_ps(new_ps);
+	struct si_ps *current_state = si_get_ps(old_ps);
+
+	if ((new_ps->vclk == old_ps->vclk) &&
+	    (new_ps->dclk == old_ps->dclk))
+		return;
+
+	if (new_state->performance_levels[new_state->performance_level_count - 1].sclk <
+	    current_state->performance_levels[current_state->performance_level_count - 1].sclk)
+		return;
+
+	amdgpu_asic_set_uvd_clocks(adev, new_ps->vclk, new_ps->dclk);
+}
+
+static u16 btc_find_voltage(struct atom_voltage_table *table, u16 voltage)
+{
+	unsigned int i;
+
+	for (i = 0; i < table->count; i++)
+		if (voltage <= table->entries[i].value)
+			return table->entries[i].value;
+
+	return table->entries[table->count - 1].value;
+}
+
+static u32 btc_find_valid_clock(struct amdgpu_clock_array *clocks,
+		                u32 max_clock, u32 requested_clock)
+{
+	unsigned int i;
+
+	if ((clocks == NULL) || (clocks->count == 0))
+		return (requested_clock < max_clock) ? requested_clock : max_clock;
+
+	for (i = 0; i < clocks->count; i++) {
+		if (clocks->values[i] >= requested_clock)
+			return (clocks->values[i] < max_clock) ? clocks->values[i] : max_clock;
+	}
+
+	return (clocks->values[clocks->count - 1] < max_clock) ?
+		clocks->values[clocks->count - 1] : max_clock;
+}
+
+static u32 btc_get_valid_mclk(struct amdgpu_device *adev,
+			      u32 max_mclk, u32 requested_mclk)
+{
+	return btc_find_valid_clock(&adev->pm.dpm.dyn_state.valid_mclk_values,
+				    max_mclk, requested_mclk);
+}
+
+static u32 btc_get_valid_sclk(struct amdgpu_device *adev,
+		              u32 max_sclk, u32 requested_sclk)
+{
+	return btc_find_valid_clock(&adev->pm.dpm.dyn_state.valid_sclk_values,
+				    max_sclk, requested_sclk);
+}
+
+static void btc_get_max_clock_from_voltage_dependency_table(struct amdgpu_clock_voltage_dependency_table *table,
+							    u32 *max_clock)
+{
+	u32 i, clock = 0;
+
+	if ((table == NULL) || (table->count == 0)) {
+		*max_clock = clock;
+		return;
+	}
+
+	for (i = 0; i < table->count; i++) {
+		if (clock < table->entries[i].clk)
+			clock = table->entries[i].clk;
+	}
+	*max_clock = clock;
+}
+
+static void btc_apply_voltage_dependency_rules(struct amdgpu_clock_voltage_dependency_table *table,
+					       u32 clock, u16 max_voltage, u16 *voltage)
+{
+	u32 i;
+
+	if ((table == NULL) || (table->count == 0))
+		return;
+
+	for (i= 0; i < table->count; i++) {
+		if (clock <= table->entries[i].clk) {
+			if (*voltage < table->entries[i].v)
+				*voltage = (u16)((table->entries[i].v < max_voltage) ?
+					   table->entries[i].v : max_voltage);
+			return;
+		}
+	}
+
+	*voltage = (*voltage > max_voltage) ? *voltage : max_voltage;
+}
+
+static void btc_adjust_clock_combinations(struct amdgpu_device *adev,
+					  const struct amdgpu_clock_and_voltage_limits *max_limits,
+					  struct rv7xx_pl *pl)
+{
+
+	if ((pl->mclk == 0) || (pl->sclk == 0))
+		return;
+
+	if (pl->mclk == pl->sclk)
+		return;
+
+	if (pl->mclk > pl->sclk) {
+		if (((pl->mclk + (pl->sclk - 1)) / pl->sclk) > adev->pm.dpm.dyn_state.mclk_sclk_ratio)
+			pl->sclk = btc_get_valid_sclk(adev,
+						      max_limits->sclk,
+						      (pl->mclk +
+						      (adev->pm.dpm.dyn_state.mclk_sclk_ratio - 1)) /
+						      adev->pm.dpm.dyn_state.mclk_sclk_ratio);
+	} else {
+		if ((pl->sclk - pl->mclk) > adev->pm.dpm.dyn_state.sclk_mclk_delta)
+			pl->mclk = btc_get_valid_mclk(adev,
+						      max_limits->mclk,
+						      pl->sclk -
+						      adev->pm.dpm.dyn_state.sclk_mclk_delta);
+	}
+}
+
+static void btc_apply_voltage_delta_rules(struct amdgpu_device *adev,
+					  u16 max_vddc, u16 max_vddci,
+					  u16 *vddc, u16 *vddci)
+{
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	u16 new_voltage;
+
+	if ((0 == *vddc) || (0 == *vddci))
+		return;
+
+	if (*vddc > *vddci) {
+		if ((*vddc - *vddci) > adev->pm.dpm.dyn_state.vddc_vddci_delta) {
+			new_voltage = btc_find_voltage(&eg_pi->vddci_voltage_table,
+						       (*vddc - adev->pm.dpm.dyn_state.vddc_vddci_delta));
+			*vddci = (new_voltage < max_vddci) ? new_voltage : max_vddci;
+		}
+	} else {
+		if ((*vddci - *vddc) > adev->pm.dpm.dyn_state.vddc_vddci_delta) {
+			new_voltage = btc_find_voltage(&eg_pi->vddc_voltage_table,
+						       (*vddci - adev->pm.dpm.dyn_state.vddc_vddci_delta));
+			*vddc = (new_voltage < max_vddc) ? new_voltage : max_vddc;
+		}
+	}
+}
+
+static void r600_calculate_u_and_p(u32 i, u32 r_c, u32 p_b,
+			    u32 *p, u32 *u)
+{
+	u32 b_c = 0;
+	u32 i_c;
+	u32 tmp;
+
+	i_c = (i * r_c) / 100;
+	tmp = i_c >> p_b;
+
+	while (tmp) {
+		b_c++;
+		tmp >>= 1;
+	}
+
+	*u = (b_c + 1) / 2;
+	*p = i_c / (1 << (2 * (*u)));
+}
+
+static int r600_calculate_at(u32 t, u32 h, u32 fh, u32 fl, u32 *tl, u32 *th)
+{
+	u32 k, a, ah, al;
+	u32 t1;
+
+	if ((fl == 0) || (fh == 0) || (fl > fh))
+		return -EINVAL;
+
+	k = (100 * fh) / fl;
+	t1 = (t * (k - 100));
+	a = (1000 * (100 * h + t1)) / (10000 + (t1 / 100));
+	a = (a + 5) / 10;
+	ah = ((a * t) + 5000) / 10000;
+	al = a - ah;
+
+	*th = t - ah;
+	*tl = t + al;
+
+	return 0;
+}
+
+static bool r600_is_uvd_state(u32 class, u32 class2)
+{
+	if (class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
+		return true;
+	if (class & ATOM_PPLIB_CLASSIFICATION_HD2STATE)
+		return true;
+	if (class & ATOM_PPLIB_CLASSIFICATION_HDSTATE)
+		return true;
+	if (class & ATOM_PPLIB_CLASSIFICATION_SDSTATE)
+		return true;
+	if (class2 & ATOM_PPLIB_CLASSIFICATION2_MVC)
+		return true;
+	return false;
+}
+
+static u8 rv770_get_memory_module_index(struct amdgpu_device *adev)
+{
+	return (u8) ((RREG32(BIOS_SCRATCH_4) >> 16) & 0xff);
+}
+
+static void rv770_get_max_vddc(struct amdgpu_device *adev)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	u16 vddc;
+
+	if (amdgpu_atombios_get_max_vddc(adev, 0, 0, &vddc))
+		pi->max_vddc = 0;
+	else
+		pi->max_vddc = vddc;
+}
+
+static void rv770_get_engine_memory_ss(struct amdgpu_device *adev)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct amdgpu_atom_ss ss;
+
+	pi->sclk_ss = amdgpu_atombios_get_asic_ss_info(adev, &ss,
+						       ASIC_INTERNAL_ENGINE_SS, 0);
+	pi->mclk_ss = amdgpu_atombios_get_asic_ss_info(adev, &ss,
+						       ASIC_INTERNAL_MEMORY_SS, 0);
+
+	if (pi->sclk_ss || pi->mclk_ss)
+		pi->dynamic_ss = true;
+	else
+		pi->dynamic_ss = false;
+}
+
+
+static void si_apply_state_adjust_rules(struct amdgpu_device *adev,
+					struct amdgpu_ps *rps)
+{
+	struct  si_ps *ps = si_get_ps(rps);
+	struct amdgpu_clock_and_voltage_limits *max_limits;
+	bool disable_mclk_switching = false;
+	bool disable_sclk_switching = false;
+	u32 mclk, sclk;
+	u16 vddc, vddci, min_vce_voltage = 0;
+	u32 max_sclk_vddc, max_mclk_vddci, max_mclk_vddc;
+	u32 max_sclk = 0, max_mclk = 0;
+	int i;
+
+	if (adev->asic_type == CHIP_HAINAN) {
+		if ((adev->pdev->revision == 0x81) ||
+		    (adev->pdev->revision == 0x83) ||
+		    (adev->pdev->revision == 0xC3) ||
+		    (adev->pdev->device == 0x6664) ||
+		    (adev->pdev->device == 0x6665) ||
+		    (adev->pdev->device == 0x6667)) {
+			max_sclk = 75000;
+		}
+		if ((adev->pdev->revision == 0xC3) ||
+		    (adev->pdev->device == 0x6665)) {
+			max_sclk = 60000;
+			max_mclk = 80000;
+		}
+	} else if (adev->asic_type == CHIP_OLAND) {
+		if ((adev->pdev->revision == 0xC7) ||
+		    (adev->pdev->revision == 0x80) ||
+		    (adev->pdev->revision == 0x81) ||
+		    (adev->pdev->revision == 0x83) ||
+		    (adev->pdev->revision == 0x87) ||
+		    (adev->pdev->device == 0x6604) ||
+		    (adev->pdev->device == 0x6605)) {
+			max_sclk = 75000;
+		}
+	}
+
+	if (rps->vce_active) {
+		rps->evclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].evclk;
+		rps->ecclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].ecclk;
+		si_get_vce_clock_voltage(adev, rps->evclk, rps->ecclk,
+					 &min_vce_voltage);
+	} else {
+		rps->evclk = 0;
+		rps->ecclk = 0;
+	}
+
+	if ((adev->pm.dpm.new_active_crtc_count > 1) ||
+	    si_dpm_vblank_too_short(adev))
+		disable_mclk_switching = true;
+
+	if (rps->vclk || rps->dclk) {
+		disable_mclk_switching = true;
+		disable_sclk_switching = true;
+	}
+
+	if (adev->pm.ac_power)
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	else
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
+
+	for (i = ps->performance_level_count - 2; i >= 0; i--) {
+		if (ps->performance_levels[i].vddc > ps->performance_levels[i+1].vddc)
+			ps->performance_levels[i].vddc = ps->performance_levels[i+1].vddc;
+	}
+	if (adev->pm.ac_power == false) {
+		for (i = 0; i < ps->performance_level_count; i++) {
+			if (ps->performance_levels[i].mclk > max_limits->mclk)
+				ps->performance_levels[i].mclk = max_limits->mclk;
+			if (ps->performance_levels[i].sclk > max_limits->sclk)
+				ps->performance_levels[i].sclk = max_limits->sclk;
+			if (ps->performance_levels[i].vddc > max_limits->vddc)
+				ps->performance_levels[i].vddc = max_limits->vddc;
+			if (ps->performance_levels[i].vddci > max_limits->vddci)
+				ps->performance_levels[i].vddci = max_limits->vddci;
+		}
+	}
+
+	/* limit clocks to max supported clocks based on voltage dependency tables */
+	btc_get_max_clock_from_voltage_dependency_table(&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk,
+							&max_sclk_vddc);
+	btc_get_max_clock_from_voltage_dependency_table(&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk,
+							&max_mclk_vddci);
+	btc_get_max_clock_from_voltage_dependency_table(&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk,
+							&max_mclk_vddc);
+
+	for (i = 0; i < ps->performance_level_count; i++) {
+		if (max_sclk_vddc) {
+			if (ps->performance_levels[i].sclk > max_sclk_vddc)
+				ps->performance_levels[i].sclk = max_sclk_vddc;
+		}
+		if (max_mclk_vddci) {
+			if (ps->performance_levels[i].mclk > max_mclk_vddci)
+				ps->performance_levels[i].mclk = max_mclk_vddci;
+		}
+		if (max_mclk_vddc) {
+			if (ps->performance_levels[i].mclk > max_mclk_vddc)
+				ps->performance_levels[i].mclk = max_mclk_vddc;
+		}
+		if (max_mclk) {
+			if (ps->performance_levels[i].mclk > max_mclk)
+				ps->performance_levels[i].mclk = max_mclk;
+		}
+		if (max_sclk) {
+			if (ps->performance_levels[i].sclk > max_sclk)
+				ps->performance_levels[i].sclk = max_sclk;
+		}
+	}
+
+	/* XXX validate the min clocks required for display */
+
+	if (disable_mclk_switching) {
+		mclk  = ps->performance_levels[ps->performance_level_count - 1].mclk;
+		vddci = ps->performance_levels[ps->performance_level_count - 1].vddci;
+	} else {
+		mclk = ps->performance_levels[0].mclk;
+		vddci = ps->performance_levels[0].vddci;
+	}
+
+	if (disable_sclk_switching) {
+		sclk = ps->performance_levels[ps->performance_level_count - 1].sclk;
+		vddc = ps->performance_levels[ps->performance_level_count - 1].vddc;
+	} else {
+		sclk = ps->performance_levels[0].sclk;
+		vddc = ps->performance_levels[0].vddc;
+	}
+
+	if (rps->vce_active) {
+		if (sclk < adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].sclk)
+			sclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].sclk;
+		if (mclk < adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].mclk)
+			mclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].mclk;
+	}
+
+	/* adjusted low state */
+	ps->performance_levels[0].sclk = sclk;
+	ps->performance_levels[0].mclk = mclk;
+	ps->performance_levels[0].vddc = vddc;
+	ps->performance_levels[0].vddci = vddci;
+
+	if (disable_sclk_switching) {
+		sclk = ps->performance_levels[0].sclk;
+		for (i = 1; i < ps->performance_level_count; i++) {
+			if (sclk < ps->performance_levels[i].sclk)
+				sclk = ps->performance_levels[i].sclk;
+		}
+		for (i = 0; i < ps->performance_level_count; i++) {
+			ps->performance_levels[i].sclk = sclk;
+			ps->performance_levels[i].vddc = vddc;
+		}
+	} else {
+		for (i = 1; i < ps->performance_level_count; i++) {
+			if (ps->performance_levels[i].sclk < ps->performance_levels[i - 1].sclk)
+				ps->performance_levels[i].sclk = ps->performance_levels[i - 1].sclk;
+			if (ps->performance_levels[i].vddc < ps->performance_levels[i - 1].vddc)
+				ps->performance_levels[i].vddc = ps->performance_levels[i - 1].vddc;
+		}
+	}
+
+	if (disable_mclk_switching) {
+		mclk = ps->performance_levels[0].mclk;
+		for (i = 1; i < ps->performance_level_count; i++) {
+			if (mclk < ps->performance_levels[i].mclk)
+				mclk = ps->performance_levels[i].mclk;
+		}
+		for (i = 0; i < ps->performance_level_count; i++) {
+			ps->performance_levels[i].mclk = mclk;
+			ps->performance_levels[i].vddci = vddci;
+		}
+	} else {
+		for (i = 1; i < ps->performance_level_count; i++) {
+			if (ps->performance_levels[i].mclk < ps->performance_levels[i - 1].mclk)
+				ps->performance_levels[i].mclk = ps->performance_levels[i - 1].mclk;
+			if (ps->performance_levels[i].vddci < ps->performance_levels[i - 1].vddci)
+				ps->performance_levels[i].vddci = ps->performance_levels[i - 1].vddci;
+		}
+	}
+
+	for (i = 0; i < ps->performance_level_count; i++)
+		btc_adjust_clock_combinations(adev, max_limits,
+					      &ps->performance_levels[i]);
+
+	for (i = 0; i < ps->performance_level_count; i++) {
+		if (ps->performance_levels[i].vddc < min_vce_voltage)
+			ps->performance_levels[i].vddc = min_vce_voltage;
+		btc_apply_voltage_dependency_rules(&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk,
+						   ps->performance_levels[i].sclk,
+						   max_limits->vddc,  &ps->performance_levels[i].vddc);
+		btc_apply_voltage_dependency_rules(&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk,
+						   ps->performance_levels[i].mclk,
+						   max_limits->vddci, &ps->performance_levels[i].vddci);
+		btc_apply_voltage_dependency_rules(&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk,
+						   ps->performance_levels[i].mclk,
+						   max_limits->vddc,  &ps->performance_levels[i].vddc);
+		btc_apply_voltage_dependency_rules(&adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk,
+						   adev->clock.current_dispclk,
+						   max_limits->vddc,  &ps->performance_levels[i].vddc);
+	}
+
+	for (i = 0; i < ps->performance_level_count; i++) {
+		btc_apply_voltage_delta_rules(adev,
+					      max_limits->vddc, max_limits->vddci,
+					      &ps->performance_levels[i].vddc,
+					      &ps->performance_levels[i].vddci);
+	}
+
+	ps->dc_compatible = true;
+	for (i = 0; i < ps->performance_level_count; i++) {
+		if (ps->performance_levels[i].vddc > adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.vddc)
+			ps->dc_compatible = false;
+	}
+}
+
+#if 0
+static int si_read_smc_soft_register(struct amdgpu_device *adev,
+				     u16 reg_offset, u32 *value)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+
+	return amdgpu_si_read_smc_sram_dword(adev,
+					     si_pi->soft_regs_start + reg_offset, value,
+					     si_pi->sram_end);
+}
+#endif
+
+static int si_write_smc_soft_register(struct amdgpu_device *adev,
+				      u16 reg_offset, u32 value)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+
+	return amdgpu_si_write_smc_sram_dword(adev,
+					      si_pi->soft_regs_start + reg_offset,
+					      value, si_pi->sram_end);
+}
+
+static bool si_is_special_1gb_platform(struct amdgpu_device *adev)
+{
+	bool ret = false;
+	u32 tmp, width, row, column, bank, density;
+	bool is_memory_gddr5, is_special;
+
+	tmp = RREG32(MC_SEQ_MISC0);
+	is_memory_gddr5 = (MC_SEQ_MISC0_GDDR5_VALUE == ((tmp & MC_SEQ_MISC0_GDDR5_MASK) >> MC_SEQ_MISC0_GDDR5_SHIFT));
+	is_special = (MC_SEQ_MISC0_REV_ID_VALUE == ((tmp & MC_SEQ_MISC0_REV_ID_MASK) >> MC_SEQ_MISC0_REV_ID_SHIFT))
+		& (MC_SEQ_MISC0_VEN_ID_VALUE == ((tmp & MC_SEQ_MISC0_VEN_ID_MASK) >> MC_SEQ_MISC0_VEN_ID_SHIFT));
+
+	WREG32(MC_SEQ_IO_DEBUG_INDEX, 0xb);
+	width = ((RREG32(MC_SEQ_IO_DEBUG_DATA) >> 1) & 1) ? 16 : 32;
+
+	tmp = RREG32(MC_ARB_RAMCFG);
+	row = ((tmp & NOOFROWS_MASK) >> NOOFROWS_SHIFT) + 10;
+	column = ((tmp & NOOFCOLS_MASK) >> NOOFCOLS_SHIFT) + 8;
+	bank = ((tmp & NOOFBANK_MASK) >> NOOFBANK_SHIFT) + 2;
+
+	density = (1 << (row + column - 20 + bank)) * width;
+
+	if ((adev->pdev->device == 0x6819) &&
+	    is_memory_gddr5 && is_special && (density == 0x400))
+		ret = true;
+
+	return ret;
+}
+
+static void si_get_leakage_vddc(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u16 vddc, count = 0;
+	int i, ret;
+
+	for (i = 0; i < SISLANDS_MAX_LEAKAGE_COUNT; i++) {
+		ret = amdgpu_atombios_get_leakage_vddc_based_on_leakage_idx(adev, &vddc, SISLANDS_LEAKAGE_INDEX0 + i);
+
+		if (!ret && (vddc > 0) && (vddc != (SISLANDS_LEAKAGE_INDEX0 + i))) {
+			si_pi->leakage_voltage.entries[count].voltage = vddc;
+			si_pi->leakage_voltage.entries[count].leakage_index =
+				SISLANDS_LEAKAGE_INDEX0 + i;
+			count++;
+		}
+	}
+	si_pi->leakage_voltage.count = count;
+}
+
+static int si_get_leakage_voltage_from_leakage_index(struct amdgpu_device *adev,
+						     u32 index, u16 *leakage_voltage)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	int i;
+
+	if (leakage_voltage == NULL)
+		return -EINVAL;
+
+	if ((index & 0xff00) != 0xff00)
+		return -EINVAL;
+
+	if ((index & 0xff) > SISLANDS_MAX_LEAKAGE_COUNT + 1)
+		return -EINVAL;
+
+	if (index < SISLANDS_LEAKAGE_INDEX0)
+		return -EINVAL;
+
+	for (i = 0; i < si_pi->leakage_voltage.count; i++) {
+		if (si_pi->leakage_voltage.entries[i].leakage_index == index) {
+			*leakage_voltage = si_pi->leakage_voltage.entries[i].voltage;
+			return 0;
+		}
+	}
+	return -EAGAIN;
+}
+
+static void si_set_dpm_event_sources(struct amdgpu_device *adev, u32 sources)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	bool want_thermal_protection;
+	enum amdgpu_dpm_event_src dpm_event_src;
+
+	switch (sources) {
+	case 0:
+	default:
+		want_thermal_protection = false;
+		break;
+	case (1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL):
+		want_thermal_protection = true;
+		dpm_event_src = AMDGPU_DPM_EVENT_SRC_DIGITAL;
+		break;
+	case (1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_EXTERNAL):
+		want_thermal_protection = true;
+		dpm_event_src = AMDGPU_DPM_EVENT_SRC_EXTERNAL;
+		break;
+	case ((1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_EXTERNAL) |
+	      (1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL)):
+		want_thermal_protection = true;
+		dpm_event_src = AMDGPU_DPM_EVENT_SRC_DIGIAL_OR_EXTERNAL;
+		break;
+	}
+
+	if (want_thermal_protection) {
+		WREG32_P(CG_THERMAL_CTRL, DPM_EVENT_SRC(dpm_event_src), ~DPM_EVENT_SRC_MASK);
+		if (pi->thermal_protection)
+			WREG32_P(GENERAL_PWRMGT, 0, ~THERMAL_PROTECTION_DIS);
+	} else {
+		WREG32_P(GENERAL_PWRMGT, THERMAL_PROTECTION_DIS, ~THERMAL_PROTECTION_DIS);
+	}
+}
+
+static void si_enable_auto_throttle_source(struct amdgpu_device *adev,
+					   enum amdgpu_dpm_auto_throttle_src source,
+					   bool enable)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+
+	if (enable) {
+		if (!(pi->active_auto_throttle_sources & (1 << source))) {
+			pi->active_auto_throttle_sources |= 1 << source;
+			si_set_dpm_event_sources(adev, pi->active_auto_throttle_sources);
+		}
+	} else {
+		if (pi->active_auto_throttle_sources & (1 << source)) {
+			pi->active_auto_throttle_sources &= ~(1 << source);
+			si_set_dpm_event_sources(adev, pi->active_auto_throttle_sources);
+		}
+	}
+}
+
+static void si_start_dpm(struct amdgpu_device *adev)
+{
+	WREG32_P(GENERAL_PWRMGT, GLOBAL_PWRMGT_EN, ~GLOBAL_PWRMGT_EN);
+}
+
+static void si_stop_dpm(struct amdgpu_device *adev)
+{
+	WREG32_P(GENERAL_PWRMGT, 0, ~GLOBAL_PWRMGT_EN);
+}
+
+static void si_enable_sclk_control(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		WREG32_P(SCLK_PWRMGT_CNTL, 0, ~SCLK_PWRMGT_OFF);
+	else
+		WREG32_P(SCLK_PWRMGT_CNTL, SCLK_PWRMGT_OFF, ~SCLK_PWRMGT_OFF);
+
+}
+
+#if 0
+static int si_notify_hardware_of_thermal_state(struct amdgpu_device *adev,
+					       u32 thermal_level)
+{
+	PPSMC_Result ret;
+
+	if (thermal_level == 0) {
+		ret = amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_EnableThermalInterrupt);
+		if (ret == PPSMC_Result_OK)
+			return 0;
+		else
+			return -EINVAL;
+	}
+	return 0;
+}
+
+static void si_notify_hardware_vpu_recovery_event(struct amdgpu_device *adev)
+{
+	si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_tdr_is_about_to_happen, true);
+}
+#endif
+
+#if 0
+static int si_notify_hw_of_powersource(struct amdgpu_device *adev, bool ac_power)
+{
+	if (ac_power)
+		return (amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_RunningOnAC) == PPSMC_Result_OK) ?
+			0 : -EINVAL;
+
+	return 0;
+}
+#endif
+
+static PPSMC_Result si_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+						      PPSMC_Msg msg, u32 parameter)
+{
+	WREG32(SMC_SCRATCH0, parameter);
+	return amdgpu_si_send_msg_to_smc(adev, msg);
+}
+
+static int si_restrict_performance_levels_before_switch(struct amdgpu_device *adev)
+{
+	if (amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_NoForcedLevel) != PPSMC_Result_OK)
+		return -EINVAL;
+
+	return (si_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SetEnabledLevels, 1) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+static int si_dpm_force_performance_level(void *handle,
+				   enum amd_dpm_forced_level level)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ps *rps = adev->pm.dpm.current_ps;
+	struct  si_ps *ps = si_get_ps(rps);
+	u32 levels = ps->performance_level_count;
+
+	if (level == AMD_DPM_FORCED_LEVEL_HIGH) {
+		if (si_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SetEnabledLevels, levels) != PPSMC_Result_OK)
+			return -EINVAL;
+
+		if (si_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SetForcedLevels, 1) != PPSMC_Result_OK)
+			return -EINVAL;
+	} else if (level == AMD_DPM_FORCED_LEVEL_LOW) {
+		if (si_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SetForcedLevels, 0) != PPSMC_Result_OK)
+			return -EINVAL;
+
+		if (si_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SetEnabledLevels, 1) != PPSMC_Result_OK)
+			return -EINVAL;
+	} else if (level == AMD_DPM_FORCED_LEVEL_AUTO) {
+		if (si_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SetForcedLevels, 0) != PPSMC_Result_OK)
+			return -EINVAL;
+
+		if (si_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SetEnabledLevels, levels) != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	adev->pm.dpm.forced_level = level;
+
+	return 0;
+}
+
+#if 0
+static int si_set_boot_state(struct amdgpu_device *adev)
+{
+	return (amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_SwitchToInitialState) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+#endif
+
+static int si_set_sw_state(struct amdgpu_device *adev)
+{
+	return (amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_SwitchToSwState) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+static int si_halt_smc(struct amdgpu_device *adev)
+{
+	if (amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_Halt) != PPSMC_Result_OK)
+		return -EINVAL;
+
+	return (amdgpu_si_wait_for_smc_inactive(adev) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+static int si_resume_smc(struct amdgpu_device *adev)
+{
+	if (amdgpu_si_send_msg_to_smc(adev, PPSMC_FlushDataCache) != PPSMC_Result_OK)
+		return -EINVAL;
+
+	return (amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_Resume) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+static void si_dpm_start_smc(struct amdgpu_device *adev)
+{
+	amdgpu_si_program_jump_on_start(adev);
+	amdgpu_si_start_smc(adev);
+	amdgpu_si_smc_clock(adev, true);
+}
+
+static void si_dpm_stop_smc(struct amdgpu_device *adev)
+{
+	amdgpu_si_reset_smc(adev);
+	amdgpu_si_smc_clock(adev, false);
+}
+
+static int si_process_firmware_header(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 tmp;
+	int ret;
+
+	ret = amdgpu_si_read_smc_sram_dword(adev,
+					    SISLANDS_SMC_FIRMWARE_HEADER_LOCATION +
+					    SISLANDS_SMC_FIRMWARE_HEADER_stateTable,
+					    &tmp, si_pi->sram_end);
+	if (ret)
+		return ret;
+
+	si_pi->state_table_start = tmp;
+
+	ret = amdgpu_si_read_smc_sram_dword(adev,
+					    SISLANDS_SMC_FIRMWARE_HEADER_LOCATION +
+					    SISLANDS_SMC_FIRMWARE_HEADER_softRegisters,
+					    &tmp, si_pi->sram_end);
+	if (ret)
+		return ret;
+
+	si_pi->soft_regs_start = tmp;
+
+	ret = amdgpu_si_read_smc_sram_dword(adev,
+					    SISLANDS_SMC_FIRMWARE_HEADER_LOCATION +
+					    SISLANDS_SMC_FIRMWARE_HEADER_mcRegisterTable,
+					    &tmp, si_pi->sram_end);
+	if (ret)
+		return ret;
+
+	si_pi->mc_reg_table_start = tmp;
+
+	ret = amdgpu_si_read_smc_sram_dword(adev,
+					    SISLANDS_SMC_FIRMWARE_HEADER_LOCATION +
+					    SISLANDS_SMC_FIRMWARE_HEADER_fanTable,
+					    &tmp, si_pi->sram_end);
+	if (ret)
+		return ret;
+
+	si_pi->fan_table_start = tmp;
+
+	ret = amdgpu_si_read_smc_sram_dword(adev,
+					    SISLANDS_SMC_FIRMWARE_HEADER_LOCATION +
+					    SISLANDS_SMC_FIRMWARE_HEADER_mcArbDramAutoRefreshTable,
+					    &tmp, si_pi->sram_end);
+	if (ret)
+		return ret;
+
+	si_pi->arb_table_start = tmp;
+
+	ret = amdgpu_si_read_smc_sram_dword(adev,
+					    SISLANDS_SMC_FIRMWARE_HEADER_LOCATION +
+					    SISLANDS_SMC_FIRMWARE_HEADER_CacConfigTable,
+					    &tmp, si_pi->sram_end);
+	if (ret)
+		return ret;
+
+	si_pi->cac_table_start = tmp;
+
+	ret = amdgpu_si_read_smc_sram_dword(adev,
+					    SISLANDS_SMC_FIRMWARE_HEADER_LOCATION +
+					    SISLANDS_SMC_FIRMWARE_HEADER_DteConfiguration,
+					    &tmp, si_pi->sram_end);
+	if (ret)
+		return ret;
+
+	si_pi->dte_table_start = tmp;
+
+	ret = amdgpu_si_read_smc_sram_dword(adev,
+					    SISLANDS_SMC_FIRMWARE_HEADER_LOCATION +
+					    SISLANDS_SMC_FIRMWARE_HEADER_spllTable,
+					    &tmp, si_pi->sram_end);
+	if (ret)
+		return ret;
+
+	si_pi->spll_table_start = tmp;
+
+	ret = amdgpu_si_read_smc_sram_dword(adev,
+					    SISLANDS_SMC_FIRMWARE_HEADER_LOCATION +
+					    SISLANDS_SMC_FIRMWARE_HEADER_PAPMParameters,
+					    &tmp, si_pi->sram_end);
+	if (ret)
+		return ret;
+
+	si_pi->papm_cfg_table_start = tmp;
+
+	return ret;
+}
+
+static void si_read_clock_registers(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+
+	si_pi->clock_registers.cg_spll_func_cntl = RREG32(CG_SPLL_FUNC_CNTL);
+	si_pi->clock_registers.cg_spll_func_cntl_2 = RREG32(CG_SPLL_FUNC_CNTL_2);
+	si_pi->clock_registers.cg_spll_func_cntl_3 = RREG32(CG_SPLL_FUNC_CNTL_3);
+	si_pi->clock_registers.cg_spll_func_cntl_4 = RREG32(CG_SPLL_FUNC_CNTL_4);
+	si_pi->clock_registers.cg_spll_spread_spectrum = RREG32(CG_SPLL_SPREAD_SPECTRUM);
+	si_pi->clock_registers.cg_spll_spread_spectrum_2 = RREG32(CG_SPLL_SPREAD_SPECTRUM_2);
+	si_pi->clock_registers.dll_cntl = RREG32(DLL_CNTL);
+	si_pi->clock_registers.mclk_pwrmgt_cntl = RREG32(MCLK_PWRMGT_CNTL);
+	si_pi->clock_registers.mpll_ad_func_cntl = RREG32(MPLL_AD_FUNC_CNTL);
+	si_pi->clock_registers.mpll_dq_func_cntl = RREG32(MPLL_DQ_FUNC_CNTL);
+	si_pi->clock_registers.mpll_func_cntl = RREG32(MPLL_FUNC_CNTL);
+	si_pi->clock_registers.mpll_func_cntl_1 = RREG32(MPLL_FUNC_CNTL_1);
+	si_pi->clock_registers.mpll_func_cntl_2 = RREG32(MPLL_FUNC_CNTL_2);
+	si_pi->clock_registers.mpll_ss1 = RREG32(MPLL_SS1);
+	si_pi->clock_registers.mpll_ss2 = RREG32(MPLL_SS2);
+}
+
+static void si_enable_thermal_protection(struct amdgpu_device *adev,
+					  bool enable)
+{
+	if (enable)
+		WREG32_P(GENERAL_PWRMGT, 0, ~THERMAL_PROTECTION_DIS);
+	else
+		WREG32_P(GENERAL_PWRMGT, THERMAL_PROTECTION_DIS, ~THERMAL_PROTECTION_DIS);
+}
+
+static void si_enable_acpi_power_management(struct amdgpu_device *adev)
+{
+	WREG32_P(GENERAL_PWRMGT, STATIC_PM_EN, ~STATIC_PM_EN);
+}
+
+#if 0
+static int si_enter_ulp_state(struct amdgpu_device *adev)
+{
+	WREG32(SMC_MESSAGE_0, PPSMC_MSG_SwitchToMinimumPower);
+
+	udelay(25000);
+
+	return 0;
+}
+
+static int si_exit_ulp_state(struct amdgpu_device *adev)
+{
+	int i;
+
+	WREG32(SMC_MESSAGE_0, PPSMC_MSG_ResumeFromMinimumPower);
+
+	udelay(7000);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(SMC_RESP_0) == 1)
+			break;
+		udelay(1000);
+	}
+
+	return 0;
+}
+#endif
+
+static int si_notify_smc_display_change(struct amdgpu_device *adev,
+				     bool has_display)
+{
+	PPSMC_Msg msg = has_display ?
+		PPSMC_MSG_HasDisplay : PPSMC_MSG_NoDisplay;
+
+	return (amdgpu_si_send_msg_to_smc(adev, msg) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+static void si_program_response_times(struct amdgpu_device *adev)
+{
+	u32 voltage_response_time, backbias_response_time, acpi_delay_time, vbi_time_out;
+	u32 vddc_dly, acpi_dly, vbi_dly;
+	u32 reference_clock;
+
+	si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_mvdd_chg_time, 1);
+
+	voltage_response_time = (u32)adev->pm.dpm.voltage_response_time;
+	backbias_response_time = (u32)adev->pm.dpm.backbias_response_time;
+
+	if (voltage_response_time == 0)
+		voltage_response_time = 1000;
+
+	acpi_delay_time = 15000;
+	vbi_time_out = 100000;
+
+	reference_clock = amdgpu_asic_get_xclk(adev);
+
+	vddc_dly = (voltage_response_time  * reference_clock) / 100;
+	acpi_dly = (acpi_delay_time * reference_clock) / 100;
+	vbi_dly  = (vbi_time_out * reference_clock) / 100;
+
+	si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_delay_vreg,  vddc_dly);
+	si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_delay_acpi,  acpi_dly);
+	si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_mclk_chg_timeout, vbi_dly);
+	si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_mc_block_delay, 0xAA);
+}
+
+static void si_program_ds_registers(struct amdgpu_device *adev)
+{
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	u32 tmp;
+
+	/* DEEP_SLEEP_CLK_SEL field should be 0x10 on tahiti A0 */
+	if (adev->asic_type == CHIP_TAHITI && adev->rev_id == 0x0)
+		tmp = 0x10;
+	else
+		tmp = 0x1;
+
+	if (eg_pi->sclk_deep_sleep) {
+		WREG32_P(MISC_CLK_CNTL, DEEP_SLEEP_CLK_SEL(tmp), ~DEEP_SLEEP_CLK_SEL_MASK);
+		WREG32_P(CG_SPLL_AUTOSCALE_CNTL, AUTOSCALE_ON_SS_CLEAR,
+			 ~AUTOSCALE_ON_SS_CLEAR);
+	}
+}
+
+static void si_program_display_gap(struct amdgpu_device *adev)
+{
+	u32 tmp, pipe;
+	int i;
+
+	tmp = RREG32(CG_DISPLAY_GAP_CNTL) & ~(DISP1_GAP_MASK | DISP2_GAP_MASK);
+	if (adev->pm.dpm.new_active_crtc_count > 0)
+		tmp |= DISP1_GAP(R600_PM_DISPLAY_GAP_VBLANK_OR_WM);
+	else
+		tmp |= DISP1_GAP(R600_PM_DISPLAY_GAP_IGNORE);
+
+	if (adev->pm.dpm.new_active_crtc_count > 1)
+		tmp |= DISP2_GAP(R600_PM_DISPLAY_GAP_VBLANK_OR_WM);
+	else
+		tmp |= DISP2_GAP(R600_PM_DISPLAY_GAP_IGNORE);
+
+	WREG32(CG_DISPLAY_GAP_CNTL, tmp);
+
+	tmp = RREG32(DCCG_DISP_SLOW_SELECT_REG);
+	pipe = (tmp & DCCG_DISP1_SLOW_SELECT_MASK) >> DCCG_DISP1_SLOW_SELECT_SHIFT;
+
+	if ((adev->pm.dpm.new_active_crtc_count > 0) &&
+	    (!(adev->pm.dpm.new_active_crtcs & (1 << pipe)))) {
+		/* find the first active crtc */
+		for (i = 0; i < adev->mode_info.num_crtc; i++) {
+			if (adev->pm.dpm.new_active_crtcs & (1 << i))
+				break;
+		}
+		if (i == adev->mode_info.num_crtc)
+			pipe = 0;
+		else
+			pipe = i;
+
+		tmp &= ~DCCG_DISP1_SLOW_SELECT_MASK;
+		tmp |= DCCG_DISP1_SLOW_SELECT(pipe);
+		WREG32(DCCG_DISP_SLOW_SELECT_REG, tmp);
+	}
+
+	/* Setting this to false forces the performance state to low if the crtcs are disabled.
+	 * This can be a problem on PowerXpress systems or if you want to use the card
+	 * for offscreen rendering or compute if there are no crtcs enabled.
+	 */
+	si_notify_smc_display_change(adev, adev->pm.dpm.new_active_crtc_count > 0);
+}
+
+static void si_enable_spread_spectrum(struct amdgpu_device *adev, bool enable)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+
+	if (enable) {
+		if (pi->sclk_ss)
+			WREG32_P(GENERAL_PWRMGT, DYN_SPREAD_SPECTRUM_EN, ~DYN_SPREAD_SPECTRUM_EN);
+	} else {
+		WREG32_P(CG_SPLL_SPREAD_SPECTRUM, 0, ~SSEN);
+		WREG32_P(GENERAL_PWRMGT, 0, ~DYN_SPREAD_SPECTRUM_EN);
+	}
+}
+
+static void si_setup_bsp(struct amdgpu_device *adev)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	u32 xclk = amdgpu_asic_get_xclk(adev);
+
+	r600_calculate_u_and_p(pi->asi,
+			       xclk,
+			       16,
+			       &pi->bsp,
+			       &pi->bsu);
+
+	r600_calculate_u_and_p(pi->pasi,
+			       xclk,
+			       16,
+			       &pi->pbsp,
+			       &pi->pbsu);
+
+
+        pi->dsp = BSP(pi->bsp) | BSU(pi->bsu);
+	pi->psp = BSP(pi->pbsp) | BSU(pi->pbsu);
+
+	WREG32(CG_BSP, pi->dsp);
+}
+
+static void si_program_git(struct amdgpu_device *adev)
+{
+	WREG32_P(CG_GIT, CG_GICST(R600_GICST_DFLT), ~CG_GICST_MASK);
+}
+
+static void si_program_tp(struct amdgpu_device *adev)
+{
+	int i;
+	enum r600_td td = R600_TD_DFLT;
+
+	for (i = 0; i < R600_PM_NUMBER_OF_TC; i++)
+		WREG32(CG_FFCT_0 + i, (UTC_0(r600_utc[i]) | DTC_0(r600_dtc[i])));
+
+	if (td == R600_TD_AUTO)
+		WREG32_P(SCLK_PWRMGT_CNTL, 0, ~FIR_FORCE_TREND_SEL);
+	else
+		WREG32_P(SCLK_PWRMGT_CNTL, FIR_FORCE_TREND_SEL, ~FIR_FORCE_TREND_SEL);
+
+	if (td == R600_TD_UP)
+		WREG32_P(SCLK_PWRMGT_CNTL, 0, ~FIR_TREND_MODE);
+
+	if (td == R600_TD_DOWN)
+		WREG32_P(SCLK_PWRMGT_CNTL, FIR_TREND_MODE, ~FIR_TREND_MODE);
+}
+
+static void si_program_tpp(struct amdgpu_device *adev)
+{
+	WREG32(CG_TPC, R600_TPC_DFLT);
+}
+
+static void si_program_sstp(struct amdgpu_device *adev)
+{
+	WREG32(CG_SSP, (SSTU(R600_SSTU_DFLT) | SST(R600_SST_DFLT)));
+}
+
+static void si_enable_display_gap(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(CG_DISPLAY_GAP_CNTL);
+
+	tmp &= ~(DISP1_GAP_MASK | DISP2_GAP_MASK);
+	tmp |= (DISP1_GAP(R600_PM_DISPLAY_GAP_IGNORE) |
+		DISP2_GAP(R600_PM_DISPLAY_GAP_IGNORE));
+
+	tmp &= ~(DISP1_GAP_MCHG_MASK | DISP2_GAP_MCHG_MASK);
+	tmp |= (DISP1_GAP_MCHG(R600_PM_DISPLAY_GAP_VBLANK) |
+		DISP2_GAP_MCHG(R600_PM_DISPLAY_GAP_IGNORE));
+	WREG32(CG_DISPLAY_GAP_CNTL, tmp);
+}
+
+static void si_program_vc(struct amdgpu_device *adev)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+
+	WREG32(CG_FTV, pi->vrc);
+}
+
+static void si_clear_vc(struct amdgpu_device *adev)
+{
+	WREG32(CG_FTV, 0);
+}
+
+static u8 si_get_ddr3_mclk_frequency_ratio(u32 memory_clock)
+{
+	u8 mc_para_index;
+
+	if (memory_clock < 10000)
+		mc_para_index = 0;
+	else if (memory_clock >= 80000)
+		mc_para_index = 0x0f;
+	else
+		mc_para_index = (u8)((memory_clock - 10000) / 5000 + 1);
+	return mc_para_index;
+}
+
+static u8 si_get_mclk_frequency_ratio(u32 memory_clock, bool strobe_mode)
+{
+	u8 mc_para_index;
+
+	if (strobe_mode) {
+		if (memory_clock < 12500)
+			mc_para_index = 0x00;
+		else if (memory_clock > 47500)
+			mc_para_index = 0x0f;
+		else
+			mc_para_index = (u8)((memory_clock - 10000) / 2500);
+	} else {
+		if (memory_clock < 65000)
+			mc_para_index = 0x00;
+		else if (memory_clock > 135000)
+			mc_para_index = 0x0f;
+		else
+			mc_para_index = (u8)((memory_clock - 60000) / 5000);
+	}
+	return mc_para_index;
+}
+
+static u8 si_get_strobe_mode_settings(struct amdgpu_device *adev, u32 mclk)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	bool strobe_mode = false;
+	u8 result = 0;
+
+	if (mclk <= pi->mclk_strobe_mode_threshold)
+		strobe_mode = true;
+
+	if (adev->gmc.vram_type == AMDGPU_VRAM_TYPE_GDDR5)
+		result = si_get_mclk_frequency_ratio(mclk, strobe_mode);
+	else
+		result = si_get_ddr3_mclk_frequency_ratio(mclk);
+
+	if (strobe_mode)
+		result |= SISLANDS_SMC_STROBE_ENABLE;
+
+	return result;
+}
+
+static int si_upload_firmware(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+
+	amdgpu_si_reset_smc(adev);
+	amdgpu_si_smc_clock(adev, false);
+
+	return amdgpu_si_load_smc_ucode(adev, si_pi->sram_end);
+}
+
+static bool si_validate_phase_shedding_tables(struct amdgpu_device *adev,
+					      const struct atom_voltage_table *table,
+					      const struct amdgpu_phase_shedding_limits_table *limits)
+{
+	u32 data, num_bits, num_levels;
+
+	if ((table == NULL) || (limits == NULL))
+		return false;
+
+	data = table->mask_low;
+
+	num_bits = hweight32(data);
+
+	if (num_bits == 0)
+		return false;
+
+	num_levels = (1 << num_bits);
+
+	if (table->count != num_levels)
+		return false;
+
+	if (limits->count != (num_levels - 1))
+		return false;
+
+	return true;
+}
+
+static void si_trim_voltage_table_to_fit_state_table(struct amdgpu_device *adev,
+					      u32 max_voltage_steps,
+					      struct atom_voltage_table *voltage_table)
+{
+	unsigned int i, diff;
+
+	if (voltage_table->count <= max_voltage_steps)
+		return;
+
+	diff = voltage_table->count - max_voltage_steps;
+
+	for (i= 0; i < max_voltage_steps; i++)
+		voltage_table->entries[i] = voltage_table->entries[i + diff];
+
+	voltage_table->count = max_voltage_steps;
+}
+
+static int si_get_svi2_voltage_table(struct amdgpu_device *adev,
+				     struct amdgpu_clock_voltage_dependency_table *voltage_dependency_table,
+				     struct atom_voltage_table *voltage_table)
+{
+	u32 i;
+
+	if (voltage_dependency_table == NULL)
+		return -EINVAL;
+
+	voltage_table->mask_low = 0;
+	voltage_table->phase_delay = 0;
+
+	voltage_table->count = voltage_dependency_table->count;
+	for (i = 0; i < voltage_table->count; i++) {
+		voltage_table->entries[i].value = voltage_dependency_table->entries[i].v;
+		voltage_table->entries[i].smio_low = 0;
+	}
+
+	return 0;
+}
+
+static int si_construct_voltage_tables(struct amdgpu_device *adev)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	int ret;
+
+	if (pi->voltage_control) {
+		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_VDDC,
+						    VOLTAGE_OBJ_GPIO_LUT, &eg_pi->vddc_voltage_table);
+		if (ret)
+			return ret;
+
+		if (eg_pi->vddc_voltage_table.count > SISLANDS_MAX_NO_VREG_STEPS)
+			si_trim_voltage_table_to_fit_state_table(adev,
+								 SISLANDS_MAX_NO_VREG_STEPS,
+								 &eg_pi->vddc_voltage_table);
+	} else if (si_pi->voltage_control_svi2) {
+		ret = si_get_svi2_voltage_table(adev,
+						&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk,
+						&eg_pi->vddc_voltage_table);
+		if (ret)
+			return ret;
+	} else {
+		return -EINVAL;
+	}
+
+	if (eg_pi->vddci_control) {
+		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_VDDCI,
+						    VOLTAGE_OBJ_GPIO_LUT, &eg_pi->vddci_voltage_table);
+		if (ret)
+			return ret;
+
+		if (eg_pi->vddci_voltage_table.count > SISLANDS_MAX_NO_VREG_STEPS)
+			si_trim_voltage_table_to_fit_state_table(adev,
+								 SISLANDS_MAX_NO_VREG_STEPS,
+								 &eg_pi->vddci_voltage_table);
+	}
+	if (si_pi->vddci_control_svi2) {
+		ret = si_get_svi2_voltage_table(adev,
+						&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk,
+						&eg_pi->vddci_voltage_table);
+		if (ret)
+			return ret;
+	}
+
+	if (pi->mvdd_control) {
+		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_MVDDC,
+						    VOLTAGE_OBJ_GPIO_LUT, &si_pi->mvdd_voltage_table);
+
+		if (ret) {
+			pi->mvdd_control = false;
+			return ret;
+		}
+
+		if (si_pi->mvdd_voltage_table.count == 0) {
+			pi->mvdd_control = false;
+			return -EINVAL;
+		}
+
+		if (si_pi->mvdd_voltage_table.count > SISLANDS_MAX_NO_VREG_STEPS)
+			si_trim_voltage_table_to_fit_state_table(adev,
+								 SISLANDS_MAX_NO_VREG_STEPS,
+								 &si_pi->mvdd_voltage_table);
+	}
+
+	if (si_pi->vddc_phase_shed_control) {
+		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_VDDC,
+						    VOLTAGE_OBJ_PHASE_LUT, &si_pi->vddc_phase_shed_table);
+		if (ret)
+			si_pi->vddc_phase_shed_control = false;
+
+		if ((si_pi->vddc_phase_shed_table.count == 0) ||
+		    (si_pi->vddc_phase_shed_table.count > SISLANDS_MAX_NO_VREG_STEPS))
+			si_pi->vddc_phase_shed_control = false;
+	}
+
+	return 0;
+}
+
+static void si_populate_smc_voltage_table(struct amdgpu_device *adev,
+					  const struct atom_voltage_table *voltage_table,
+					  SISLANDS_SMC_STATETABLE *table)
+{
+	unsigned int i;
+
+	for (i = 0; i < voltage_table->count; i++)
+		table->lowSMIO[i] |= cpu_to_be32(voltage_table->entries[i].smio_low);
+}
+
+static int si_populate_smc_voltage_tables(struct amdgpu_device *adev,
+					  SISLANDS_SMC_STATETABLE *table)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u8 i;
+
+	if (si_pi->voltage_control_svi2) {
+		si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_svi_rework_gpio_id_svc,
+			si_pi->svc_gpio_id);
+		si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_svi_rework_gpio_id_svd,
+			si_pi->svd_gpio_id);
+		si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_svi_rework_plat_type,
+					   2);
+	} else {
+		if (eg_pi->vddc_voltage_table.count) {
+			si_populate_smc_voltage_table(adev, &eg_pi->vddc_voltage_table, table);
+			table->voltageMaskTable.lowMask[SISLANDS_SMC_VOLTAGEMASK_VDDC] =
+				cpu_to_be32(eg_pi->vddc_voltage_table.mask_low);
+
+			for (i = 0; i < eg_pi->vddc_voltage_table.count; i++) {
+				if (pi->max_vddc_in_table <= eg_pi->vddc_voltage_table.entries[i].value) {
+					table->maxVDDCIndexInPPTable = i;
+					break;
+				}
+			}
+		}
+
+		if (eg_pi->vddci_voltage_table.count) {
+			si_populate_smc_voltage_table(adev, &eg_pi->vddci_voltage_table, table);
+
+			table->voltageMaskTable.lowMask[SISLANDS_SMC_VOLTAGEMASK_VDDCI] =
+				cpu_to_be32(eg_pi->vddci_voltage_table.mask_low);
+		}
+
+
+		if (si_pi->mvdd_voltage_table.count) {
+			si_populate_smc_voltage_table(adev, &si_pi->mvdd_voltage_table, table);
+
+			table->voltageMaskTable.lowMask[SISLANDS_SMC_VOLTAGEMASK_MVDD] =
+				cpu_to_be32(si_pi->mvdd_voltage_table.mask_low);
+		}
+
+		if (si_pi->vddc_phase_shed_control) {
+			if (si_validate_phase_shedding_tables(adev, &si_pi->vddc_phase_shed_table,
+							      &adev->pm.dpm.dyn_state.phase_shedding_limits_table)) {
+				si_populate_smc_voltage_table(adev, &si_pi->vddc_phase_shed_table, table);
+
+				table->phaseMaskTable.lowMask[SISLANDS_SMC_VOLTAGEMASK_VDDC_PHASE_SHEDDING] =
+					cpu_to_be32(si_pi->vddc_phase_shed_table.mask_low);
+
+				si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_phase_shedding_delay,
+							   (u32)si_pi->vddc_phase_shed_table.phase_delay);
+			} else {
+				si_pi->vddc_phase_shed_control = false;
+			}
+		}
+	}
+
+	return 0;
+}
+
+static int si_populate_voltage_value(struct amdgpu_device *adev,
+				     const struct atom_voltage_table *table,
+				     u16 value, SISLANDS_SMC_VOLTAGE_VALUE *voltage)
+{
+	unsigned int i;
+
+	for (i = 0; i < table->count; i++) {
+		if (value <= table->entries[i].value) {
+			voltage->index = (u8)i;
+			voltage->value = cpu_to_be16(table->entries[i].value);
+			break;
+		}
+	}
+
+	if (i >= table->count)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int si_populate_mvdd_value(struct amdgpu_device *adev, u32 mclk,
+				  SISLANDS_SMC_VOLTAGE_VALUE *voltage)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+
+	if (pi->mvdd_control) {
+		if (mclk <= pi->mvdd_split_frequency)
+			voltage->index = 0;
+		else
+			voltage->index = (u8)(si_pi->mvdd_voltage_table.count) - 1;
+
+		voltage->value = cpu_to_be16(si_pi->mvdd_voltage_table.entries[voltage->index].value);
+	}
+	return 0;
+}
+
+static int si_get_std_voltage_value(struct amdgpu_device *adev,
+				    SISLANDS_SMC_VOLTAGE_VALUE *voltage,
+				    u16 *std_voltage)
+{
+	u16 v_index;
+	bool voltage_found = false;
+	*std_voltage = be16_to_cpu(voltage->value);
+
+	if (adev->pm.dpm.dyn_state.cac_leakage_table.entries) {
+		if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_NEW_CAC_VOLTAGE) {
+			if (adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries == NULL)
+				return -EINVAL;
+
+			for (v_index = 0; (u32)v_index < adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count; v_index++) {
+				if (be16_to_cpu(voltage->value) ==
+				    (u16)adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[v_index].v) {
+					voltage_found = true;
+					if ((u32)v_index < adev->pm.dpm.dyn_state.cac_leakage_table.count)
+						*std_voltage =
+							adev->pm.dpm.dyn_state.cac_leakage_table.entries[v_index].vddc;
+					else
+						*std_voltage =
+							adev->pm.dpm.dyn_state.cac_leakage_table.entries[adev->pm.dpm.dyn_state.cac_leakage_table.count-1].vddc;
+					break;
+				}
+			}
+
+			if (!voltage_found) {
+				for (v_index = 0; (u32)v_index < adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count; v_index++) {
+					if (be16_to_cpu(voltage->value) <=
+					    (u16)adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[v_index].v) {
+						voltage_found = true;
+						if ((u32)v_index < adev->pm.dpm.dyn_state.cac_leakage_table.count)
+							*std_voltage =
+								adev->pm.dpm.dyn_state.cac_leakage_table.entries[v_index].vddc;
+						else
+							*std_voltage =
+								adev->pm.dpm.dyn_state.cac_leakage_table.entries[adev->pm.dpm.dyn_state.cac_leakage_table.count-1].vddc;
+						break;
+					}
+				}
+			}
+		} else {
+			if ((u32)voltage->index < adev->pm.dpm.dyn_state.cac_leakage_table.count)
+				*std_voltage = adev->pm.dpm.dyn_state.cac_leakage_table.entries[voltage->index].vddc;
+		}
+	}
+
+	return 0;
+}
+
+static int si_populate_std_voltage_value(struct amdgpu_device *adev,
+					 u16 value, u8 index,
+					 SISLANDS_SMC_VOLTAGE_VALUE *voltage)
+{
+	voltage->index = index;
+	voltage->value = cpu_to_be16(value);
+
+	return 0;
+}
+
+static int si_populate_phase_shedding_value(struct amdgpu_device *adev,
+					    const struct amdgpu_phase_shedding_limits_table *limits,
+					    u16 voltage, u32 sclk, u32 mclk,
+					    SISLANDS_SMC_VOLTAGE_VALUE *smc_voltage)
+{
+	unsigned int i;
+
+	for (i = 0; i < limits->count; i++) {
+		if ((voltage <= limits->entries[i].voltage) &&
+		    (sclk <= limits->entries[i].sclk) &&
+		    (mclk <= limits->entries[i].mclk))
+			break;
+	}
+
+	smc_voltage->phase_settings = (u8)i;
+
+	return 0;
+}
+
+static int si_init_arb_table_index(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 tmp;
+	int ret;
+
+	ret = amdgpu_si_read_smc_sram_dword(adev, si_pi->arb_table_start,
+					    &tmp, si_pi->sram_end);
+	if (ret)
+		return ret;
+
+	tmp &= 0x00FFFFFF;
+	tmp |= MC_CG_ARB_FREQ_F1 << 24;
+
+	return amdgpu_si_write_smc_sram_dword(adev, si_pi->arb_table_start,
+					      tmp, si_pi->sram_end);
+}
+
+static int si_initial_switch_from_arb_f0_to_f1(struct amdgpu_device *adev)
+{
+	return ni_copy_and_switch_arb_sets(adev, MC_CG_ARB_FREQ_F0, MC_CG_ARB_FREQ_F1);
+}
+
+static int si_reset_to_default(struct amdgpu_device *adev)
+{
+	return (amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_ResetToDefaults) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+static int si_force_switch_to_arb_f0(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 tmp;
+	int ret;
+
+	ret = amdgpu_si_read_smc_sram_dword(adev, si_pi->arb_table_start,
+					    &tmp, si_pi->sram_end);
+	if (ret)
+		return ret;
+
+	tmp = (tmp >> 24) & 0xff;
+
+	if (tmp == MC_CG_ARB_FREQ_F0)
+		return 0;
+
+	return ni_copy_and_switch_arb_sets(adev, tmp, MC_CG_ARB_FREQ_F0);
+}
+
+static u32 si_calculate_memory_refresh_rate(struct amdgpu_device *adev,
+					    u32 engine_clock)
+{
+	u32 dram_rows;
+	u32 dram_refresh_rate;
+	u32 mc_arb_rfsh_rate;
+	u32 tmp = (RREG32(MC_ARB_RAMCFG) & NOOFROWS_MASK) >> NOOFROWS_SHIFT;
+
+	if (tmp >= 4)
+		dram_rows = 16384;
+	else
+		dram_rows = 1 << (tmp + 10);
+
+	dram_refresh_rate = 1 << ((RREG32(MC_SEQ_MISC0) & 0x3) + 3);
+	mc_arb_rfsh_rate = ((engine_clock * 10) * dram_refresh_rate / dram_rows - 32) / 64;
+
+	return mc_arb_rfsh_rate;
+}
+
+static int si_populate_memory_timing_parameters(struct amdgpu_device *adev,
+						struct rv7xx_pl *pl,
+						SMC_SIslands_MCArbDramTimingRegisterSet *arb_regs)
+{
+	u32 dram_timing;
+	u32 dram_timing2;
+	u32 burst_time;
+
+	arb_regs->mc_arb_rfsh_rate =
+		(u8)si_calculate_memory_refresh_rate(adev, pl->sclk);
+
+	amdgpu_atombios_set_engine_dram_timings(adev,
+					    pl->sclk,
+		                            pl->mclk);
+
+	dram_timing  = RREG32(MC_ARB_DRAM_TIMING);
+	dram_timing2 = RREG32(MC_ARB_DRAM_TIMING2);
+	burst_time = RREG32(MC_ARB_BURST_TIME) & STATE0_MASK;
+
+	arb_regs->mc_arb_dram_timing  = cpu_to_be32(dram_timing);
+	arb_regs->mc_arb_dram_timing2 = cpu_to_be32(dram_timing2);
+	arb_regs->mc_arb_burst_time = (u8)burst_time;
+
+	return 0;
+}
+
+static int si_do_program_memory_timing_parameters(struct amdgpu_device *adev,
+						  struct amdgpu_ps *amdgpu_state,
+						  unsigned int first_arb_set)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct  si_ps *state = si_get_ps(amdgpu_state);
+	SMC_SIslands_MCArbDramTimingRegisterSet arb_regs = { 0 };
+	int i, ret = 0;
+
+	for (i = 0; i < state->performance_level_count; i++) {
+		ret = si_populate_memory_timing_parameters(adev, &state->performance_levels[i], &arb_regs);
+		if (ret)
+			break;
+		ret = amdgpu_si_copy_bytes_to_smc(adev,
+						  si_pi->arb_table_start +
+						  offsetof(SMC_SIslands_MCArbDramTimingRegisters, data) +
+						  sizeof(SMC_SIslands_MCArbDramTimingRegisterSet) * (first_arb_set + i),
+						  (u8 *)&arb_regs,
+						  sizeof(SMC_SIslands_MCArbDramTimingRegisterSet),
+						  si_pi->sram_end);
+		if (ret)
+			break;
+	}
+
+	return ret;
+}
+
+static int si_program_memory_timing_parameters(struct amdgpu_device *adev,
+					       struct amdgpu_ps *amdgpu_new_state)
+{
+	return si_do_program_memory_timing_parameters(adev, amdgpu_new_state,
+						      SISLANDS_DRIVER_STATE_ARB_INDEX);
+}
+
+static int si_populate_initial_mvdd_value(struct amdgpu_device *adev,
+					  struct SISLANDS_SMC_VOLTAGE_VALUE *voltage)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+
+	if (pi->mvdd_control)
+		return si_populate_voltage_value(adev, &si_pi->mvdd_voltage_table,
+						 si_pi->mvdd_bootup_value, voltage);
+
+	return 0;
+}
+
+static int si_populate_smc_initial_state(struct amdgpu_device *adev,
+					 struct amdgpu_ps *amdgpu_initial_state,
+					 SISLANDS_SMC_STATETABLE *table)
+{
+	struct  si_ps *initial_state = si_get_ps(amdgpu_initial_state);
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 reg;
+	int ret;
+
+	table->initialState.levels[0].mclk.vDLL_CNTL =
+		cpu_to_be32(si_pi->clock_registers.dll_cntl);
+	table->initialState.levels[0].mclk.vMCLK_PWRMGT_CNTL =
+		cpu_to_be32(si_pi->clock_registers.mclk_pwrmgt_cntl);
+	table->initialState.levels[0].mclk.vMPLL_AD_FUNC_CNTL =
+		cpu_to_be32(si_pi->clock_registers.mpll_ad_func_cntl);
+	table->initialState.levels[0].mclk.vMPLL_DQ_FUNC_CNTL =
+		cpu_to_be32(si_pi->clock_registers.mpll_dq_func_cntl);
+	table->initialState.levels[0].mclk.vMPLL_FUNC_CNTL =
+		cpu_to_be32(si_pi->clock_registers.mpll_func_cntl);
+	table->initialState.levels[0].mclk.vMPLL_FUNC_CNTL_1 =
+		cpu_to_be32(si_pi->clock_registers.mpll_func_cntl_1);
+	table->initialState.levels[0].mclk.vMPLL_FUNC_CNTL_2 =
+		cpu_to_be32(si_pi->clock_registers.mpll_func_cntl_2);
+	table->initialState.levels[0].mclk.vMPLL_SS =
+		cpu_to_be32(si_pi->clock_registers.mpll_ss1);
+	table->initialState.levels[0].mclk.vMPLL_SS2 =
+		cpu_to_be32(si_pi->clock_registers.mpll_ss2);
+
+	table->initialState.levels[0].mclk.mclk_value =
+		cpu_to_be32(initial_state->performance_levels[0].mclk);
+
+	table->initialState.levels[0].sclk.vCG_SPLL_FUNC_CNTL =
+		cpu_to_be32(si_pi->clock_registers.cg_spll_func_cntl);
+	table->initialState.levels[0].sclk.vCG_SPLL_FUNC_CNTL_2 =
+		cpu_to_be32(si_pi->clock_registers.cg_spll_func_cntl_2);
+	table->initialState.levels[0].sclk.vCG_SPLL_FUNC_CNTL_3 =
+		cpu_to_be32(si_pi->clock_registers.cg_spll_func_cntl_3);
+	table->initialState.levels[0].sclk.vCG_SPLL_FUNC_CNTL_4 =
+		cpu_to_be32(si_pi->clock_registers.cg_spll_func_cntl_4);
+	table->initialState.levels[0].sclk.vCG_SPLL_SPREAD_SPECTRUM =
+		cpu_to_be32(si_pi->clock_registers.cg_spll_spread_spectrum);
+	table->initialState.levels[0].sclk.vCG_SPLL_SPREAD_SPECTRUM_2  =
+		cpu_to_be32(si_pi->clock_registers.cg_spll_spread_spectrum_2);
+
+	table->initialState.levels[0].sclk.sclk_value =
+		cpu_to_be32(initial_state->performance_levels[0].sclk);
+
+	table->initialState.levels[0].arbRefreshState =
+		SISLANDS_INITIAL_STATE_ARB_INDEX;
+
+	table->initialState.levels[0].ACIndex = 0;
+
+	ret = si_populate_voltage_value(adev, &eg_pi->vddc_voltage_table,
+					initial_state->performance_levels[0].vddc,
+					&table->initialState.levels[0].vddc);
+
+	if (!ret) {
+		u16 std_vddc;
+
+		ret = si_get_std_voltage_value(adev,
+					       &table->initialState.levels[0].vddc,
+					       &std_vddc);
+		if (!ret)
+			si_populate_std_voltage_value(adev, std_vddc,
+						      table->initialState.levels[0].vddc.index,
+						      &table->initialState.levels[0].std_vddc);
+	}
+
+	if (eg_pi->vddci_control)
+		si_populate_voltage_value(adev,
+					  &eg_pi->vddci_voltage_table,
+					  initial_state->performance_levels[0].vddci,
+					  &table->initialState.levels[0].vddci);
+
+	if (si_pi->vddc_phase_shed_control)
+		si_populate_phase_shedding_value(adev,
+						 &adev->pm.dpm.dyn_state.phase_shedding_limits_table,
+						 initial_state->performance_levels[0].vddc,
+						 initial_state->performance_levels[0].sclk,
+						 initial_state->performance_levels[0].mclk,
+						 &table->initialState.levels[0].vddc);
+
+	si_populate_initial_mvdd_value(adev, &table->initialState.levels[0].mvdd);
+
+	reg = CG_R(0xffff) | CG_L(0);
+	table->initialState.levels[0].aT = cpu_to_be32(reg);
+	table->initialState.levels[0].bSP = cpu_to_be32(pi->dsp);
+	table->initialState.levels[0].gen2PCIE = (u8)si_pi->boot_pcie_gen;
+
+	if (adev->gmc.vram_type == AMDGPU_VRAM_TYPE_GDDR5) {
+		table->initialState.levels[0].strobeMode =
+			si_get_strobe_mode_settings(adev,
+						    initial_state->performance_levels[0].mclk);
+
+		if (initial_state->performance_levels[0].mclk > pi->mclk_edc_enable_threshold)
+			table->initialState.levels[0].mcFlags = SISLANDS_SMC_MC_EDC_RD_FLAG | SISLANDS_SMC_MC_EDC_WR_FLAG;
+		else
+			table->initialState.levels[0].mcFlags =  0;
+	}
+
+	table->initialState.levelCount = 1;
+
+	table->initialState.flags |= PPSMC_SWSTATE_FLAG_DC;
+
+	table->initialState.levels[0].dpm2.MaxPS = 0;
+	table->initialState.levels[0].dpm2.NearTDPDec = 0;
+	table->initialState.levels[0].dpm2.AboveSafeInc = 0;
+	table->initialState.levels[0].dpm2.BelowSafeInc = 0;
+	table->initialState.levels[0].dpm2.PwrEfficiencyRatio = 0;
+
+	reg = MIN_POWER_MASK | MAX_POWER_MASK;
+	table->initialState.levels[0].SQPowerThrottle = cpu_to_be32(reg);
+
+	reg = MAX_POWER_DELTA_MASK | STI_SIZE_MASK | LTI_RATIO_MASK;
+	table->initialState.levels[0].SQPowerThrottle_2 = cpu_to_be32(reg);
+
+	return 0;
+}
+
+static int si_populate_smc_acpi_state(struct amdgpu_device *adev,
+				      SISLANDS_SMC_STATETABLE *table)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 spll_func_cntl = si_pi->clock_registers.cg_spll_func_cntl;
+	u32 spll_func_cntl_2 = si_pi->clock_registers.cg_spll_func_cntl_2;
+	u32 spll_func_cntl_3 = si_pi->clock_registers.cg_spll_func_cntl_3;
+	u32 spll_func_cntl_4 = si_pi->clock_registers.cg_spll_func_cntl_4;
+	u32 dll_cntl = si_pi->clock_registers.dll_cntl;
+	u32 mclk_pwrmgt_cntl = si_pi->clock_registers.mclk_pwrmgt_cntl;
+	u32 mpll_ad_func_cntl = si_pi->clock_registers.mpll_ad_func_cntl;
+	u32 mpll_dq_func_cntl = si_pi->clock_registers.mpll_dq_func_cntl;
+	u32 mpll_func_cntl = si_pi->clock_registers.mpll_func_cntl;
+	u32 mpll_func_cntl_1 = si_pi->clock_registers.mpll_func_cntl_1;
+	u32 mpll_func_cntl_2 = si_pi->clock_registers.mpll_func_cntl_2;
+	u32 reg;
+	int ret;
+
+	table->ACPIState = table->initialState;
+
+	table->ACPIState.flags &= ~PPSMC_SWSTATE_FLAG_DC;
+
+	if (pi->acpi_vddc) {
+		ret = si_populate_voltage_value(adev, &eg_pi->vddc_voltage_table,
+						pi->acpi_vddc, &table->ACPIState.levels[0].vddc);
+		if (!ret) {
+			u16 std_vddc;
+
+			ret = si_get_std_voltage_value(adev,
+						       &table->ACPIState.levels[0].vddc, &std_vddc);
+			if (!ret)
+				si_populate_std_voltage_value(adev, std_vddc,
+							      table->ACPIState.levels[0].vddc.index,
+							      &table->ACPIState.levels[0].std_vddc);
+		}
+		table->ACPIState.levels[0].gen2PCIE = si_pi->acpi_pcie_gen;
+
+		if (si_pi->vddc_phase_shed_control) {
+			si_populate_phase_shedding_value(adev,
+							 &adev->pm.dpm.dyn_state.phase_shedding_limits_table,
+							 pi->acpi_vddc,
+							 0,
+							 0,
+							 &table->ACPIState.levels[0].vddc);
+		}
+	} else {
+		ret = si_populate_voltage_value(adev, &eg_pi->vddc_voltage_table,
+						pi->min_vddc_in_table, &table->ACPIState.levels[0].vddc);
+		if (!ret) {
+			u16 std_vddc;
+
+			ret = si_get_std_voltage_value(adev,
+						       &table->ACPIState.levels[0].vddc, &std_vddc);
+
+			if (!ret)
+				si_populate_std_voltage_value(adev, std_vddc,
+							      table->ACPIState.levels[0].vddc.index,
+							      &table->ACPIState.levels[0].std_vddc);
+		}
+		table->ACPIState.levels[0].gen2PCIE =
+			(u8)amdgpu_get_pcie_gen_support(adev,
+							si_pi->sys_pcie_mask,
+							si_pi->boot_pcie_gen,
+							AMDGPU_PCIE_GEN1);
+
+		if (si_pi->vddc_phase_shed_control)
+			si_populate_phase_shedding_value(adev,
+							 &adev->pm.dpm.dyn_state.phase_shedding_limits_table,
+							 pi->min_vddc_in_table,
+							 0,
+							 0,
+							 &table->ACPIState.levels[0].vddc);
+	}
+
+	if (pi->acpi_vddc) {
+		if (eg_pi->acpi_vddci)
+			si_populate_voltage_value(adev, &eg_pi->vddci_voltage_table,
+						  eg_pi->acpi_vddci,
+						  &table->ACPIState.levels[0].vddci);
+	}
+
+	mclk_pwrmgt_cntl |= MRDCK0_RESET | MRDCK1_RESET;
+	mclk_pwrmgt_cntl &= ~(MRDCK0_PDNB | MRDCK1_PDNB);
+
+	dll_cntl &= ~(MRDCK0_BYPASS | MRDCK1_BYPASS);
+
+	spll_func_cntl_2 &= ~SCLK_MUX_SEL_MASK;
+	spll_func_cntl_2 |= SCLK_MUX_SEL(4);
+
+	table->ACPIState.levels[0].mclk.vDLL_CNTL =
+		cpu_to_be32(dll_cntl);
+	table->ACPIState.levels[0].mclk.vMCLK_PWRMGT_CNTL =
+		cpu_to_be32(mclk_pwrmgt_cntl);
+	table->ACPIState.levels[0].mclk.vMPLL_AD_FUNC_CNTL =
+		cpu_to_be32(mpll_ad_func_cntl);
+	table->ACPIState.levels[0].mclk.vMPLL_DQ_FUNC_CNTL =
+		cpu_to_be32(mpll_dq_func_cntl);
+	table->ACPIState.levels[0].mclk.vMPLL_FUNC_CNTL =
+		cpu_to_be32(mpll_func_cntl);
+	table->ACPIState.levels[0].mclk.vMPLL_FUNC_CNTL_1 =
+		cpu_to_be32(mpll_func_cntl_1);
+	table->ACPIState.levels[0].mclk.vMPLL_FUNC_CNTL_2 =
+		cpu_to_be32(mpll_func_cntl_2);
+	table->ACPIState.levels[0].mclk.vMPLL_SS =
+		cpu_to_be32(si_pi->clock_registers.mpll_ss1);
+	table->ACPIState.levels[0].mclk.vMPLL_SS2 =
+		cpu_to_be32(si_pi->clock_registers.mpll_ss2);
+
+	table->ACPIState.levels[0].sclk.vCG_SPLL_FUNC_CNTL =
+		cpu_to_be32(spll_func_cntl);
+	table->ACPIState.levels[0].sclk.vCG_SPLL_FUNC_CNTL_2 =
+		cpu_to_be32(spll_func_cntl_2);
+	table->ACPIState.levels[0].sclk.vCG_SPLL_FUNC_CNTL_3 =
+		cpu_to_be32(spll_func_cntl_3);
+	table->ACPIState.levels[0].sclk.vCG_SPLL_FUNC_CNTL_4 =
+		cpu_to_be32(spll_func_cntl_4);
+
+	table->ACPIState.levels[0].mclk.mclk_value = 0;
+	table->ACPIState.levels[0].sclk.sclk_value = 0;
+
+	si_populate_mvdd_value(adev, 0, &table->ACPIState.levels[0].mvdd);
+
+	if (eg_pi->dynamic_ac_timing)
+		table->ACPIState.levels[0].ACIndex = 0;
+
+	table->ACPIState.levels[0].dpm2.MaxPS = 0;
+	table->ACPIState.levels[0].dpm2.NearTDPDec = 0;
+	table->ACPIState.levels[0].dpm2.AboveSafeInc = 0;
+	table->ACPIState.levels[0].dpm2.BelowSafeInc = 0;
+	table->ACPIState.levels[0].dpm2.PwrEfficiencyRatio = 0;
+
+	reg = MIN_POWER_MASK | MAX_POWER_MASK;
+	table->ACPIState.levels[0].SQPowerThrottle = cpu_to_be32(reg);
+
+	reg = MAX_POWER_DELTA_MASK | STI_SIZE_MASK | LTI_RATIO_MASK;
+	table->ACPIState.levels[0].SQPowerThrottle_2 = cpu_to_be32(reg);
+
+	return 0;
+}
+
+static int si_populate_ulv_state(struct amdgpu_device *adev,
+				 SISLANDS_SMC_SWSTATE *state)
+{
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct si_ulv_param *ulv = &si_pi->ulv;
+	u32 sclk_in_sr = 1350; /* ??? */
+	int ret;
+
+	ret = si_convert_power_level_to_smc(adev, &ulv->pl,
+					    &state->levels[0]);
+	if (!ret) {
+		if (eg_pi->sclk_deep_sleep) {
+			if (sclk_in_sr <= SCLK_MIN_DEEPSLEEP_FREQ)
+				state->levels[0].stateFlags |= PPSMC_STATEFLAG_DEEPSLEEP_BYPASS;
+			else
+				state->levels[0].stateFlags |= PPSMC_STATEFLAG_DEEPSLEEP_THROTTLE;
+		}
+		if (ulv->one_pcie_lane_in_ulv)
+			state->flags |= PPSMC_SWSTATE_FLAG_PCIE_X1;
+		state->levels[0].arbRefreshState = (u8)(SISLANDS_ULV_STATE_ARB_INDEX);
+		state->levels[0].ACIndex = 1;
+		state->levels[0].std_vddc = state->levels[0].vddc;
+		state->levelCount = 1;
+
+		state->flags |= PPSMC_SWSTATE_FLAG_DC;
+	}
+
+	return ret;
+}
+
+static int si_program_ulv_memory_timing_parameters(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct si_ulv_param *ulv = &si_pi->ulv;
+	SMC_SIslands_MCArbDramTimingRegisterSet arb_regs = { 0 };
+	int ret;
+
+	ret = si_populate_memory_timing_parameters(adev, &ulv->pl,
+						   &arb_regs);
+	if (ret)
+		return ret;
+
+	si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_ulv_volt_change_delay,
+				   ulv->volt_change_delay);
+
+	ret = amdgpu_si_copy_bytes_to_smc(adev,
+					  si_pi->arb_table_start +
+					  offsetof(SMC_SIslands_MCArbDramTimingRegisters, data) +
+					  sizeof(SMC_SIslands_MCArbDramTimingRegisterSet) * SISLANDS_ULV_STATE_ARB_INDEX,
+					  (u8 *)&arb_regs,
+					  sizeof(SMC_SIslands_MCArbDramTimingRegisterSet),
+					  si_pi->sram_end);
+
+	return ret;
+}
+
+static void si_get_mvdd_configuration(struct amdgpu_device *adev)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+
+	pi->mvdd_split_frequency = 30000;
+}
+
+static int si_init_smc_table(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct amdgpu_ps *amdgpu_boot_state = adev->pm.dpm.boot_ps;
+	const struct si_ulv_param *ulv = &si_pi->ulv;
+	SISLANDS_SMC_STATETABLE  *table = &si_pi->smc_statetable;
+	int ret;
+	u32 lane_width;
+	u32 vr_hot_gpio;
+
+	si_populate_smc_voltage_tables(adev, table);
+
+	switch (adev->pm.int_thermal_type) {
+	case THERMAL_TYPE_SI:
+	case THERMAL_TYPE_EMC2103_WITH_INTERNAL:
+		table->thermalProtectType = PPSMC_THERMAL_PROTECT_TYPE_INTERNAL;
+		break;
+	case THERMAL_TYPE_NONE:
+		table->thermalProtectType = PPSMC_THERMAL_PROTECT_TYPE_NONE;
+		break;
+	default:
+		table->thermalProtectType = PPSMC_THERMAL_PROTECT_TYPE_EXTERNAL;
+		break;
+	}
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_HARDWAREDC)
+		table->systemFlags |= PPSMC_SYSTEMFLAG_GPIO_DC;
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_REGULATOR_HOT) {
+		if ((adev->pdev->device != 0x6818) && (adev->pdev->device != 0x6819))
+			table->systemFlags |= PPSMC_SYSTEMFLAG_REGULATOR_HOT;
+	}
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_STEPVDDC)
+		table->systemFlags |= PPSMC_SYSTEMFLAG_STEPVDDC;
+
+	if (adev->gmc.vram_type == AMDGPU_VRAM_TYPE_GDDR5)
+		table->systemFlags |= PPSMC_SYSTEMFLAG_GDDR5;
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_REVERT_GPIO5_POLARITY)
+		table->extraFlags |= PPSMC_EXTRAFLAGS_AC2DC_GPIO5_POLARITY_HIGH;
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_VRHOT_GPIO_CONFIGURABLE) {
+		table->systemFlags |= PPSMC_SYSTEMFLAG_REGULATOR_HOT_PROG_GPIO;
+		vr_hot_gpio = adev->pm.dpm.backbias_response_time;
+		si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_vr_hot_gpio,
+					   vr_hot_gpio);
+	}
+
+	ret = si_populate_smc_initial_state(adev, amdgpu_boot_state, table);
+	if (ret)
+		return ret;
+
+	ret = si_populate_smc_acpi_state(adev, table);
+	if (ret)
+		return ret;
+
+	table->driverState = table->initialState;
+
+	ret = si_do_program_memory_timing_parameters(adev, amdgpu_boot_state,
+						     SISLANDS_INITIAL_STATE_ARB_INDEX);
+	if (ret)
+		return ret;
+
+	if (ulv->supported && ulv->pl.vddc) {
+		ret = si_populate_ulv_state(adev, &table->ULVState);
+		if (ret)
+			return ret;
+
+		ret = si_program_ulv_memory_timing_parameters(adev);
+		if (ret)
+			return ret;
+
+		WREG32(CG_ULV_CONTROL, ulv->cg_ulv_control);
+		WREG32(CG_ULV_PARAMETER, ulv->cg_ulv_parameter);
+
+		lane_width = amdgpu_get_pcie_lanes(adev);
+		si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_non_ulv_pcie_link_width, lane_width);
+	} else {
+		table->ULVState = table->initialState;
+	}
+
+	return amdgpu_si_copy_bytes_to_smc(adev, si_pi->state_table_start,
+					   (u8 *)table, sizeof(SISLANDS_SMC_STATETABLE),
+					   si_pi->sram_end);
+}
+
+static int si_calculate_sclk_params(struct amdgpu_device *adev,
+				    u32 engine_clock,
+				    SISLANDS_SMC_SCLK_VALUE *sclk)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct atom_clock_dividers dividers;
+	u32 spll_func_cntl = si_pi->clock_registers.cg_spll_func_cntl;
+	u32 spll_func_cntl_2 = si_pi->clock_registers.cg_spll_func_cntl_2;
+	u32 spll_func_cntl_3 = si_pi->clock_registers.cg_spll_func_cntl_3;
+	u32 spll_func_cntl_4 = si_pi->clock_registers.cg_spll_func_cntl_4;
+	u32 cg_spll_spread_spectrum = si_pi->clock_registers.cg_spll_spread_spectrum;
+	u32 cg_spll_spread_spectrum_2 = si_pi->clock_registers.cg_spll_spread_spectrum_2;
+	u64 tmp;
+	u32 reference_clock = adev->clock.spll.reference_freq;
+	u32 reference_divider;
+	u32 fbdiv;
+	int ret;
+
+	ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+					     engine_clock, false, &dividers);
+	if (ret)
+		return ret;
+
+	reference_divider = 1 + dividers.ref_div;
+
+	tmp = (u64) engine_clock * reference_divider * dividers.post_div * 16384;
+	do_div(tmp, reference_clock);
+	fbdiv = (u32) tmp;
+
+	spll_func_cntl &= ~(SPLL_PDIV_A_MASK | SPLL_REF_DIV_MASK);
+	spll_func_cntl |= SPLL_REF_DIV(dividers.ref_div);
+	spll_func_cntl |= SPLL_PDIV_A(dividers.post_div);
+
+	spll_func_cntl_2 &= ~SCLK_MUX_SEL_MASK;
+	spll_func_cntl_2 |= SCLK_MUX_SEL(2);
+
+	spll_func_cntl_3 &= ~SPLL_FB_DIV_MASK;
+	spll_func_cntl_3 |= SPLL_FB_DIV(fbdiv);
+	spll_func_cntl_3 |= SPLL_DITHEN;
+
+	if (pi->sclk_ss) {
+		struct amdgpu_atom_ss ss;
+		u32 vco_freq = engine_clock * dividers.post_div;
+
+		if (amdgpu_atombios_get_asic_ss_info(adev, &ss,
+						     ASIC_INTERNAL_ENGINE_SS, vco_freq)) {
+			u32 clk_s = reference_clock * 5 / (reference_divider * ss.rate);
+			u32 clk_v = 4 * ss.percentage * fbdiv / (clk_s * 10000);
+
+			cg_spll_spread_spectrum &= ~CLK_S_MASK;
+			cg_spll_spread_spectrum |= CLK_S(clk_s);
+			cg_spll_spread_spectrum |= SSEN;
+
+			cg_spll_spread_spectrum_2 &= ~CLK_V_MASK;
+			cg_spll_spread_spectrum_2 |= CLK_V(clk_v);
+		}
+	}
+
+	sclk->sclk_value = engine_clock;
+	sclk->vCG_SPLL_FUNC_CNTL = spll_func_cntl;
+	sclk->vCG_SPLL_FUNC_CNTL_2 = spll_func_cntl_2;
+	sclk->vCG_SPLL_FUNC_CNTL_3 = spll_func_cntl_3;
+	sclk->vCG_SPLL_FUNC_CNTL_4 = spll_func_cntl_4;
+	sclk->vCG_SPLL_SPREAD_SPECTRUM = cg_spll_spread_spectrum;
+	sclk->vCG_SPLL_SPREAD_SPECTRUM_2 = cg_spll_spread_spectrum_2;
+
+	return 0;
+}
+
+static int si_populate_sclk_value(struct amdgpu_device *adev,
+				  u32 engine_clock,
+				  SISLANDS_SMC_SCLK_VALUE *sclk)
+{
+	SISLANDS_SMC_SCLK_VALUE sclk_tmp;
+	int ret;
+
+	ret = si_calculate_sclk_params(adev, engine_clock, &sclk_tmp);
+	if (!ret) {
+		sclk->sclk_value = cpu_to_be32(sclk_tmp.sclk_value);
+		sclk->vCG_SPLL_FUNC_CNTL = cpu_to_be32(sclk_tmp.vCG_SPLL_FUNC_CNTL);
+		sclk->vCG_SPLL_FUNC_CNTL_2 = cpu_to_be32(sclk_tmp.vCG_SPLL_FUNC_CNTL_2);
+		sclk->vCG_SPLL_FUNC_CNTL_3 = cpu_to_be32(sclk_tmp.vCG_SPLL_FUNC_CNTL_3);
+		sclk->vCG_SPLL_FUNC_CNTL_4 = cpu_to_be32(sclk_tmp.vCG_SPLL_FUNC_CNTL_4);
+		sclk->vCG_SPLL_SPREAD_SPECTRUM = cpu_to_be32(sclk_tmp.vCG_SPLL_SPREAD_SPECTRUM);
+		sclk->vCG_SPLL_SPREAD_SPECTRUM_2 = cpu_to_be32(sclk_tmp.vCG_SPLL_SPREAD_SPECTRUM_2);
+	}
+
+	return ret;
+}
+
+static int si_populate_mclk_value(struct amdgpu_device *adev,
+				  u32 engine_clock,
+				  u32 memory_clock,
+				  SISLANDS_SMC_MCLK_VALUE *mclk,
+				  bool strobe_mode,
+				  bool dll_state_on)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32  dll_cntl = si_pi->clock_registers.dll_cntl;
+	u32  mclk_pwrmgt_cntl = si_pi->clock_registers.mclk_pwrmgt_cntl;
+	u32  mpll_ad_func_cntl = si_pi->clock_registers.mpll_ad_func_cntl;
+	u32  mpll_dq_func_cntl = si_pi->clock_registers.mpll_dq_func_cntl;
+	u32  mpll_func_cntl = si_pi->clock_registers.mpll_func_cntl;
+	u32  mpll_func_cntl_1 = si_pi->clock_registers.mpll_func_cntl_1;
+	u32  mpll_func_cntl_2 = si_pi->clock_registers.mpll_func_cntl_2;
+	u32  mpll_ss1 = si_pi->clock_registers.mpll_ss1;
+	u32  mpll_ss2 = si_pi->clock_registers.mpll_ss2;
+	struct atom_mpll_param mpll_param;
+	int ret;
+
+	ret = amdgpu_atombios_get_memory_pll_dividers(adev, memory_clock, strobe_mode, &mpll_param);
+	if (ret)
+		return ret;
+
+	mpll_func_cntl &= ~BWCTRL_MASK;
+	mpll_func_cntl |= BWCTRL(mpll_param.bwcntl);
+
+	mpll_func_cntl_1 &= ~(CLKF_MASK | CLKFRAC_MASK | VCO_MODE_MASK);
+	mpll_func_cntl_1 |= CLKF(mpll_param.clkf) |
+		CLKFRAC(mpll_param.clkfrac) | VCO_MODE(mpll_param.vco_mode);
+
+	mpll_ad_func_cntl &= ~YCLK_POST_DIV_MASK;
+	mpll_ad_func_cntl |= YCLK_POST_DIV(mpll_param.post_div);
+
+	if (adev->gmc.vram_type == AMDGPU_VRAM_TYPE_GDDR5) {
+		mpll_dq_func_cntl &= ~(YCLK_SEL_MASK | YCLK_POST_DIV_MASK);
+		mpll_dq_func_cntl |= YCLK_SEL(mpll_param.yclk_sel) |
+			YCLK_POST_DIV(mpll_param.post_div);
+	}
+
+	if (pi->mclk_ss) {
+		struct amdgpu_atom_ss ss;
+		u32 freq_nom;
+		u32 tmp;
+		u32 reference_clock = adev->clock.mpll.reference_freq;
+
+		if (adev->gmc.vram_type == AMDGPU_VRAM_TYPE_GDDR5)
+			freq_nom = memory_clock * 4;
+		else
+			freq_nom = memory_clock * 2;
+
+		tmp = freq_nom / reference_clock;
+		tmp = tmp * tmp;
+		if (amdgpu_atombios_get_asic_ss_info(adev, &ss,
+		                                     ASIC_INTERNAL_MEMORY_SS, freq_nom)) {
+			u32 clks = reference_clock * 5 / ss.rate;
+			u32 clkv = (u32)((((131 * ss.percentage * ss.rate) / 100) * tmp) / freq_nom);
+
+		        mpll_ss1 &= ~CLKV_MASK;
+		        mpll_ss1 |= CLKV(clkv);
+
+		        mpll_ss2 &= ~CLKS_MASK;
+		        mpll_ss2 |= CLKS(clks);
+		}
+	}
+
+	mclk_pwrmgt_cntl &= ~DLL_SPEED_MASK;
+	mclk_pwrmgt_cntl |= DLL_SPEED(mpll_param.dll_speed);
+
+	if (dll_state_on)
+		mclk_pwrmgt_cntl |= MRDCK0_PDNB | MRDCK1_PDNB;
+	else
+		mclk_pwrmgt_cntl &= ~(MRDCK0_PDNB | MRDCK1_PDNB);
+
+	mclk->mclk_value = cpu_to_be32(memory_clock);
+	mclk->vMPLL_FUNC_CNTL = cpu_to_be32(mpll_func_cntl);
+	mclk->vMPLL_FUNC_CNTL_1 = cpu_to_be32(mpll_func_cntl_1);
+	mclk->vMPLL_FUNC_CNTL_2 = cpu_to_be32(mpll_func_cntl_2);
+	mclk->vMPLL_AD_FUNC_CNTL = cpu_to_be32(mpll_ad_func_cntl);
+	mclk->vMPLL_DQ_FUNC_CNTL = cpu_to_be32(mpll_dq_func_cntl);
+	mclk->vMCLK_PWRMGT_CNTL = cpu_to_be32(mclk_pwrmgt_cntl);
+	mclk->vDLL_CNTL = cpu_to_be32(dll_cntl);
+	mclk->vMPLL_SS = cpu_to_be32(mpll_ss1);
+	mclk->vMPLL_SS2 = cpu_to_be32(mpll_ss2);
+
+	return 0;
+}
+
+static void si_populate_smc_sp(struct amdgpu_device *adev,
+			       struct amdgpu_ps *amdgpu_state,
+			       SISLANDS_SMC_SWSTATE *smc_state)
+{
+	struct  si_ps *ps = si_get_ps(amdgpu_state);
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	int i;
+
+	for (i = 0; i < ps->performance_level_count - 1; i++)
+		smc_state->levels[i].bSP = cpu_to_be32(pi->dsp);
+
+	smc_state->levels[ps->performance_level_count - 1].bSP =
+		cpu_to_be32(pi->psp);
+}
+
+static int si_convert_power_level_to_smc(struct amdgpu_device *adev,
+					 struct rv7xx_pl *pl,
+					 SISLANDS_SMC_HW_PERFORMANCE_LEVEL *level)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	int ret;
+	bool dll_state_on;
+	u16 std_vddc;
+	bool gmc_pg = false;
+
+	if (eg_pi->pcie_performance_request &&
+	    (si_pi->force_pcie_gen != AMDGPU_PCIE_GEN_INVALID))
+		level->gen2PCIE = (u8)si_pi->force_pcie_gen;
+	else
+		level->gen2PCIE = (u8)pl->pcie_gen;
+
+	ret = si_populate_sclk_value(adev, pl->sclk, &level->sclk);
+	if (ret)
+		return ret;
+
+	level->mcFlags =  0;
+
+	if (pi->mclk_stutter_mode_threshold &&
+	    (pl->mclk <= pi->mclk_stutter_mode_threshold) &&
+	    !eg_pi->uvd_enabled &&
+	    (RREG32(DPG_PIPE_STUTTER_CONTROL) & STUTTER_ENABLE) &&
+	    (adev->pm.dpm.new_active_crtc_count <= 2)) {
+		level->mcFlags |= SISLANDS_SMC_MC_STUTTER_EN;
+
+		if (gmc_pg)
+			level->mcFlags |= SISLANDS_SMC_MC_PG_EN;
+	}
+
+	if (adev->gmc.vram_type == AMDGPU_VRAM_TYPE_GDDR5) {
+		if (pl->mclk > pi->mclk_edc_enable_threshold)
+			level->mcFlags |= SISLANDS_SMC_MC_EDC_RD_FLAG;
+
+		if (pl->mclk > eg_pi->mclk_edc_wr_enable_threshold)
+			level->mcFlags |= SISLANDS_SMC_MC_EDC_WR_FLAG;
+
+		level->strobeMode = si_get_strobe_mode_settings(adev, pl->mclk);
+
+		if (level->strobeMode & SISLANDS_SMC_STROBE_ENABLE) {
+			if (si_get_mclk_frequency_ratio(pl->mclk, true) >=
+			    ((RREG32(MC_SEQ_MISC7) >> 16) & 0xf))
+				dll_state_on = ((RREG32(MC_SEQ_MISC5) >> 1) & 0x1) ? true : false;
+			else
+				dll_state_on = ((RREG32(MC_SEQ_MISC6) >> 1) & 0x1) ? true : false;
+		} else {
+			dll_state_on = false;
+		}
+	} else {
+		level->strobeMode = si_get_strobe_mode_settings(adev,
+								pl->mclk);
+
+		dll_state_on = ((RREG32(MC_SEQ_MISC5) >> 1) & 0x1) ? true : false;
+	}
+
+	ret = si_populate_mclk_value(adev,
+				     pl->sclk,
+				     pl->mclk,
+				     &level->mclk,
+				     (level->strobeMode & SISLANDS_SMC_STROBE_ENABLE) != 0, dll_state_on);
+	if (ret)
+		return ret;
+
+	ret = si_populate_voltage_value(adev,
+					&eg_pi->vddc_voltage_table,
+					pl->vddc, &level->vddc);
+	if (ret)
+		return ret;
+
+
+	ret = si_get_std_voltage_value(adev, &level->vddc, &std_vddc);
+	if (ret)
+		return ret;
+
+	ret = si_populate_std_voltage_value(adev, std_vddc,
+					    level->vddc.index, &level->std_vddc);
+	if (ret)
+		return ret;
+
+	if (eg_pi->vddci_control) {
+		ret = si_populate_voltage_value(adev, &eg_pi->vddci_voltage_table,
+						pl->vddci, &level->vddci);
+		if (ret)
+			return ret;
+	}
+
+	if (si_pi->vddc_phase_shed_control) {
+		ret = si_populate_phase_shedding_value(adev,
+						       &adev->pm.dpm.dyn_state.phase_shedding_limits_table,
+						       pl->vddc,
+						       pl->sclk,
+						       pl->mclk,
+						       &level->vddc);
+		if (ret)
+			return ret;
+	}
+
+	level->MaxPoweredUpCU = si_pi->max_cu;
+
+	ret = si_populate_mvdd_value(adev, pl->mclk, &level->mvdd);
+
+	return ret;
+}
+
+static int si_populate_smc_t(struct amdgpu_device *adev,
+			     struct amdgpu_ps *amdgpu_state,
+			     SISLANDS_SMC_SWSTATE *smc_state)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct  si_ps *state = si_get_ps(amdgpu_state);
+	u32 a_t;
+	u32 t_l, t_h;
+	u32 high_bsp;
+	int i, ret;
+
+	if (state->performance_level_count >= 9)
+		return -EINVAL;
+
+	if (state->performance_level_count < 2) {
+		a_t = CG_R(0xffff) | CG_L(0);
+		smc_state->levels[0].aT = cpu_to_be32(a_t);
+		return 0;
+	}
+
+	smc_state->levels[0].aT = cpu_to_be32(0);
+
+	for (i = 0; i <= state->performance_level_count - 2; i++) {
+		ret = r600_calculate_at(
+			(50 / SISLANDS_MAX_HARDWARE_POWERLEVELS) * 100 * (i + 1),
+			100 * R600_AH_DFLT,
+			state->performance_levels[i + 1].sclk,
+			state->performance_levels[i].sclk,
+			&t_l,
+			&t_h);
+
+		if (ret) {
+			t_h = (i + 1) * 1000 - 50 * R600_AH_DFLT;
+			t_l = (i + 1) * 1000 + 50 * R600_AH_DFLT;
+		}
+
+		a_t = be32_to_cpu(smc_state->levels[i].aT) & ~CG_R_MASK;
+		a_t |= CG_R(t_l * pi->bsp / 20000);
+		smc_state->levels[i].aT = cpu_to_be32(a_t);
+
+		high_bsp = (i == state->performance_level_count - 2) ?
+			pi->pbsp : pi->bsp;
+		a_t = CG_R(0xffff) | CG_L(t_h * high_bsp / 20000);
+		smc_state->levels[i + 1].aT = cpu_to_be32(a_t);
+	}
+
+	return 0;
+}
+
+static int si_disable_ulv(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct si_ulv_param *ulv = &si_pi->ulv;
+
+	if (ulv->supported)
+		return (amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_DisableULV) == PPSMC_Result_OK) ?
+			0 : -EINVAL;
+
+	return 0;
+}
+
+static bool si_is_state_ulv_compatible(struct amdgpu_device *adev,
+				       struct amdgpu_ps *amdgpu_state)
+{
+	const struct si_power_info *si_pi = si_get_pi(adev);
+	const struct si_ulv_param *ulv = &si_pi->ulv;
+	const struct  si_ps *state = si_get_ps(amdgpu_state);
+	int i;
+
+	if (state->performance_levels[0].mclk != ulv->pl.mclk)
+		return false;
+
+	/* XXX validate against display requirements! */
+
+	for (i = 0; i < adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.count; i++) {
+		if (adev->clock.current_dispclk <=
+		    adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[i].clk) {
+			if (ulv->pl.vddc <
+			    adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[i].v)
+				return false;
+		}
+	}
+
+	if ((amdgpu_state->vclk != 0) || (amdgpu_state->dclk != 0))
+		return false;
+
+	return true;
+}
+
+static int si_set_power_state_conditionally_enable_ulv(struct amdgpu_device *adev,
+						       struct amdgpu_ps *amdgpu_new_state)
+{
+	const struct si_power_info *si_pi = si_get_pi(adev);
+	const struct si_ulv_param *ulv = &si_pi->ulv;
+
+	if (ulv->supported) {
+		if (si_is_state_ulv_compatible(adev, amdgpu_new_state))
+			return (amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_EnableULV) == PPSMC_Result_OK) ?
+				0 : -EINVAL;
+	}
+	return 0;
+}
+
+static int si_convert_power_state_to_smc(struct amdgpu_device *adev,
+					 struct amdgpu_ps *amdgpu_state,
+					 SISLANDS_SMC_SWSTATE *smc_state)
+{
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct ni_power_info *ni_pi = ni_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct  si_ps *state = si_get_ps(amdgpu_state);
+	int i, ret;
+	u32 threshold;
+	u32 sclk_in_sr = 1350; /* ??? */
+
+	if (state->performance_level_count > SISLANDS_MAX_HARDWARE_POWERLEVELS)
+		return -EINVAL;
+
+	threshold = state->performance_levels[state->performance_level_count-1].sclk * 100 / 100;
+
+	if (amdgpu_state->vclk && amdgpu_state->dclk) {
+		eg_pi->uvd_enabled = true;
+		if (eg_pi->smu_uvd_hs)
+			smc_state->flags |= PPSMC_SWSTATE_FLAG_UVD;
+	} else {
+		eg_pi->uvd_enabled = false;
+	}
+
+	if (state->dc_compatible)
+		smc_state->flags |= PPSMC_SWSTATE_FLAG_DC;
+
+	smc_state->levelCount = 0;
+	for (i = 0; i < state->performance_level_count; i++) {
+		if (eg_pi->sclk_deep_sleep) {
+			if ((i == 0) || si_pi->sclk_deep_sleep_above_low) {
+				if (sclk_in_sr <= SCLK_MIN_DEEPSLEEP_FREQ)
+					smc_state->levels[i].stateFlags |= PPSMC_STATEFLAG_DEEPSLEEP_BYPASS;
+				else
+					smc_state->levels[i].stateFlags |= PPSMC_STATEFLAG_DEEPSLEEP_THROTTLE;
+			}
+		}
+
+		ret = si_convert_power_level_to_smc(adev, &state->performance_levels[i],
+						    &smc_state->levels[i]);
+		smc_state->levels[i].arbRefreshState =
+			(u8)(SISLANDS_DRIVER_STATE_ARB_INDEX + i);
+
+		if (ret)
+			return ret;
+
+		if (ni_pi->enable_power_containment)
+			smc_state->levels[i].displayWatermark =
+				(state->performance_levels[i].sclk < threshold) ?
+				PPSMC_DISPLAY_WATERMARK_LOW : PPSMC_DISPLAY_WATERMARK_HIGH;
+		else
+			smc_state->levels[i].displayWatermark = (i < 2) ?
+				PPSMC_DISPLAY_WATERMARK_LOW : PPSMC_DISPLAY_WATERMARK_HIGH;
+
+		if (eg_pi->dynamic_ac_timing)
+			smc_state->levels[i].ACIndex = SISLANDS_MCREGISTERTABLE_FIRST_DRIVERSTATE_SLOT + i;
+		else
+			smc_state->levels[i].ACIndex = 0;
+
+		smc_state->levelCount++;
+	}
+
+	si_write_smc_soft_register(adev,
+				   SI_SMC_SOFT_REGISTER_watermark_threshold,
+				   threshold / 512);
+
+	si_populate_smc_sp(adev, amdgpu_state, smc_state);
+
+	ret = si_populate_power_containment_values(adev, amdgpu_state, smc_state);
+	if (ret)
+		ni_pi->enable_power_containment = false;
+
+	ret = si_populate_sq_ramping_values(adev, amdgpu_state, smc_state);
+	if (ret)
+		ni_pi->enable_sq_ramping = false;
+
+	return si_populate_smc_t(adev, amdgpu_state, smc_state);
+}
+
+static int si_upload_sw_state(struct amdgpu_device *adev,
+			      struct amdgpu_ps *amdgpu_new_state)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct  si_ps *new_state = si_get_ps(amdgpu_new_state);
+	int ret;
+	u32 address = si_pi->state_table_start +
+		offsetof(SISLANDS_SMC_STATETABLE, driverState);
+	u32 state_size = sizeof(SISLANDS_SMC_SWSTATE) +
+		((new_state->performance_level_count - 1) *
+		 sizeof(SISLANDS_SMC_HW_PERFORMANCE_LEVEL));
+	SISLANDS_SMC_SWSTATE *smc_state = &si_pi->smc_statetable.driverState;
+
+	memset(smc_state, 0, state_size);
+
+	ret = si_convert_power_state_to_smc(adev, amdgpu_new_state, smc_state);
+	if (ret)
+		return ret;
+
+	return amdgpu_si_copy_bytes_to_smc(adev, address, (u8 *)smc_state,
+					   state_size, si_pi->sram_end);
+}
+
+static int si_upload_ulv_state(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct si_ulv_param *ulv = &si_pi->ulv;
+	int ret = 0;
+
+	if (ulv->supported && ulv->pl.vddc) {
+		u32 address = si_pi->state_table_start +
+			offsetof(SISLANDS_SMC_STATETABLE, ULVState);
+		SISLANDS_SMC_SWSTATE *smc_state = &si_pi->smc_statetable.ULVState;
+		u32 state_size = sizeof(SISLANDS_SMC_SWSTATE);
+
+		memset(smc_state, 0, state_size);
+
+		ret = si_populate_ulv_state(adev, smc_state);
+		if (!ret)
+			ret = amdgpu_si_copy_bytes_to_smc(adev, address, (u8 *)smc_state,
+							  state_size, si_pi->sram_end);
+	}
+
+	return ret;
+}
+
+static int si_upload_smc_data(struct amdgpu_device *adev)
+{
+	struct amdgpu_crtc *amdgpu_crtc = NULL;
+	int i;
+
+	if (adev->pm.dpm.new_active_crtc_count == 0)
+		return 0;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->pm.dpm.new_active_crtcs & (1 << i)) {
+			amdgpu_crtc = adev->mode_info.crtcs[i];
+			break;
+		}
+	}
+
+	if (amdgpu_crtc == NULL)
+		return 0;
+
+	if (amdgpu_crtc->line_time <= 0)
+		return 0;
+
+	if (si_write_smc_soft_register(adev,
+				       SI_SMC_SOFT_REGISTER_crtc_index,
+				       amdgpu_crtc->crtc_id) != PPSMC_Result_OK)
+		return 0;
+
+	if (si_write_smc_soft_register(adev,
+				       SI_SMC_SOFT_REGISTER_mclk_change_block_cp_min,
+				       amdgpu_crtc->wm_high / amdgpu_crtc->line_time) != PPSMC_Result_OK)
+		return 0;
+
+	if (si_write_smc_soft_register(adev,
+				       SI_SMC_SOFT_REGISTER_mclk_change_block_cp_max,
+				       amdgpu_crtc->wm_low / amdgpu_crtc->line_time) != PPSMC_Result_OK)
+		return 0;
+
+	return 0;
+}
+
+static int si_set_mc_special_registers(struct amdgpu_device *adev,
+				       struct si_mc_reg_table *table)
+{
+	u8 i, j, k;
+	u32 temp_reg;
+
+	for (i = 0, j = table->last; i < table->last; i++) {
+		if (j >= SMC_SISLANDS_MC_REGISTER_ARRAY_SIZE)
+			return -EINVAL;
+		switch (table->mc_reg_address[i].s1) {
+		case MC_SEQ_MISC1:
+			temp_reg = RREG32(MC_PMG_CMD_EMRS);
+			table->mc_reg_address[j].s1 = MC_PMG_CMD_EMRS;
+			table->mc_reg_address[j].s0 = MC_SEQ_PMG_CMD_EMRS_LP;
+			for (k = 0; k < table->num_entries; k++)
+				table->mc_reg_table_entry[k].mc_data[j] =
+					((temp_reg & 0xffff0000)) |
+					((table->mc_reg_table_entry[k].mc_data[i] & 0xffff0000) >> 16);
+			j++;
+
+			if (j >= SMC_SISLANDS_MC_REGISTER_ARRAY_SIZE)
+				return -EINVAL;
+			temp_reg = RREG32(MC_PMG_CMD_MRS);
+			table->mc_reg_address[j].s1 = MC_PMG_CMD_MRS;
+			table->mc_reg_address[j].s0 = MC_SEQ_PMG_CMD_MRS_LP;
+			for (k = 0; k < table->num_entries; k++) {
+				table->mc_reg_table_entry[k].mc_data[j] =
+					(temp_reg & 0xffff0000) |
+					(table->mc_reg_table_entry[k].mc_data[i] & 0x0000ffff);
+				if (adev->gmc.vram_type != AMDGPU_VRAM_TYPE_GDDR5)
+					table->mc_reg_table_entry[k].mc_data[j] |= 0x100;
+			}
+			j++;
+
+			if (adev->gmc.vram_type != AMDGPU_VRAM_TYPE_GDDR5) {
+				if (j >= SMC_SISLANDS_MC_REGISTER_ARRAY_SIZE)
+					return -EINVAL;
+				table->mc_reg_address[j].s1 = MC_PMG_AUTO_CMD;
+				table->mc_reg_address[j].s0 = MC_PMG_AUTO_CMD;
+				for (k = 0; k < table->num_entries; k++)
+					table->mc_reg_table_entry[k].mc_data[j] =
+						(table->mc_reg_table_entry[k].mc_data[i] & 0xffff0000) >> 16;
+				j++;
+			}
+			break;
+		case MC_SEQ_RESERVE_M:
+			temp_reg = RREG32(MC_PMG_CMD_MRS1);
+			table->mc_reg_address[j].s1 = MC_PMG_CMD_MRS1;
+			table->mc_reg_address[j].s0 = MC_SEQ_PMG_CMD_MRS1_LP;
+			for(k = 0; k < table->num_entries; k++)
+				table->mc_reg_table_entry[k].mc_data[j] =
+					(temp_reg & 0xffff0000) |
+					(table->mc_reg_table_entry[k].mc_data[i] & 0x0000ffff);
+			j++;
+			break;
+		default:
+			break;
+		}
+	}
+
+	table->last = j;
+
+	return 0;
+}
+
+static bool si_check_s0_mc_reg_index(u16 in_reg, u16 *out_reg)
+{
+	bool result = true;
+	switch (in_reg) {
+	case  MC_SEQ_RAS_TIMING:
+		*out_reg = MC_SEQ_RAS_TIMING_LP;
+		break;
+	case MC_SEQ_CAS_TIMING:
+		*out_reg = MC_SEQ_CAS_TIMING_LP;
+		break;
+	case MC_SEQ_MISC_TIMING:
+		*out_reg = MC_SEQ_MISC_TIMING_LP;
+		break;
+	case MC_SEQ_MISC_TIMING2:
+		*out_reg = MC_SEQ_MISC_TIMING2_LP;
+		break;
+	case MC_SEQ_RD_CTL_D0:
+		*out_reg = MC_SEQ_RD_CTL_D0_LP;
+		break;
+	case MC_SEQ_RD_CTL_D1:
+		*out_reg = MC_SEQ_RD_CTL_D1_LP;
+		break;
+	case MC_SEQ_WR_CTL_D0:
+		*out_reg = MC_SEQ_WR_CTL_D0_LP;
+		break;
+	case MC_SEQ_WR_CTL_D1:
+		*out_reg = MC_SEQ_WR_CTL_D1_LP;
+		break;
+	case MC_PMG_CMD_EMRS:
+		*out_reg = MC_SEQ_PMG_CMD_EMRS_LP;
+		break;
+	case MC_PMG_CMD_MRS:
+		*out_reg = MC_SEQ_PMG_CMD_MRS_LP;
+		break;
+	case MC_PMG_CMD_MRS1:
+		*out_reg = MC_SEQ_PMG_CMD_MRS1_LP;
+		break;
+	case MC_SEQ_PMG_TIMING:
+		*out_reg = MC_SEQ_PMG_TIMING_LP;
+		break;
+	case MC_PMG_CMD_MRS2:
+		*out_reg = MC_SEQ_PMG_CMD_MRS2_LP;
+		break;
+	case MC_SEQ_WR_CTL_2:
+		*out_reg = MC_SEQ_WR_CTL_2_LP;
+		break;
+	default:
+		result = false;
+		break;
+	}
+
+	return result;
+}
+
+static void si_set_valid_flag(struct si_mc_reg_table *table)
+{
+	u8 i, j;
+
+	for (i = 0; i < table->last; i++) {
+		for (j = 1; j < table->num_entries; j++) {
+			if (table->mc_reg_table_entry[j-1].mc_data[i] != table->mc_reg_table_entry[j].mc_data[i]) {
+				table->valid_flag |= 1 << i;
+				break;
+			}
+		}
+	}
+}
+
+static void si_set_s0_mc_reg_index(struct si_mc_reg_table *table)
+{
+	u32 i;
+	u16 address;
+
+	for (i = 0; i < table->last; i++)
+		table->mc_reg_address[i].s0 = si_check_s0_mc_reg_index(table->mc_reg_address[i].s1, &address) ?
+			address : table->mc_reg_address[i].s1;
+
+}
+
+static int si_copy_vbios_mc_reg_table(struct atom_mc_reg_table *table,
+				      struct si_mc_reg_table *si_table)
+{
+	u8 i, j;
+
+	if (table->last > SMC_SISLANDS_MC_REGISTER_ARRAY_SIZE)
+		return -EINVAL;
+	if (table->num_entries > MAX_AC_TIMING_ENTRIES)
+		return -EINVAL;
+
+	for (i = 0; i < table->last; i++)
+		si_table->mc_reg_address[i].s1 = table->mc_reg_address[i].s1;
+	si_table->last = table->last;
+
+	for (i = 0; i < table->num_entries; i++) {
+		si_table->mc_reg_table_entry[i].mclk_max =
+			table->mc_reg_table_entry[i].mclk_max;
+		for (j = 0; j < table->last; j++) {
+			si_table->mc_reg_table_entry[i].mc_data[j] =
+				table->mc_reg_table_entry[i].mc_data[j];
+		}
+	}
+	si_table->num_entries = table->num_entries;
+
+	return 0;
+}
+
+static int si_initialize_mc_reg_table(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct atom_mc_reg_table *table;
+	struct si_mc_reg_table *si_table = &si_pi->mc_reg_table;
+	u8 module_index = rv770_get_memory_module_index(adev);
+	int ret;
+
+	table = kzalloc(sizeof(struct atom_mc_reg_table), GFP_KERNEL);
+	if (!table)
+		return -ENOMEM;
+
+	WREG32(MC_SEQ_RAS_TIMING_LP, RREG32(MC_SEQ_RAS_TIMING));
+	WREG32(MC_SEQ_CAS_TIMING_LP, RREG32(MC_SEQ_CAS_TIMING));
+	WREG32(MC_SEQ_MISC_TIMING_LP, RREG32(MC_SEQ_MISC_TIMING));
+	WREG32(MC_SEQ_MISC_TIMING2_LP, RREG32(MC_SEQ_MISC_TIMING2));
+	WREG32(MC_SEQ_PMG_CMD_EMRS_LP, RREG32(MC_PMG_CMD_EMRS));
+	WREG32(MC_SEQ_PMG_CMD_MRS_LP, RREG32(MC_PMG_CMD_MRS));
+	WREG32(MC_SEQ_PMG_CMD_MRS1_LP, RREG32(MC_PMG_CMD_MRS1));
+	WREG32(MC_SEQ_WR_CTL_D0_LP, RREG32(MC_SEQ_WR_CTL_D0));
+	WREG32(MC_SEQ_WR_CTL_D1_LP, RREG32(MC_SEQ_WR_CTL_D1));
+	WREG32(MC_SEQ_RD_CTL_D0_LP, RREG32(MC_SEQ_RD_CTL_D0));
+	WREG32(MC_SEQ_RD_CTL_D1_LP, RREG32(MC_SEQ_RD_CTL_D1));
+	WREG32(MC_SEQ_PMG_TIMING_LP, RREG32(MC_SEQ_PMG_TIMING));
+	WREG32(MC_SEQ_PMG_CMD_MRS2_LP, RREG32(MC_PMG_CMD_MRS2));
+	WREG32(MC_SEQ_WR_CTL_2_LP, RREG32(MC_SEQ_WR_CTL_2));
+
+	ret = amdgpu_atombios_init_mc_reg_table(adev, module_index, table);
+	if (ret)
+		goto init_mc_done;
+
+	ret = si_copy_vbios_mc_reg_table(table, si_table);
+	if (ret)
+		goto init_mc_done;
+
+	si_set_s0_mc_reg_index(si_table);
+
+	ret = si_set_mc_special_registers(adev, si_table);
+	if (ret)
+		goto init_mc_done;
+
+	si_set_valid_flag(si_table);
+
+init_mc_done:
+	kfree(table);
+
+	return ret;
+
+}
+
+static void si_populate_mc_reg_addresses(struct amdgpu_device *adev,
+					 SMC_SIslands_MCRegisters *mc_reg_table)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 i, j;
+
+	for (i = 0, j = 0; j < si_pi->mc_reg_table.last; j++) {
+		if (si_pi->mc_reg_table.valid_flag & (1 << j)) {
+			if (i >= SMC_SISLANDS_MC_REGISTER_ARRAY_SIZE)
+				break;
+			mc_reg_table->address[i].s0 =
+				cpu_to_be16(si_pi->mc_reg_table.mc_reg_address[j].s0);
+			mc_reg_table->address[i].s1 =
+				cpu_to_be16(si_pi->mc_reg_table.mc_reg_address[j].s1);
+			i++;
+		}
+	}
+	mc_reg_table->last = (u8)i;
+}
+
+static void si_convert_mc_registers(const struct si_mc_reg_entry *entry,
+				    SMC_SIslands_MCRegisterSet *data,
+				    u32 num_entries, u32 valid_flag)
+{
+	u32 i, j;
+
+	for(i = 0, j = 0; j < num_entries; j++) {
+		if (valid_flag & (1 << j)) {
+			data->value[i] = cpu_to_be32(entry->mc_data[j]);
+			i++;
+		}
+	}
+}
+
+static void si_convert_mc_reg_table_entry_to_smc(struct amdgpu_device *adev,
+						 struct rv7xx_pl *pl,
+						 SMC_SIslands_MCRegisterSet *mc_reg_table_data)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 i = 0;
+
+	for (i = 0; i < si_pi->mc_reg_table.num_entries; i++) {
+		if (pl->mclk <= si_pi->mc_reg_table.mc_reg_table_entry[i].mclk_max)
+			break;
+	}
+
+	if ((i == si_pi->mc_reg_table.num_entries) && (i > 0))
+		--i;
+
+	si_convert_mc_registers(&si_pi->mc_reg_table.mc_reg_table_entry[i],
+				mc_reg_table_data, si_pi->mc_reg_table.last,
+				si_pi->mc_reg_table.valid_flag);
+}
+
+static void si_convert_mc_reg_table_to_smc(struct amdgpu_device *adev,
+					   struct amdgpu_ps *amdgpu_state,
+					   SMC_SIslands_MCRegisters *mc_reg_table)
+{
+	struct si_ps *state = si_get_ps(amdgpu_state);
+	int i;
+
+	for (i = 0; i < state->performance_level_count; i++) {
+		si_convert_mc_reg_table_entry_to_smc(adev,
+						     &state->performance_levels[i],
+						     &mc_reg_table->data[SISLANDS_MCREGISTERTABLE_FIRST_DRIVERSTATE_SLOT + i]);
+	}
+}
+
+static int si_populate_mc_reg_table(struct amdgpu_device *adev,
+				    struct amdgpu_ps *amdgpu_boot_state)
+{
+	struct  si_ps *boot_state = si_get_ps(amdgpu_boot_state);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct si_ulv_param *ulv = &si_pi->ulv;
+	SMC_SIslands_MCRegisters *smc_mc_reg_table = &si_pi->smc_mc_reg_table;
+
+	memset(smc_mc_reg_table, 0, sizeof(SMC_SIslands_MCRegisters));
+
+	si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_seq_index, 1);
+
+	si_populate_mc_reg_addresses(adev, smc_mc_reg_table);
+
+	si_convert_mc_reg_table_entry_to_smc(adev, &boot_state->performance_levels[0],
+					     &smc_mc_reg_table->data[SISLANDS_MCREGISTERTABLE_INITIAL_SLOT]);
+
+	si_convert_mc_registers(&si_pi->mc_reg_table.mc_reg_table_entry[0],
+				&smc_mc_reg_table->data[SISLANDS_MCREGISTERTABLE_ACPI_SLOT],
+				si_pi->mc_reg_table.last,
+				si_pi->mc_reg_table.valid_flag);
+
+	if (ulv->supported && ulv->pl.vddc != 0)
+		si_convert_mc_reg_table_entry_to_smc(adev, &ulv->pl,
+						     &smc_mc_reg_table->data[SISLANDS_MCREGISTERTABLE_ULV_SLOT]);
+	else
+		si_convert_mc_registers(&si_pi->mc_reg_table.mc_reg_table_entry[0],
+					&smc_mc_reg_table->data[SISLANDS_MCREGISTERTABLE_ULV_SLOT],
+					si_pi->mc_reg_table.last,
+					si_pi->mc_reg_table.valid_flag);
+
+	si_convert_mc_reg_table_to_smc(adev, amdgpu_boot_state, smc_mc_reg_table);
+
+	return amdgpu_si_copy_bytes_to_smc(adev, si_pi->mc_reg_table_start,
+					   (u8 *)smc_mc_reg_table,
+					   sizeof(SMC_SIslands_MCRegisters), si_pi->sram_end);
+}
+
+static int si_upload_mc_reg_table(struct amdgpu_device *adev,
+				  struct amdgpu_ps *amdgpu_new_state)
+{
+	struct si_ps *new_state = si_get_ps(amdgpu_new_state);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 address = si_pi->mc_reg_table_start +
+		offsetof(SMC_SIslands_MCRegisters,
+			 data[SISLANDS_MCREGISTERTABLE_FIRST_DRIVERSTATE_SLOT]);
+	SMC_SIslands_MCRegisters *smc_mc_reg_table = &si_pi->smc_mc_reg_table;
+
+	memset(smc_mc_reg_table, 0, sizeof(SMC_SIslands_MCRegisters));
+
+	si_convert_mc_reg_table_to_smc(adev, amdgpu_new_state, smc_mc_reg_table);
+
+	return amdgpu_si_copy_bytes_to_smc(adev, address,
+					   (u8 *)&smc_mc_reg_table->data[SISLANDS_MCREGISTERTABLE_FIRST_DRIVERSTATE_SLOT],
+					   sizeof(SMC_SIslands_MCRegisterSet) * new_state->performance_level_count,
+					   si_pi->sram_end);
+}
+
+static void si_enable_voltage_control(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		WREG32_P(GENERAL_PWRMGT, VOLT_PWRMGT_EN, ~VOLT_PWRMGT_EN);
+	else
+		WREG32_P(GENERAL_PWRMGT, 0, ~VOLT_PWRMGT_EN);
+}
+
+static enum amdgpu_pcie_gen si_get_maximum_link_speed(struct amdgpu_device *adev,
+						      struct amdgpu_ps *amdgpu_state)
+{
+	struct si_ps *state = si_get_ps(amdgpu_state);
+	int i;
+	u16 pcie_speed, max_speed = 0;
+
+	for (i = 0; i < state->performance_level_count; i++) {
+		pcie_speed = state->performance_levels[i].pcie_gen;
+		if (max_speed < pcie_speed)
+			max_speed = pcie_speed;
+	}
+	return max_speed;
+}
+
+static u16 si_get_current_pcie_speed(struct amdgpu_device *adev)
+{
+	u32 speed_cntl;
+
+	speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL) & LC_CURRENT_DATA_RATE_MASK;
+	speed_cntl >>= LC_CURRENT_DATA_RATE_SHIFT;
+
+	return (u16)speed_cntl;
+}
+
+static void si_request_link_speed_change_before_state_change(struct amdgpu_device *adev,
+							     struct amdgpu_ps *amdgpu_new_state,
+							     struct amdgpu_ps *amdgpu_current_state)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	enum amdgpu_pcie_gen target_link_speed = si_get_maximum_link_speed(adev, amdgpu_new_state);
+	enum amdgpu_pcie_gen current_link_speed;
+
+	if (si_pi->force_pcie_gen == AMDGPU_PCIE_GEN_INVALID)
+		current_link_speed = si_get_maximum_link_speed(adev, amdgpu_current_state);
+	else
+		current_link_speed = si_pi->force_pcie_gen;
+
+	si_pi->force_pcie_gen = AMDGPU_PCIE_GEN_INVALID;
+	si_pi->pspp_notify_required = false;
+	if (target_link_speed > current_link_speed) {
+		switch (target_link_speed) {
+#if defined(CONFIG_ACPI)
+		case AMDGPU_PCIE_GEN3:
+			if (amdgpu_acpi_pcie_performance_request(adev, PCIE_PERF_REQ_PECI_GEN3, false) == 0)
+				break;
+			si_pi->force_pcie_gen = AMDGPU_PCIE_GEN2;
+			if (current_link_speed == AMDGPU_PCIE_GEN2)
+				break;
+		case AMDGPU_PCIE_GEN2:
+			if (amdgpu_acpi_pcie_performance_request(adev, PCIE_PERF_REQ_PECI_GEN2, false) == 0)
+				break;
+#endif
+		default:
+			si_pi->force_pcie_gen = si_get_current_pcie_speed(adev);
+			break;
+		}
+	} else {
+		if (target_link_speed < current_link_speed)
+			si_pi->pspp_notify_required = true;
+	}
+}
+
+static void si_notify_link_speed_change_after_state_change(struct amdgpu_device *adev,
+							   struct amdgpu_ps *amdgpu_new_state,
+							   struct amdgpu_ps *amdgpu_current_state)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	enum amdgpu_pcie_gen target_link_speed = si_get_maximum_link_speed(adev, amdgpu_new_state);
+	u8 request;
+
+	if (si_pi->pspp_notify_required) {
+		if (target_link_speed == AMDGPU_PCIE_GEN3)
+			request = PCIE_PERF_REQ_PECI_GEN3;
+		else if (target_link_speed == AMDGPU_PCIE_GEN2)
+			request = PCIE_PERF_REQ_PECI_GEN2;
+		else
+			request = PCIE_PERF_REQ_PECI_GEN1;
+
+		if ((request == PCIE_PERF_REQ_PECI_GEN1) &&
+		    (si_get_current_pcie_speed(adev) > 0))
+			return;
+
+#if defined(CONFIG_ACPI)
+		amdgpu_acpi_pcie_performance_request(adev, request, false);
+#endif
+	}
+}
+
+#if 0
+static int si_ds_request(struct amdgpu_device *adev,
+			 bool ds_status_on, u32 count_write)
+{
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+
+	if (eg_pi->sclk_deep_sleep) {
+		if (ds_status_on)
+			return (amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_CancelThrottleOVRDSCLKDS) ==
+				PPSMC_Result_OK) ?
+				0 : -EINVAL;
+		else
+			return (amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_ThrottleOVRDSCLKDS) ==
+				PPSMC_Result_OK) ? 0 : -EINVAL;
+	}
+	return 0;
+}
+#endif
+
+static void si_set_max_cu_value(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+
+	if (adev->asic_type == CHIP_VERDE) {
+		switch (adev->pdev->device) {
+		case 0x6820:
+		case 0x6825:
+		case 0x6821:
+		case 0x6823:
+		case 0x6827:
+			si_pi->max_cu = 10;
+			break;
+		case 0x682D:
+		case 0x6824:
+		case 0x682F:
+		case 0x6826:
+			si_pi->max_cu = 8;
+			break;
+		case 0x6828:
+		case 0x6830:
+		case 0x6831:
+		case 0x6838:
+		case 0x6839:
+		case 0x683D:
+			si_pi->max_cu = 10;
+			break;
+		case 0x683B:
+		case 0x683F:
+		case 0x6829:
+			si_pi->max_cu = 8;
+			break;
+		default:
+			si_pi->max_cu = 0;
+			break;
+		}
+	} else {
+		si_pi->max_cu = 0;
+	}
+}
+
+static int si_patch_single_dependency_table_based_on_leakage(struct amdgpu_device *adev,
+							     struct amdgpu_clock_voltage_dependency_table *table)
+{
+	u32 i;
+	int j;
+	u16 leakage_voltage;
+
+	if (table) {
+		for (i = 0; i < table->count; i++) {
+			switch (si_get_leakage_voltage_from_leakage_index(adev,
+									  table->entries[i].v,
+									  &leakage_voltage)) {
+			case 0:
+				table->entries[i].v = leakage_voltage;
+				break;
+			case -EAGAIN:
+				return -EINVAL;
+			case -EINVAL:
+			default:
+				break;
+			}
+		}
+
+		for (j = (table->count - 2); j >= 0; j--) {
+			table->entries[j].v = (table->entries[j].v <= table->entries[j + 1].v) ?
+				table->entries[j].v : table->entries[j + 1].v;
+		}
+	}
+	return 0;
+}
+
+static int si_patch_dependency_tables_based_on_leakage(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	ret = si_patch_single_dependency_table_based_on_leakage(adev,
+								&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk);
+	if (ret)
+		DRM_ERROR("Could not patch vddc_on_sclk leakage table\n");
+	ret = si_patch_single_dependency_table_based_on_leakage(adev,
+								&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk);
+	if (ret)
+		DRM_ERROR("Could not patch vddc_on_mclk leakage table\n");
+	ret = si_patch_single_dependency_table_based_on_leakage(adev,
+								&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk);
+	if (ret)
+		DRM_ERROR("Could not patch vddci_on_mclk leakage table\n");
+	return ret;
+}
+
+static void si_set_pcie_lane_width_in_smc(struct amdgpu_device *adev,
+					  struct amdgpu_ps *amdgpu_new_state,
+					  struct amdgpu_ps *amdgpu_current_state)
+{
+	u32 lane_width;
+	u32 new_lane_width =
+		((amdgpu_new_state->caps & ATOM_PPLIB_PCIE_LINK_WIDTH_MASK) >> ATOM_PPLIB_PCIE_LINK_WIDTH_SHIFT) + 1;
+	u32 current_lane_width =
+		((amdgpu_current_state->caps & ATOM_PPLIB_PCIE_LINK_WIDTH_MASK) >> ATOM_PPLIB_PCIE_LINK_WIDTH_SHIFT) + 1;
+
+	if (new_lane_width != current_lane_width) {
+		amdgpu_set_pcie_lanes(adev, new_lane_width);
+		lane_width = amdgpu_get_pcie_lanes(adev);
+		si_write_smc_soft_register(adev, SI_SMC_SOFT_REGISTER_non_ulv_pcie_link_width, lane_width);
+	}
+}
+
+static void si_dpm_setup_asic(struct amdgpu_device *adev)
+{
+	si_read_clock_registers(adev);
+	si_enable_acpi_power_management(adev);
+}
+
+static int si_thermal_enable_alert(struct amdgpu_device *adev,
+				   bool enable)
+{
+	u32 thermal_int = RREG32(CG_THERMAL_INT);
+
+	if (enable) {
+		PPSMC_Result result;
+
+		thermal_int &= ~(THERM_INT_MASK_HIGH | THERM_INT_MASK_LOW);
+		WREG32(CG_THERMAL_INT, thermal_int);
+		result = amdgpu_si_send_msg_to_smc(adev, PPSMC_MSG_EnableThermalInterrupt);
+		if (result != PPSMC_Result_OK) {
+			DRM_DEBUG_KMS("Could not enable thermal interrupts.\n");
+			return -EINVAL;
+		}
+	} else {
+		thermal_int |= THERM_INT_MASK_HIGH | THERM_INT_MASK_LOW;
+		WREG32(CG_THERMAL_INT, thermal_int);
+	}
+
+	return 0;
+}
+
+static int si_thermal_set_temperature_range(struct amdgpu_device *adev,
+					    int min_temp, int max_temp)
+{
+	int low_temp = 0 * 1000;
+	int high_temp = 255 * 1000;
+
+	if (low_temp < min_temp)
+		low_temp = min_temp;
+	if (high_temp > max_temp)
+		high_temp = max_temp;
+	if (high_temp < low_temp) {
+		DRM_ERROR("invalid thermal range: %d - %d\n", low_temp, high_temp);
+		return -EINVAL;
+	}
+
+	WREG32_P(CG_THERMAL_INT, DIG_THERM_INTH(high_temp / 1000), ~DIG_THERM_INTH_MASK);
+	WREG32_P(CG_THERMAL_INT, DIG_THERM_INTL(low_temp / 1000), ~DIG_THERM_INTL_MASK);
+	WREG32_P(CG_THERMAL_CTRL, DIG_THERM_DPM(high_temp / 1000), ~DIG_THERM_DPM_MASK);
+
+	adev->pm.dpm.thermal.min_temp = low_temp;
+	adev->pm.dpm.thermal.max_temp = high_temp;
+
+	return 0;
+}
+
+static void si_fan_ctrl_set_static_mode(struct amdgpu_device *adev, u32 mode)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 tmp;
+
+	if (si_pi->fan_ctrl_is_in_default_mode) {
+		tmp = (RREG32(CG_FDO_CTRL2) & FDO_PWM_MODE_MASK) >> FDO_PWM_MODE_SHIFT;
+		si_pi->fan_ctrl_default_mode = tmp;
+		tmp = (RREG32(CG_FDO_CTRL2) & TMIN_MASK) >> TMIN_SHIFT;
+		si_pi->t_min = tmp;
+		si_pi->fan_ctrl_is_in_default_mode = false;
+	}
+
+	tmp = RREG32(CG_FDO_CTRL2) & ~TMIN_MASK;
+	tmp |= TMIN(0);
+	WREG32(CG_FDO_CTRL2, tmp);
+
+	tmp = RREG32(CG_FDO_CTRL2) & ~FDO_PWM_MODE_MASK;
+	tmp |= FDO_PWM_MODE(mode);
+	WREG32(CG_FDO_CTRL2, tmp);
+}
+
+static int si_thermal_setup_fan_table(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	PP_SIslands_FanTable fan_table = { FDO_MODE_HARDWARE };
+	u32 duty100;
+	u32 t_diff1, t_diff2, pwm_diff1, pwm_diff2;
+	u16 fdo_min, slope1, slope2;
+	u32 reference_clock, tmp;
+	int ret;
+	u64 tmp64;
+
+	if (!si_pi->fan_table_start) {
+		adev->pm.dpm.fan.ucode_fan_control = false;
+		return 0;
+	}
+
+	duty100 = (RREG32(CG_FDO_CTRL1) & FMAX_DUTY100_MASK) >> FMAX_DUTY100_SHIFT;
+
+	if (duty100 == 0) {
+		adev->pm.dpm.fan.ucode_fan_control = false;
+		return 0;
+	}
+
+	tmp64 = (u64)adev->pm.dpm.fan.pwm_min * duty100;
+	do_div(tmp64, 10000);
+	fdo_min = (u16)tmp64;
+
+	t_diff1 = adev->pm.dpm.fan.t_med - adev->pm.dpm.fan.t_min;
+	t_diff2 = adev->pm.dpm.fan.t_high - adev->pm.dpm.fan.t_med;
+
+	pwm_diff1 = adev->pm.dpm.fan.pwm_med - adev->pm.dpm.fan.pwm_min;
+	pwm_diff2 = adev->pm.dpm.fan.pwm_high - adev->pm.dpm.fan.pwm_med;
+
+	slope1 = (u16)((50 + ((16 * duty100 * pwm_diff1) / t_diff1)) / 100);
+	slope2 = (u16)((50 + ((16 * duty100 * pwm_diff2) / t_diff2)) / 100);
+
+	fan_table.temp_min = cpu_to_be16((50 + adev->pm.dpm.fan.t_min) / 100);
+	fan_table.temp_med = cpu_to_be16((50 + adev->pm.dpm.fan.t_med) / 100);
+	fan_table.temp_max = cpu_to_be16((50 + adev->pm.dpm.fan.t_max) / 100);
+	fan_table.slope1 = cpu_to_be16(slope1);
+	fan_table.slope2 = cpu_to_be16(slope2);
+	fan_table.fdo_min = cpu_to_be16(fdo_min);
+	fan_table.hys_down = cpu_to_be16(adev->pm.dpm.fan.t_hyst);
+	fan_table.hys_up = cpu_to_be16(1);
+	fan_table.hys_slope = cpu_to_be16(1);
+	fan_table.temp_resp_lim = cpu_to_be16(5);
+	reference_clock = amdgpu_asic_get_xclk(adev);
+
+	fan_table.refresh_period = cpu_to_be32((adev->pm.dpm.fan.cycle_delay *
+						reference_clock) / 1600);
+	fan_table.fdo_max = cpu_to_be16((u16)duty100);
+
+	tmp = (RREG32(CG_MULT_THERMAL_CTRL) & TEMP_SEL_MASK) >> TEMP_SEL_SHIFT;
+	fan_table.temp_src = (uint8_t)tmp;
+
+	ret = amdgpu_si_copy_bytes_to_smc(adev,
+					  si_pi->fan_table_start,
+					  (u8 *)(&fan_table),
+					  sizeof(fan_table),
+					  si_pi->sram_end);
+
+	if (ret) {
+		DRM_ERROR("Failed to load fan table to the SMC.");
+		adev->pm.dpm.fan.ucode_fan_control = false;
+	}
+
+	return ret;
+}
+
+static int si_fan_ctrl_start_smc_fan_control(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	PPSMC_Result ret;
+
+	ret = amdgpu_si_send_msg_to_smc(adev, PPSMC_StartFanControl);
+	if (ret == PPSMC_Result_OK) {
+		si_pi->fan_is_controlled_by_smc = true;
+		return 0;
+	} else {
+		return -EINVAL;
+	}
+}
+
+static int si_fan_ctrl_stop_smc_fan_control(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	PPSMC_Result ret;
+
+	ret = amdgpu_si_send_msg_to_smc(adev, PPSMC_StopFanControl);
+
+	if (ret == PPSMC_Result_OK) {
+		si_pi->fan_is_controlled_by_smc = false;
+		return 0;
+	} else {
+		return -EINVAL;
+	}
+}
+
+static int si_dpm_get_fan_speed_percent(void *handle,
+				      u32 *speed)
+{
+	u32 duty, duty100;
+	u64 tmp64;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.no_fan)
+		return -ENOENT;
+
+	duty100 = (RREG32(CG_FDO_CTRL1) & FMAX_DUTY100_MASK) >> FMAX_DUTY100_SHIFT;
+	duty = (RREG32(CG_THERMAL_STATUS) & FDO_PWM_DUTY_MASK) >> FDO_PWM_DUTY_SHIFT;
+
+	if (duty100 == 0)
+		return -EINVAL;
+
+	tmp64 = (u64)duty * 100;
+	do_div(tmp64, duty100);
+	*speed = (u32)tmp64;
+
+	if (*speed > 100)
+		*speed = 100;
+
+	return 0;
+}
+
+static int si_dpm_set_fan_speed_percent(void *handle,
+				      u32 speed)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 tmp;
+	u32 duty, duty100;
+	u64 tmp64;
+
+	if (adev->pm.no_fan)
+		return -ENOENT;
+
+	if (si_pi->fan_is_controlled_by_smc)
+		return -EINVAL;
+
+	if (speed > 100)
+		return -EINVAL;
+
+	duty100 = (RREG32(CG_FDO_CTRL1) & FMAX_DUTY100_MASK) >> FMAX_DUTY100_SHIFT;
+
+	if (duty100 == 0)
+		return -EINVAL;
+
+	tmp64 = (u64)speed * duty100;
+	do_div(tmp64, 100);
+	duty = (u32)tmp64;
+
+	tmp = RREG32(CG_FDO_CTRL0) & ~FDO_STATIC_DUTY_MASK;
+	tmp |= FDO_STATIC_DUTY(duty);
+	WREG32(CG_FDO_CTRL0, tmp);
+
+	return 0;
+}
+
+static void si_dpm_set_fan_control_mode(void *handle, u32 mode)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (mode) {
+		/* stop auto-manage */
+		if (adev->pm.dpm.fan.ucode_fan_control)
+			si_fan_ctrl_stop_smc_fan_control(adev);
+		si_fan_ctrl_set_static_mode(adev, mode);
+	} else {
+		/* restart auto-manage */
+		if (adev->pm.dpm.fan.ucode_fan_control)
+			si_thermal_start_smc_fan_control(adev);
+		else
+			si_fan_ctrl_set_default_mode(adev);
+	}
+}
+
+static u32 si_dpm_get_fan_control_mode(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 tmp;
+
+	if (si_pi->fan_is_controlled_by_smc)
+		return 0;
+
+	tmp = RREG32(CG_FDO_CTRL2) & FDO_PWM_MODE_MASK;
+	return (tmp >> FDO_PWM_MODE_SHIFT);
+}
+
+#if 0
+static int si_fan_ctrl_get_fan_speed_rpm(struct amdgpu_device *adev,
+					 u32 *speed)
+{
+	u32 tach_period;
+	u32 xclk = amdgpu_asic_get_xclk(adev);
+
+	if (adev->pm.no_fan)
+		return -ENOENT;
+
+	if (adev->pm.fan_pulses_per_revolution == 0)
+		return -ENOENT;
+
+	tach_period = (RREG32(CG_TACH_STATUS) & TACH_PERIOD_MASK) >> TACH_PERIOD_SHIFT;
+	if (tach_period == 0)
+		return -ENOENT;
+
+	*speed = 60 * xclk * 10000 / tach_period;
+
+	return 0;
+}
+
+static int si_fan_ctrl_set_fan_speed_rpm(struct amdgpu_device *adev,
+					 u32 speed)
+{
+	u32 tach_period, tmp;
+	u32 xclk = amdgpu_asic_get_xclk(adev);
+
+	if (adev->pm.no_fan)
+		return -ENOENT;
+
+	if (adev->pm.fan_pulses_per_revolution == 0)
+		return -ENOENT;
+
+	if ((speed < adev->pm.fan_min_rpm) ||
+	    (speed > adev->pm.fan_max_rpm))
+		return -EINVAL;
+
+	if (adev->pm.dpm.fan.ucode_fan_control)
+		si_fan_ctrl_stop_smc_fan_control(adev);
+
+	tach_period = 60 * xclk * 10000 / (8 * speed);
+	tmp = RREG32(CG_TACH_CTRL) & ~TARGET_PERIOD_MASK;
+	tmp |= TARGET_PERIOD(tach_period);
+	WREG32(CG_TACH_CTRL, tmp);
+
+	si_fan_ctrl_set_static_mode(adev, FDO_PWM_MODE_STATIC_RPM);
+
+	return 0;
+}
+#endif
+
+static void si_fan_ctrl_set_default_mode(struct amdgpu_device *adev)
+{
+	struct si_power_info *si_pi = si_get_pi(adev);
+	u32 tmp;
+
+	if (!si_pi->fan_ctrl_is_in_default_mode) {
+		tmp = RREG32(CG_FDO_CTRL2) & ~FDO_PWM_MODE_MASK;
+		tmp |= FDO_PWM_MODE(si_pi->fan_ctrl_default_mode);
+		WREG32(CG_FDO_CTRL2, tmp);
+
+		tmp = RREG32(CG_FDO_CTRL2) & ~TMIN_MASK;
+		tmp |= TMIN(si_pi->t_min);
+		WREG32(CG_FDO_CTRL2, tmp);
+		si_pi->fan_ctrl_is_in_default_mode = true;
+	}
+}
+
+static void si_thermal_start_smc_fan_control(struct amdgpu_device *adev)
+{
+	if (adev->pm.dpm.fan.ucode_fan_control) {
+		si_fan_ctrl_start_smc_fan_control(adev);
+		si_fan_ctrl_set_static_mode(adev, FDO_PWM_MODE_STATIC);
+	}
+}
+
+static void si_thermal_initialize(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	if (adev->pm.fan_pulses_per_revolution) {
+		tmp = RREG32(CG_TACH_CTRL) & ~EDGE_PER_REV_MASK;
+		tmp |= EDGE_PER_REV(adev->pm.fan_pulses_per_revolution -1);
+		WREG32(CG_TACH_CTRL, tmp);
+	}
+
+	tmp = RREG32(CG_FDO_CTRL2) & ~TACH_PWM_RESP_RATE_MASK;
+	tmp |= TACH_PWM_RESP_RATE(0x28);
+	WREG32(CG_FDO_CTRL2, tmp);
+}
+
+static int si_thermal_start_thermal_controller(struct amdgpu_device *adev)
+{
+	int ret;
+
+	si_thermal_initialize(adev);
+	ret = si_thermal_set_temperature_range(adev, R600_TEMP_RANGE_MIN, R600_TEMP_RANGE_MAX);
+	if (ret)
+		return ret;
+	ret = si_thermal_enable_alert(adev, true);
+	if (ret)
+		return ret;
+	if (adev->pm.dpm.fan.ucode_fan_control) {
+		ret = si_halt_smc(adev);
+		if (ret)
+			return ret;
+		ret = si_thermal_setup_fan_table(adev);
+		if (ret)
+			return ret;
+		ret = si_resume_smc(adev);
+		if (ret)
+			return ret;
+		si_thermal_start_smc_fan_control(adev);
+	}
+
+	return 0;
+}
+
+static void si_thermal_stop_thermal_controller(struct amdgpu_device *adev)
+{
+	if (!adev->pm.no_fan) {
+		si_fan_ctrl_set_default_mode(adev);
+		si_fan_ctrl_stop_smc_fan_control(adev);
+	}
+}
+
+static int si_dpm_enable(struct amdgpu_device *adev)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct amdgpu_ps *boot_ps = adev->pm.dpm.boot_ps;
+	int ret;
+
+	if (amdgpu_si_is_smc_running(adev))
+		return -EINVAL;
+	if (pi->voltage_control || si_pi->voltage_control_svi2)
+		si_enable_voltage_control(adev, true);
+	if (pi->mvdd_control)
+		si_get_mvdd_configuration(adev);
+	if (pi->voltage_control || si_pi->voltage_control_svi2) {
+		ret = si_construct_voltage_tables(adev);
+		if (ret) {
+			DRM_ERROR("si_construct_voltage_tables failed\n");
+			return ret;
+		}
+	}
+	if (eg_pi->dynamic_ac_timing) {
+		ret = si_initialize_mc_reg_table(adev);
+		if (ret)
+			eg_pi->dynamic_ac_timing = false;
+	}
+	if (pi->dynamic_ss)
+		si_enable_spread_spectrum(adev, true);
+	if (pi->thermal_protection)
+		si_enable_thermal_protection(adev, true);
+	si_setup_bsp(adev);
+	si_program_git(adev);
+	si_program_tp(adev);
+	si_program_tpp(adev);
+	si_program_sstp(adev);
+	si_enable_display_gap(adev);
+	si_program_vc(adev);
+	ret = si_upload_firmware(adev);
+	if (ret) {
+		DRM_ERROR("si_upload_firmware failed\n");
+		return ret;
+	}
+	ret = si_process_firmware_header(adev);
+	if (ret) {
+		DRM_ERROR("si_process_firmware_header failed\n");
+		return ret;
+	}
+	ret = si_initial_switch_from_arb_f0_to_f1(adev);
+	if (ret) {
+		DRM_ERROR("si_initial_switch_from_arb_f0_to_f1 failed\n");
+		return ret;
+	}
+	ret = si_init_smc_table(adev);
+	if (ret) {
+		DRM_ERROR("si_init_smc_table failed\n");
+		return ret;
+	}
+	ret = si_init_smc_spll_table(adev);
+	if (ret) {
+		DRM_ERROR("si_init_smc_spll_table failed\n");
+		return ret;
+	}
+	ret = si_init_arb_table_index(adev);
+	if (ret) {
+		DRM_ERROR("si_init_arb_table_index failed\n");
+		return ret;
+	}
+	if (eg_pi->dynamic_ac_timing) {
+		ret = si_populate_mc_reg_table(adev, boot_ps);
+		if (ret) {
+			DRM_ERROR("si_populate_mc_reg_table failed\n");
+			return ret;
+		}
+	}
+	ret = si_initialize_smc_cac_tables(adev);
+	if (ret) {
+		DRM_ERROR("si_initialize_smc_cac_tables failed\n");
+		return ret;
+	}
+	ret = si_initialize_hardware_cac_manager(adev);
+	if (ret) {
+		DRM_ERROR("si_initialize_hardware_cac_manager failed\n");
+		return ret;
+	}
+	ret = si_initialize_smc_dte_tables(adev);
+	if (ret) {
+		DRM_ERROR("si_initialize_smc_dte_tables failed\n");
+		return ret;
+	}
+	ret = si_populate_smc_tdp_limits(adev, boot_ps);
+	if (ret) {
+		DRM_ERROR("si_populate_smc_tdp_limits failed\n");
+		return ret;
+	}
+	ret = si_populate_smc_tdp_limits_2(adev, boot_ps);
+	if (ret) {
+		DRM_ERROR("si_populate_smc_tdp_limits_2 failed\n");
+		return ret;
+	}
+	si_program_response_times(adev);
+	si_program_ds_registers(adev);
+	si_dpm_start_smc(adev);
+	ret = si_notify_smc_display_change(adev, false);
+	if (ret) {
+		DRM_ERROR("si_notify_smc_display_change failed\n");
+		return ret;
+	}
+	si_enable_sclk_control(adev, true);
+	si_start_dpm(adev);
+
+	si_enable_auto_throttle_source(adev, AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL, true);
+	si_thermal_start_thermal_controller(adev);
+
+	return 0;
+}
+
+static int si_set_temperature_range(struct amdgpu_device *adev)
+{
+	int ret;
+
+	ret = si_thermal_enable_alert(adev, false);
+	if (ret)
+		return ret;
+	ret = si_thermal_set_temperature_range(adev, R600_TEMP_RANGE_MIN, R600_TEMP_RANGE_MAX);
+	if (ret)
+		return ret;
+	ret = si_thermal_enable_alert(adev, true);
+	if (ret)
+		return ret;
+
+	return ret;
+}
+
+static void si_dpm_disable(struct amdgpu_device *adev)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct amdgpu_ps *boot_ps = adev->pm.dpm.boot_ps;
+
+	if (!amdgpu_si_is_smc_running(adev))
+		return;
+	si_thermal_stop_thermal_controller(adev);
+	si_disable_ulv(adev);
+	si_clear_vc(adev);
+	if (pi->thermal_protection)
+		si_enable_thermal_protection(adev, false);
+	si_enable_power_containment(adev, boot_ps, false);
+	si_enable_smc_cac(adev, boot_ps, false);
+	si_enable_spread_spectrum(adev, false);
+	si_enable_auto_throttle_source(adev, AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL, false);
+	si_stop_dpm(adev);
+	si_reset_to_default(adev);
+	si_dpm_stop_smc(adev);
+	si_force_switch_to_arb_f0(adev);
+
+	ni_update_current_ps(adev, boot_ps);
+}
+
+static int si_dpm_pre_set_power_state(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct amdgpu_ps requested_ps = *adev->pm.dpm.requested_ps;
+	struct amdgpu_ps *new_ps = &requested_ps;
+
+	ni_update_requested_ps(adev, new_ps);
+	si_apply_state_adjust_rules(adev, &eg_pi->requested_rps);
+
+	return 0;
+}
+
+static int si_power_control_set_level(struct amdgpu_device *adev)
+{
+	struct amdgpu_ps *new_ps = adev->pm.dpm.requested_ps;
+	int ret;
+
+	ret = si_restrict_performance_levels_before_switch(adev);
+	if (ret)
+		return ret;
+	ret = si_halt_smc(adev);
+	if (ret)
+		return ret;
+	ret = si_populate_smc_tdp_limits(adev, new_ps);
+	if (ret)
+		return ret;
+	ret = si_populate_smc_tdp_limits_2(adev, new_ps);
+	if (ret)
+		return ret;
+	ret = si_resume_smc(adev);
+	if (ret)
+		return ret;
+	ret = si_set_sw_state(adev);
+	if (ret)
+		return ret;
+	return 0;
+}
+
+static int si_dpm_set_power_state(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct amdgpu_ps *new_ps = &eg_pi->requested_rps;
+	struct amdgpu_ps *old_ps = &eg_pi->current_rps;
+	int ret;
+
+	ret = si_disable_ulv(adev);
+	if (ret) {
+		DRM_ERROR("si_disable_ulv failed\n");
+		return ret;
+	}
+	ret = si_restrict_performance_levels_before_switch(adev);
+	if (ret) {
+		DRM_ERROR("si_restrict_performance_levels_before_switch failed\n");
+		return ret;
+	}
+	if (eg_pi->pcie_performance_request)
+		si_request_link_speed_change_before_state_change(adev, new_ps, old_ps);
+	ni_set_uvd_clock_before_set_eng_clock(adev, new_ps, old_ps);
+	ret = si_enable_power_containment(adev, new_ps, false);
+	if (ret) {
+		DRM_ERROR("si_enable_power_containment failed\n");
+		return ret;
+	}
+	ret = si_enable_smc_cac(adev, new_ps, false);
+	if (ret) {
+		DRM_ERROR("si_enable_smc_cac failed\n");
+		return ret;
+	}
+	ret = si_halt_smc(adev);
+	if (ret) {
+		DRM_ERROR("si_halt_smc failed\n");
+		return ret;
+	}
+	ret = si_upload_sw_state(adev, new_ps);
+	if (ret) {
+		DRM_ERROR("si_upload_sw_state failed\n");
+		return ret;
+	}
+	ret = si_upload_smc_data(adev);
+	if (ret) {
+		DRM_ERROR("si_upload_smc_data failed\n");
+		return ret;
+	}
+	ret = si_upload_ulv_state(adev);
+	if (ret) {
+		DRM_ERROR("si_upload_ulv_state failed\n");
+		return ret;
+	}
+	if (eg_pi->dynamic_ac_timing) {
+		ret = si_upload_mc_reg_table(adev, new_ps);
+		if (ret) {
+			DRM_ERROR("si_upload_mc_reg_table failed\n");
+			return ret;
+		}
+	}
+	ret = si_program_memory_timing_parameters(adev, new_ps);
+	if (ret) {
+		DRM_ERROR("si_program_memory_timing_parameters failed\n");
+		return ret;
+	}
+	si_set_pcie_lane_width_in_smc(adev, new_ps, old_ps);
+
+	ret = si_resume_smc(adev);
+	if (ret) {
+		DRM_ERROR("si_resume_smc failed\n");
+		return ret;
+	}
+	ret = si_set_sw_state(adev);
+	if (ret) {
+		DRM_ERROR("si_set_sw_state failed\n");
+		return ret;
+	}
+	ni_set_uvd_clock_after_set_eng_clock(adev, new_ps, old_ps);
+	if (eg_pi->pcie_performance_request)
+		si_notify_link_speed_change_after_state_change(adev, new_ps, old_ps);
+	ret = si_set_power_state_conditionally_enable_ulv(adev, new_ps);
+	if (ret) {
+		DRM_ERROR("si_set_power_state_conditionally_enable_ulv failed\n");
+		return ret;
+	}
+	ret = si_enable_smc_cac(adev, new_ps, true);
+	if (ret) {
+		DRM_ERROR("si_enable_smc_cac failed\n");
+		return ret;
+	}
+	ret = si_enable_power_containment(adev, new_ps, true);
+	if (ret) {
+		DRM_ERROR("si_enable_power_containment failed\n");
+		return ret;
+	}
+
+	ret = si_power_control_set_level(adev);
+	if (ret) {
+		DRM_ERROR("si_power_control_set_level failed\n");
+		return ret;
+	}
+
+	return 0;
+}
+
+static void si_dpm_post_set_power_state(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct amdgpu_ps *new_ps = &eg_pi->requested_rps;
+
+	ni_update_current_ps(adev, new_ps);
+}
+
+#if 0
+void si_dpm_reset_asic(struct amdgpu_device *adev)
+{
+	si_restrict_performance_levels_before_switch(adev);
+	si_disable_ulv(adev);
+	si_set_boot_state(adev);
+}
+#endif
+
+static void si_dpm_display_configuration_changed(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	si_program_display_gap(adev);
+}
+
+
+static void si_parse_pplib_non_clock_info(struct amdgpu_device *adev,
+					  struct amdgpu_ps *rps,
+					  struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info,
+					  u8 table_rev)
+{
+	rps->caps = le32_to_cpu(non_clock_info->ulCapsAndSettings);
+	rps->class = le16_to_cpu(non_clock_info->usClassification);
+	rps->class2 = le16_to_cpu(non_clock_info->usClassification2);
+
+	if (ATOM_PPLIB_NONCLOCKINFO_VER1 < table_rev) {
+		rps->vclk = le32_to_cpu(non_clock_info->ulVCLK);
+		rps->dclk = le32_to_cpu(non_clock_info->ulDCLK);
+	} else if (r600_is_uvd_state(rps->class, rps->class2)) {
+		rps->vclk = RV770_DEFAULT_VCLK_FREQ;
+		rps->dclk = RV770_DEFAULT_DCLK_FREQ;
+	} else {
+		rps->vclk = 0;
+		rps->dclk = 0;
+	}
+
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_BOOT)
+		adev->pm.dpm.boot_ps = rps;
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
+		adev->pm.dpm.uvd_ps = rps;
+}
+
+static void si_parse_pplib_clock_info(struct amdgpu_device *adev,
+				      struct amdgpu_ps *rps, int index,
+				      union pplib_clock_info *clock_info)
+{
+	struct rv7xx_power_info *pi = rv770_get_pi(adev);
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct si_power_info *si_pi = si_get_pi(adev);
+	struct  si_ps *ps = si_get_ps(rps);
+	u16 leakage_voltage;
+	struct rv7xx_pl *pl = &ps->performance_levels[index];
+	int ret;
+
+	ps->performance_level_count = index + 1;
+
+	pl->sclk = le16_to_cpu(clock_info->si.usEngineClockLow);
+	pl->sclk |= clock_info->si.ucEngineClockHigh << 16;
+	pl->mclk = le16_to_cpu(clock_info->si.usMemoryClockLow);
+	pl->mclk |= clock_info->si.ucMemoryClockHigh << 16;
+
+	pl->vddc = le16_to_cpu(clock_info->si.usVDDC);
+	pl->vddci = le16_to_cpu(clock_info->si.usVDDCI);
+	pl->flags = le32_to_cpu(clock_info->si.ulFlags);
+	pl->pcie_gen = amdgpu_get_pcie_gen_support(adev,
+						   si_pi->sys_pcie_mask,
+						   si_pi->boot_pcie_gen,
+						   clock_info->si.ucPCIEGen);
+
+	/* patch up vddc if necessary */
+	ret = si_get_leakage_voltage_from_leakage_index(adev, pl->vddc,
+							&leakage_voltage);
+	if (ret == 0)
+		pl->vddc = leakage_voltage;
+
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_ACPI) {
+		pi->acpi_vddc = pl->vddc;
+		eg_pi->acpi_vddci = pl->vddci;
+		si_pi->acpi_pcie_gen = pl->pcie_gen;
+	}
+
+	if ((rps->class2 & ATOM_PPLIB_CLASSIFICATION2_ULV) &&
+	    index == 0) {
+		/* XXX disable for A0 tahiti */
+		si_pi->ulv.supported = false;
+		si_pi->ulv.pl = *pl;
+		si_pi->ulv.one_pcie_lane_in_ulv = false;
+		si_pi->ulv.volt_change_delay = SISLANDS_ULVVOLTAGECHANGEDELAY_DFLT;
+		si_pi->ulv.cg_ulv_parameter = SISLANDS_CGULVPARAMETER_DFLT;
+		si_pi->ulv.cg_ulv_control = SISLANDS_CGULVCONTROL_DFLT;
+	}
+
+	if (pi->min_vddc_in_table > pl->vddc)
+		pi->min_vddc_in_table = pl->vddc;
+
+	if (pi->max_vddc_in_table < pl->vddc)
+		pi->max_vddc_in_table = pl->vddc;
+
+	/* patch up boot state */
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_BOOT) {
+		u16 vddc, vddci, mvdd;
+		amdgpu_atombios_get_default_voltages(adev, &vddc, &vddci, &mvdd);
+		pl->mclk = adev->clock.default_mclk;
+		pl->sclk = adev->clock.default_sclk;
+		pl->vddc = vddc;
+		pl->vddci = vddci;
+		si_pi->mvdd_bootup_value = mvdd;
+	}
+
+	if ((rps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) ==
+	    ATOM_PPLIB_CLASSIFICATION_UI_PERFORMANCE) {
+		adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.sclk = pl->sclk;
+		adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.mclk = pl->mclk;
+		adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.vddc = pl->vddc;
+		adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.vddci = pl->vddci;
+	}
+}
+
+union pplib_power_state {
+	struct _ATOM_PPLIB_STATE v1;
+	struct _ATOM_PPLIB_STATE_V2 v2;
+};
+
+static int si_parse_power_table(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info;
+	union pplib_power_state *power_state;
+	int i, j, k, non_clock_array_index, clock_array_index;
+	union pplib_clock_info *clock_info;
+	struct _StateArray *state_array;
+	struct _ClockInfoArray *clock_info_array;
+	struct _NonClockInfoArray *non_clock_info_array;
+	union power_info *power_info;
+	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
+	u16 data_offset;
+	u8 frev, crev;
+	u8 *power_state_offset;
+	struct  si_ps *ps;
+
+	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset))
+		return -EINVAL;
+	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
+
+	amdgpu_add_thermal_controller(adev);
+
+	state_array = (struct _StateArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usStateArrayOffset));
+	clock_info_array = (struct _ClockInfoArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usClockInfoArrayOffset));
+	non_clock_info_array = (struct _NonClockInfoArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usNonClockInfoArrayOffset));
+
+	adev->pm.dpm.ps = kcalloc(state_array->ucNumEntries,
+				  sizeof(struct amdgpu_ps),
+				  GFP_KERNEL);
+	if (!adev->pm.dpm.ps)
+		return -ENOMEM;
+	power_state_offset = (u8 *)state_array->states;
+	for (i = 0; i < state_array->ucNumEntries; i++) {
+		u8 *idx;
+		power_state = (union pplib_power_state *)power_state_offset;
+		non_clock_array_index = power_state->v2.nonClockInfoIndex;
+		non_clock_info = (struct _ATOM_PPLIB_NONCLOCK_INFO *)
+			&non_clock_info_array->nonClockInfo[non_clock_array_index];
+		ps = kzalloc(sizeof(struct  si_ps), GFP_KERNEL);
+		if (ps == NULL) {
+			kfree(adev->pm.dpm.ps);
+			return -ENOMEM;
+		}
+		adev->pm.dpm.ps[i].ps_priv = ps;
+		si_parse_pplib_non_clock_info(adev, &adev->pm.dpm.ps[i],
+					      non_clock_info,
+					      non_clock_info_array->ucEntrySize);
+		k = 0;
+		idx = (u8 *)&power_state->v2.clockInfoIndex[0];
+		for (j = 0; j < power_state->v2.ucNumDPMLevels; j++) {
+			clock_array_index = idx[j];
+			if (clock_array_index >= clock_info_array->ucNumEntries)
+				continue;
+			if (k >= SISLANDS_MAX_HARDWARE_POWERLEVELS)
+				break;
+			clock_info = (union pplib_clock_info *)
+				((u8 *)&clock_info_array->clockInfo[0] +
+				 (clock_array_index * clock_info_array->ucEntrySize));
+			si_parse_pplib_clock_info(adev,
+						  &adev->pm.dpm.ps[i], k,
+						  clock_info);
+			k++;
+		}
+		power_state_offset += 2 + power_state->v2.ucNumDPMLevels;
+	}
+	adev->pm.dpm.num_ps = state_array->ucNumEntries;
+
+	/* fill in the vce power states */
+	for (i = 0; i < adev->pm.dpm.num_of_vce_states; i++) {
+		u32 sclk, mclk;
+		clock_array_index = adev->pm.dpm.vce_states[i].clk_idx;
+		clock_info = (union pplib_clock_info *)
+			&clock_info_array->clockInfo[clock_array_index * clock_info_array->ucEntrySize];
+		sclk = le16_to_cpu(clock_info->si.usEngineClockLow);
+		sclk |= clock_info->si.ucEngineClockHigh << 16;
+		mclk = le16_to_cpu(clock_info->si.usMemoryClockLow);
+		mclk |= clock_info->si.ucMemoryClockHigh << 16;
+		adev->pm.dpm.vce_states[i].sclk = sclk;
+		adev->pm.dpm.vce_states[i].mclk = mclk;
+	}
+
+	return 0;
+}
+
+static int si_dpm_init(struct amdgpu_device *adev)
+{
+	struct rv7xx_power_info *pi;
+	struct evergreen_power_info *eg_pi;
+	struct ni_power_info *ni_pi;
+	struct si_power_info *si_pi;
+	struct atom_clock_dividers dividers;
+	int ret;
+
+	si_pi = kzalloc(sizeof(struct si_power_info), GFP_KERNEL);
+	if (si_pi == NULL)
+		return -ENOMEM;
+	adev->pm.dpm.priv = si_pi;
+	ni_pi = &si_pi->ni;
+	eg_pi = &ni_pi->eg;
+	pi = &eg_pi->rv7xx;
+
+	si_pi->sys_pcie_mask =
+		adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_MASK;
+	si_pi->force_pcie_gen = AMDGPU_PCIE_GEN_INVALID;
+	si_pi->boot_pcie_gen = si_get_current_pcie_speed(adev);
+
+	si_set_max_cu_value(adev);
+
+	rv770_get_max_vddc(adev);
+	si_get_leakage_vddc(adev);
+	si_patch_dependency_tables_based_on_leakage(adev);
+
+	pi->acpi_vddc = 0;
+	eg_pi->acpi_vddci = 0;
+	pi->min_vddc_in_table = 0;
+	pi->max_vddc_in_table = 0;
+
+	ret = amdgpu_get_platform_caps(adev);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_parse_extended_power_table(adev);
+	if (ret)
+		return ret;
+
+	ret = si_parse_power_table(adev);
+	if (ret)
+		return ret;
+
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries =
+		kcalloc(4,
+			sizeof(struct amdgpu_clock_voltage_dependency_entry),
+			GFP_KERNEL);
+	if (!adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries) {
+		amdgpu_free_extended_power_table(adev);
+		return -ENOMEM;
+	}
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.count = 4;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[0].clk = 0;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[0].v = 0;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[1].clk = 36000;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[1].v = 720;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[2].clk = 54000;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[2].v = 810;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[3].clk = 72000;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[3].v = 900;
+
+	if (adev->pm.dpm.voltage_response_time == 0)
+		adev->pm.dpm.voltage_response_time = R600_VOLTAGERESPONSETIME_DFLT;
+	if (adev->pm.dpm.backbias_response_time == 0)
+		adev->pm.dpm.backbias_response_time = R600_BACKBIASRESPONSETIME_DFLT;
+
+	ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+					     0, false, &dividers);
+	if (ret)
+		pi->ref_div = dividers.ref_div + 1;
+	else
+		pi->ref_div = R600_REFERENCEDIVIDER_DFLT;
+
+	eg_pi->smu_uvd_hs = false;
+
+	pi->mclk_strobe_mode_threshold = 40000;
+	if (si_is_special_1gb_platform(adev))
+		pi->mclk_stutter_mode_threshold = 0;
+	else
+		pi->mclk_stutter_mode_threshold = pi->mclk_strobe_mode_threshold;
+	pi->mclk_edc_enable_threshold = 40000;
+	eg_pi->mclk_edc_wr_enable_threshold = 40000;
+
+	ni_pi->mclk_rtt_mode_threshold = eg_pi->mclk_edc_wr_enable_threshold;
+
+	pi->voltage_control =
+		amdgpu_atombios_is_voltage_gpio(adev, SET_VOLTAGE_TYPE_ASIC_VDDC,
+					    VOLTAGE_OBJ_GPIO_LUT);
+	if (!pi->voltage_control) {
+		si_pi->voltage_control_svi2 =
+			amdgpu_atombios_is_voltage_gpio(adev, SET_VOLTAGE_TYPE_ASIC_VDDC,
+						    VOLTAGE_OBJ_SVID2);
+		if (si_pi->voltage_control_svi2)
+			amdgpu_atombios_get_svi2_info(adev, SET_VOLTAGE_TYPE_ASIC_VDDC,
+						  &si_pi->svd_gpio_id, &si_pi->svc_gpio_id);
+	}
+
+	pi->mvdd_control =
+		amdgpu_atombios_is_voltage_gpio(adev, SET_VOLTAGE_TYPE_ASIC_MVDDC,
+					    VOLTAGE_OBJ_GPIO_LUT);
+
+	eg_pi->vddci_control =
+		amdgpu_atombios_is_voltage_gpio(adev, SET_VOLTAGE_TYPE_ASIC_VDDCI,
+					    VOLTAGE_OBJ_GPIO_LUT);
+	if (!eg_pi->vddci_control)
+		si_pi->vddci_control_svi2 =
+			amdgpu_atombios_is_voltage_gpio(adev, SET_VOLTAGE_TYPE_ASIC_VDDCI,
+						    VOLTAGE_OBJ_SVID2);
+
+	si_pi->vddc_phase_shed_control =
+		amdgpu_atombios_is_voltage_gpio(adev, SET_VOLTAGE_TYPE_ASIC_VDDC,
+					    VOLTAGE_OBJ_PHASE_LUT);
+
+	rv770_get_engine_memory_ss(adev);
+
+	pi->asi = RV770_ASI_DFLT;
+	pi->pasi = CYPRESS_HASI_DFLT;
+	pi->vrc = SISLANDS_VRC_DFLT;
+
+	pi->gfx_clock_gating = true;
+
+	eg_pi->sclk_deep_sleep = true;
+	si_pi->sclk_deep_sleep_above_low = false;
+
+	if (adev->pm.int_thermal_type != THERMAL_TYPE_NONE)
+		pi->thermal_protection = true;
+	else
+		pi->thermal_protection = false;
+
+	eg_pi->dynamic_ac_timing = true;
+
+	eg_pi->light_sleep = true;
+#if defined(CONFIG_ACPI)
+	eg_pi->pcie_performance_request =
+		amdgpu_acpi_is_pcie_performance_request_supported(adev);
+#else
+	eg_pi->pcie_performance_request = false;
+#endif
+
+	si_pi->sram_end = SMC_RAM_END;
+
+	adev->pm.dpm.dyn_state.mclk_sclk_ratio = 4;
+	adev->pm.dpm.dyn_state.sclk_mclk_delta = 15000;
+	adev->pm.dpm.dyn_state.vddc_vddci_delta = 200;
+	adev->pm.dpm.dyn_state.valid_sclk_values.count = 0;
+	adev->pm.dpm.dyn_state.valid_sclk_values.values = NULL;
+	adev->pm.dpm.dyn_state.valid_mclk_values.count = 0;
+	adev->pm.dpm.dyn_state.valid_mclk_values.values = NULL;
+
+	si_initialize_powertune_defaults(adev);
+
+	/* make sure dc limits are valid */
+	if ((adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.sclk == 0) ||
+	    (adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.mclk == 0))
+		adev->pm.dpm.dyn_state.max_clock_voltage_on_dc =
+			adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+
+	si_pi->fan_ctrl_is_in_default_mode = true;
+
+	return 0;
+}
+
+static void si_dpm_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (adev->pm.dpm.ps)
+		for (i = 0; i < adev->pm.dpm.num_ps; i++)
+			kfree(adev->pm.dpm.ps[i].ps_priv);
+	kfree(adev->pm.dpm.ps);
+	kfree(adev->pm.dpm.priv);
+	kfree(adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries);
+	amdgpu_free_extended_power_table(adev);
+}
+
+static void si_dpm_debugfs_print_current_performance_level(void *handle,
+						    struct seq_file *m)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct amdgpu_ps *rps = &eg_pi->current_rps;
+	struct  si_ps *ps = si_get_ps(rps);
+	struct rv7xx_pl *pl;
+	u32 current_index =
+		(RREG32(TARGET_AND_CURRENT_PROFILE_INDEX) & CURRENT_STATE_INDEX_MASK) >>
+		CURRENT_STATE_INDEX_SHIFT;
+
+	if (current_index >= ps->performance_level_count) {
+		seq_printf(m, "invalid dpm profile %d\n", current_index);
+	} else {
+		pl = &ps->performance_levels[current_index];
+		seq_printf(m, "uvd    vclk: %d dclk: %d\n", rps->vclk, rps->dclk);
+		seq_printf(m, "power level %d    sclk: %u mclk: %u vddc: %u vddci: %u pcie gen: %u\n",
+			   current_index, pl->sclk, pl->mclk, pl->vddc, pl->vddci, pl->pcie_gen + 1);
+	}
+}
+
+static int si_dpm_set_interrupt_state(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      unsigned type,
+				      enum amdgpu_interrupt_state state)
+{
+	u32 cg_thermal_int;
+
+	switch (type) {
+	case AMDGPU_THERMAL_IRQ_LOW_TO_HIGH:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			cg_thermal_int = RREG32_SMC(CG_THERMAL_INT);
+			cg_thermal_int |= THERM_INT_MASK_HIGH;
+			WREG32_SMC(CG_THERMAL_INT, cg_thermal_int);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			cg_thermal_int = RREG32_SMC(CG_THERMAL_INT);
+			cg_thermal_int &= ~THERM_INT_MASK_HIGH;
+			WREG32_SMC(CG_THERMAL_INT, cg_thermal_int);
+			break;
+		default:
+			break;
+		}
+		break;
+
+	case AMDGPU_THERMAL_IRQ_HIGH_TO_LOW:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			cg_thermal_int = RREG32_SMC(CG_THERMAL_INT);
+			cg_thermal_int |= THERM_INT_MASK_LOW;
+			WREG32_SMC(CG_THERMAL_INT, cg_thermal_int);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			cg_thermal_int = RREG32_SMC(CG_THERMAL_INT);
+			cg_thermal_int &= ~THERM_INT_MASK_LOW;
+			WREG32_SMC(CG_THERMAL_INT, cg_thermal_int);
+			break;
+		default:
+			break;
+		}
+		break;
+
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int si_dpm_process_interrupt(struct amdgpu_device *adev,
+				    struct amdgpu_irq_src *source,
+				    struct amdgpu_iv_entry *entry)
+{
+	bool queue_thermal = false;
+
+	if (entry == NULL)
+		return -EINVAL;
+
+	switch (entry->src_id) {
+	case 230: /* thermal low to high */
+		DRM_DEBUG("IH: thermal low to high\n");
+		adev->pm.dpm.thermal.high_to_low = false;
+		queue_thermal = true;
+		break;
+	case 231: /* thermal high to low */
+		DRM_DEBUG("IH: thermal high to low\n");
+		adev->pm.dpm.thermal.high_to_low = true;
+		queue_thermal = true;
+		break;
+	default:
+		break;
+	}
+
+	if (queue_thermal)
+		schedule_work(&adev->pm.dpm.thermal.work);
+
+	return 0;
+}
+
+static int si_dpm_late_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->pm.dpm_enabled)
+		return 0;
+
+	ret = si_set_temperature_range(adev);
+	if (ret)
+		return ret;
+#if 0 //TODO ?
+	si_dpm_powergate_uvd(adev, true);
+#endif
+	return 0;
+}
+
+/**
+ * si_dpm_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int si_dpm_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+
+	DRM_DEBUG("\n");
+	switch (adev->asic_type) {
+	case CHIP_TAHITI:
+		chip_name = "tahiti";
+		break;
+	case CHIP_PITCAIRN:
+		if ((adev->pdev->revision == 0x81) &&
+		    ((adev->pdev->device == 0x6810) ||
+		    (adev->pdev->device == 0x6811)))
+			chip_name = "pitcairn_k";
+		else
+			chip_name = "pitcairn";
+		break;
+	case CHIP_VERDE:
+		if (((adev->pdev->device == 0x6820) &&
+			((adev->pdev->revision == 0x81) ||
+			(adev->pdev->revision == 0x83))) ||
+		    ((adev->pdev->device == 0x6821) &&
+			((adev->pdev->revision == 0x83) ||
+			(adev->pdev->revision == 0x87))) ||
+		    ((adev->pdev->revision == 0x87) &&
+			((adev->pdev->device == 0x6823) ||
+			(adev->pdev->device == 0x682b))))
+			chip_name = "verde_k";
+		else
+			chip_name = "verde";
+		break;
+	case CHIP_OLAND:
+		if (((adev->pdev->revision == 0x81) &&
+			((adev->pdev->device == 0x6600) ||
+			(adev->pdev->device == 0x6604) ||
+			(adev->pdev->device == 0x6605) ||
+			(adev->pdev->device == 0x6610))) ||
+		    ((adev->pdev->revision == 0x83) &&
+			(adev->pdev->device == 0x6610)))
+			chip_name = "oland_k";
+		else
+			chip_name = "oland";
+		break;
+	case CHIP_HAINAN:
+		if (((adev->pdev->revision == 0x81) &&
+			(adev->pdev->device == 0x6660)) ||
+		    ((adev->pdev->revision == 0x83) &&
+			((adev->pdev->device == 0x6660) ||
+			(adev->pdev->device == 0x6663) ||
+			(adev->pdev->device == 0x6665) ||
+			 (adev->pdev->device == 0x6667))))
+			chip_name = "hainan_k";
+		else if ((adev->pdev->revision == 0xc3) &&
+			 (adev->pdev->device == 0x6665))
+			chip_name = "banks_k_2";
+		else
+			chip_name = "hainan";
+		break;
+	default: BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_smc.bin", chip_name);
+	err = request_firmware(&adev->pm.fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->pm.fw);
+
+out:
+	if (err) {
+		DRM_ERROR("si_smc: Failed to load firmware. err = %d\"%s\"\n",
+			  err, fw_name);
+		release_firmware(adev->pm.fw);
+		adev->pm.fw = NULL;
+	}
+	return err;
+
+}
+
+static int si_dpm_sw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 230, &adev->pm.dpm.thermal.irq);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 231, &adev->pm.dpm.thermal.irq);
+	if (ret)
+		return ret;
+
+	/* default to balanced state */
+	adev->pm.dpm.state = POWER_STATE_TYPE_BALANCED;
+	adev->pm.dpm.user_state = POWER_STATE_TYPE_BALANCED;
+	adev->pm.dpm.forced_level = AMD_DPM_FORCED_LEVEL_AUTO;
+	adev->pm.default_sclk = adev->clock.default_sclk;
+	adev->pm.default_mclk = adev->clock.default_mclk;
+	adev->pm.current_sclk = adev->clock.default_sclk;
+	adev->pm.current_mclk = adev->clock.default_mclk;
+	adev->pm.int_thermal_type = THERMAL_TYPE_NONE;
+
+	if (amdgpu_dpm == 0)
+		return 0;
+
+	ret = si_dpm_init_microcode(adev);
+	if (ret)
+		return ret;
+
+	INIT_WORK(&adev->pm.dpm.thermal.work, amdgpu_dpm_thermal_work_handler);
+	mutex_lock(&adev->pm.mutex);
+	ret = si_dpm_init(adev);
+	if (ret)
+		goto dpm_failed;
+	adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
+	if (amdgpu_dpm == 1)
+		amdgpu_pm_print_power_states(adev);
+	mutex_unlock(&adev->pm.mutex);
+	DRM_INFO("amdgpu: dpm initialized\n");
+
+	return 0;
+
+dpm_failed:
+	si_dpm_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+	DRM_ERROR("amdgpu: dpm initialization failed\n");
+	return ret;
+}
+
+static int si_dpm_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	flush_work(&adev->pm.dpm.thermal.work);
+
+	mutex_lock(&adev->pm.mutex);
+	si_dpm_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+
+	return 0;
+}
+
+static int si_dpm_hw_init(void *handle)
+{
+	int ret;
+
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!amdgpu_dpm)
+		return 0;
+
+	mutex_lock(&adev->pm.mutex);
+	si_dpm_setup_asic(adev);
+	ret = si_dpm_enable(adev);
+	if (ret)
+		adev->pm.dpm_enabled = false;
+	else
+		adev->pm.dpm_enabled = true;
+	mutex_unlock(&adev->pm.mutex);
+	amdgpu_pm_compute_clocks(adev);
+	return ret;
+}
+
+static int si_dpm_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		si_dpm_disable(adev);
+		mutex_unlock(&adev->pm.mutex);
+	}
+
+	return 0;
+}
+
+static int si_dpm_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		/* disable dpm */
+		si_dpm_disable(adev);
+		/* reset the power state */
+		adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
+		mutex_unlock(&adev->pm.mutex);
+	}
+	return 0;
+}
+
+static int si_dpm_resume(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		/* asic init will reset to the boot state */
+		mutex_lock(&adev->pm.mutex);
+		si_dpm_setup_asic(adev);
+		ret = si_dpm_enable(adev);
+		if (ret)
+			adev->pm.dpm_enabled = false;
+		else
+			adev->pm.dpm_enabled = true;
+		mutex_unlock(&adev->pm.mutex);
+		if (adev->pm.dpm_enabled)
+			amdgpu_pm_compute_clocks(adev);
+	}
+	return 0;
+}
+
+static bool si_dpm_is_idle(void *handle)
+{
+	/* XXX */
+	return true;
+}
+
+static int si_dpm_wait_for_idle(void *handle)
+{
+	/* XXX */
+	return 0;
+}
+
+static int si_dpm_soft_reset(void *handle)
+{
+	return 0;
+}
+
+static int si_dpm_set_clockgating_state(void *handle,
+					enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int si_dpm_set_powergating_state(void *handle,
+					enum amd_powergating_state state)
+{
+	return 0;
+}
+
+/* get temperature in millidegrees */
+static int si_dpm_get_temp(void *handle)
+{
+	u32 temp;
+	int actual_temp = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	temp = (RREG32(CG_MULT_THERMAL_STATUS) & CTF_TEMP_MASK) >>
+		CTF_TEMP_SHIFT;
+
+	if (temp & 0x200)
+		actual_temp = 255;
+	else
+		actual_temp = temp & 0x1ff;
+
+	actual_temp = (actual_temp * 1000);
+
+	return actual_temp;
+}
+
+static u32 si_dpm_get_sclk(void *handle, bool low)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct  si_ps *requested_state = si_get_ps(&eg_pi->requested_rps);
+
+	if (low)
+		return requested_state->performance_levels[0].sclk;
+	else
+		return requested_state->performance_levels[requested_state->performance_level_count - 1].sclk;
+}
+
+static u32 si_dpm_get_mclk(void *handle, bool low)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct  si_ps *requested_state = si_get_ps(&eg_pi->requested_rps);
+
+	if (low)
+		return requested_state->performance_levels[0].mclk;
+	else
+		return requested_state->performance_levels[requested_state->performance_level_count - 1].mclk;
+}
+
+static void si_dpm_print_power_state(void *handle,
+				     void *current_ps)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ps *rps = (struct amdgpu_ps *)current_ps;
+	struct  si_ps *ps = si_get_ps(rps);
+	struct rv7xx_pl *pl;
+	int i;
+
+	amdgpu_dpm_print_class_info(rps->class, rps->class2);
+	amdgpu_dpm_print_cap_info(rps->caps);
+	DRM_INFO("\tuvd    vclk: %d dclk: %d\n", rps->vclk, rps->dclk);
+	for (i = 0; i < ps->performance_level_count; i++) {
+		pl = &ps->performance_levels[i];
+		if (adev->asic_type >= CHIP_TAHITI)
+			DRM_INFO("\t\tpower level %d    sclk: %u mclk: %u vddc: %u vddci: %u pcie gen: %u\n",
+				 i, pl->sclk, pl->mclk, pl->vddc, pl->vddci, pl->pcie_gen + 1);
+		else
+			DRM_INFO("\t\tpower level %d    sclk: %u mclk: %u vddc: %u vddci: %u\n",
+				 i, pl->sclk, pl->mclk, pl->vddc, pl->vddci);
+	}
+	amdgpu_dpm_print_ps_status(adev, rps);
+}
+
+static int si_dpm_early_init(void *handle)
+{
+
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->powerplay.pp_funcs = &si_dpm_funcs;
+	adev->powerplay.pp_handle = adev;
+	si_dpm_set_irq_funcs(adev);
+	return 0;
+}
+
+static inline bool si_are_power_levels_equal(const struct rv7xx_pl  *si_cpl1,
+						const struct rv7xx_pl *si_cpl2)
+{
+	return ((si_cpl1->mclk == si_cpl2->mclk) &&
+		  (si_cpl1->sclk == si_cpl2->sclk) &&
+		  (si_cpl1->pcie_gen == si_cpl2->pcie_gen) &&
+		  (si_cpl1->vddc == si_cpl2->vddc) &&
+		  (si_cpl1->vddci == si_cpl2->vddci));
+}
+
+static int si_check_state_equal(void *handle,
+				void *current_ps,
+				void *request_ps,
+				bool *equal)
+{
+	struct si_ps *si_cps;
+	struct si_ps *si_rps;
+	int i;
+	struct amdgpu_ps *cps = (struct amdgpu_ps *)current_ps;
+	struct amdgpu_ps *rps = (struct amdgpu_ps *)request_ps;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev == NULL || cps == NULL || rps == NULL || equal == NULL)
+		return -EINVAL;
+
+	si_cps = si_get_ps((struct amdgpu_ps *)cps);
+	si_rps = si_get_ps((struct amdgpu_ps *)rps);
+
+	if (si_cps == NULL) {
+		printk("si_cps is NULL\n");
+		*equal = false;
+		return 0;
+	}
+
+	if (si_cps->performance_level_count != si_rps->performance_level_count) {
+		*equal = false;
+		return 0;
+	}
+
+	for (i = 0; i < si_cps->performance_level_count; i++) {
+		if (!si_are_power_levels_equal(&(si_cps->performance_levels[i]),
+					&(si_rps->performance_levels[i]))) {
+			*equal = false;
+			return 0;
+		}
+	}
+
+	/* If all performance levels are the same try to use the UVD clocks to break the tie.*/
+	*equal = ((cps->vclk == rps->vclk) && (cps->dclk == rps->dclk));
+	*equal &= ((cps->evclk == rps->evclk) && (cps->ecclk == rps->ecclk));
+
+	return 0;
+}
+
+static int si_dpm_read_sensor(void *handle, int idx,
+			      void *value, int *size)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct evergreen_power_info *eg_pi = evergreen_get_pi(adev);
+	struct amdgpu_ps *rps = &eg_pi->current_rps;
+	struct  si_ps *ps = si_get_ps(rps);
+	uint32_t sclk, mclk;
+	u32 pl_index =
+		(RREG32(TARGET_AND_CURRENT_PROFILE_INDEX) & CURRENT_STATE_INDEX_MASK) >>
+		CURRENT_STATE_INDEX_SHIFT;
+
+	/* size must be at least 4 bytes for all sensors */
+	if (*size < 4)
+		return -EINVAL;
+
+	switch (idx) {
+	case AMDGPU_PP_SENSOR_GFX_SCLK:
+		if (pl_index < ps->performance_level_count) {
+			sclk = ps->performance_levels[pl_index].sclk;
+			*((uint32_t *)value) = sclk;
+			*size = 4;
+			return 0;
+		}
+		return -EINVAL;
+	case AMDGPU_PP_SENSOR_GFX_MCLK:
+		if (pl_index < ps->performance_level_count) {
+			mclk = ps->performance_levels[pl_index].mclk;
+			*((uint32_t *)value) = mclk;
+			*size = 4;
+			return 0;
+		}
+		return -EINVAL;
+	case AMDGPU_PP_SENSOR_GPU_TEMP:
+		*((uint32_t *)value) = si_dpm_get_temp(adev);
+		*size = 4;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
+static const struct amd_ip_funcs si_dpm_ip_funcs = {
+	.name = "si_dpm",
+	.early_init = si_dpm_early_init,
+	.late_init = si_dpm_late_init,
+	.sw_init = si_dpm_sw_init,
+	.sw_fini = si_dpm_sw_fini,
+	.hw_init = si_dpm_hw_init,
+	.hw_fini = si_dpm_hw_fini,
+	.suspend = si_dpm_suspend,
+	.resume = si_dpm_resume,
+	.is_idle = si_dpm_is_idle,
+	.wait_for_idle = si_dpm_wait_for_idle,
+	.soft_reset = si_dpm_soft_reset,
+	.set_clockgating_state = si_dpm_set_clockgating_state,
+	.set_powergating_state = si_dpm_set_powergating_state,
+};
+
+const struct amdgpu_ip_block_version si_smu_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_SMC,
+	.major = 6,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &si_dpm_ip_funcs,
+};
+
+static const struct amd_pm_funcs si_dpm_funcs = {
+	.pre_set_power_state = &si_dpm_pre_set_power_state,
+	.set_power_state = &si_dpm_set_power_state,
+	.post_set_power_state = &si_dpm_post_set_power_state,
+	.display_configuration_changed = &si_dpm_display_configuration_changed,
+	.get_sclk = &si_dpm_get_sclk,
+	.get_mclk = &si_dpm_get_mclk,
+	.print_power_state = &si_dpm_print_power_state,
+	.debugfs_print_current_performance_level = &si_dpm_debugfs_print_current_performance_level,
+	.force_performance_level = &si_dpm_force_performance_level,
+	.vblank_too_short = &si_dpm_vblank_too_short,
+	.set_fan_control_mode = &si_dpm_set_fan_control_mode,
+	.get_fan_control_mode = &si_dpm_get_fan_control_mode,
+	.set_fan_speed_percent = &si_dpm_set_fan_speed_percent,
+	.get_fan_speed_percent = &si_dpm_get_fan_speed_percent,
+	.check_state_equal = &si_check_state_equal,
+	.get_vce_clock_state = amdgpu_get_vce_clock_state,
+	.read_sensor = &si_dpm_read_sensor,
+};
+
+static const struct amdgpu_irq_src_funcs si_dpm_irq_funcs = {
+	.set = si_dpm_set_interrupt_state,
+	.process = si_dpm_process_interrupt,
+};
+
+static void si_dpm_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->pm.dpm.thermal.irq.num_types = AMDGPU_THERMAL_IRQ_LAST;
+	adev->pm.dpm.thermal.irq.funcs = &si_dpm_irq_funcs;
+}
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_si_smc.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_si_smc.c
new file mode 100644
index 00000000000..4a2fd8b6194
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_si_smc.c
@@ -0,0 +1,273 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "sid.h"
+#include "ppsmc.h"
+#include "amdgpu_ucode.h"
+#include "sislands_smc.h"
+
+static int si_set_smc_sram_address(struct amdgpu_device *adev,
+				   u32 smc_address, u32 limit)
+{
+	if (smc_address & 3)
+		return -EINVAL;
+	if ((smc_address + 3) > limit)
+		return -EINVAL;
+
+	WREG32(SMC_IND_INDEX_0, smc_address);
+	WREG32_P(SMC_IND_ACCESS_CNTL, 0, ~AUTO_INCREMENT_IND_0);
+
+	return 0;
+}
+
+int amdgpu_si_copy_bytes_to_smc(struct amdgpu_device *adev,
+				u32 smc_start_address,
+				const u8 *src, u32 byte_count, u32 limit)
+{
+	unsigned long flags;
+	int ret = 0;
+	u32 data, original_data, addr, extra_shift;
+
+	if (smc_start_address & 3)
+		return -EINVAL;
+	if ((smc_start_address + byte_count) > limit)
+		return -EINVAL;
+
+	addr = smc_start_address;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	while (byte_count >= 4) {
+		/* SMC address space is BE */
+		data = (src[0] << 24) | (src[1] << 16) | (src[2] << 8) | src[3];
+
+		ret = si_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			goto done;
+
+		WREG32(SMC_IND_DATA_0, data);
+
+		src += 4;
+		byte_count -= 4;
+		addr += 4;
+	}
+
+	/* RMW for the final bytes */
+	if (byte_count > 0) {
+		data = 0;
+
+		ret = si_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			goto done;
+
+		original_data = RREG32(SMC_IND_DATA_0);
+		extra_shift = 8 * (4 - byte_count);
+
+		while (byte_count > 0) {
+			/* SMC address space is BE */
+			data = (data << 8) + *src++;
+			byte_count--;
+		}
+
+		data <<= extra_shift;
+		data |= (original_data & ~((~0UL) << extra_shift));
+
+		ret = si_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			goto done;
+
+		WREG32(SMC_IND_DATA_0, data);
+	}
+
+done:
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return ret;
+}
+
+void amdgpu_si_start_smc(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(SMC_SYSCON_RESET_CNTL);
+
+	tmp &= ~RST_REG;
+
+	WREG32_SMC(SMC_SYSCON_RESET_CNTL, tmp);
+}
+
+void amdgpu_si_reset_smc(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	RREG32(CB_CGTT_SCLK_CTRL);
+	RREG32(CB_CGTT_SCLK_CTRL);
+	RREG32(CB_CGTT_SCLK_CTRL);
+	RREG32(CB_CGTT_SCLK_CTRL);
+
+	tmp = RREG32_SMC(SMC_SYSCON_RESET_CNTL) |
+	      RST_REG;
+	WREG32_SMC(SMC_SYSCON_RESET_CNTL, tmp);
+}
+
+int amdgpu_si_program_jump_on_start(struct amdgpu_device *adev)
+{
+	static const u8 data[] = { 0x0E, 0x00, 0x40, 0x40 };
+
+	return amdgpu_si_copy_bytes_to_smc(adev, 0x0, data, 4, sizeof(data)+1);
+}
+
+void amdgpu_si_smc_clock(struct amdgpu_device *adev, bool enable)
+{
+	u32 tmp = RREG32_SMC(SMC_SYSCON_CLOCK_CNTL_0);
+
+	if (enable)
+		tmp &= ~CK_DISABLE;
+	else
+		tmp |= CK_DISABLE;
+
+	WREG32_SMC(SMC_SYSCON_CLOCK_CNTL_0, tmp);
+}
+
+bool amdgpu_si_is_smc_running(struct amdgpu_device *adev)
+{
+	u32 rst = RREG32_SMC(SMC_SYSCON_RESET_CNTL);
+	u32 clk = RREG32_SMC(SMC_SYSCON_CLOCK_CNTL_0);
+
+	if (!(rst & RST_REG) && !(clk & CK_DISABLE))
+		return true;
+
+	return false;
+}
+
+PPSMC_Result amdgpu_si_send_msg_to_smc(struct amdgpu_device *adev,
+				       PPSMC_Msg msg)
+{
+	u32 tmp;
+	int i;
+
+	if (!amdgpu_si_is_smc_running(adev))
+		return PPSMC_Result_Failed;
+
+	WREG32(SMC_MESSAGE_0, msg);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(SMC_RESP_0);
+		if (tmp != 0)
+			break;
+		udelay(1);
+	}
+
+	return (PPSMC_Result)RREG32(SMC_RESP_0);
+}
+
+PPSMC_Result amdgpu_si_wait_for_smc_inactive(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	int i;
+
+	if (!amdgpu_si_is_smc_running(adev))
+		return PPSMC_Result_OK;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32_SMC(SMC_SYSCON_CLOCK_CNTL_0);
+		if ((tmp & CKEN) == 0)
+			break;
+		udelay(1);
+	}
+
+	return PPSMC_Result_OK;
+}
+
+int amdgpu_si_load_smc_ucode(struct amdgpu_device *adev, u32 limit)
+{
+	const struct smc_firmware_header_v1_0 *hdr;
+	unsigned long flags;
+	u32 ucode_start_address;
+	u32 ucode_size;
+	const u8 *src;
+	u32 data;
+
+	if (!adev->pm.fw)
+		return -EINVAL;
+
+	hdr = (const struct smc_firmware_header_v1_0 *)adev->pm.fw->data;
+
+	amdgpu_ucode_print_smc_hdr(&hdr->header);
+
+	adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	ucode_start_address = le32_to_cpu(hdr->ucode_start_addr);
+	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
+	src = (const u8 *)
+		(adev->pm.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	if (ucode_size & 3)
+		return -EINVAL;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(SMC_IND_INDEX_0, ucode_start_address);
+	WREG32_P(SMC_IND_ACCESS_CNTL, AUTO_INCREMENT_IND_0, ~AUTO_INCREMENT_IND_0);
+	while (ucode_size >= 4) {
+		/* SMC address space is BE */
+		data = (src[0] << 24) | (src[1] << 16) | (src[2] << 8) | src[3];
+
+		WREG32(SMC_IND_DATA_0, data);
+
+		src += 4;
+		ucode_size -= 4;
+	}
+	WREG32_P(SMC_IND_ACCESS_CNTL, 0, ~AUTO_INCREMENT_IND_0);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return 0;
+}
+
+int amdgpu_si_read_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
+				  u32 *value, u32 limit)
+{
+	unsigned long flags;
+	int ret;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	ret = si_set_smc_sram_address(adev, smc_address, limit);
+	if (ret == 0)
+		*value = RREG32(SMC_IND_DATA_0);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return ret;
+}
+
+int amdgpu_si_write_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
+				   u32 value, u32 limit)
+{
+	unsigned long flags;
+	int ret;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	ret = si_set_smc_sram_address(adev, smc_address, limit);
+	if (ret == 0)
+		WREG32(SMC_IND_DATA_0, value);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return ret;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_sync.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_sync.c
new file mode 100644
index 00000000000..ff8bb577032
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_sync.c
@@ -0,0 +1,458 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Christian König <christian.koenig@amd.com>
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+#include "amdgpu_amdkfd.h"
+
+struct amdgpu_sync_entry {
+	struct hlist_node	node;
+	struct dma_fence	*fence;
+	bool	explicit;
+};
+
+static struct pool amdgpu_sync_slab;
+
+/**
+ * amdgpu_sync_create - zero init sync object
+ *
+ * @sync: sync object to initialize
+ *
+ * Just clear the sync object for now.
+ */
+void amdgpu_sync_create(struct amdgpu_sync *sync)
+{
+	hash_init(sync->fences);
+	sync->last_vm_update = NULL;
+}
+
+/**
+ * amdgpu_sync_same_dev - test if fence belong to us
+ *
+ * @adev: amdgpu device to use for the test
+ * @f: fence to test
+ *
+ * Test if the fence was issued by us.
+ */
+static bool amdgpu_sync_same_dev(struct amdgpu_device *adev,
+				 struct dma_fence *f)
+{
+	struct drm_sched_fence *s_fence = to_drm_sched_fence(f);
+
+	if (s_fence) {
+		struct amdgpu_ring *ring;
+
+		ring = container_of(s_fence->sched, struct amdgpu_ring, sched);
+		return ring->adev == adev;
+	}
+
+	return false;
+}
+
+/**
+ * amdgpu_sync_get_owner - extract the owner of a fence
+ *
+ * @fence: fence get the owner from
+ *
+ * Extract who originally created the fence.
+ */
+static void *amdgpu_sync_get_owner(struct dma_fence *f)
+{
+	struct drm_sched_fence *s_fence;
+	struct amdgpu_amdkfd_fence *kfd_fence;
+
+	if (!f)
+		return AMDGPU_FENCE_OWNER_UNDEFINED;
+
+	s_fence = to_drm_sched_fence(f);
+	if (s_fence)
+		return s_fence->owner;
+
+	kfd_fence = to_amdgpu_amdkfd_fence(f);
+	if (kfd_fence)
+		return AMDGPU_FENCE_OWNER_KFD;
+
+	return AMDGPU_FENCE_OWNER_UNDEFINED;
+}
+
+/**
+ * amdgpu_sync_keep_later - Keep the later fence
+ *
+ * @keep: existing fence to test
+ * @fence: new fence
+ *
+ * Either keep the existing fence or the new one, depending which one is later.
+ */
+static void amdgpu_sync_keep_later(struct dma_fence **keep,
+				   struct dma_fence *fence)
+{
+	if (*keep && dma_fence_is_later(*keep, fence))
+		return;
+
+	dma_fence_put(*keep);
+	*keep = dma_fence_get(fence);
+}
+
+/**
+ * amdgpu_sync_add_later - add the fence to the hash
+ *
+ * @sync: sync object to add the fence to
+ * @f: fence to add
+ *
+ * Tries to add the fence to an existing hash entry. Returns true when an entry
+ * was found, false otherwise.
+ */
+static bool amdgpu_sync_add_later(struct amdgpu_sync *sync, struct dma_fence *f, bool explicit)
+{
+	struct amdgpu_sync_entry *e;
+
+	hash_for_each_possible(sync->fences, e, node, f->context) {
+		if (unlikely(e->fence->context != f->context))
+			continue;
+
+		amdgpu_sync_keep_later(&e->fence, f);
+
+		/* Preserve eplicit flag to not loose pipe line sync */
+		e->explicit |= explicit;
+
+		return true;
+	}
+	return false;
+}
+
+/**
+ * amdgpu_sync_fence - remember to sync to this fence
+ *
+ * @sync: sync object to add fence to
+ * @fence: fence to sync to
+ *
+ */
+int amdgpu_sync_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync,
+		      struct dma_fence *f, bool explicit)
+{
+	struct amdgpu_sync_entry *e;
+
+	if (!f)
+		return 0;
+	if (amdgpu_sync_same_dev(adev, f) &&
+	    amdgpu_sync_get_owner(f) == AMDGPU_FENCE_OWNER_VM)
+		amdgpu_sync_keep_later(&sync->last_vm_update, f);
+
+	if (amdgpu_sync_add_later(sync, f, explicit))
+		return 0;
+
+#ifdef __linux__
+	e = kmem_cache_alloc(amdgpu_sync_slab, GFP_KERNEL);
+#else
+	e = pool_get(&amdgpu_sync_slab, PR_WAITOK);
+#endif
+	if (!e)
+		return -ENOMEM;
+
+	e->explicit = explicit;
+
+	hash_add(sync->fences, &e->node, f->context);
+	e->fence = dma_fence_get(f);
+	return 0;
+}
+
+/**
+ * amdgpu_sync_resv - sync to a reservation object
+ *
+ * @sync: sync object to add fences from reservation object to
+ * @resv: reservation object with embedded fence
+ * @explicit_sync: true if we should only sync to the exclusive fence
+ *
+ * Sync to the fence
+ */
+int amdgpu_sync_resv(struct amdgpu_device *adev,
+		     struct amdgpu_sync *sync,
+		     struct reservation_object *resv,
+		     void *owner, bool explicit_sync)
+{
+	struct reservation_object_list *flist;
+	struct dma_fence *f;
+	void *fence_owner;
+	unsigned i;
+	int r = 0;
+
+	if (resv == NULL)
+		return -EINVAL;
+
+	/* always sync to the exclusive fence */
+	f = reservation_object_get_excl(resv);
+	r = amdgpu_sync_fence(adev, sync, f, false);
+
+	flist = reservation_object_get_list(resv);
+	if (!flist || r)
+		return r;
+
+	for (i = 0; i < flist->shared_count; ++i) {
+		f = rcu_dereference_protected(flist->shared[i],
+					      reservation_object_held(resv));
+		/* We only want to trigger KFD eviction fences on
+		 * evict or move jobs. Skip KFD fences otherwise.
+		 */
+		fence_owner = amdgpu_sync_get_owner(f);
+		if (fence_owner == AMDGPU_FENCE_OWNER_KFD &&
+		    owner != AMDGPU_FENCE_OWNER_UNDEFINED)
+			continue;
+
+		if (amdgpu_sync_same_dev(adev, f)) {
+			/* VM updates are only interesting
+			 * for other VM updates and moves.
+			 */
+			if ((owner != AMDGPU_FENCE_OWNER_UNDEFINED) &&
+			    (fence_owner != AMDGPU_FENCE_OWNER_UNDEFINED) &&
+			    ((owner == AMDGPU_FENCE_OWNER_VM) !=
+			     (fence_owner == AMDGPU_FENCE_OWNER_VM)))
+				continue;
+
+			/* Ignore fence from the same owner and explicit one as
+			 * long as it isn't undefined.
+			 */
+			if (owner != AMDGPU_FENCE_OWNER_UNDEFINED &&
+			    (fence_owner == owner || explicit_sync))
+				continue;
+		}
+
+		r = amdgpu_sync_fence(adev, sync, f, false);
+		if (r)
+			break;
+	}
+	return r;
+}
+
+/**
+ * amdgpu_sync_peek_fence - get the next fence not signaled yet
+ *
+ * @sync: the sync object
+ * @ring: optional ring to use for test
+ *
+ * Returns the next fence not signaled yet without removing it from the sync
+ * object.
+ */
+struct dma_fence *amdgpu_sync_peek_fence(struct amdgpu_sync *sync,
+					 struct amdgpu_ring *ring)
+{
+	struct amdgpu_sync_entry *e;
+	struct hlist_node *tmp;
+	int i;
+
+	hash_for_each_safe(sync->fences, i, tmp, e, node) {
+		struct dma_fence *f = e->fence;
+		struct drm_sched_fence *s_fence = to_drm_sched_fence(f);
+
+		if (dma_fence_is_signaled(f)) {
+			hash_del(&e->node);
+			dma_fence_put(f);
+#ifdef __linux__
+			kmem_cache_free(amdgpu_sync_slab, e);
+#else
+			pool_put(&amdgpu_sync_slab, e);
+#endif
+			continue;
+		}
+		if (ring && s_fence) {
+			/* For fences from the same ring it is sufficient
+			 * when they are scheduled.
+			 */
+			if (s_fence->sched == &ring->sched) {
+				if (dma_fence_is_signaled(&s_fence->scheduled))
+					continue;
+
+				return &s_fence->scheduled;
+			}
+		}
+
+		return f;
+	}
+
+	return NULL;
+}
+
+/**
+ * amdgpu_sync_get_fence - get the next fence from the sync object
+ *
+ * @sync: sync object to use
+ * @explicit: true if the next fence is explicit
+ *
+ * Get and removes the next fence from the sync object not signaled yet.
+ */
+struct dma_fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync, bool *explicit)
+{
+	struct amdgpu_sync_entry *e;
+	struct hlist_node *tmp;
+	struct dma_fence *f;
+	int i;
+	hash_for_each_safe(sync->fences, i, tmp, e, node) {
+
+		f = e->fence;
+		if (explicit)
+			*explicit = e->explicit;
+
+		hash_del(&e->node);
+#ifdef __linux__
+		kmem_cache_free(amdgpu_sync_slab, e);
+#else
+		pool_put(&amdgpu_sync_slab, e);
+#endif
+
+		if (!dma_fence_is_signaled(f))
+			return f;
+
+		dma_fence_put(f);
+	}
+	return NULL;
+}
+
+/**
+ * amdgpu_sync_clone - clone a sync object
+ *
+ * @source: sync object to clone
+ * @clone: pointer to destination sync object
+ *
+ * Adds references to all unsignaled fences in @source to @clone. Also
+ * removes signaled fences from @source while at it.
+ */
+int amdgpu_sync_clone(struct amdgpu_sync *source, struct amdgpu_sync *clone)
+{
+	struct amdgpu_sync_entry *e;
+	struct hlist_node *tmp;
+	struct dma_fence *f;
+	int i, r;
+
+	hash_for_each_safe(source->fences, i, tmp, e, node) {
+		f = e->fence;
+		if (!dma_fence_is_signaled(f)) {
+			r = amdgpu_sync_fence(NULL, clone, f, e->explicit);
+			if (r)
+				return r;
+		} else {
+			hash_del(&e->node);
+			dma_fence_put(f);
+#ifdef __linux__
+			kmem_cache_free(amdgpu_sync_slab, e);
+#else
+			pool_put(&amdgpu_sync_slab, e);
+#endif
+		}
+	}
+
+	dma_fence_put(clone->last_vm_update);
+	clone->last_vm_update = dma_fence_get(source->last_vm_update);
+
+	return 0;
+}
+
+int amdgpu_sync_wait(struct amdgpu_sync *sync, bool intr)
+{
+	struct amdgpu_sync_entry *e;
+	struct hlist_node *tmp;
+	int i, r;
+
+	hash_for_each_safe(sync->fences, i, tmp, e, node) {
+		r = dma_fence_wait(e->fence, intr);
+		if (r)
+			return r;
+
+		hash_del(&e->node);
+		dma_fence_put(e->fence);
+#ifdef __linux__
+		kmem_cache_free(amdgpu_sync_slab, e);
+#else
+		pool_put(&amdgpu_sync_slab, e);
+#endif
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_sync_free - free the sync object
+ *
+ * @sync: sync object to use
+ *
+ * Free the sync object.
+ */
+void amdgpu_sync_free(struct amdgpu_sync *sync)
+{
+	struct amdgpu_sync_entry *e;
+	struct hlist_node *tmp;
+	unsigned i;
+
+	hash_for_each_safe(sync->fences, i, tmp, e, node) {
+		hash_del(&e->node);
+		dma_fence_put(e->fence);
+#ifdef __linux__
+		kmem_cache_free(amdgpu_sync_slab, e);
+#else
+		pool_put(&amdgpu_sync_slab, e);
+#endif
+	}
+
+	dma_fence_put(sync->last_vm_update);
+}
+
+/**
+ * amdgpu_sync_init - init sync object subsystem
+ *
+ * Allocate the slab allocator.
+ */
+int amdgpu_sync_init(void)
+{
+#ifdef __linux__
+	amdgpu_sync_slab = kmem_cache_create(
+		"amdgpu_sync", sizeof(struct amdgpu_sync_entry), 0,
+		SLAB_HWCACHE_ALIGN, NULL);
+	if (!amdgpu_sync_slab)
+		return -ENOMEM;
+#else
+	pool_init(&amdgpu_sync_slab, sizeof(struct amdgpu_sync_entry),
+	    0, IPL_TTY, 0, "amdgpu_sync", NULL);
+#endif
+
+	return 0;
+}
+
+/**
+ * amdgpu_sync_fini - fini sync object subsystem
+ *
+ * Free the slab allocator.
+ */
+void amdgpu_sync_fini(void)
+{
+#ifdef __linux__
+	kmem_cache_destroy(amdgpu_sync_slab);
+#else
+	pool_destroy(&amdgpu_sync_slab);
+#endif
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_sync.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_sync.h
new file mode 100644
index 00000000000..10cf23a57f1
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_sync.h
@@ -0,0 +1,59 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König
+ */
+#ifndef __AMDGPU_SYNC_H__
+#define __AMDGPU_SYNC_H__
+
+#include <linux/hashtable.h>
+
+struct dma_fence;
+struct reservation_object;
+struct amdgpu_device;
+struct amdgpu_ring;
+
+/*
+ * Container for fences used to sync command submissions.
+ */
+struct amdgpu_sync {
+	DECLARE_HASHTABLE(fences, 4);
+	struct dma_fence	*last_vm_update;
+};
+
+void amdgpu_sync_create(struct amdgpu_sync *sync);
+int amdgpu_sync_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync,
+		      struct dma_fence *f, bool explicit);
+int amdgpu_sync_resv(struct amdgpu_device *adev,
+		     struct amdgpu_sync *sync,
+		     struct reservation_object *resv,
+		     void *owner,
+		     bool explicit_sync);
+struct dma_fence *amdgpu_sync_peek_fence(struct amdgpu_sync *sync,
+				     struct amdgpu_ring *ring);
+struct dma_fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync, bool *explicit);
+int amdgpu_sync_clone(struct amdgpu_sync *source, struct amdgpu_sync *clone);
+int amdgpu_sync_wait(struct amdgpu_sync *sync, bool intr);
+void amdgpu_sync_free(struct amdgpu_sync *sync);
+int amdgpu_sync_init(void);
+void amdgpu_sync_fini(void);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_test.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_test.c
new file mode 100644
index 00000000000..8904e62dca7
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_test.c
@@ -0,0 +1,252 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright 2009 VMware, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Michel Dänzer
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+
+/* Test BO GTT->VRAM and VRAM->GTT GPU copies across the whole GTT aperture */
+static void amdgpu_do_test_moves(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+	struct amdgpu_bo *vram_obj = NULL;
+	struct amdgpu_bo **gtt_obj = NULL;
+	struct amdgpu_bo_param bp;
+	uint64_t gart_addr, vram_addr;
+	unsigned n, size;
+	int i, r;
+
+	size = 1024 * 1024;
+
+	/* Number of tests =
+	 * (Total GTT - IB pool - writeback page - ring buffers) / test size
+	 */
+	n = adev->gmc.gart_size - AMDGPU_IB_POOL_SIZE*64*1024;
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
+		if (adev->rings[i])
+			n -= adev->rings[i]->ring_size;
+	if (adev->wb.wb_obj)
+		n -= AMDGPU_GPU_PAGE_SIZE;
+	if (adev->irq.ih.ring_obj)
+		n -= adev->irq.ih.ring_size;
+	n /= size;
+
+	gtt_obj = kcalloc(n, sizeof(*gtt_obj), GFP_KERNEL);
+	if (!gtt_obj) {
+		DRM_ERROR("Failed to allocate %d pointers\n", n);
+		r = 1;
+		goto out_cleanup;
+	}
+	memset(&bp, 0, sizeof(bp));
+	bp.size = size;
+	bp.byte_align = PAGE_SIZE;
+	bp.domain = AMDGPU_GEM_DOMAIN_VRAM;
+	bp.flags = 0;
+	bp.type = ttm_bo_type_kernel;
+	bp.resv = NULL;
+
+	r = amdgpu_bo_create(adev, &bp, &vram_obj);
+	if (r) {
+		DRM_ERROR("Failed to create VRAM object\n");
+		goto out_cleanup;
+	}
+	r = amdgpu_bo_reserve(vram_obj, false);
+	if (unlikely(r != 0))
+		goto out_unref;
+	r = amdgpu_bo_pin(vram_obj, AMDGPU_GEM_DOMAIN_VRAM);
+	if (r) {
+		DRM_ERROR("Failed to pin VRAM object\n");
+		goto out_unres;
+	}
+	vram_addr = amdgpu_bo_gpu_offset(vram_obj);
+	for (i = 0; i < n; i++) {
+		void *gtt_map, *vram_map;
+		void **gart_start, **gart_end;
+		void **vram_start, **vram_end;
+		struct dma_fence *fence = NULL;
+
+		bp.domain = AMDGPU_GEM_DOMAIN_GTT;
+		r = amdgpu_bo_create(adev, &bp, gtt_obj + i);
+		if (r) {
+			DRM_ERROR("Failed to create GTT object %d\n", i);
+			goto out_lclean;
+		}
+
+		r = amdgpu_bo_reserve(gtt_obj[i], false);
+		if (unlikely(r != 0))
+			goto out_lclean_unref;
+		r = amdgpu_bo_pin(gtt_obj[i], AMDGPU_GEM_DOMAIN_GTT);
+		if (r) {
+			DRM_ERROR("Failed to pin GTT object %d\n", i);
+			goto out_lclean_unres;
+		}
+		r = amdgpu_ttm_alloc_gart(&gtt_obj[i]->tbo);
+		if (r) {
+			DRM_ERROR("%p bind failed\n", gtt_obj[i]);
+			goto out_lclean_unpin;
+		}
+		gart_addr = amdgpu_bo_gpu_offset(gtt_obj[i]);
+
+		r = amdgpu_bo_kmap(gtt_obj[i], &gtt_map);
+		if (r) {
+			DRM_ERROR("Failed to map GTT object %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		for (gart_start = gtt_map, gart_end = gtt_map + size;
+		     gart_start < gart_end;
+		     gart_start++)
+			*gart_start = gart_start;
+
+		amdgpu_bo_kunmap(gtt_obj[i]);
+
+		r = amdgpu_copy_buffer(ring, gart_addr, vram_addr,
+				       size, NULL, &fence, false, false);
+
+		if (r) {
+			DRM_ERROR("Failed GTT->VRAM copy %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		r = dma_fence_wait(fence, false);
+		if (r) {
+			DRM_ERROR("Failed to wait for GTT->VRAM fence %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		dma_fence_put(fence);
+
+		r = amdgpu_bo_kmap(vram_obj, &vram_map);
+		if (r) {
+			DRM_ERROR("Failed to map VRAM object after copy %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		for (gart_start = gtt_map, gart_end = gtt_map + size,
+		     vram_start = vram_map, vram_end = vram_map + size;
+		     vram_start < vram_end;
+		     gart_start++, vram_start++) {
+			if (*vram_start != gart_start) {
+				DRM_ERROR("Incorrect GTT->VRAM copy %d: Got 0x%p, "
+					  "expected 0x%p (GTT/VRAM offset "
+					  "0x%16llx/0x%16llx)\n",
+					  i, *vram_start, gart_start,
+					  (unsigned long long)
+					  (gart_addr - adev->gmc.gart_start +
+					   (void*)gart_start - gtt_map),
+					  (unsigned long long)
+					  (vram_addr - adev->gmc.vram_start +
+					   (void*)gart_start - gtt_map));
+				amdgpu_bo_kunmap(vram_obj);
+				goto out_lclean_unpin;
+			}
+			*vram_start = vram_start;
+		}
+
+		amdgpu_bo_kunmap(vram_obj);
+
+		r = amdgpu_copy_buffer(ring, vram_addr, gart_addr,
+				       size, NULL, &fence, false, false);
+
+		if (r) {
+			DRM_ERROR("Failed VRAM->GTT copy %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		r = dma_fence_wait(fence, false);
+		if (r) {
+			DRM_ERROR("Failed to wait for VRAM->GTT fence %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		dma_fence_put(fence);
+
+		r = amdgpu_bo_kmap(gtt_obj[i], &gtt_map);
+		if (r) {
+			DRM_ERROR("Failed to map GTT object after copy %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		for (gart_start = gtt_map, gart_end = gtt_map + size,
+		     vram_start = vram_map, vram_end = vram_map + size;
+		     gart_start < gart_end;
+		     gart_start++, vram_start++) {
+			if (*gart_start != vram_start) {
+				DRM_ERROR("Incorrect VRAM->GTT copy %d: Got 0x%p, "
+					  "expected 0x%p (VRAM/GTT offset "
+					  "0x%16llx/0x%16llx)\n",
+					  i, *gart_start, vram_start,
+					  (unsigned long long)
+					  (vram_addr - adev->gmc.vram_start +
+					   (void*)vram_start - vram_map),
+					  (unsigned long long)
+					  (gart_addr - adev->gmc.gart_start +
+					   (void*)vram_start - vram_map));
+				amdgpu_bo_kunmap(gtt_obj[i]);
+				goto out_lclean_unpin;
+			}
+		}
+
+		amdgpu_bo_kunmap(gtt_obj[i]);
+
+		DRM_INFO("Tested GTT->VRAM and VRAM->GTT copy for GTT offset 0x%llx\n",
+			 gart_addr - adev->gmc.gart_start);
+		continue;
+
+out_lclean_unpin:
+		amdgpu_bo_unpin(gtt_obj[i]);
+out_lclean_unres:
+		amdgpu_bo_unreserve(gtt_obj[i]);
+out_lclean_unref:
+		amdgpu_bo_unref(&gtt_obj[i]);
+out_lclean:
+		for (--i; i >= 0; --i) {
+			amdgpu_bo_unpin(gtt_obj[i]);
+			amdgpu_bo_unreserve(gtt_obj[i]);
+			amdgpu_bo_unref(&gtt_obj[i]);
+		}
+		if (fence)
+			dma_fence_put(fence);
+		break;
+	}
+
+	amdgpu_bo_unpin(vram_obj);
+out_unres:
+	amdgpu_bo_unreserve(vram_obj);
+out_unref:
+	amdgpu_bo_unref(&vram_obj);
+out_cleanup:
+	kfree(gtt_obj);
+	if (r) {
+		pr_warn("Error while testing BO move\n");
+	}
+}
+
+void amdgpu_test_moves(struct amdgpu_device *adev)
+{
+	if (adev->mman.buffer_funcs)
+		amdgpu_do_test_moves(adev);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_trace.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_trace.h
new file mode 100644
index 00000000000..e9693af77f5
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_trace.h
@@ -0,0 +1,473 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#if !defined(_AMDGPU_TRACE_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _AMDGPU_TRACE_H_
+
+#include <linux/stringify.h>
+#include <linux/types.h>
+#include <linux/tracepoint.h>
+
+#include <drm/drmP.h>
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM amdgpu
+#define TRACE_INCLUDE_FILE amdgpu_trace
+
+#define AMDGPU_JOB_GET_TIMELINE_NAME(job) \
+	 job->base.s_fence->finished.ops->get_timeline_name(&job->base.s_fence->finished)
+
+TRACE_EVENT(amdgpu_mm_rreg,
+	    TP_PROTO(unsigned did, uint32_t reg, uint32_t value),
+	    TP_ARGS(did, reg, value),
+	    TP_STRUCT__entry(
+				__field(unsigned, did)
+				__field(uint32_t, reg)
+				__field(uint32_t, value)
+			    ),
+	    TP_fast_assign(
+			   __entry->did = did;
+			   __entry->reg = reg;
+			   __entry->value = value;
+			   ),
+	    TP_printk("0x%04lx, 0x%08lx, 0x%08lx",
+		      (unsigned long)__entry->did,
+		      (unsigned long)__entry->reg,
+		      (unsigned long)__entry->value)
+);
+
+TRACE_EVENT(amdgpu_mm_wreg,
+	    TP_PROTO(unsigned did, uint32_t reg, uint32_t value),
+	    TP_ARGS(did, reg, value),
+	    TP_STRUCT__entry(
+				__field(unsigned, did)
+				__field(uint32_t, reg)
+				__field(uint32_t, value)
+			    ),
+	    TP_fast_assign(
+			   __entry->did = did;
+			   __entry->reg = reg;
+			   __entry->value = value;
+			   ),
+	    TP_printk("0x%04lx, 0x%08lx, 0x%08lx",
+		      (unsigned long)__entry->did,
+		      (unsigned long)__entry->reg,
+		      (unsigned long)__entry->value)
+);
+
+TRACE_EVENT(amdgpu_iv,
+	    TP_PROTO(struct amdgpu_iv_entry *iv),
+	    TP_ARGS(iv),
+	    TP_STRUCT__entry(
+			     __field(unsigned, client_id)
+			     __field(unsigned, src_id)
+			     __field(unsigned, ring_id)
+			     __field(unsigned, vmid)
+			     __field(unsigned, vmid_src)
+			     __field(uint64_t, timestamp)
+			     __field(unsigned, timestamp_src)
+			     __field(unsigned, pasid)
+			     __array(unsigned, src_data, 4)
+			    ),
+	    TP_fast_assign(
+			   __entry->client_id = iv->client_id;
+			   __entry->src_id = iv->src_id;
+			   __entry->ring_id = iv->ring_id;
+			   __entry->vmid = iv->vmid;
+			   __entry->vmid_src = iv->vmid_src;
+			   __entry->timestamp = iv->timestamp;
+			   __entry->timestamp_src = iv->timestamp_src;
+			   __entry->pasid = iv->pasid;
+			   __entry->src_data[0] = iv->src_data[0];
+			   __entry->src_data[1] = iv->src_data[1];
+			   __entry->src_data[2] = iv->src_data[2];
+			   __entry->src_data[3] = iv->src_data[3];
+			   ),
+	    TP_printk("client_id:%u src_id:%u ring:%u vmid:%u timestamp: %llu pasid:%u src_data: %08x %08x %08x %08x\n",
+		      __entry->client_id, __entry->src_id,
+		      __entry->ring_id, __entry->vmid,
+		      __entry->timestamp, __entry->pasid,
+		      __entry->src_data[0], __entry->src_data[1],
+		      __entry->src_data[2], __entry->src_data[3])
+);
+
+
+TRACE_EVENT(amdgpu_bo_create,
+	    TP_PROTO(struct amdgpu_bo *bo),
+	    TP_ARGS(bo),
+	    TP_STRUCT__entry(
+			     __field(struct amdgpu_bo *, bo)
+			     __field(u32, pages)
+			     __field(u32, type)
+			     __field(u32, prefer)
+			     __field(u32, allow)
+			     __field(u32, visible)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->bo = bo;
+			   __entry->pages = bo->tbo.num_pages;
+			   __entry->type = bo->tbo.mem.mem_type;
+			   __entry->prefer = bo->preferred_domains;
+			   __entry->allow = bo->allowed_domains;
+			   __entry->visible = bo->flags;
+			   ),
+
+	    TP_printk("bo=%p, pages=%u, type=%d, preferred=%d, allowed=%d, visible=%d",
+		       __entry->bo, __entry->pages, __entry->type,
+		       __entry->prefer, __entry->allow, __entry->visible)
+);
+
+TRACE_EVENT(amdgpu_cs,
+	    TP_PROTO(struct amdgpu_cs_parser *p, int i),
+	    TP_ARGS(p, i),
+	    TP_STRUCT__entry(
+			     __field(struct amdgpu_bo_list *, bo_list)
+			     __field(u32, ring)
+			     __field(u32, dw)
+			     __field(u32, fences)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->bo_list = p->bo_list;
+			   __entry->ring = p->ring->idx;
+			   __entry->dw = p->job->ibs[i].length_dw;
+			   __entry->fences = amdgpu_fence_count_emitted(
+				p->ring);
+			   ),
+	    TP_printk("bo_list=%p, ring=%u, dw=%u, fences=%u",
+		      __entry->bo_list, __entry->ring, __entry->dw,
+		      __entry->fences)
+);
+
+TRACE_EVENT(amdgpu_cs_ioctl,
+	    TP_PROTO(struct amdgpu_job *job),
+	    TP_ARGS(job),
+	    TP_STRUCT__entry(
+			     __field(uint64_t, sched_job_id)
+			     __string(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job))
+			     __field(unsigned int, context)
+			     __field(unsigned int, seqno)
+			     __field(struct dma_fence *, fence)
+			     __field(char *, ring_name)
+			     __field(u32, num_ibs)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->sched_job_id = job->base.id;
+			   __assign_str(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job))
+			   __entry->context = job->base.s_fence->finished.context;
+			   __entry->seqno = job->base.s_fence->finished.seqno;
+			   __entry->ring_name = to_amdgpu_ring(job->base.sched)->name;
+			   __entry->num_ibs = job->num_ibs;
+			   ),
+	    TP_printk("sched_job=%llu, timeline=%s, context=%u, seqno=%u, ring_name=%s, num_ibs=%u",
+		      __entry->sched_job_id, __get_str(timeline), __entry->context,
+		      __entry->seqno, __entry->ring_name, __entry->num_ibs)
+);
+
+TRACE_EVENT(amdgpu_sched_run_job,
+	    TP_PROTO(struct amdgpu_job *job),
+	    TP_ARGS(job),
+	    TP_STRUCT__entry(
+			     __field(uint64_t, sched_job_id)
+			     __string(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job))
+			     __field(unsigned int, context)
+			     __field(unsigned int, seqno)
+			     __field(char *, ring_name)
+			     __field(u32, num_ibs)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->sched_job_id = job->base.id;
+			   __assign_str(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job))
+			   __entry->context = job->base.s_fence->finished.context;
+			   __entry->seqno = job->base.s_fence->finished.seqno;
+			   __entry->ring_name = to_amdgpu_ring(job->base.sched)->name;
+			   __entry->num_ibs = job->num_ibs;
+			   ),
+	    TP_printk("sched_job=%llu, timeline=%s, context=%u, seqno=%u, ring_name=%s, num_ibs=%u",
+		      __entry->sched_job_id, __get_str(timeline), __entry->context,
+		      __entry->seqno, __entry->ring_name, __entry->num_ibs)
+);
+
+
+TRACE_EVENT(amdgpu_vm_grab_id,
+	    TP_PROTO(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
+		     struct amdgpu_job *job),
+	    TP_ARGS(vm, ring, job),
+	    TP_STRUCT__entry(
+			     __field(u32, pasid)
+			     __field(u32, ring)
+			     __field(u32, vmid)
+			     __field(u32, vm_hub)
+			     __field(u64, pd_addr)
+			     __field(u32, needs_flush)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->pasid = vm->pasid;
+			   __entry->ring = ring->idx;
+			   __entry->vmid = job->vmid;
+			   __entry->vm_hub = ring->funcs->vmhub,
+			   __entry->pd_addr = job->vm_pd_addr;
+			   __entry->needs_flush = job->vm_needs_flush;
+			   ),
+	    TP_printk("pasid=%d, ring=%u, id=%u, hub=%u, pd_addr=%010Lx needs_flush=%u",
+		      __entry->pasid, __entry->ring, __entry->vmid,
+		      __entry->vm_hub, __entry->pd_addr, __entry->needs_flush)
+);
+
+TRACE_EVENT(amdgpu_vm_bo_map,
+	    TP_PROTO(struct amdgpu_bo_va *bo_va,
+		     struct amdgpu_bo_va_mapping *mapping),
+	    TP_ARGS(bo_va, mapping),
+	    TP_STRUCT__entry(
+			     __field(struct amdgpu_bo *, bo)
+			     __field(long, start)
+			     __field(long, last)
+			     __field(u64, offset)
+			     __field(u64, flags)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->bo = bo_va ? bo_va->base.bo : NULL;
+			   __entry->start = mapping->start;
+			   __entry->last = mapping->last;
+			   __entry->offset = mapping->offset;
+			   __entry->flags = mapping->flags;
+			   ),
+	    TP_printk("bo=%p, start=%lx, last=%lx, offset=%010llx, flags=%llx",
+		      __entry->bo, __entry->start, __entry->last,
+		      __entry->offset, __entry->flags)
+);
+
+TRACE_EVENT(amdgpu_vm_bo_unmap,
+	    TP_PROTO(struct amdgpu_bo_va *bo_va,
+		     struct amdgpu_bo_va_mapping *mapping),
+	    TP_ARGS(bo_va, mapping),
+	    TP_STRUCT__entry(
+			     __field(struct amdgpu_bo *, bo)
+			     __field(long, start)
+			     __field(long, last)
+			     __field(u64, offset)
+			     __field(u64, flags)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->bo = bo_va ? bo_va->base.bo : NULL;
+			   __entry->start = mapping->start;
+			   __entry->last = mapping->last;
+			   __entry->offset = mapping->offset;
+			   __entry->flags = mapping->flags;
+			   ),
+	    TP_printk("bo=%p, start=%lx, last=%lx, offset=%010llx, flags=%llx",
+		      __entry->bo, __entry->start, __entry->last,
+		      __entry->offset, __entry->flags)
+);
+
+DECLARE_EVENT_CLASS(amdgpu_vm_mapping,
+	    TP_PROTO(struct amdgpu_bo_va_mapping *mapping),
+	    TP_ARGS(mapping),
+	    TP_STRUCT__entry(
+			     __field(u64, soffset)
+			     __field(u64, eoffset)
+			     __field(u64, flags)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->soffset = mapping->start;
+			   __entry->eoffset = mapping->last + 1;
+			   __entry->flags = mapping->flags;
+			   ),
+	    TP_printk("soffs=%010llx, eoffs=%010llx, flags=%llx",
+		      __entry->soffset, __entry->eoffset, __entry->flags)
+);
+
+DEFINE_EVENT(amdgpu_vm_mapping, amdgpu_vm_bo_update,
+	    TP_PROTO(struct amdgpu_bo_va_mapping *mapping),
+	    TP_ARGS(mapping)
+);
+
+DEFINE_EVENT(amdgpu_vm_mapping, amdgpu_vm_bo_mapping,
+	    TP_PROTO(struct amdgpu_bo_va_mapping *mapping),
+	    TP_ARGS(mapping)
+);
+
+DEFINE_EVENT(amdgpu_vm_mapping, amdgpu_vm_bo_cs,
+	    TP_PROTO(struct amdgpu_bo_va_mapping *mapping),
+	    TP_ARGS(mapping)
+);
+
+TRACE_EVENT(amdgpu_vm_set_ptes,
+	    TP_PROTO(uint64_t pe, uint64_t addr, unsigned count,
+		     uint32_t incr, uint64_t flags),
+	    TP_ARGS(pe, addr, count, incr, flags),
+	    TP_STRUCT__entry(
+			     __field(u64, pe)
+			     __field(u64, addr)
+			     __field(u32, count)
+			     __field(u32, incr)
+			     __field(u64, flags)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->pe = pe;
+			   __entry->addr = addr;
+			   __entry->count = count;
+			   __entry->incr = incr;
+			   __entry->flags = flags;
+			   ),
+	    TP_printk("pe=%010Lx, addr=%010Lx, incr=%u, flags=%llx, count=%u",
+		      __entry->pe, __entry->addr, __entry->incr,
+		      __entry->flags, __entry->count)
+);
+
+TRACE_EVENT(amdgpu_vm_copy_ptes,
+	    TP_PROTO(uint64_t pe, uint64_t src, unsigned count),
+	    TP_ARGS(pe, src, count),
+	    TP_STRUCT__entry(
+			     __field(u64, pe)
+			     __field(u64, src)
+			     __field(u32, count)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->pe = pe;
+			   __entry->src = src;
+			   __entry->count = count;
+			   ),
+	    TP_printk("pe=%010Lx, src=%010Lx, count=%u",
+		      __entry->pe, __entry->src, __entry->count)
+);
+
+TRACE_EVENT(amdgpu_vm_flush,
+	    TP_PROTO(struct amdgpu_ring *ring, unsigned vmid,
+		     uint64_t pd_addr),
+	    TP_ARGS(ring, vmid, pd_addr),
+	    TP_STRUCT__entry(
+			     __field(u32, ring)
+			     __field(u32, vmid)
+			     __field(u32, vm_hub)
+			     __field(u64, pd_addr)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->ring = ring->idx;
+			   __entry->vmid = vmid;
+			   __entry->vm_hub = ring->funcs->vmhub;
+			   __entry->pd_addr = pd_addr;
+			   ),
+	    TP_printk("ring=%u, id=%u, hub=%u, pd_addr=%010Lx",
+		      __entry->ring, __entry->vmid,
+		      __entry->vm_hub,__entry->pd_addr)
+);
+
+DECLARE_EVENT_CLASS(amdgpu_pasid,
+	    TP_PROTO(unsigned pasid),
+	    TP_ARGS(pasid),
+	    TP_STRUCT__entry(
+			     __field(unsigned, pasid)
+			     ),
+	    TP_fast_assign(
+			   __entry->pasid = pasid;
+			   ),
+	    TP_printk("pasid=%u", __entry->pasid)
+);
+
+DEFINE_EVENT(amdgpu_pasid, amdgpu_pasid_allocated,
+	    TP_PROTO(unsigned pasid),
+	    TP_ARGS(pasid)
+);
+
+DEFINE_EVENT(amdgpu_pasid, amdgpu_pasid_freed,
+	    TP_PROTO(unsigned pasid),
+	    TP_ARGS(pasid)
+);
+
+TRACE_EVENT(amdgpu_bo_list_set,
+	    TP_PROTO(struct amdgpu_bo_list *list, struct amdgpu_bo *bo),
+	    TP_ARGS(list, bo),
+	    TP_STRUCT__entry(
+			     __field(struct amdgpu_bo_list *, list)
+			     __field(struct amdgpu_bo *, bo)
+			     __field(u64, bo_size)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->list = list;
+			   __entry->bo = bo;
+			   __entry->bo_size = amdgpu_bo_size(bo);
+			   ),
+	    TP_printk("list=%p, bo=%p, bo_size=%Ld",
+		      __entry->list,
+		      __entry->bo,
+		      __entry->bo_size)
+);
+
+TRACE_EVENT(amdgpu_cs_bo_status,
+	    TP_PROTO(uint64_t total_bo, uint64_t total_size),
+	    TP_ARGS(total_bo, total_size),
+	    TP_STRUCT__entry(
+			__field(u64, total_bo)
+			__field(u64, total_size)
+			),
+
+	    TP_fast_assign(
+			__entry->total_bo = total_bo;
+			__entry->total_size = total_size;
+			),
+	    TP_printk("total_bo_size=%Ld, total_bo_count=%Ld",
+			__entry->total_bo, __entry->total_size)
+);
+
+TRACE_EVENT(amdgpu_bo_move,
+	    TP_PROTO(struct amdgpu_bo* bo, uint32_t new_placement, uint32_t old_placement),
+	    TP_ARGS(bo, new_placement, old_placement),
+	    TP_STRUCT__entry(
+			__field(struct amdgpu_bo *, bo)
+			__field(u64, bo_size)
+			__field(u32, new_placement)
+			__field(u32, old_placement)
+			),
+
+	    TP_fast_assign(
+			__entry->bo      = bo;
+			__entry->bo_size = amdgpu_bo_size(bo);
+			__entry->new_placement = new_placement;
+			__entry->old_placement = old_placement;
+			),
+	    TP_printk("bo=%p, from=%d, to=%d, size=%Ld",
+			__entry->bo, __entry->old_placement,
+			__entry->new_placement, __entry->bo_size)
+);
+
+#undef AMDGPU_JOB_GET_TIMELINE_NAME
+#endif
+
+/* This part must be outside protection */
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH ../../drivers/gpu/drm/amd/amdgpu
+#ifdef __linux__
+#include <trace/define_trace.h>
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_trace_points.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_trace_points.c
new file mode 100644
index 00000000000..b160b958e5f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_trace_points.c
@@ -0,0 +1,29 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright Red Hat Inc 2010.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author : Dave Airlie <airlied@redhat.com>
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+#define CREATE_TRACE_POINTS
+#include "amdgpu_trace.h"
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ttm.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ttm.c
new file mode 100644
index 00000000000..7fe4c457120
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ttm.c
@@ -0,0 +1,2626 @@
+/*
+ * Copyright 2009 Jerome Glisse.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Jerome Glisse <glisse@freedesktop.org>
+ *    Thomas Hellstrom <thomas-at-tungstengraphics-dot-com>
+ *    Dave Airlie
+ */
+#include <drm/ttm/ttm_bo_api.h>
+#include <drm/ttm/ttm_bo_driver.h>
+#include <drm/ttm/ttm_placement.h>
+#include <drm/ttm/ttm_module.h>
+#include <drm/ttm/ttm_page_alloc.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include <linux/seq_file.h>
+#include <linux/slab.h>
+#include <linux/swiotlb.h>
+#include <linux/swap.h>
+#include <linux/pagemap.h>
+#include <linux/debugfs.h>
+#include <linux/iommu.h>
+#include "amdgpu.h"
+#include "amdgpu_object.h"
+#include "amdgpu_trace.h"
+#include "amdgpu_amdkfd.h"
+#include "bif/bif_4_1_d.h"
+
+#define DRM_FILE_PAGE_OFFSET (0x100000000ULL >> PAGE_SHIFT)
+
+static int amdgpu_map_buffer(struct ttm_buffer_object *bo,
+			     struct ttm_mem_reg *mem, unsigned num_pages,
+			     uint64_t offset, unsigned window,
+			     struct amdgpu_ring *ring,
+			     uint64_t *addr);
+
+static int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev);
+static void amdgpu_ttm_debugfs_fini(struct amdgpu_device *adev);
+
+/*
+ * Global memory.
+ */
+
+/**
+ * amdgpu_ttm_mem_global_init - Initialize and acquire reference to
+ * memory object
+ *
+ * @ref: Object for initialization.
+ *
+ * This is called by drm_global_item_ref() when an object is being
+ * initialized.
+ */
+static int amdgpu_ttm_mem_global_init(struct drm_global_reference *ref)
+{
+	return ttm_mem_global_init(ref->object);
+}
+
+/**
+ * amdgpu_ttm_mem_global_release - Drop reference to a memory object
+ *
+ * @ref: Object being removed
+ *
+ * This is called by drm_global_item_unref() when an object is being
+ * released.
+ */
+static void amdgpu_ttm_mem_global_release(struct drm_global_reference *ref)
+{
+	ttm_mem_global_release(ref->object);
+}
+
+/**
+ * amdgpu_ttm_global_init - Initialize global TTM memory reference structures.
+ *
+ * @adev: AMDGPU device for which the global structures need to be registered.
+ *
+ * This is called as part of the AMDGPU ttm init from amdgpu_ttm_init()
+ * during bring up.
+ */
+static int amdgpu_ttm_global_init(struct amdgpu_device *adev)
+{
+	struct drm_global_reference *global_ref;
+	int r;
+
+	/* ensure reference is false in case init fails */
+	adev->mman.mem_global_referenced = false;
+
+	global_ref = &adev->mman.mem_global_ref;
+	global_ref->global_type = DRM_GLOBAL_TTM_MEM;
+	global_ref->size = sizeof(struct ttm_mem_global);
+	global_ref->init = &amdgpu_ttm_mem_global_init;
+	global_ref->release = &amdgpu_ttm_mem_global_release;
+	r = drm_global_item_ref(global_ref);
+	if (r) {
+		DRM_ERROR("Failed setting up TTM memory accounting "
+			  "subsystem.\n");
+		goto error_mem;
+	}
+
+	adev->mman.bo_global_ref.mem_glob =
+		adev->mman.mem_global_ref.object;
+	global_ref = &adev->mman.bo_global_ref.ref;
+	global_ref->global_type = DRM_GLOBAL_TTM_BO;
+	global_ref->size = sizeof(struct ttm_bo_global);
+	global_ref->init = &ttm_bo_global_init;
+	global_ref->release = &ttm_bo_global_release;
+	r = drm_global_item_ref(global_ref);
+	if (r) {
+		DRM_ERROR("Failed setting up TTM BO subsystem.\n");
+		goto error_bo;
+	}
+
+	rw_init(&adev->mman.gtt_window_lock, "gttwin");
+
+	adev->mman.mem_global_referenced = true;
+
+	return 0;
+
+error_bo:
+	drm_global_item_unref(&adev->mman.mem_global_ref);
+error_mem:
+	return r;
+}
+
+static void amdgpu_ttm_global_fini(struct amdgpu_device *adev)
+{
+	if (adev->mman.mem_global_referenced) {
+		mutex_destroy(&adev->mman.gtt_window_lock);
+		drm_global_item_unref(&adev->mman.bo_global_ref.ref);
+		drm_global_item_unref(&adev->mman.mem_global_ref);
+		adev->mman.mem_global_referenced = false;
+	}
+}
+
+static int amdgpu_invalidate_caches(struct ttm_bo_device *bdev, uint32_t flags)
+{
+	return 0;
+}
+
+/**
+ * amdgpu_init_mem_type - Initialize a memory manager for a specific type of
+ * memory request.
+ *
+ * @bdev: The TTM BO device object (contains a reference to amdgpu_device)
+ * @type: The type of memory requested
+ * @man: The memory type manager for each domain
+ *
+ * This is called by ttm_bo_init_mm() when a buffer object is being
+ * initialized.
+ */
+static int amdgpu_init_mem_type(struct ttm_bo_device *bdev, uint32_t type,
+				struct ttm_mem_type_manager *man)
+{
+	struct amdgpu_device *adev;
+
+	adev = amdgpu_ttm_adev(bdev);
+
+	switch (type) {
+	case TTM_PL_SYSTEM:
+		/* System memory */
+		man->flags = TTM_MEMTYPE_FLAG_MAPPABLE;
+		man->available_caching = TTM_PL_MASK_CACHING;
+		man->default_caching = TTM_PL_FLAG_CACHED;
+		break;
+	case TTM_PL_TT:
+		/* GTT memory  */
+		man->func = &amdgpu_gtt_mgr_func;
+		man->gpu_offset = adev->gmc.gart_start;
+		man->available_caching = TTM_PL_MASK_CACHING;
+		man->default_caching = TTM_PL_FLAG_CACHED;
+		man->flags = TTM_MEMTYPE_FLAG_MAPPABLE | TTM_MEMTYPE_FLAG_CMA;
+		break;
+	case TTM_PL_VRAM:
+		/* "On-card" video ram */
+		man->func = &amdgpu_vram_mgr_func;
+		man->gpu_offset = adev->gmc.vram_start;
+		man->flags = TTM_MEMTYPE_FLAG_FIXED |
+			     TTM_MEMTYPE_FLAG_MAPPABLE;
+		man->available_caching = TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_WC;
+		man->default_caching = TTM_PL_FLAG_WC;
+		break;
+	case AMDGPU_PL_GDS:
+	case AMDGPU_PL_GWS:
+	case AMDGPU_PL_OA:
+		/* On-chip GDS memory*/
+		man->func = &ttm_bo_manager_func;
+		man->gpu_offset = 0;
+		man->flags = TTM_MEMTYPE_FLAG_FIXED | TTM_MEMTYPE_FLAG_CMA;
+		man->available_caching = TTM_PL_FLAG_UNCACHED;
+		man->default_caching = TTM_PL_FLAG_UNCACHED;
+		break;
+	default:
+		DRM_ERROR("Unsupported memory type %u\n", (unsigned)type);
+		return -EINVAL;
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_evict_flags - Compute placement flags
+ *
+ * @bo: The buffer object to evict
+ * @placement: Possible destination(s) for evicted BO
+ *
+ * Fill in placement data when ttm_bo_evict() is called
+ */
+static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
+				struct ttm_placement *placement)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
+	struct amdgpu_bo *abo;
+	static const struct ttm_place placements = {
+		.fpfn = 0,
+		.lpfn = 0,
+		.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM
+	};
+
+	/* Don't handle scatter gather BOs */
+	if (bo->type == ttm_bo_type_sg) {
+		placement->num_placement = 0;
+		placement->num_busy_placement = 0;
+		return;
+	}
+
+	/* Object isn't an AMDGPU object so ignore */
+	if (!amdgpu_bo_is_amdgpu_bo(bo)) {
+		placement->placement = &placements;
+		placement->busy_placement = &placements;
+		placement->num_placement = 1;
+		placement->num_busy_placement = 1;
+		return;
+	}
+
+	abo = ttm_to_amdgpu_bo(bo);
+	switch (bo->mem.mem_type) {
+	case TTM_PL_VRAM:
+		if (!adev->mman.buffer_funcs_enabled) {
+			/* Move to system memory */
+			amdgpu_bo_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_CPU);
+		} else if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+			   !(abo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) &&
+			   amdgpu_bo_in_cpu_visible_vram(abo)) {
+
+			/* Try evicting to the CPU inaccessible part of VRAM
+			 * first, but only set GTT as busy placement, so this
+			 * BO will be evicted to GTT rather than causing other
+			 * BOs to be evicted from VRAM
+			 */
+			amdgpu_bo_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM |
+							 AMDGPU_GEM_DOMAIN_GTT);
+			abo->placements[0].fpfn = adev->gmc.visible_vram_size >> PAGE_SHIFT;
+			abo->placements[0].lpfn = 0;
+			abo->placement.busy_placement = &abo->placements[1];
+			abo->placement.num_busy_placement = 1;
+		} else {
+			/* Move to GTT memory */
+			amdgpu_bo_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_GTT);
+		}
+		break;
+	case TTM_PL_TT:
+	default:
+		amdgpu_bo_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_CPU);
+	}
+	*placement = abo->placement;
+}
+
+/**
+ * amdgpu_verify_access - Verify access for a mmap call
+ *
+ * @bo:	The buffer object to map
+ * @filp: The file pointer from the process performing the mmap
+ *
+ * This is called by ttm_bo_mmap() to verify whether a process
+ * has the right to mmap a BO to their process space.
+ */
+static int amdgpu_verify_access(struct ttm_buffer_object *bo, struct file *filp)
+{
+	struct amdgpu_bo *abo = ttm_to_amdgpu_bo(bo);
+
+	/*
+	 * Don't verify access for KFD BOs. They don't have a GEM
+	 * object associated with them.
+	 */
+	if (abo->kfd_bo)
+		return 0;
+
+	if (amdgpu_ttm_tt_get_usermm(bo->ttm))
+		return -EPERM;
+	return drm_vma_node_verify_access(&abo->gem_base.vma_node, filp);
+}
+
+/**
+ * amdgpu_move_null - Register memory for a buffer object
+ *
+ * @bo: The bo to assign the memory to
+ * @new_mem: The memory to be assigned.
+ *
+ * Assign the memory from new_mem to the memory of the buffer object bo.
+ */
+static void amdgpu_move_null(struct ttm_buffer_object *bo,
+			     struct ttm_mem_reg *new_mem)
+{
+	struct ttm_mem_reg *old_mem = &bo->mem;
+
+	BUG_ON(old_mem->mm_node != NULL);
+	*old_mem = *new_mem;
+	new_mem->mm_node = NULL;
+}
+
+/**
+ * amdgpu_mm_node_addr - Compute the GPU relative offset of a GTT buffer.
+ *
+ * @bo: The bo to assign the memory to.
+ * @mm_node: Memory manager node for drm allocator.
+ * @mem: The region where the bo resides.
+ *
+ */
+static uint64_t amdgpu_mm_node_addr(struct ttm_buffer_object *bo,
+				    struct drm_mm_node *mm_node,
+				    struct ttm_mem_reg *mem)
+{
+	uint64_t addr = 0;
+
+	if (mem->mem_type != TTM_PL_TT || amdgpu_gtt_mgr_has_gart_addr(mem)) {
+		addr = mm_node->start << PAGE_SHIFT;
+		addr += bo->bdev->man[mem->mem_type].gpu_offset;
+	}
+	return addr;
+}
+
+/**
+ * amdgpu_find_mm_node - Helper function finds the drm_mm_node corresponding to
+ * @offset. It also modifies the offset to be within the drm_mm_node returned
+ *
+ * @mem: The region where the bo resides.
+ * @offset: The offset that drm_mm_node is used for finding.
+ *
+ */
+static struct drm_mm_node *amdgpu_find_mm_node(struct ttm_mem_reg *mem,
+					       unsigned long *offset)
+{
+	struct drm_mm_node *mm_node = mem->mm_node;
+
+	while (*offset >= (mm_node->size << PAGE_SHIFT)) {
+		*offset -= (mm_node->size << PAGE_SHIFT);
+		++mm_node;
+	}
+	return mm_node;
+}
+
+/**
+ * amdgpu_copy_ttm_mem_to_mem - Helper function for copy
+ *
+ * The function copies @size bytes from {src->mem + src->offset} to
+ * {dst->mem + dst->offset}. src->bo and dst->bo could be same BO for a
+ * move and different for a BO to BO copy.
+ *
+ * @f: Returns the last fence if multiple jobs are submitted.
+ */
+int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
+			       struct amdgpu_copy_mem *src,
+			       struct amdgpu_copy_mem *dst,
+			       uint64_t size,
+			       struct reservation_object *resv,
+			       struct dma_fence **f)
+{
+	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+	struct drm_mm_node *src_mm, *dst_mm;
+	uint64_t src_node_start, dst_node_start, src_node_size,
+		 dst_node_size, src_page_offset, dst_page_offset;
+	struct dma_fence *fence = NULL;
+	int r = 0;
+	const uint64_t GTT_MAX_BYTES = (AMDGPU_GTT_MAX_TRANSFER_SIZE *
+					AMDGPU_GPU_PAGE_SIZE);
+
+	if (!adev->mman.buffer_funcs_enabled) {
+		DRM_ERROR("Trying to move memory with ring turned off.\n");
+		return -EINVAL;
+	}
+
+	src_mm = amdgpu_find_mm_node(src->mem, &src->offset);
+	src_node_start = amdgpu_mm_node_addr(src->bo, src_mm, src->mem) +
+					     src->offset;
+	src_node_size = (src_mm->size << PAGE_SHIFT) - src->offset;
+	src_page_offset = src_node_start & (PAGE_SIZE - 1);
+
+	dst_mm = amdgpu_find_mm_node(dst->mem, &dst->offset);
+	dst_node_start = amdgpu_mm_node_addr(dst->bo, dst_mm, dst->mem) +
+					     dst->offset;
+	dst_node_size = (dst_mm->size << PAGE_SHIFT) - dst->offset;
+	dst_page_offset = dst_node_start & (PAGE_SIZE - 1);
+
+	mutex_lock(&adev->mman.gtt_window_lock);
+
+	while (size) {
+		unsigned long cur_size;
+		uint64_t from = src_node_start, to = dst_node_start;
+		struct dma_fence *next;
+
+		/* Copy size cannot exceed GTT_MAX_BYTES. So if src or dst
+		 * begins at an offset, then adjust the size accordingly
+		 */
+		cur_size = min3(min(src_node_size, dst_node_size), size,
+				GTT_MAX_BYTES);
+		if (cur_size + src_page_offset > GTT_MAX_BYTES ||
+		    cur_size + dst_page_offset > GTT_MAX_BYTES)
+			cur_size -= max(src_page_offset, dst_page_offset);
+
+		/* Map only what needs to be accessed. Map src to window 0 and
+		 * dst to window 1
+		 */
+		if (src->mem->mem_type == TTM_PL_TT &&
+		    !amdgpu_gtt_mgr_has_gart_addr(src->mem)) {
+			r = amdgpu_map_buffer(src->bo, src->mem,
+					PFN_UP(cur_size + src_page_offset),
+					src_node_start, 0, ring,
+					&from);
+			if (r)
+				goto error;
+			/* Adjust the offset because amdgpu_map_buffer returns
+			 * start of mapped page
+			 */
+			from += src_page_offset;
+		}
+
+		if (dst->mem->mem_type == TTM_PL_TT &&
+		    !amdgpu_gtt_mgr_has_gart_addr(dst->mem)) {
+			r = amdgpu_map_buffer(dst->bo, dst->mem,
+					PFN_UP(cur_size + dst_page_offset),
+					dst_node_start, 1, ring,
+					&to);
+			if (r)
+				goto error;
+			to += dst_page_offset;
+		}
+
+		r = amdgpu_copy_buffer(ring, from, to, cur_size,
+				       resv, &next, false, true);
+		if (r)
+			goto error;
+
+		dma_fence_put(fence);
+		fence = next;
+
+		size -= cur_size;
+		if (!size)
+			break;
+
+		src_node_size -= cur_size;
+		if (!src_node_size) {
+			src_node_start = amdgpu_mm_node_addr(src->bo, ++src_mm,
+							     src->mem);
+			src_node_size = (src_mm->size << PAGE_SHIFT);
+		} else {
+			src_node_start += cur_size;
+			src_page_offset = src_node_start & (PAGE_SIZE - 1);
+		}
+		dst_node_size -= cur_size;
+		if (!dst_node_size) {
+			dst_node_start = amdgpu_mm_node_addr(dst->bo, ++dst_mm,
+							     dst->mem);
+			dst_node_size = (dst_mm->size << PAGE_SHIFT);
+		} else {
+			dst_node_start += cur_size;
+			dst_page_offset = dst_node_start & (PAGE_SIZE - 1);
+		}
+	}
+error:
+	mutex_unlock(&adev->mman.gtt_window_lock);
+	if (f)
+		*f = dma_fence_get(fence);
+	dma_fence_put(fence);
+	return r;
+}
+
+/**
+ * amdgpu_move_blit - Copy an entire buffer to another buffer
+ *
+ * This is a helper called by amdgpu_bo_move() and amdgpu_move_vram_ram() to
+ * help move buffers to and from VRAM.
+ */
+static int amdgpu_move_blit(struct ttm_buffer_object *bo,
+			    bool evict, bool no_wait_gpu,
+			    struct ttm_mem_reg *new_mem,
+			    struct ttm_mem_reg *old_mem)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
+	struct amdgpu_copy_mem src, dst;
+	struct dma_fence *fence = NULL;
+	int r;
+
+	src.bo = bo;
+	dst.bo = bo;
+	src.mem = old_mem;
+	dst.mem = new_mem;
+	src.offset = 0;
+	dst.offset = 0;
+
+	r = amdgpu_ttm_copy_mem_to_mem(adev, &src, &dst,
+				       new_mem->num_pages << PAGE_SHIFT,
+				       bo->resv, &fence);
+	if (r)
+		goto error;
+
+	r = ttm_bo_pipeline_move(bo, fence, evict, new_mem);
+	dma_fence_put(fence);
+	return r;
+
+error:
+	if (fence)
+		dma_fence_wait(fence, false);
+	dma_fence_put(fence);
+	return r;
+}
+
+/**
+ * amdgpu_move_vram_ram - Copy VRAM buffer to RAM buffer
+ *
+ * Called by amdgpu_bo_move().
+ */
+static int amdgpu_move_vram_ram(struct ttm_buffer_object *bo, bool evict,
+				struct ttm_operation_ctx *ctx,
+				struct ttm_mem_reg *new_mem)
+{
+	struct amdgpu_device *adev;
+	struct ttm_mem_reg *old_mem = &bo->mem;
+	struct ttm_mem_reg tmp_mem;
+	struct ttm_place placements;
+	struct ttm_placement placement;
+	int r;
+
+	adev = amdgpu_ttm_adev(bo->bdev);
+
+	/* create space/pages for new_mem in GTT space */
+	tmp_mem = *new_mem;
+	tmp_mem.mm_node = NULL;
+	placement.num_placement = 1;
+	placement.placement = &placements;
+	placement.num_busy_placement = 1;
+	placement.busy_placement = &placements;
+	placements.fpfn = 0;
+	placements.lpfn = 0;
+	placements.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
+	r = ttm_bo_mem_space(bo, &placement, &tmp_mem, ctx);
+	if (unlikely(r)) {
+		return r;
+	}
+
+	/* set caching flags */
+	r = ttm_tt_set_placement_caching(bo->ttm, tmp_mem.placement);
+	if (unlikely(r)) {
+		goto out_cleanup;
+	}
+
+	/* Bind the memory to the GTT space */
+	r = ttm_tt_bind(bo->ttm, &tmp_mem, ctx);
+	if (unlikely(r)) {
+		goto out_cleanup;
+	}
+
+	/* blit VRAM to GTT */
+	r = amdgpu_move_blit(bo, evict, ctx->no_wait_gpu, &tmp_mem, old_mem);
+	if (unlikely(r)) {
+		goto out_cleanup;
+	}
+
+	/* move BO (in tmp_mem) to new_mem */
+	r = ttm_bo_move_ttm(bo, ctx, new_mem);
+out_cleanup:
+	ttm_bo_mem_put(bo, &tmp_mem);
+	return r;
+}
+
+/**
+ * amdgpu_move_ram_vram - Copy buffer from RAM to VRAM
+ *
+ * Called by amdgpu_bo_move().
+ */
+static int amdgpu_move_ram_vram(struct ttm_buffer_object *bo, bool evict,
+				struct ttm_operation_ctx *ctx,
+				struct ttm_mem_reg *new_mem)
+{
+	struct amdgpu_device *adev;
+	struct ttm_mem_reg *old_mem = &bo->mem;
+	struct ttm_mem_reg tmp_mem;
+	struct ttm_placement placement;
+	struct ttm_place placements;
+	int r;
+
+	adev = amdgpu_ttm_adev(bo->bdev);
+
+	/* make space in GTT for old_mem buffer */
+	tmp_mem = *new_mem;
+	tmp_mem.mm_node = NULL;
+	placement.num_placement = 1;
+	placement.placement = &placements;
+	placement.num_busy_placement = 1;
+	placement.busy_placement = &placements;
+	placements.fpfn = 0;
+	placements.lpfn = 0;
+	placements.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
+	r = ttm_bo_mem_space(bo, &placement, &tmp_mem, ctx);
+	if (unlikely(r)) {
+		return r;
+	}
+
+	/* move/bind old memory to GTT space */
+	r = ttm_bo_move_ttm(bo, ctx, &tmp_mem);
+	if (unlikely(r)) {
+		goto out_cleanup;
+	}
+
+	/* copy to VRAM */
+	r = amdgpu_move_blit(bo, evict, ctx->no_wait_gpu, new_mem, old_mem);
+	if (unlikely(r)) {
+		goto out_cleanup;
+	}
+out_cleanup:
+	ttm_bo_mem_put(bo, &tmp_mem);
+	return r;
+}
+
+/**
+ * amdgpu_bo_move - Move a buffer object to a new memory location
+ *
+ * Called by ttm_bo_handle_move_mem()
+ */
+static int amdgpu_bo_move(struct ttm_buffer_object *bo, bool evict,
+			  struct ttm_operation_ctx *ctx,
+			  struct ttm_mem_reg *new_mem)
+{
+	struct amdgpu_device *adev;
+	struct amdgpu_bo *abo;
+	struct ttm_mem_reg *old_mem = &bo->mem;
+	int r;
+
+	/* Can't move a pinned BO */
+	abo = ttm_to_amdgpu_bo(bo);
+	if (WARN_ON_ONCE(abo->pin_count > 0))
+		return -EINVAL;
+
+	adev = amdgpu_ttm_adev(bo->bdev);
+
+	if (old_mem->mem_type == TTM_PL_SYSTEM && bo->ttm == NULL) {
+		amdgpu_move_null(bo, new_mem);
+		return 0;
+	}
+	if ((old_mem->mem_type == TTM_PL_TT &&
+	     new_mem->mem_type == TTM_PL_SYSTEM) ||
+	    (old_mem->mem_type == TTM_PL_SYSTEM &&
+	     new_mem->mem_type == TTM_PL_TT)) {
+		/* bind is enough */
+		amdgpu_move_null(bo, new_mem);
+		return 0;
+	}
+
+	if (!adev->mman.buffer_funcs_enabled)
+		goto memcpy;
+
+	if (old_mem->mem_type == TTM_PL_VRAM &&
+	    new_mem->mem_type == TTM_PL_SYSTEM) {
+		r = amdgpu_move_vram_ram(bo, evict, ctx, new_mem);
+	} else if (old_mem->mem_type == TTM_PL_SYSTEM &&
+		   new_mem->mem_type == TTM_PL_VRAM) {
+		r = amdgpu_move_ram_vram(bo, evict, ctx, new_mem);
+	} else {
+		r = amdgpu_move_blit(bo, evict, ctx->no_wait_gpu,
+				     new_mem, old_mem);
+	}
+
+	if (r) {
+memcpy:
+		r = ttm_bo_move_memcpy(bo, ctx, new_mem);
+		if (r) {
+			return r;
+		}
+	}
+
+	if (bo->type == ttm_bo_type_device &&
+	    new_mem->mem_type == TTM_PL_VRAM &&
+	    old_mem->mem_type != TTM_PL_VRAM) {
+		/* amdgpu_bo_fault_reserve_notify will re-set this if the CPU
+		 * accesses the BO after it's moved.
+		 */
+		abo->flags &= ~AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+	}
+
+	/* update statistics */
+	atomic64_add((u64)bo->num_pages << PAGE_SHIFT, &adev->num_bytes_moved);
+	return 0;
+}
+
+/**
+ * amdgpu_ttm_io_mem_reserve - Reserve a block of memory during a fault
+ *
+ * Called by ttm_mem_io_reserve() ultimately via ttm_bo_vm_fault()
+ */
+static int amdgpu_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem)
+{
+	struct ttm_mem_type_manager *man = &bdev->man[mem->mem_type];
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bdev);
+	struct drm_mm_node *mm_node = mem->mm_node;
+
+	mem->bus.addr = NULL;
+	mem->bus.offset = 0;
+	mem->bus.size = mem->num_pages << PAGE_SHIFT;
+	mem->bus.base = 0;
+	mem->bus.is_iomem = false;
+	if (!(man->flags & TTM_MEMTYPE_FLAG_MAPPABLE))
+		return -EINVAL;
+	switch (mem->mem_type) {
+	case TTM_PL_SYSTEM:
+		/* system memory */
+		return 0;
+	case TTM_PL_TT:
+		break;
+	case TTM_PL_VRAM:
+		mem->bus.offset = mem->start << PAGE_SHIFT;
+		/* check if it's visible */
+		if ((mem->bus.offset + mem->bus.size) > adev->gmc.visible_vram_size)
+			return -EINVAL;
+		/* Only physically contiguous buffers apply. In a contiguous
+		 * buffer, size of the first mm_node would match the number of
+		 * pages in ttm_mem_reg.
+		 */
+		if (adev->mman.aper_base_kaddr &&
+		    (mm_node->size == mem->num_pages))
+			mem->bus.addr = (u8 *)adev->mman.aper_base_kaddr +
+					mem->bus.offset;
+
+		mem->bus.base = adev->gmc.aper_base;
+		mem->bus.is_iomem = true;
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static void amdgpu_ttm_io_mem_free(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem)
+{
+}
+
+static unsigned long amdgpu_ttm_io_mem_pfn(struct ttm_buffer_object *bo,
+					   unsigned long page_offset)
+{
+	struct drm_mm_node *mm;
+	unsigned long offset = (page_offset << PAGE_SHIFT);
+
+	mm = amdgpu_find_mm_node(&bo->mem, &offset);
+	return (bo->mem.bus.base >> PAGE_SHIFT) + mm->start +
+		(offset >> PAGE_SHIFT);
+}
+
+/*
+ * TTM backend functions.
+ */
+struct amdgpu_ttm_gup_task_list {
+	struct list_head	list;
+	struct task_struct	*task;
+};
+
+struct amdgpu_ttm_tt {
+	struct ttm_dma_tt	ttm;
+	struct amdgpu_device	*adev;
+	u64			offset;
+	uint64_t		userptr;
+	struct task_struct	*usertask;
+	uint32_t		userflags;
+	spinlock_t              guptasklock;
+	struct list_head        guptasks;
+	atomic_t		mmu_invalidations;
+	uint32_t		last_set_pages;
+
+	bus_dmamap_t		map;
+	bus_dma_segment_t	*segs;
+};
+
+/**
+ * amdgpu_ttm_tt_get_user_pages - Pin pages of memory pointed to by a USERPTR
+ * pointer to memory
+ *
+ * Called by amdgpu_gem_userptr_ioctl() and amdgpu_cs_parser_bos().
+ * This provides a wrapper around the get_user_pages() call to provide
+ * device accessible pages that back user memory.
+ */
+int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct vm_page **pages)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	struct mm_struct *mm = gtt->usertask->mm;
+	unsigned int flags = 0;
+	unsigned pinned = 0;
+	int r;
+
+	if (!mm) /* Happens during process shutdown */
+		return -ESRCH;
+
+	if (!(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY))
+		flags |= FOLL_WRITE;
+
+	down_read(&mm->mmap_sem);
+
+	if (gtt->userflags & AMDGPU_GEM_USERPTR_ANONONLY) {
+		/*
+		 * check that we only use anonymous memory to prevent problems
+		 * with writeback
+		 */
+		unsigned long end = gtt->userptr + ttm->num_pages * PAGE_SIZE;
+		struct vm_area_struct *vma;
+
+		vma = find_vma(mm, gtt->userptr);
+		if (!vma || vma->vm_file || vma->vm_end < end) {
+			up_read(&mm->mmap_sem);
+			return -EPERM;
+		}
+	}
+
+	/* loop enough times using contiguous pages of memory */
+	do {
+		unsigned num_pages = ttm->num_pages - pinned;
+		uint64_t userptr = gtt->userptr + pinned * PAGE_SIZE;
+		struct vm_page **p = pages + pinned;
+		struct amdgpu_ttm_gup_task_list guptask;
+
+		guptask.task = current;
+		spin_lock(&gtt->guptasklock);
+		list_add(&guptask.list, &gtt->guptasks);
+		spin_unlock(&gtt->guptasklock);
+
+		if (mm == current->mm)
+			r = get_user_pages(userptr, num_pages, flags, p, NULL);
+		else
+			r = get_user_pages_remote(gtt->usertask,
+					mm, userptr, num_pages,
+					flags, p, NULL, NULL);
+
+		spin_lock(&gtt->guptasklock);
+		list_del(&guptask.list);
+		spin_unlock(&gtt->guptasklock);
+
+		if (r < 0)
+			goto release_pages;
+
+		pinned += r;
+
+	} while (pinned < ttm->num_pages);
+
+	up_read(&mm->mmap_sem);
+	return 0;
+
+release_pages:
+	release_pages(pages, pinned);
+	up_read(&mm->mmap_sem);
+	return r;
+#endif
+}
+
+/**
+ * amdgpu_ttm_tt_set_user_pages - Copy pages in, putting old pages as necessary.
+ *
+ * Called by amdgpu_cs_list_validate(). This creates the page list
+ * that backs user memory and will ultimately be mapped into the device
+ * address space.
+ */
+void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct vm_page **pages)
+{
+	STUB();
+#if 0
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	unsigned i;
+
+	gtt->last_set_pages = atomic_read(&gtt->mmu_invalidations);
+	for (i = 0; i < ttm->num_pages; ++i) {
+		if (ttm->pages[i])
+			put_page(ttm->pages[i]);
+
+		ttm->pages[i] = pages ? pages[i] : NULL;
+	}
+#endif
+}
+
+/**
+ * amdgpu_ttm_tt_mark_user_page - Mark pages as dirty
+ *
+ * Called while unpinning userptr pages
+ */
+void amdgpu_ttm_tt_mark_user_pages(struct ttm_tt *ttm)
+{
+	STUB();
+#if 0
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	unsigned i;
+
+	for (i = 0; i < ttm->num_pages; ++i) {
+		struct vm_page *page = ttm->pages[i];
+
+		if (!page)
+			continue;
+
+		if (!(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY))
+			set_page_dirty(page);
+
+		mark_page_accessed(page);
+	}
+#endif
+}
+
+/**
+ * amdgpu_ttm_tt_pin_userptr - 	prepare the sg table with the user pages
+ *
+ * Called by amdgpu_ttm_backend_bind()
+ **/
+static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
+{
+	STUB();
+	return -ENOSYS;
+#if 0
+	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	unsigned nents;
+	int r;
+
+	int write = !(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY);
+	enum dma_data_direction direction = write ?
+		DMA_BIDIRECTIONAL : DMA_TO_DEVICE;
+
+	/* Allocate an SG array and squash pages into it */
+	r = sg_alloc_table_from_pages(ttm->sg, ttm->pages, ttm->num_pages, 0,
+				      ttm->num_pages << PAGE_SHIFT,
+				      GFP_KERNEL);
+	if (r)
+		goto release_sg;
+
+	/* Map SG to device */
+	r = -ENOMEM;
+	nents = dma_map_sg(adev->dev, ttm->sg->sgl, ttm->sg->nents, direction);
+	if (nents != ttm->sg->nents)
+		goto release_sg;
+
+	/* convert SG to linear array of pages and dma addresses */
+	drm_prime_sg_to_page_addr_arrays(ttm->sg, ttm->pages,
+					 gtt->ttm.dma_address, ttm->num_pages);
+
+	return 0;
+
+release_sg:
+	kfree(ttm->sg);
+	return r;
+#endif
+}
+
+/**
+ * amdgpu_ttm_tt_unpin_userptr - Unpin and unmap userptr pages
+ */
+static void amdgpu_ttm_tt_unpin_userptr(struct ttm_tt *ttm)
+{
+	STUB();
+#if 0
+	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+
+	int write = !(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY);
+	enum dma_data_direction direction = write ?
+		DMA_BIDIRECTIONAL : DMA_TO_DEVICE;
+
+	/* double check that we don't free the table twice */
+	if (!ttm->sg->sgl)
+		return;
+
+	/* unmap the pages mapped to the device */
+	dma_unmap_sg(adev->dev, ttm->sg->sgl, ttm->sg->nents, direction);
+
+	/* mark the pages as dirty */
+	amdgpu_ttm_tt_mark_user_pages(ttm);
+
+	sg_free_table(ttm->sg);
+#endif
+}
+
+int amdgpu_ttm_gart_bind(struct amdgpu_device *adev,
+				struct ttm_buffer_object *tbo,
+				uint64_t flags)
+{
+	struct amdgpu_bo *abo = ttm_to_amdgpu_bo(tbo);
+	struct ttm_tt *ttm = tbo->ttm;
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	int r;
+
+	if (abo->flags & AMDGPU_GEM_CREATE_MQD_GFX9) {
+		uint64_t page_idx = 1;
+
+		r = amdgpu_gart_bind(adev, gtt->offset, page_idx,
+				ttm->pages, gtt->ttm.dma_address, flags);
+		if (r)
+			goto gart_bind_fail;
+
+		/* Patch mtype of the second part BO */
+		flags &=  ~AMDGPU_PTE_MTYPE_MASK;
+		flags |= AMDGPU_PTE_MTYPE(AMDGPU_MTYPE_NC);
+
+		r = amdgpu_gart_bind(adev,
+				gtt->offset + (page_idx << PAGE_SHIFT),
+				ttm->num_pages - page_idx,
+				&ttm->pages[page_idx],
+				&(gtt->ttm.dma_address[page_idx]), flags);
+	} else {
+		r = amdgpu_gart_bind(adev, gtt->offset, ttm->num_pages,
+				     ttm->pages, gtt->ttm.dma_address, flags);
+	}
+
+gart_bind_fail:
+	if (r)
+		DRM_ERROR("failed to bind %lu pages at 0x%08llX\n",
+			  ttm->num_pages, gtt->offset);
+
+	return r;
+}
+
+/**
+ * amdgpu_ttm_backend_bind - Bind GTT memory
+ *
+ * Called by ttm_tt_bind() on behalf of ttm_bo_handle_move_mem().
+ * This handles binding GTT memory to the device address space.
+ */
+static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
+				   struct ttm_mem_reg *bo_mem)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
+	struct amdgpu_ttm_tt *gtt = (void*)ttm;
+	uint64_t flags;
+	int r = 0;
+
+	if (gtt->userptr) {
+		r = amdgpu_ttm_tt_pin_userptr(ttm);
+		if (r) {
+			DRM_ERROR("failed to pin userptr\n");
+			return r;
+		}
+	}
+	if (!ttm->num_pages) {
+		WARN(1, "nothing to bind %lu pages for mreg %p back %p!\n",
+		     ttm->num_pages, bo_mem, ttm);
+	}
+
+	if (bo_mem->mem_type == AMDGPU_PL_GDS ||
+	    bo_mem->mem_type == AMDGPU_PL_GWS ||
+	    bo_mem->mem_type == AMDGPU_PL_OA)
+		return -EINVAL;
+
+	if (!amdgpu_gtt_mgr_has_gart_addr(bo_mem)) {
+		gtt->offset = AMDGPU_BO_INVALID_OFFSET;
+		return 0;
+	}
+
+	/* compute PTE flags relevant to this BO memory */
+	flags = amdgpu_ttm_tt_pte_flags(adev, ttm, bo_mem);
+
+	/* bind pages into GART page tables */
+	gtt->offset = (u64)bo_mem->start << PAGE_SHIFT;
+	r = amdgpu_gart_bind(adev, gtt->offset, ttm->num_pages,
+		ttm->pages, gtt->ttm.dma_address, flags);
+
+	if (r)
+		DRM_ERROR("failed to bind %lu pages at 0x%08llX\n",
+			  ttm->num_pages, gtt->offset);
+	return r;
+}
+
+/**
+ * amdgpu_ttm_alloc_gart - Allocate GART memory for buffer object
+ */
+int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_ttm_tt *gtt = (void*)bo->ttm;
+	struct ttm_mem_reg tmp;
+	struct ttm_placement placement;
+	struct ttm_place placements;
+	uint64_t flags;
+	int r;
+
+	if (bo->mem.mem_type != TTM_PL_TT ||
+	    amdgpu_gtt_mgr_has_gart_addr(&bo->mem))
+		return 0;
+
+	/* allocate GTT space */
+	tmp = bo->mem;
+	tmp.mm_node = NULL;
+	placement.num_placement = 1;
+	placement.placement = &placements;
+	placement.num_busy_placement = 1;
+	placement.busy_placement = &placements;
+	placements.fpfn = 0;
+	placements.lpfn = adev->gmc.gart_size >> PAGE_SHIFT;
+	placements.flags = (bo->mem.placement & ~TTM_PL_MASK_MEM) |
+		TTM_PL_FLAG_TT;
+
+	r = ttm_bo_mem_space(bo, &placement, &tmp, &ctx);
+	if (unlikely(r))
+		return r;
+
+	/* compute PTE flags for this buffer object */
+	flags = amdgpu_ttm_tt_pte_flags(adev, bo->ttm, &tmp);
+
+	/* Bind pages */
+	gtt->offset = (u64)tmp.start << PAGE_SHIFT;
+	r = amdgpu_ttm_gart_bind(adev, bo, flags);
+	if (unlikely(r)) {
+		ttm_bo_mem_put(bo, &tmp);
+		return r;
+	}
+
+	ttm_bo_mem_put(bo, &bo->mem);
+	bo->mem = tmp;
+	bo->offset = (bo->mem.start << PAGE_SHIFT) +
+		bo->bdev->man[bo->mem.mem_type].gpu_offset;
+
+	return 0;
+}
+
+/**
+ * amdgpu_ttm_recover_gart - Rebind GTT pages
+ *
+ * Called by amdgpu_gtt_mgr_recover() from amdgpu_device_reset() to
+ * rebind GTT pages during a GPU reset.
+ */
+int amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(tbo->bdev);
+	uint64_t flags;
+	int r;
+
+	if (!tbo->ttm)
+		return 0;
+
+	flags = amdgpu_ttm_tt_pte_flags(adev, tbo->ttm, &tbo->mem);
+	r = amdgpu_ttm_gart_bind(adev, tbo, flags);
+
+	return r;
+}
+
+/**
+ * amdgpu_ttm_backend_unbind - Unbind GTT mapped pages
+ *
+ * Called by ttm_tt_unbind() on behalf of ttm_bo_move_ttm() and
+ * ttm_tt_destroy().
+ */
+static int amdgpu_ttm_backend_unbind(struct ttm_tt *ttm)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	int r;
+
+	/* if the pages have userptr pinning then clear that first */
+	if (gtt->userptr)
+		amdgpu_ttm_tt_unpin_userptr(ttm);
+
+	if (gtt->offset == AMDGPU_BO_INVALID_OFFSET)
+		return 0;
+
+	/* unbind shouldn't be done for GDS/GWS/OA in ttm_bo_clean_mm */
+	r = amdgpu_gart_unbind(adev, gtt->offset, ttm->num_pages);
+	if (r)
+		DRM_ERROR("failed to unbind %lu pages at 0x%08llX\n",
+			  gtt->ttm.ttm.num_pages, gtt->offset);
+	return r;
+}
+
+static void amdgpu_ttm_backend_destroy(struct ttm_tt *ttm)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+
+#ifdef notyet
+	if (gtt->usertask)
+		put_task_struct(gtt->usertask);
+#endif
+
+	bus_dmamap_destroy(gtt->adev->dmat, gtt->map);
+	free(gtt->segs, M_DRM, 0);
+
+	ttm_dma_tt_fini(&gtt->ttm);
+	kfree(gtt);
+}
+
+static struct ttm_backend_func amdgpu_backend_func = {
+	.bind = &amdgpu_ttm_backend_bind,
+	.unbind = &amdgpu_ttm_backend_unbind,
+	.destroy = &amdgpu_ttm_backend_destroy,
+};
+
+/**
+ * amdgpu_ttm_tt_create - Create a ttm_tt object for a given BO
+ *
+ * @bo: The buffer object to create a GTT ttm_tt object around
+ *
+ * Called by ttm_tt_create().
+ */
+static struct ttm_tt *amdgpu_ttm_tt_create(struct ttm_buffer_object *bo,
+					   uint32_t page_flags)
+{
+	struct amdgpu_device *adev;
+	struct amdgpu_ttm_tt *gtt;
+	unsigned long size = bo->num_pages << PAGE_SHIFT;
+
+	adev = amdgpu_ttm_adev(bo->bdev);
+
+	gtt = kzalloc(sizeof(struct amdgpu_ttm_tt), GFP_KERNEL);
+	if (gtt == NULL) {
+		return NULL;
+	}
+	gtt->ttm.ttm.func = &amdgpu_backend_func;
+	gtt->adev = adev;
+
+	/* allocate space for the uninitialized page entries */
+	if (ttm_sg_tt_init(&gtt->ttm, bo, page_flags)) {
+		kfree(gtt);
+		return NULL;
+	}
+
+	gtt->segs = mallocarray(gtt->ttm.ttm.num_pages,
+	    sizeof(bus_dma_segment_t), M_DRM, M_WAITOK | M_ZERO);
+
+	if (bus_dmamap_create(adev->dmat, size, gtt->ttm.ttm.num_pages, size,
+	    0, BUS_DMA_WAITOK, &gtt->map)) {
+		free(gtt->segs, M_DRM, 0);
+		ttm_dma_tt_fini(&gtt->ttm);
+		free(gtt, M_DRM, 0);
+		return NULL;
+	}
+
+	return &gtt->ttm.ttm;
+}
+
+/**
+ * amdgpu_ttm_tt_populate - Map GTT pages visible to the device
+ *
+ * Map the pages of a ttm_tt object to an address space visible
+ * to the underlying device.
+ */
+static int amdgpu_ttm_tt_populate(struct ttm_tt *ttm,
+			struct ttm_operation_ctx *ctx)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	bool slave = !!(ttm->page_flags & TTM_PAGE_FLAG_SG);
+
+	/* user pages are bound by amdgpu_ttm_tt_pin_userptr() */
+	if (gtt && gtt->userptr) {
+		ttm->sg = kzalloc(sizeof(struct sg_table), GFP_KERNEL);
+		if (!ttm->sg)
+			return -ENOMEM;
+
+		ttm->page_flags |= TTM_PAGE_FLAG_SG;
+		ttm->state = tt_unbound;
+		return 0;
+	}
+
+	if (slave && ttm->sg) {
+#ifdef notyet
+		drm_prime_sg_to_page_addr_arrays(ttm->sg, ttm->pages,
+						 gtt->ttm.dma_address,
+						 ttm->num_pages);
+#endif
+		ttm->state = tt_unbound;
+		return 0;
+	}
+
+#ifdef CONFIG_SWIOTLB
+	if (adev->need_swiotlb && swiotlb_nr_tbl()) {
+		return ttm_dma_populate(&gtt->ttm, adev->dev, ctx);
+	}
+#endif
+
+	/* fall back to generic helper to populate the page array
+	 * and map them to the device */
+	return ttm_populate_and_map_pages(adev->dev, &gtt->ttm, ctx);
+}
+
+/**
+ * amdgpu_ttm_tt_unpopulate - unmap GTT pages and unpopulate page arrays
+ *
+ * Unmaps pages of a ttm_tt object from the device address space and
+ * unpopulates the page array backing it.
+ */
+static void amdgpu_ttm_tt_unpopulate(struct ttm_tt *ttm)
+{
+	struct amdgpu_device *adev;
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	bool slave = !!(ttm->page_flags & TTM_PAGE_FLAG_SG);
+
+	if (gtt && gtt->userptr) {
+		amdgpu_ttm_tt_set_user_pages(ttm, NULL);
+		kfree(ttm->sg);
+		ttm->page_flags &= ~TTM_PAGE_FLAG_SG;
+		return;
+	}
+
+	if (slave)
+		return;
+
+	adev = amdgpu_ttm_adev(ttm->bdev);
+
+#ifdef CONFIG_SWIOTLB
+	if (adev->need_swiotlb && swiotlb_nr_tbl()) {
+		ttm_dma_unpopulate(&gtt->ttm, adev->dev);
+		return;
+	}
+#endif
+
+	/* fall back to generic helper to unmap and unpopulate array */
+	ttm_unmap_and_unpopulate_pages(adev->dev, &gtt->ttm);
+}
+
+/**
+ * amdgpu_ttm_tt_set_userptr - Initialize userptr GTT ttm_tt for the current
+ * task
+ *
+ * @ttm: The ttm_tt object to bind this userptr object to
+ * @addr:  The address in the current tasks VM space to use
+ * @flags: Requirements of userptr object.
+ *
+ * Called by amdgpu_gem_userptr_ioctl() to bind userptr pages
+ * to current task
+ */
+int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
+			      uint32_t flags)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+
+	if (gtt == NULL)
+		return -EINVAL;
+
+	gtt->userptr = addr;
+	gtt->userflags = flags;
+
+#ifdef notyet
+	if (gtt->usertask)
+		put_task_struct(gtt->usertask);
+	gtt->usertask = current->group_leader;
+	get_task_struct(gtt->usertask);
+#endif
+
+	mtx_init(&gtt->guptasklock, IPL_TTY);
+	INIT_LIST_HEAD(&gtt->guptasks);
+	atomic_set(&gtt->mmu_invalidations, 0);
+	gtt->last_set_pages = 0;
+
+	return 0;
+}
+
+/**
+ * amdgpu_ttm_tt_get_usermm - Return memory manager for ttm_tt object
+ */
+struct mm_struct *amdgpu_ttm_tt_get_usermm(struct ttm_tt *ttm)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+
+	if (gtt == NULL)
+		return NULL;
+
+	if (gtt->usertask == NULL)
+		return NULL;
+
+#if 0
+	return gtt->usertask->mm;
+#else
+	STUB();
+	return NULL;
+#endif
+}
+
+/**
+ * amdgpu_ttm_tt_affect_userptr - Determine if a ttm_tt object lays inside an
+ * address range for the current task.
+ *
+ */
+bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
+				  unsigned long end)
+{
+	STUB();
+	return false;
+#if 0
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	struct amdgpu_ttm_gup_task_list *entry;
+	unsigned long size;
+
+	if (gtt == NULL || !gtt->userptr)
+		return false;
+
+	/* Return false if no part of the ttm_tt object lies within
+	 * the range
+	 */
+	size = (unsigned long)gtt->ttm.ttm.num_pages * PAGE_SIZE;
+	if (gtt->userptr > end || gtt->userptr + size <= start)
+		return false;
+
+	/* Search the lists of tasks that hold this mapping and see
+	 * if current is one of them.  If it is return false.
+	 */
+	spin_lock(&gtt->guptasklock);
+	list_for_each_entry(entry, &gtt->guptasks, list) {
+		if (entry->task == current) {
+			spin_unlock(&gtt->guptasklock);
+			return false;
+		}
+	}
+	spin_unlock(&gtt->guptasklock);
+
+	atomic_inc(&gtt->mmu_invalidations);
+
+	return true;
+#endif
+}
+
+/**
+ * amdgpu_ttm_tt_userptr_invalidated - Has the ttm_tt object been invalidated?
+ */
+bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm,
+				       int *last_invalidated)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	int prev_invalidated = *last_invalidated;
+
+	*last_invalidated = atomic_read(&gtt->mmu_invalidations);
+	return prev_invalidated != *last_invalidated;
+}
+
+/**
+ * amdgpu_ttm_tt_userptr_needs_pages - Have the pages backing this ttm_tt object
+ * been invalidated since the last time they've been set?
+ */
+bool amdgpu_ttm_tt_userptr_needs_pages(struct ttm_tt *ttm)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+
+	if (gtt == NULL || !gtt->userptr)
+		return false;
+
+	return atomic_read(&gtt->mmu_invalidations) != gtt->last_set_pages;
+}
+
+/**
+ * amdgpu_ttm_tt_is_readonly - Is the ttm_tt object read only?
+ */
+bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+
+	if (gtt == NULL)
+		return false;
+
+	return !!(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY);
+}
+
+/**
+ * amdgpu_ttm_tt_pte_flags - Compute PTE flags for ttm_tt object
+ *
+ * @ttm: The ttm_tt object to compute the flags for
+ * @mem: The memory registry backing this ttm_tt object
+ */
+uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
+				 struct ttm_mem_reg *mem)
+{
+	uint64_t flags = 0;
+
+	if (mem && mem->mem_type != TTM_PL_SYSTEM)
+		flags |= AMDGPU_PTE_VALID;
+
+	if (mem && mem->mem_type == TTM_PL_TT) {
+		flags |= AMDGPU_PTE_SYSTEM;
+
+		if (ttm->caching_state == tt_cached)
+			flags |= AMDGPU_PTE_SNOOPED;
+	}
+
+	flags |= adev->gart.gart_pte_flags;
+	flags |= AMDGPU_PTE_READABLE;
+
+	if (!amdgpu_ttm_tt_is_readonly(ttm))
+		flags |= AMDGPU_PTE_WRITEABLE;
+
+	return flags;
+}
+
+/**
+ * amdgpu_ttm_bo_eviction_valuable - Check to see if we can evict a buffer
+ * object.
+ *
+ * Return true if eviction is sensible. Called by ttm_mem_evict_first() on
+ * behalf of ttm_bo_mem_force_space() which tries to evict buffer objects until
+ * it can find space for a new object and by ttm_bo_force_list_clean() which is
+ * used to clean out a memory space.
+ */
+static bool amdgpu_ttm_bo_eviction_valuable(struct ttm_buffer_object *bo,
+					    const struct ttm_place *place)
+{
+	STUB();
+	unsigned long num_pages = bo->mem.num_pages;
+	struct drm_mm_node *node = bo->mem.mm_node;
+	struct reservation_object_list *flist;
+	struct dma_fence *f;
+	int i;
+
+	/* If bo is a KFD BO, check if the bo belongs to the current process.
+	 * If true, then return false as any KFD process needs all its BOs to
+	 * be resident to run successfully
+	 */
+	flist = reservation_object_get_list(bo->resv);
+	if (flist) {
+		for (i = 0; i < flist->shared_count; ++i) {
+			f = rcu_dereference_protected(flist->shared[i],
+				reservation_object_held(bo->resv));
+#ifdef notyet
+			if (amdkfd_fence_check_mm(f, current->mm))
+				return false;
+#endif
+		}
+	}
+
+	switch (bo->mem.mem_type) {
+	case TTM_PL_TT:
+		return true;
+
+	case TTM_PL_VRAM:
+		/* Check each drm MM node individually */
+		while (num_pages) {
+			if (place->fpfn < (node->start + node->size) &&
+			    !(place->lpfn && place->lpfn <= node->start))
+				return true;
+
+			num_pages -= node->size;
+			++node;
+		}
+		return false;
+
+	default:
+		break;
+	}
+
+	return ttm_bo_eviction_valuable(bo, place);
+}
+
+/**
+ * amdgpu_ttm_access_memory - Read or Write memory that backs a buffer object.
+ *
+ * @bo:  The buffer object to read/write
+ * @offset:  Offset into buffer object
+ * @buf:  Secondary buffer to write/read from
+ * @len: Length in bytes of access
+ * @write:  true if writing
+ *
+ * This is used to access VRAM that backs a buffer object via MMIO
+ * access for debugging purposes.
+ */
+static int amdgpu_ttm_access_memory(struct ttm_buffer_object *bo,
+				    unsigned long offset,
+				    void *buf, int len, int write)
+{
+	struct amdgpu_bo *abo = ttm_to_amdgpu_bo(bo);
+	struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev);
+	struct drm_mm_node *nodes;
+	uint32_t value = 0;
+	int ret = 0;
+	uint64_t pos;
+	unsigned long flags;
+
+	if (bo->mem.mem_type != TTM_PL_VRAM)
+		return -EIO;
+
+	nodes = amdgpu_find_mm_node(&abo->tbo.mem, &offset);
+	pos = (nodes->start << PAGE_SHIFT) + offset;
+
+	while (len && pos < adev->gmc.mc_vram_size) {
+		uint64_t aligned_pos = pos & ~(uint64_t)3;
+		uint32_t bytes = 4 - (pos & 3);
+		uint32_t shift = (pos & 3) * 8;
+		uint32_t mask = 0xffffffff << shift;
+
+		if (len < bytes) {
+			mask &= 0xffffffff >> (bytes - len) * 8;
+			bytes = len;
+		}
+
+		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
+		WREG32_NO_KIQ(mmMM_INDEX, ((uint32_t)aligned_pos) | 0x80000000);
+		WREG32_NO_KIQ(mmMM_INDEX_HI, aligned_pos >> 31);
+		if (!write || mask != 0xffffffff)
+			value = RREG32_NO_KIQ(mmMM_DATA);
+		if (write) {
+			value &= ~mask;
+			value |= (*(uint32_t *)buf << shift) & mask;
+			WREG32_NO_KIQ(mmMM_DATA, value);
+		}
+		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
+		if (!write) {
+			value = (value & mask) >> shift;
+			memcpy(buf, &value, bytes);
+		}
+
+		ret += bytes;
+		buf = (uint8_t *)buf + bytes;
+		pos += bytes;
+		len -= bytes;
+		if (pos >= (nodes->start + nodes->size) << PAGE_SHIFT) {
+			++nodes;
+			pos = (nodes->start << PAGE_SHIFT);
+		}
+	}
+
+	return ret;
+}
+
+static struct ttm_bo_driver amdgpu_bo_driver = {
+	.ttm_tt_create = &amdgpu_ttm_tt_create,
+	.ttm_tt_populate = &amdgpu_ttm_tt_populate,
+	.ttm_tt_unpopulate = &amdgpu_ttm_tt_unpopulate,
+	.invalidate_caches = &amdgpu_invalidate_caches,
+	.init_mem_type = &amdgpu_init_mem_type,
+	.eviction_valuable = amdgpu_ttm_bo_eviction_valuable,
+	.evict_flags = &amdgpu_evict_flags,
+	.move = &amdgpu_bo_move,
+	.verify_access = &amdgpu_verify_access,
+	.move_notify = &amdgpu_bo_move_notify,
+	.fault_reserve_notify = &amdgpu_bo_fault_reserve_notify,
+	.io_mem_reserve = &amdgpu_ttm_io_mem_reserve,
+	.io_mem_free = &amdgpu_ttm_io_mem_free,
+	.io_mem_pfn = amdgpu_ttm_io_mem_pfn,
+	.access_memory = &amdgpu_ttm_access_memory
+};
+
+/*
+ * Firmware Reservation functions
+ */
+/**
+ * amdgpu_ttm_fw_reserve_vram_fini - free fw reserved vram
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * free fw reserved vram if it has been reserved.
+ */
+static void amdgpu_ttm_fw_reserve_vram_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->fw_vram_usage.reserved_bo,
+		NULL, &adev->fw_vram_usage.va);
+}
+
+/**
+ * amdgpu_ttm_fw_reserve_vram_init - create bo vram reservation from fw
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * create bo vram reservation from fw.
+ */
+static int amdgpu_ttm_fw_reserve_vram_init(struct amdgpu_device *adev)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_bo_param bp;
+	int r = 0;
+	int i;
+	u64 vram_size = adev->gmc.visible_vram_size;
+	u64 offset = adev->fw_vram_usage.start_offset;
+	u64 size = adev->fw_vram_usage.size;
+	struct amdgpu_bo *bo;
+
+	memset(&bp, 0, sizeof(bp));
+	bp.size = adev->fw_vram_usage.size;
+	bp.byte_align = PAGE_SIZE;
+	bp.domain = AMDGPU_GEM_DOMAIN_VRAM;
+	bp.flags = AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
+		AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	bp.type = ttm_bo_type_kernel;
+	bp.resv = NULL;
+	adev->fw_vram_usage.va = NULL;
+	adev->fw_vram_usage.reserved_bo = NULL;
+
+	if (adev->fw_vram_usage.size > 0 &&
+		adev->fw_vram_usage.size <= vram_size) {
+
+		r = amdgpu_bo_create(adev, &bp,
+				     &adev->fw_vram_usage.reserved_bo);
+		if (r)
+			goto error_create;
+
+		r = amdgpu_bo_reserve(adev->fw_vram_usage.reserved_bo, false);
+		if (r)
+			goto error_reserve;
+
+		/* remove the original mem node and create a new one at the
+		 * request position
+		 */
+		bo = adev->fw_vram_usage.reserved_bo;
+		offset = roundup2(offset, PAGE_SIZE);
+		for (i = 0; i < bo->placement.num_placement; ++i) {
+			bo->placements[i].fpfn = offset >> PAGE_SHIFT;
+			bo->placements[i].lpfn = (offset + size) >> PAGE_SHIFT;
+		}
+
+		ttm_bo_mem_put(&bo->tbo, &bo->tbo.mem);
+		r = ttm_bo_mem_space(&bo->tbo, &bo->placement,
+				     &bo->tbo.mem, &ctx);
+		if (r)
+			goto error_pin;
+
+		r = amdgpu_bo_pin_restricted(adev->fw_vram_usage.reserved_bo,
+			AMDGPU_GEM_DOMAIN_VRAM,
+			adev->fw_vram_usage.start_offset,
+			(adev->fw_vram_usage.start_offset +
+			adev->fw_vram_usage.size));
+		if (r)
+			goto error_pin;
+		r = amdgpu_bo_kmap(adev->fw_vram_usage.reserved_bo,
+			&adev->fw_vram_usage.va);
+		if (r)
+			goto error_kmap;
+
+		amdgpu_bo_unreserve(adev->fw_vram_usage.reserved_bo);
+	}
+	return r;
+
+error_kmap:
+	amdgpu_bo_unpin(adev->fw_vram_usage.reserved_bo);
+error_pin:
+	amdgpu_bo_unreserve(adev->fw_vram_usage.reserved_bo);
+error_reserve:
+	amdgpu_bo_unref(&adev->fw_vram_usage.reserved_bo);
+error_create:
+	adev->fw_vram_usage.va = NULL;
+	adev->fw_vram_usage.reserved_bo = NULL;
+	return r;
+}
+/**
+ * amdgpu_ttm_init - Init the memory management (ttm) as well as various
+ * gtt/vram related fields.
+ *
+ * This initializes all of the memory space pools that the TTM layer
+ * will need such as the GTT space (system memory mapped to the device),
+ * VRAM (on-board memory), and on-chip memories (GDS, GWS, OA) which
+ * can be mapped per VMID.
+ */
+int amdgpu_ttm_init(struct amdgpu_device *adev)
+{
+	uint64_t gtt_size;
+	int r;
+	u64 vis_vram_limit;
+
+	/* initialize global references for vram/gtt */
+	r = amdgpu_ttm_global_init(adev);
+	if (r) {
+		return r;
+	}
+	/* No others user of address space so set it to 0 */
+#ifdef notyet
+	r = ttm_bo_device_init(&adev->mman.bdev,
+			       adev->mman.bo_global_ref.ref.object,
+			       &amdgpu_bo_driver,
+			       adev->ddev->anon_inode->i_mapping,
+			       DRM_FILE_PAGE_OFFSET,
+			       adev->need_dma32);
+#else
+	r = ttm_bo_device_init(&adev->mman.bdev,
+			       adev->mman.bo_global_ref.ref.object,
+			       &amdgpu_bo_driver,
+			       /*adev->ddev->anon_inode->i_mapping*/ NULL,
+			       DRM_FILE_PAGE_OFFSET,
+			       adev->need_dma32);
+#endif
+	if (r) {
+		DRM_ERROR("failed initializing buffer object driver(%d).\n", r);
+		return r;
+	}
+	adev->mman.bdev.iot = adev->iot;
+	adev->mman.bdev.memt = adev->memt;
+	adev->mman.bdev.dmat = adev->dmat;
+	adev->mman.initialized = true;
+
+	/* We opt to avoid OOM on system pages allocations */
+	adev->mman.bdev.no_retry = true;
+
+	/* Initialize VRAM pool with all of VRAM divided into pages */
+	r = ttm_bo_init_mm(&adev->mman.bdev, TTM_PL_VRAM,
+				adev->gmc.real_vram_size >> PAGE_SHIFT);
+	if (r) {
+		DRM_ERROR("Failed initializing VRAM heap.\n");
+		return r;
+	}
+
+	/* Reduce size of CPU-visible VRAM if requested */
+	vis_vram_limit = (u64)amdgpu_vis_vram_limit * 1024 * 1024;
+	if (amdgpu_vis_vram_limit > 0 &&
+	    vis_vram_limit <= adev->gmc.visible_vram_size)
+		adev->gmc.visible_vram_size = vis_vram_limit;
+
+	/* Change the size here instead of the init above so only lpfn is affected */
+	amdgpu_ttm_set_buffer_funcs_status(adev, false);
+#ifdef CONFIG_64BIT
+	adev->mman.aper_base_kaddr = ioremap_wc(adev->gmc.aper_base,
+						adev->gmc.visible_vram_size);
+#endif
+
+	/*
+	 *The reserved vram for firmware must be pinned to the specified
+	 *place on the VRAM, so reserve it early.
+	 */
+	r = amdgpu_ttm_fw_reserve_vram_init(adev);
+	if (r) {
+		return r;
+	}
+
+	/* allocate memory as required for VGA
+	 * This is used for VGA emulation and pre-OS scanout buffers to
+	 * avoid display artifacts while transitioning between pre-OS
+	 * and driver.  */
+	if (adev->gmc.stolen_size) {
+		r = amdgpu_bo_create_kernel(adev, adev->gmc.stolen_size, PAGE_SIZE,
+					    AMDGPU_GEM_DOMAIN_VRAM,
+					    &adev->stolen_vga_memory,
+					    NULL, NULL);
+		if (r)
+			return r;
+	}
+	DRM_INFO("amdgpu: %uM of VRAM memory ready\n",
+		 (unsigned) (adev->gmc.real_vram_size / (1024 * 1024)));
+
+	/* Compute GTT size, either bsaed on 3/4th the size of RAM size
+	 * or whatever the user passed on module init */
+	if (amdgpu_gtt_size == -1) {
+#ifdef __linux__
+		struct sysinfo si;
+
+		si_meminfo(&si);
+		gtt_size = min(max((AMDGPU_DEFAULT_GTT_SIZE_MB << 20),
+			       adev->gmc.mc_vram_size),
+			       ((uint64_t)si.totalram * si.mem_unit * 3/4));
+#else
+		gtt_size = min(max((AMDGPU_DEFAULT_GTT_SIZE_MB << 20),
+			       adev->gmc.mc_vram_size),
+			       ((uint64_t)ptoa(physmem) * 3/4));
+#endif
+	}
+	else
+		gtt_size = (uint64_t)amdgpu_gtt_size << 20;
+
+	/* Initialize GTT memory pool */
+	r = ttm_bo_init_mm(&adev->mman.bdev, TTM_PL_TT, gtt_size >> PAGE_SHIFT);
+	if (r) {
+		DRM_ERROR("Failed initializing GTT heap.\n");
+		return r;
+	}
+	DRM_INFO("amdgpu: %uM of GTT memory ready.\n",
+		 (unsigned)(gtt_size / (1024 * 1024)));
+
+	/* Initialize various on-chip memory pools */
+	adev->gds.mem.total_size = adev->gds.mem.total_size << AMDGPU_GDS_SHIFT;
+	adev->gds.mem.gfx_partition_size = adev->gds.mem.gfx_partition_size << AMDGPU_GDS_SHIFT;
+	adev->gds.mem.cs_partition_size = adev->gds.mem.cs_partition_size << AMDGPU_GDS_SHIFT;
+	adev->gds.gws.total_size = adev->gds.gws.total_size << AMDGPU_GWS_SHIFT;
+	adev->gds.gws.gfx_partition_size = adev->gds.gws.gfx_partition_size << AMDGPU_GWS_SHIFT;
+	adev->gds.gws.cs_partition_size = adev->gds.gws.cs_partition_size << AMDGPU_GWS_SHIFT;
+	adev->gds.oa.total_size = adev->gds.oa.total_size << AMDGPU_OA_SHIFT;
+	adev->gds.oa.gfx_partition_size = adev->gds.oa.gfx_partition_size << AMDGPU_OA_SHIFT;
+	adev->gds.oa.cs_partition_size = adev->gds.oa.cs_partition_size << AMDGPU_OA_SHIFT;
+	/* GDS Memory */
+	if (adev->gds.mem.total_size) {
+		r = ttm_bo_init_mm(&adev->mman.bdev, AMDGPU_PL_GDS,
+				   adev->gds.mem.total_size >> PAGE_SHIFT);
+		if (r) {
+			DRM_ERROR("Failed initializing GDS heap.\n");
+			return r;
+		}
+	}
+
+	/* GWS */
+	if (adev->gds.gws.total_size) {
+		r = ttm_bo_init_mm(&adev->mman.bdev, AMDGPU_PL_GWS,
+				   adev->gds.gws.total_size >> PAGE_SHIFT);
+		if (r) {
+			DRM_ERROR("Failed initializing gws heap.\n");
+			return r;
+		}
+	}
+
+	/* OA */
+	if (adev->gds.oa.total_size) {
+		r = ttm_bo_init_mm(&adev->mman.bdev, AMDGPU_PL_OA,
+				   adev->gds.oa.total_size >> PAGE_SHIFT);
+		if (r) {
+			DRM_ERROR("Failed initializing oa heap.\n");
+			return r;
+		}
+	}
+
+	/* Register debugfs entries for amdgpu_ttm */
+	r = amdgpu_ttm_debugfs_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init debugfs\n");
+		return r;
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_ttm_late_init - Handle any late initialization for amdgpu_ttm
+ */
+void amdgpu_ttm_late_init(struct amdgpu_device *adev)
+{
+	/* return the VGA stolen memory (if any) back to VRAM */
+	amdgpu_bo_free_kernel(&adev->stolen_vga_memory, NULL, NULL);
+}
+
+/**
+ * amdgpu_ttm_fini - De-initialize the TTM memory pools
+ */
+void amdgpu_ttm_fini(struct amdgpu_device *adev)
+{
+	if (!adev->mman.initialized)
+		return;
+
+	amdgpu_ttm_debugfs_fini(adev);
+	amdgpu_ttm_fw_reserve_vram_fini(adev);
+#ifdef notyet
+	if (adev->mman.aper_base_kaddr)
+		iounmap(adev->mman.aper_base_kaddr);
+#endif
+	adev->mman.aper_base_kaddr = NULL;
+
+	ttm_bo_clean_mm(&adev->mman.bdev, TTM_PL_VRAM);
+	ttm_bo_clean_mm(&adev->mman.bdev, TTM_PL_TT);
+	if (adev->gds.mem.total_size)
+		ttm_bo_clean_mm(&adev->mman.bdev, AMDGPU_PL_GDS);
+	if (adev->gds.gws.total_size)
+		ttm_bo_clean_mm(&adev->mman.bdev, AMDGPU_PL_GWS);
+	if (adev->gds.oa.total_size)
+		ttm_bo_clean_mm(&adev->mman.bdev, AMDGPU_PL_OA);
+	ttm_bo_device_release(&adev->mman.bdev);
+	amdgpu_ttm_global_fini(adev);
+	adev->mman.initialized = false;
+	DRM_INFO("amdgpu: ttm finalized\n");
+}
+
+/**
+ * amdgpu_ttm_set_buffer_funcs_status - enable/disable use of buffer functions
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: true when we can use buffer functions.
+ *
+ * Enable/disable use of buffer functions during suspend/resume. This should
+ * only be called at bootup or when userspace isn't running.
+ */
+void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable)
+{
+	struct ttm_mem_type_manager *man = &adev->mman.bdev.man[TTM_PL_VRAM];
+	uint64_t size;
+	int r;
+
+	if (!adev->mman.initialized || adev->in_gpu_reset ||
+	    adev->mman.buffer_funcs_enabled == enable)
+		return;
+
+	if (enable) {
+		struct amdgpu_ring *ring;
+		struct drm_sched_rq *rq;
+
+		ring = adev->mman.buffer_funcs_ring;
+		rq = &ring->sched.sched_rq[DRM_SCHED_PRIORITY_KERNEL];
+		r = drm_sched_entity_init(&adev->mman.entity, &rq, 1, NULL);
+		if (r) {
+			DRM_ERROR("Failed setting up TTM BO move entity (%d)\n",
+				  r);
+			return;
+		}
+	} else {
+		drm_sched_entity_destroy(&adev->mman.entity);
+		dma_fence_put(man->move);
+		man->move = NULL;
+	}
+
+	/* this just adjusts TTM size idea, which sets lpfn to the correct value */
+	if (enable)
+		size = adev->gmc.real_vram_size;
+	else
+		size = adev->gmc.visible_vram_size;
+	man->size = size >> PAGE_SHIFT;
+	adev->mman.buffer_funcs_enabled = enable;
+}
+
+#ifdef __linux__
+
+int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma)
+{
+	struct drm_file *file_priv;
+	struct amdgpu_device *adev;
+
+	if (unlikely(vma->vm_pgoff < DRM_FILE_PAGE_OFFSET))
+		return -EINVAL;
+
+	file_priv = filp->private_data;
+	adev = file_priv->minor->dev->dev_private;
+	if (adev == NULL)
+		return -EINVAL;
+
+	return ttm_bo_mmap(filp, vma, &adev->mman.bdev);
+}
+
+#else
+
+struct uvm_object *
+amdgpu_mmap(struct drm_device *dev, voff_t off, vsize_t size)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (unlikely(off < DRM_FILE_PAGE_OFFSET))
+		return NULL;
+
+#if 0
+	file_priv = filp->private_data;
+	adev = file_priv->minor->dev->dev_private;
+	if (adev == NULL)
+		return -EINVAL;
+#endif
+
+	return ttm_bo_mmap(off, size, &adev->mman.bdev);
+}
+
+#endif
+
+static int amdgpu_map_buffer(struct ttm_buffer_object *bo,
+			     struct ttm_mem_reg *mem, unsigned num_pages,
+			     uint64_t offset, unsigned window,
+			     struct amdgpu_ring *ring,
+			     uint64_t *addr)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)bo->ttm;
+	struct amdgpu_device *adev = ring->adev;
+	struct ttm_tt *ttm = bo->ttm;
+	struct amdgpu_job *job;
+	unsigned num_dw, num_bytes;
+	dma_addr_t *dma_address;
+	struct dma_fence *fence;
+	uint64_t src_addr, dst_addr;
+	uint64_t flags;
+	int r;
+
+	BUG_ON(adev->mman.buffer_funcs->copy_max_bytes <
+	       AMDGPU_GTT_MAX_TRANSFER_SIZE * 8);
+
+	*addr = adev->gmc.gart_start;
+	*addr += (u64)window * AMDGPU_GTT_MAX_TRANSFER_SIZE *
+		AMDGPU_GPU_PAGE_SIZE;
+
+	num_dw = adev->mman.buffer_funcs->copy_num_dw;
+	while (num_dw & 0x7)
+		num_dw++;
+
+	num_bytes = num_pages * 8;
+
+	r = amdgpu_job_alloc_with_ib(adev, num_dw * 4 + num_bytes, &job);
+	if (r)
+		return r;
+
+	src_addr = num_dw * 4;
+	src_addr += job->ibs[0].gpu_addr;
+
+	dst_addr = adev->gart.table_addr;
+	dst_addr += window * AMDGPU_GTT_MAX_TRANSFER_SIZE * 8;
+	amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr,
+				dst_addr, num_bytes);
+
+	amdgpu_ring_pad_ib(ring, &job->ibs[0]);
+	WARN_ON(job->ibs[0].length_dw > num_dw);
+
+	dma_address = &gtt->ttm.dma_address[offset >> PAGE_SHIFT];
+	flags = amdgpu_ttm_tt_pte_flags(adev, ttm, mem);
+	r = amdgpu_gart_map(adev, 0, num_pages, dma_address, flags,
+			    &job->ibs[0].ptr[num_dw]);
+	if (r)
+		goto error_free;
+
+	r = amdgpu_job_submit(job, &adev->mman.entity,
+			      AMDGPU_FENCE_OWNER_UNDEFINED, &fence);
+	if (r)
+		goto error_free;
+
+	dma_fence_put(fence);
+
+	return r;
+
+error_free:
+	amdgpu_job_free(job);
+	return r;
+}
+
+int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
+		       uint64_t dst_offset, uint32_t byte_count,
+		       struct reservation_object *resv,
+		       struct dma_fence **fence, bool direct_submit,
+		       bool vm_needs_flush)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_job *job;
+
+	uint32_t max_bytes;
+	unsigned num_loops, num_dw;
+	unsigned i;
+	int r;
+
+	if (direct_submit && !ring->ready) {
+		DRM_ERROR("Trying to move memory with ring turned off.\n");
+		return -EINVAL;
+	}
+
+	max_bytes = adev->mman.buffer_funcs->copy_max_bytes;
+	num_loops = DIV_ROUND_UP(byte_count, max_bytes);
+	num_dw = num_loops * adev->mman.buffer_funcs->copy_num_dw;
+
+	/* for IB padding */
+	while (num_dw & 0x7)
+		num_dw++;
+
+	r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, &job);
+	if (r)
+		return r;
+
+	job->vm_needs_flush = vm_needs_flush;
+	if (resv) {
+		r = amdgpu_sync_resv(adev, &job->sync, resv,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     false);
+		if (r) {
+			DRM_ERROR("sync failed (%d).\n", r);
+			goto error_free;
+		}
+	}
+
+	for (i = 0; i < num_loops; i++) {
+		uint32_t cur_size_in_bytes = min(byte_count, max_bytes);
+
+		amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_offset,
+					dst_offset, cur_size_in_bytes);
+
+		src_offset += cur_size_in_bytes;
+		dst_offset += cur_size_in_bytes;
+		byte_count -= cur_size_in_bytes;
+	}
+
+	amdgpu_ring_pad_ib(ring, &job->ibs[0]);
+	WARN_ON(job->ibs[0].length_dw > num_dw);
+	if (direct_submit)
+		r = amdgpu_job_submit_direct(job, ring, fence);
+	else
+		r = amdgpu_job_submit(job, &adev->mman.entity,
+				      AMDGPU_FENCE_OWNER_UNDEFINED, fence);
+	if (r)
+		goto error_free;
+
+	return r;
+
+error_free:
+	amdgpu_job_free(job);
+	DRM_ERROR("Error scheduling IBs (%d)\n", r);
+	return r;
+}
+
+int amdgpu_fill_buffer(struct amdgpu_bo *bo,
+		       uint32_t src_data,
+		       struct reservation_object *resv,
+		       struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	uint32_t max_bytes = adev->mman.buffer_funcs->fill_max_bytes;
+	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+
+	struct drm_mm_node *mm_node;
+	unsigned long num_pages;
+	unsigned int num_loops, num_dw;
+
+	struct amdgpu_job *job;
+	int r;
+
+	if (!adev->mman.buffer_funcs_enabled) {
+		DRM_ERROR("Trying to clear memory with ring turned off.\n");
+		return -EINVAL;
+	}
+
+	if (bo->tbo.mem.mem_type == TTM_PL_TT) {
+		r = amdgpu_ttm_alloc_gart(&bo->tbo);
+		if (r)
+			return r;
+	}
+
+	num_pages = bo->tbo.num_pages;
+	mm_node = bo->tbo.mem.mm_node;
+	num_loops = 0;
+	while (num_pages) {
+		uint32_t byte_count = mm_node->size << PAGE_SHIFT;
+
+		num_loops += DIV_ROUND_UP(byte_count, max_bytes);
+		num_pages -= mm_node->size;
+		++mm_node;
+	}
+	num_dw = num_loops * adev->mman.buffer_funcs->fill_num_dw;
+
+	/* for IB padding */
+	num_dw += 64;
+
+	r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, &job);
+	if (r)
+		return r;
+
+	if (resv) {
+		r = amdgpu_sync_resv(adev, &job->sync, resv,
+				     AMDGPU_FENCE_OWNER_UNDEFINED, false);
+		if (r) {
+			DRM_ERROR("sync failed (%d).\n", r);
+			goto error_free;
+		}
+	}
+
+	num_pages = bo->tbo.num_pages;
+	mm_node = bo->tbo.mem.mm_node;
+
+	while (num_pages) {
+		uint32_t byte_count = mm_node->size << PAGE_SHIFT;
+		uint64_t dst_addr;
+
+		dst_addr = amdgpu_mm_node_addr(&bo->tbo, mm_node, &bo->tbo.mem);
+		while (byte_count) {
+			uint32_t cur_size_in_bytes = min(byte_count, max_bytes);
+
+			amdgpu_emit_fill_buffer(adev, &job->ibs[0], src_data,
+						dst_addr, cur_size_in_bytes);
+
+			dst_addr += cur_size_in_bytes;
+			byte_count -= cur_size_in_bytes;
+		}
+
+		num_pages -= mm_node->size;
+		++mm_node;
+	}
+
+	amdgpu_ring_pad_ib(ring, &job->ibs[0]);
+	WARN_ON(job->ibs[0].length_dw > num_dw);
+	r = amdgpu_job_submit(job, &adev->mman.entity,
+			      AMDGPU_FENCE_OWNER_UNDEFINED, fence);
+	if (r)
+		goto error_free;
+
+	return 0;
+
+error_free:
+	amdgpu_job_free(job);
+	return r;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+static int amdgpu_mm_dump_table(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *)m->private;
+	unsigned ttm_pl = *(int *)node->info_ent->data;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct ttm_mem_type_manager *man = &adev->mman.bdev.man[ttm_pl];
+	struct drm_printer p = drm_seq_file_printer(m);
+
+	man->func->debug(man, &p);
+	return 0;
+}
+
+static int ttm_pl_vram = TTM_PL_VRAM;
+static int ttm_pl_tt = TTM_PL_TT;
+
+static const struct drm_info_list amdgpu_ttm_debugfs_list[] = {
+	{"amdgpu_vram_mm", amdgpu_mm_dump_table, 0, &ttm_pl_vram},
+	{"amdgpu_gtt_mm", amdgpu_mm_dump_table, 0, &ttm_pl_tt},
+	{"ttm_page_pool", ttm_page_alloc_debugfs, 0, NULL},
+#ifdef CONFIG_SWIOTLB
+	{"ttm_dma_page_pool", ttm_dma_page_alloc_debugfs, 0, NULL}
+#endif
+};
+
+/**
+ * amdgpu_ttm_vram_read - Linear read access to VRAM
+ *
+ * Accesses VRAM via MMIO for debugging purposes.
+ */
+static ssize_t amdgpu_ttm_vram_read(struct file *f, char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	if (*pos >= adev->gmc.mc_vram_size)
+		return -ENXIO;
+
+	while (size) {
+		unsigned long flags;
+		uint32_t value;
+
+		if (*pos >= adev->gmc.mc_vram_size)
+			return result;
+
+		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
+		WREG32_NO_KIQ(mmMM_INDEX, ((uint32_t)*pos) | 0x80000000);
+		WREG32_NO_KIQ(mmMM_INDEX_HI, *pos >> 31);
+		value = RREG32_NO_KIQ(mmMM_DATA);
+		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
+
+		r = put_user(value, (uint32_t *)buf);
+		if (r)
+			return r;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	return result;
+}
+
+/**
+ * amdgpu_ttm_vram_write - Linear write access to VRAM
+ *
+ * Accesses VRAM via MMIO for debugging purposes.
+ */
+static ssize_t amdgpu_ttm_vram_write(struct file *f, const char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	if (*pos >= adev->gmc.mc_vram_size)
+		return -ENXIO;
+
+	while (size) {
+		unsigned long flags;
+		uint32_t value;
+
+		if (*pos >= adev->gmc.mc_vram_size)
+			return result;
+
+		r = get_user(value, (uint32_t *)buf);
+		if (r)
+			return r;
+
+		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
+		WREG32_NO_KIQ(mmMM_INDEX, ((uint32_t)*pos) | 0x80000000);
+		WREG32_NO_KIQ(mmMM_INDEX_HI, *pos >> 31);
+		WREG32_NO_KIQ(mmMM_DATA, value);
+		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	return result;
+}
+
+static const struct file_operations amdgpu_ttm_vram_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_ttm_vram_read,
+	.write = amdgpu_ttm_vram_write,
+	.llseek = default_llseek,
+};
+
+#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
+
+/**
+ * amdgpu_ttm_gtt_read - Linear read access to GTT memory
+ */
+static ssize_t amdgpu_ttm_gtt_read(struct file *f, char __user *buf,
+				   size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	ssize_t result = 0;
+	int r;
+
+	while (size) {
+		loff_t p = *pos / PAGE_SIZE;
+		unsigned off = *pos & PAGE_MASK;
+		size_t cur_size = min_t(size_t, size, PAGE_SIZE - off);
+		struct vm_page *page;
+		void *ptr;
+
+		if (p >= adev->gart.num_cpu_pages)
+			return result;
+
+		page = adev->gart.pages[p];
+		if (page) {
+			ptr = kmap(page);
+			ptr += off;
+
+			r = copy_to_user(buf, ptr, cur_size);
+			kunmap(ptr);
+		} else
+			r = clear_user(buf, cur_size);
+
+		if (r)
+			return -EFAULT;
+
+		result += cur_size;
+		buf += cur_size;
+		*pos += cur_size;
+		size -= cur_size;
+	}
+
+	return result;
+}
+
+static const struct file_operations amdgpu_ttm_gtt_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_ttm_gtt_read,
+	.llseek = default_llseek
+};
+
+#endif
+
+/**
+ * amdgpu_iomem_read - Virtual read access to GPU mapped memory
+ *
+ * This function is used to read memory that has been mapped to the
+ * GPU and the known addresses are not physical addresses but instead
+ * bus addresses (e.g., what you'd put in an IB or ring buffer).
+ */
+static ssize_t amdgpu_iomem_read(struct file *f, char __user *buf,
+				 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	struct iommu_domain *dom;
+	ssize_t result = 0;
+	int r;
+
+	/* retrieve the IOMMU domain if any for this device */
+	dom = iommu_get_domain_for_dev(adev->dev);
+
+	while (size) {
+		phys_addr_t addr = *pos & ~PAGE_MASK;
+		loff_t off = *pos & PAGE_MASK;
+		size_t bytes = PAGE_SIZE - off;
+		unsigned long pfn;
+		struct vm_page *p;
+		void *ptr;
+
+		bytes = bytes < size ? bytes : size;
+
+		/* Translate the bus address to a physical address.  If
+		 * the domain is NULL it means there is no IOMMU active
+		 * and the address translation is the identity
+		 */
+		addr = dom ? iommu_iova_to_phys(dom, addr) : addr;
+
+		pfn = addr >> PAGE_SHIFT;
+		if (!pfn_valid(pfn))
+			return -EPERM;
+
+		p = pfn_to_page(pfn);
+		if (p->mapping != adev->mman.bdev.dev_mapping)
+			return -EPERM;
+
+		ptr = kmap(p);
+		r = copy_to_user(buf, ptr + off, bytes);
+		kunmap(ptr);
+		if (r)
+			return -EFAULT;
+
+		size -= bytes;
+		*pos += bytes;
+		result += bytes;
+	}
+
+	return result;
+}
+
+/**
+ * amdgpu_iomem_write - Virtual write access to GPU mapped memory
+ *
+ * This function is used to write memory that has been mapped to the
+ * GPU and the known addresses are not physical addresses but instead
+ * bus addresses (e.g., what you'd put in an IB or ring buffer).
+ */
+static ssize_t amdgpu_iomem_write(struct file *f, const char __user *buf,
+				 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = file_inode(f)->i_private;
+	struct iommu_domain *dom;
+	ssize_t result = 0;
+	int r;
+
+	dom = iommu_get_domain_for_dev(adev->dev);
+
+	while (size) {
+		phys_addr_t addr = *pos & ~PAGE_MASK;
+		loff_t off = *pos & PAGE_MASK;
+		size_t bytes = PAGE_SIZE - off;
+		unsigned long pfn;
+		struct vm_page *p;
+		void *ptr;
+
+		bytes = bytes < size ? bytes : size;
+
+		addr = dom ? iommu_iova_to_phys(dom, addr) : addr;
+
+		pfn = addr >> PAGE_SHIFT;
+		if (!pfn_valid(pfn))
+			return -EPERM;
+
+		p = pfn_to_page(pfn);
+		if (p->mapping != adev->mman.bdev.dev_mapping)
+			return -EPERM;
+
+		ptr = kmap(p);
+		r = copy_from_user(ptr + off, buf, bytes);
+		kunmap(ptr);
+		if (r)
+			return -EFAULT;
+
+		size -= bytes;
+		*pos += bytes;
+		result += bytes;
+	}
+
+	return result;
+}
+
+static const struct file_operations amdgpu_ttm_iomem_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_iomem_read,
+	.write = amdgpu_iomem_write,
+	.llseek = default_llseek
+};
+
+static const struct {
+	char *name;
+	const struct file_operations *fops;
+	int domain;
+} ttm_debugfs_entries[] = {
+	{ "amdgpu_vram", &amdgpu_ttm_vram_fops, TTM_PL_VRAM },
+#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
+	{ "amdgpu_gtt", &amdgpu_ttm_gtt_fops, TTM_PL_TT },
+#endif
+	{ "amdgpu_iomem", &amdgpu_ttm_iomem_fops, TTM_PL_SYSTEM },
+};
+
+#endif
+
+static int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	unsigned count;
+
+	struct drm_minor *minor = adev->ddev->primary;
+	struct dentry *ent, *root = minor->debugfs_root;
+
+	for (count = 0; count < ARRAY_SIZE(ttm_debugfs_entries); count++) {
+		ent = debugfs_create_file(
+				ttm_debugfs_entries[count].name,
+				S_IFREG | S_IRUGO, root,
+				adev,
+				ttm_debugfs_entries[count].fops);
+		if (IS_ERR(ent))
+			return PTR_ERR(ent);
+		if (ttm_debugfs_entries[count].domain == TTM_PL_VRAM)
+			i_size_write(ent->d_inode, adev->gmc.mc_vram_size);
+		else if (ttm_debugfs_entries[count].domain == TTM_PL_TT)
+			i_size_write(ent->d_inode, adev->gmc.gart_size);
+		adev->mman.debugfs_entries[count] = ent;
+	}
+
+	count = ARRAY_SIZE(amdgpu_ttm_debugfs_list);
+
+#ifdef CONFIG_SWIOTLB
+	if (!(adev->need_swiotlb && swiotlb_nr_tbl()))
+		--count;
+#endif
+
+	return amdgpu_debugfs_add_files(adev, amdgpu_ttm_debugfs_list, count);
+#else
+	return 0;
+#endif
+}
+
+static void amdgpu_ttm_debugfs_fini(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	unsigned i;
+
+	for (i = 0; i < ARRAY_SIZE(ttm_debugfs_entries); i++)
+		debugfs_remove(adev->mman.debugfs_entries[i]);
+#endif
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ttm.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ttm.h
new file mode 100644
index 00000000000..7abcdd298d2
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ttm.h
@@ -0,0 +1,126 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_TTM_H__
+#define __AMDGPU_TTM_H__
+
+#include "amdgpu.h"
+#include <drm/gpu_scheduler.h>
+
+#define AMDGPU_PL_GDS		(TTM_PL_PRIV + 0)
+#define AMDGPU_PL_GWS		(TTM_PL_PRIV + 1)
+#define AMDGPU_PL_OA		(TTM_PL_PRIV + 2)
+
+#define AMDGPU_PL_FLAG_GDS		(TTM_PL_FLAG_PRIV << 0)
+#define AMDGPU_PL_FLAG_GWS		(TTM_PL_FLAG_PRIV << 1)
+#define AMDGPU_PL_FLAG_OA		(TTM_PL_FLAG_PRIV << 2)
+
+#define AMDGPU_GTT_MAX_TRANSFER_SIZE	512
+#define AMDGPU_GTT_NUM_TRANSFER_WINDOWS	2
+
+struct amdgpu_mman {
+	struct ttm_bo_global_ref        bo_global_ref;
+	struct drm_global_reference	mem_global_ref;
+	struct ttm_bo_device		bdev;
+	bool				mem_global_referenced;
+	bool				initialized;
+	void __iomem			*aper_base_kaddr;
+
+#if defined(CONFIG_DEBUG_FS)
+	struct dentry			*debugfs_entries[8];
+#endif
+
+	/* buffer handling */
+	const struct amdgpu_buffer_funcs	*buffer_funcs;
+	struct amdgpu_ring			*buffer_funcs_ring;
+	bool					buffer_funcs_enabled;
+
+	struct rwlock				gtt_window_lock;
+	/* Scheduler entity for buffer moves */
+	struct drm_sched_entity			entity;
+};
+
+struct amdgpu_copy_mem {
+	struct ttm_buffer_object	*bo;
+	struct ttm_mem_reg		*mem;
+	unsigned long			offset;
+};
+
+extern const struct ttm_mem_type_manager_func amdgpu_gtt_mgr_func;
+extern const struct ttm_mem_type_manager_func amdgpu_vram_mgr_func;
+
+bool amdgpu_gtt_mgr_has_gart_addr(struct ttm_mem_reg *mem);
+uint64_t amdgpu_gtt_mgr_usage(struct ttm_mem_type_manager *man);
+int amdgpu_gtt_mgr_recover(struct ttm_mem_type_manager *man);
+
+u64 amdgpu_vram_mgr_bo_visible_size(struct amdgpu_bo *bo);
+uint64_t amdgpu_vram_mgr_usage(struct ttm_mem_type_manager *man);
+uint64_t amdgpu_vram_mgr_vis_usage(struct ttm_mem_type_manager *man);
+
+int amdgpu_ttm_init(struct amdgpu_device *adev);
+void amdgpu_ttm_late_init(struct amdgpu_device *adev);
+void amdgpu_ttm_fini(struct amdgpu_device *adev);
+void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev,
+					bool enable);
+
+int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
+		       uint64_t dst_offset, uint32_t byte_count,
+		       struct reservation_object *resv,
+		       struct dma_fence **fence, bool direct_submit,
+		       bool vm_needs_flush);
+int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
+			       struct amdgpu_copy_mem *src,
+			       struct amdgpu_copy_mem *dst,
+			       uint64_t size,
+			       struct reservation_object *resv,
+			       struct dma_fence **f);
+int amdgpu_fill_buffer(struct amdgpu_bo *bo,
+			uint32_t src_data,
+			struct reservation_object *resv,
+			struct dma_fence **fence);
+
+#ifdef __linux__
+int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma);
+#else
+struct uvm_object *amdgpu_mmap(struct drm_device *, voff_t, vsize_t);
+#endif
+int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo);
+int amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo);
+
+int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct vm_page **pages);
+void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct vm_page **pages);
+void amdgpu_ttm_tt_mark_user_pages(struct ttm_tt *ttm);
+int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
+				     uint32_t flags);
+bool amdgpu_ttm_tt_has_userptr(struct ttm_tt *ttm);
+struct mm_struct *amdgpu_ttm_tt_get_usermm(struct ttm_tt *ttm);
+bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
+				  unsigned long end);
+bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm,
+				       int *last_invalidated);
+bool amdgpu_ttm_tt_userptr_needs_pages(struct ttm_tt *ttm);
+bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm);
+uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
+				 struct ttm_mem_reg *mem);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ucode.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ucode.c
new file mode 100644
index 00000000000..ae53fa8130a
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ucode.c
@@ -0,0 +1,496 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+
+static void amdgpu_ucode_print_common_hdr(const struct common_firmware_header *hdr)
+{
+	DRM_DEBUG("size_bytes: %u\n", le32_to_cpu(hdr->size_bytes));
+	DRM_DEBUG("header_size_bytes: %u\n", le32_to_cpu(hdr->header_size_bytes));
+	DRM_DEBUG("header_version_major: %u\n", le16_to_cpu(hdr->header_version_major));
+	DRM_DEBUG("header_version_minor: %u\n", le16_to_cpu(hdr->header_version_minor));
+	DRM_DEBUG("ip_version_major: %u\n", le16_to_cpu(hdr->ip_version_major));
+	DRM_DEBUG("ip_version_minor: %u\n", le16_to_cpu(hdr->ip_version_minor));
+	DRM_DEBUG("ucode_version: 0x%08x\n", le32_to_cpu(hdr->ucode_version));
+	DRM_DEBUG("ucode_size_bytes: %u\n", le32_to_cpu(hdr->ucode_size_bytes));
+	DRM_DEBUG("ucode_array_offset_bytes: %u\n",
+		  le32_to_cpu(hdr->ucode_array_offset_bytes));
+	DRM_DEBUG("crc32: 0x%08x\n", le32_to_cpu(hdr->crc32));
+}
+
+void amdgpu_ucode_print_mc_hdr(const struct common_firmware_header *hdr)
+{
+	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
+	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+
+	DRM_DEBUG("MC\n");
+	amdgpu_ucode_print_common_hdr(hdr);
+
+	if (version_major == 1) {
+		const struct mc_firmware_header_v1_0 *mc_hdr =
+			container_of(hdr, struct mc_firmware_header_v1_0, header);
+
+		DRM_DEBUG("io_debug_size_bytes: %u\n",
+			  le32_to_cpu(mc_hdr->io_debug_size_bytes));
+		DRM_DEBUG("io_debug_array_offset_bytes: %u\n",
+			  le32_to_cpu(mc_hdr->io_debug_array_offset_bytes));
+	} else {
+		DRM_ERROR("Unknown MC ucode version: %u.%u\n", version_major, version_minor);
+	}
+}
+
+void amdgpu_ucode_print_smc_hdr(const struct common_firmware_header *hdr)
+{
+	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
+	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+
+	DRM_DEBUG("SMC\n");
+	amdgpu_ucode_print_common_hdr(hdr);
+
+	if (version_major == 1) {
+		const struct smc_firmware_header_v1_0 *smc_hdr =
+			container_of(hdr, struct smc_firmware_header_v1_0, header);
+
+		DRM_DEBUG("ucode_start_addr: %u\n", le32_to_cpu(smc_hdr->ucode_start_addr));
+	} else {
+		DRM_ERROR("Unknown SMC ucode version: %u.%u\n", version_major, version_minor);
+	}
+}
+
+void amdgpu_ucode_print_gfx_hdr(const struct common_firmware_header *hdr)
+{
+	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
+	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+
+	DRM_DEBUG("GFX\n");
+	amdgpu_ucode_print_common_hdr(hdr);
+
+	if (version_major == 1) {
+		const struct gfx_firmware_header_v1_0 *gfx_hdr =
+			container_of(hdr, struct gfx_firmware_header_v1_0, header);
+
+		DRM_DEBUG("ucode_feature_version: %u\n",
+			  le32_to_cpu(gfx_hdr->ucode_feature_version));
+		DRM_DEBUG("jt_offset: %u\n", le32_to_cpu(gfx_hdr->jt_offset));
+		DRM_DEBUG("jt_size: %u\n", le32_to_cpu(gfx_hdr->jt_size));
+	} else {
+		DRM_ERROR("Unknown GFX ucode version: %u.%u\n", version_major, version_minor);
+	}
+}
+
+void amdgpu_ucode_print_rlc_hdr(const struct common_firmware_header *hdr)
+{
+	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
+	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+
+	DRM_DEBUG("RLC\n");
+	amdgpu_ucode_print_common_hdr(hdr);
+
+	if (version_major == 1) {
+		const struct rlc_firmware_header_v1_0 *rlc_hdr =
+			container_of(hdr, struct rlc_firmware_header_v1_0, header);
+
+		DRM_DEBUG("ucode_feature_version: %u\n",
+			  le32_to_cpu(rlc_hdr->ucode_feature_version));
+		DRM_DEBUG("save_and_restore_offset: %u\n",
+			  le32_to_cpu(rlc_hdr->save_and_restore_offset));
+		DRM_DEBUG("clear_state_descriptor_offset: %u\n",
+			  le32_to_cpu(rlc_hdr->clear_state_descriptor_offset));
+		DRM_DEBUG("avail_scratch_ram_locations: %u\n",
+			  le32_to_cpu(rlc_hdr->avail_scratch_ram_locations));
+		DRM_DEBUG("master_pkt_description_offset: %u\n",
+			  le32_to_cpu(rlc_hdr->master_pkt_description_offset));
+	} else if (version_major == 2) {
+		const struct rlc_firmware_header_v2_0 *rlc_hdr =
+			container_of(hdr, struct rlc_firmware_header_v2_0, header);
+
+		DRM_DEBUG("ucode_feature_version: %u\n",
+			  le32_to_cpu(rlc_hdr->ucode_feature_version));
+		DRM_DEBUG("jt_offset: %u\n", le32_to_cpu(rlc_hdr->jt_offset));
+		DRM_DEBUG("jt_size: %u\n", le32_to_cpu(rlc_hdr->jt_size));
+		DRM_DEBUG("save_and_restore_offset: %u\n",
+			  le32_to_cpu(rlc_hdr->save_and_restore_offset));
+		DRM_DEBUG("clear_state_descriptor_offset: %u\n",
+			  le32_to_cpu(rlc_hdr->clear_state_descriptor_offset));
+		DRM_DEBUG("avail_scratch_ram_locations: %u\n",
+			  le32_to_cpu(rlc_hdr->avail_scratch_ram_locations));
+		DRM_DEBUG("reg_restore_list_size: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_restore_list_size));
+		DRM_DEBUG("reg_list_format_start: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_format_start));
+		DRM_DEBUG("reg_list_format_separate_start: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_format_separate_start));
+		DRM_DEBUG("starting_offsets_start: %u\n",
+			  le32_to_cpu(rlc_hdr->starting_offsets_start));
+		DRM_DEBUG("reg_list_format_size_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_format_size_bytes));
+		DRM_DEBUG("reg_list_format_array_offset_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_format_array_offset_bytes));
+		DRM_DEBUG("reg_list_size_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_size_bytes));
+		DRM_DEBUG("reg_list_array_offset_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_array_offset_bytes));
+		DRM_DEBUG("reg_list_format_separate_size_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_format_separate_size_bytes));
+		DRM_DEBUG("reg_list_format_separate_array_offset_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_format_separate_array_offset_bytes));
+		DRM_DEBUG("reg_list_separate_size_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_separate_size_bytes));
+		DRM_DEBUG("reg_list_separate_array_offset_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_separate_array_offset_bytes));
+		if (version_minor == 1) {
+			const struct rlc_firmware_header_v2_1 *v2_1 =
+				container_of(rlc_hdr, struct rlc_firmware_header_v2_1, v2_0);
+			DRM_DEBUG("reg_list_format_direct_reg_list_length: %u\n",
+				  le32_to_cpu(v2_1->reg_list_format_direct_reg_list_length));
+			DRM_DEBUG("save_restore_list_cntl_ucode_ver: %u\n",
+				  le32_to_cpu(v2_1->save_restore_list_cntl_ucode_ver));
+			DRM_DEBUG("save_restore_list_cntl_feature_ver: %u\n",
+				  le32_to_cpu(v2_1->save_restore_list_cntl_feature_ver));
+			DRM_DEBUG("save_restore_list_cntl_size_bytes %u\n",
+				  le32_to_cpu(v2_1->save_restore_list_cntl_size_bytes));
+			DRM_DEBUG("save_restore_list_cntl_offset_bytes: %u\n",
+				  le32_to_cpu(v2_1->save_restore_list_cntl_offset_bytes));
+			DRM_DEBUG("save_restore_list_gpm_ucode_ver: %u\n",
+				  le32_to_cpu(v2_1->save_restore_list_gpm_ucode_ver));
+			DRM_DEBUG("save_restore_list_gpm_feature_ver: %u\n",
+				  le32_to_cpu(v2_1->save_restore_list_gpm_feature_ver));
+			DRM_DEBUG("save_restore_list_gpm_size_bytes %u\n",
+				  le32_to_cpu(v2_1->save_restore_list_gpm_size_bytes));
+			DRM_DEBUG("save_restore_list_gpm_offset_bytes: %u\n",
+				  le32_to_cpu(v2_1->save_restore_list_gpm_offset_bytes));
+			DRM_DEBUG("save_restore_list_srm_ucode_ver: %u\n",
+				  le32_to_cpu(v2_1->save_restore_list_srm_ucode_ver));
+			DRM_DEBUG("save_restore_list_srm_feature_ver: %u\n",
+				  le32_to_cpu(v2_1->save_restore_list_srm_feature_ver));
+			DRM_DEBUG("save_restore_list_srm_size_bytes %u\n",
+				  le32_to_cpu(v2_1->save_restore_list_srm_size_bytes));
+			DRM_DEBUG("save_restore_list_srm_offset_bytes: %u\n",
+				  le32_to_cpu(v2_1->save_restore_list_srm_offset_bytes));
+		}
+	} else {
+		DRM_ERROR("Unknown RLC ucode version: %u.%u\n", version_major, version_minor);
+	}
+}
+
+void amdgpu_ucode_print_sdma_hdr(const struct common_firmware_header *hdr)
+{
+	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
+	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+
+	DRM_DEBUG("SDMA\n");
+	amdgpu_ucode_print_common_hdr(hdr);
+
+	if (version_major == 1) {
+		const struct sdma_firmware_header_v1_0 *sdma_hdr =
+			container_of(hdr, struct sdma_firmware_header_v1_0, header);
+
+		DRM_DEBUG("ucode_feature_version: %u\n",
+			  le32_to_cpu(sdma_hdr->ucode_feature_version));
+		DRM_DEBUG("ucode_change_version: %u\n",
+			  le32_to_cpu(sdma_hdr->ucode_change_version));
+		DRM_DEBUG("jt_offset: %u\n", le32_to_cpu(sdma_hdr->jt_offset));
+		DRM_DEBUG("jt_size: %u\n", le32_to_cpu(sdma_hdr->jt_size));
+		if (version_minor >= 1) {
+			const struct sdma_firmware_header_v1_1 *sdma_v1_1_hdr =
+				container_of(sdma_hdr, struct sdma_firmware_header_v1_1, v1_0);
+			DRM_DEBUG("digest_size: %u\n", le32_to_cpu(sdma_v1_1_hdr->digest_size));
+		}
+	} else {
+		DRM_ERROR("Unknown SDMA ucode version: %u.%u\n",
+			  version_major, version_minor);
+	}
+}
+
+void amdgpu_ucode_print_gpu_info_hdr(const struct common_firmware_header *hdr)
+{
+	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
+	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+
+	DRM_DEBUG("GPU_INFO\n");
+	amdgpu_ucode_print_common_hdr(hdr);
+
+	if (version_major == 1) {
+		const struct gpu_info_firmware_header_v1_0 *gpu_info_hdr =
+			container_of(hdr, struct gpu_info_firmware_header_v1_0, header);
+
+		DRM_DEBUG("version_major: %u\n",
+			  le16_to_cpu(gpu_info_hdr->version_major));
+		DRM_DEBUG("version_minor: %u\n",
+			  le16_to_cpu(gpu_info_hdr->version_minor));
+	} else {
+		DRM_ERROR("Unknown gpu_info ucode version: %u.%u\n", version_major, version_minor);
+	}
+}
+
+int amdgpu_ucode_validate(const struct firmware *fw)
+{
+	const struct common_firmware_header *hdr =
+		(const struct common_firmware_header *)fw->data;
+
+	if (fw->size == le32_to_cpu(hdr->size_bytes))
+		return 0;
+
+	return -EINVAL;
+}
+
+bool amdgpu_ucode_hdr_version(union amdgpu_firmware_header *hdr,
+				uint16_t hdr_major, uint16_t hdr_minor)
+{
+	if ((hdr->common.header_version_major == hdr_major) &&
+		(hdr->common.header_version_minor == hdr_minor))
+		return false;
+	return true;
+}
+
+enum amdgpu_firmware_load_type
+amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type)
+{
+	switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+	case CHIP_OLAND:
+	case CHIP_HAINAN:
+		return AMDGPU_FW_LOAD_DIRECT;
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_BONAIRE:
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_HAWAII:
+	case CHIP_MULLINS:
+		return AMDGPU_FW_LOAD_DIRECT;
+#endif
+	case CHIP_TOPAZ:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		if (!load_type)
+			return AMDGPU_FW_LOAD_DIRECT;
+		else
+			return AMDGPU_FW_LOAD_SMU;
+	case CHIP_VEGA10:
+	case CHIP_RAVEN:
+	case CHIP_VEGA12:
+		if (!load_type)
+			return AMDGPU_FW_LOAD_DIRECT;
+		else
+			return AMDGPU_FW_LOAD_PSP;
+	case CHIP_VEGA20:
+		return AMDGPU_FW_LOAD_DIRECT;
+	default:
+		DRM_ERROR("Unknown firmware load type\n");
+	}
+
+	return AMDGPU_FW_LOAD_DIRECT;
+}
+
+static int amdgpu_ucode_init_single_fw(struct amdgpu_device *adev,
+				       struct amdgpu_firmware_info *ucode,
+				       uint64_t mc_addr, void *kptr)
+{
+	const struct common_firmware_header *header = NULL;
+	const struct gfx_firmware_header_v1_0 *cp_hdr = NULL;
+
+	if (NULL == ucode->fw)
+		return 0;
+
+	ucode->mc_addr = mc_addr;
+	ucode->kaddr = kptr;
+
+	if (ucode->ucode_id == AMDGPU_UCODE_ID_STORAGE)
+		return 0;
+
+	header = (const struct common_firmware_header *)ucode->fw->data;
+
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)ucode->fw->data;
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP ||
+	    (ucode->ucode_id != AMDGPU_UCODE_ID_CP_MEC1 &&
+	     ucode->ucode_id != AMDGPU_UCODE_ID_CP_MEC2 &&
+	     ucode->ucode_id != AMDGPU_UCODE_ID_CP_MEC1_JT &&
+	     ucode->ucode_id != AMDGPU_UCODE_ID_CP_MEC2_JT &&
+	     ucode->ucode_id != AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL &&
+	     ucode->ucode_id != AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM &&
+	     ucode->ucode_id != AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM)) {
+		ucode->ucode_size = le32_to_cpu(header->ucode_size_bytes);
+
+		memcpy(ucode->kaddr, (void *)((uint8_t *)ucode->fw->data +
+					      le32_to_cpu(header->ucode_array_offset_bytes)),
+		       ucode->ucode_size);
+	} else if (ucode->ucode_id == AMDGPU_UCODE_ID_CP_MEC1 ||
+		   ucode->ucode_id == AMDGPU_UCODE_ID_CP_MEC2) {
+		ucode->ucode_size = le32_to_cpu(header->ucode_size_bytes) -
+			le32_to_cpu(cp_hdr->jt_size) * 4;
+
+		memcpy(ucode->kaddr, (void *)((uint8_t *)ucode->fw->data +
+					      le32_to_cpu(header->ucode_array_offset_bytes)),
+		       ucode->ucode_size);
+	} else if (ucode->ucode_id == AMDGPU_UCODE_ID_CP_MEC1_JT ||
+		   ucode->ucode_id == AMDGPU_UCODE_ID_CP_MEC2_JT) {
+		ucode->ucode_size = le32_to_cpu(cp_hdr->jt_size) * 4;
+
+		memcpy(ucode->kaddr, (void *)((uint8_t *)ucode->fw->data +
+					      le32_to_cpu(header->ucode_array_offset_bytes) +
+					      le32_to_cpu(cp_hdr->jt_offset) * 4),
+		       ucode->ucode_size);
+	} else if (ucode->ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL) {
+		ucode->ucode_size = adev->gfx.rlc.save_restore_list_cntl_size_bytes;
+		memcpy(ucode->kaddr, adev->gfx.rlc.save_restore_list_cntl,
+		       ucode->ucode_size);
+	} else if (ucode->ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM) {
+		ucode->ucode_size = adev->gfx.rlc.save_restore_list_gpm_size_bytes;
+		memcpy(ucode->kaddr, adev->gfx.rlc.save_restore_list_gpm,
+		       ucode->ucode_size);
+	} else if (ucode->ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM) {
+		ucode->ucode_size = adev->gfx.rlc.save_restore_list_srm_size_bytes;
+		memcpy(ucode->kaddr, adev->gfx.rlc.save_restore_list_srm,
+		       ucode->ucode_size);
+	}
+
+	return 0;
+}
+
+static int amdgpu_ucode_patch_jt(struct amdgpu_firmware_info *ucode,
+				uint64_t mc_addr, void *kptr)
+{
+	const struct gfx_firmware_header_v1_0 *header = NULL;
+	const struct common_firmware_header *comm_hdr = NULL;
+	uint8_t* src_addr = NULL;
+	uint8_t* dst_addr = NULL;
+
+	if (NULL == ucode->fw)
+		return 0;
+
+	comm_hdr = (const struct common_firmware_header *)ucode->fw->data;
+	header = (const struct gfx_firmware_header_v1_0 *)ucode->fw->data;
+	dst_addr = ucode->kaddr +
+			   roundup2(le32_to_cpu(comm_hdr->ucode_size_bytes),
+			   PAGE_SIZE);
+	src_addr = (uint8_t *)ucode->fw->data +
+			   le32_to_cpu(comm_hdr->ucode_array_offset_bytes) +
+			   (le32_to_cpu(header->jt_offset) * 4);
+	memcpy(dst_addr, src_addr, le32_to_cpu(header->jt_size) * 4);
+
+	return 0;
+}
+
+int amdgpu_ucode_init_bo(struct amdgpu_device *adev)
+{
+	uint64_t fw_offset = 0;
+	int i, err;
+	struct amdgpu_firmware_info *ucode = NULL;
+	const struct common_firmware_header *header = NULL;
+
+	if (!adev->firmware.fw_size) {
+		dev_warn(adev->dev, "No ip firmware need to load\n");
+		return 0;
+	}
+
+	if (!adev->in_gpu_reset) {
+		err = amdgpu_bo_create_kernel(adev, adev->firmware.fw_size, PAGE_SIZE,
+					amdgpu_sriov_vf(adev) ? AMDGPU_GEM_DOMAIN_VRAM : AMDGPU_GEM_DOMAIN_GTT,
+					&adev->firmware.fw_buf,
+					&adev->firmware.fw_buf_mc,
+					&adev->firmware.fw_buf_ptr);
+		if (err) {
+			dev_err(adev->dev, "failed to create kernel buffer for firmware.fw_buf\n");
+			goto failed;
+		}
+	}
+
+	memset(adev->firmware.fw_buf_ptr, 0, adev->firmware.fw_size);
+
+	/*
+	 * if SMU loaded firmware, it needn't add SMC, UVD, and VCE
+	 * ucode info here
+	 */
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+		if (amdgpu_sriov_vf(adev))
+			adev->firmware.max_ucodes = AMDGPU_UCODE_ID_MAXIMUM - 3;
+		else
+			adev->firmware.max_ucodes = AMDGPU_UCODE_ID_MAXIMUM - 4;
+	} else {
+		adev->firmware.max_ucodes = AMDGPU_UCODE_ID_MAXIMUM;
+	}
+
+	for (i = 0; i < adev->firmware.max_ucodes; i++) {
+		ucode = &adev->firmware.ucode[i];
+		if (ucode->fw) {
+			header = (const struct common_firmware_header *)ucode->fw->data;
+			amdgpu_ucode_init_single_fw(adev, ucode, adev->firmware.fw_buf_mc + fw_offset,
+						    adev->firmware.fw_buf_ptr + fw_offset);
+			if (i == AMDGPU_UCODE_ID_CP_MEC1 &&
+			    adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+				const struct gfx_firmware_header_v1_0 *cp_hdr;
+				cp_hdr = (const struct gfx_firmware_header_v1_0 *)ucode->fw->data;
+				amdgpu_ucode_patch_jt(ucode,  adev->firmware.fw_buf_mc + fw_offset,
+						    adev->firmware.fw_buf_ptr + fw_offset);
+				fw_offset += roundup2(le32_to_cpu(cp_hdr->jt_size) << 2, PAGE_SIZE);
+			}
+			fw_offset += roundup2(ucode->ucode_size, PAGE_SIZE);
+		}
+	}
+	return 0;
+
+failed:
+	if (err)
+		adev->firmware.load_type = AMDGPU_FW_LOAD_DIRECT;
+
+	return err;
+}
+
+int amdgpu_ucode_fini_bo(struct amdgpu_device *adev)
+{
+	int i;
+	struct amdgpu_firmware_info *ucode = NULL;
+
+	if (!adev->firmware.fw_size)
+		return 0;
+
+	for (i = 0; i < adev->firmware.max_ucodes; i++) {
+		ucode = &adev->firmware.ucode[i];
+		if (ucode->fw) {
+			ucode->mc_addr = 0;
+			ucode->kaddr = NULL;
+		}
+	}
+
+	amdgpu_bo_free_kernel(&adev->firmware.fw_buf,
+				&adev->firmware.fw_buf_mc,
+				&adev->firmware.fw_buf_ptr);
+
+	return 0;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_ucode.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ucode.h
new file mode 100644
index 00000000000..bdc472b6e64
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_ucode.h
@@ -0,0 +1,250 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_UCODE_H__
+#define __AMDGPU_UCODE_H__
+
+struct common_firmware_header {
+	uint32_t size_bytes; /* size of the entire header+image(s) in bytes */
+	uint32_t header_size_bytes; /* size of just the header in bytes */
+	uint16_t header_version_major; /* header version */
+	uint16_t header_version_minor; /* header version */
+	uint16_t ip_version_major; /* IP version */
+	uint16_t ip_version_minor; /* IP version */
+	uint32_t ucode_version;
+	uint32_t ucode_size_bytes; /* size of ucode in bytes */
+	uint32_t ucode_array_offset_bytes; /* payload offset from the start of the header */
+	uint32_t crc32;  /* crc32 checksum of the payload */
+};
+
+/* version_major=1, version_minor=0 */
+struct mc_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t io_debug_size_bytes; /* size of debug array in dwords */
+	uint32_t io_debug_array_offset_bytes; /* payload offset from the start of the header */
+};
+
+/* version_major=1, version_minor=0 */
+struct smc_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_start_addr;
+};
+
+/* version_major=1, version_minor=0 */
+struct psp_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_feature_version;
+	uint32_t sos_offset_bytes;
+	uint32_t sos_size_bytes;
+};
+
+/* version_major=1, version_minor=0 */
+struct gfx_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_feature_version;
+	uint32_t jt_offset; /* jt location */
+	uint32_t jt_size;  /* size of jt */
+};
+
+/* version_major=1, version_minor=0 */
+struct rlc_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_feature_version;
+	uint32_t save_and_restore_offset;
+	uint32_t clear_state_descriptor_offset;
+	uint32_t avail_scratch_ram_locations;
+	uint32_t master_pkt_description_offset;
+};
+
+/* version_major=2, version_minor=0 */
+struct rlc_firmware_header_v2_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_feature_version;
+	uint32_t jt_offset; /* jt location */
+	uint32_t jt_size;  /* size of jt */
+	uint32_t save_and_restore_offset;
+	uint32_t clear_state_descriptor_offset;
+	uint32_t avail_scratch_ram_locations;
+	uint32_t reg_restore_list_size;
+	uint32_t reg_list_format_start;
+	uint32_t reg_list_format_separate_start;
+	uint32_t starting_offsets_start;
+	uint32_t reg_list_format_size_bytes; /* size of reg list format array in bytes */
+	uint32_t reg_list_format_array_offset_bytes; /* payload offset from the start of the header */
+	uint32_t reg_list_size_bytes; /* size of reg list array in bytes */
+	uint32_t reg_list_array_offset_bytes; /* payload offset from the start of the header */
+	uint32_t reg_list_format_separate_size_bytes; /* size of reg list format array in bytes */
+	uint32_t reg_list_format_separate_array_offset_bytes; /* payload offset from the start of the header */
+	uint32_t reg_list_separate_size_bytes; /* size of reg list array in bytes */
+	uint32_t reg_list_separate_array_offset_bytes; /* payload offset from the start of the header */
+};
+
+/* version_major=2, version_minor=1 */
+struct rlc_firmware_header_v2_1 {
+	struct rlc_firmware_header_v2_0 v2_0;
+	uint32_t reg_list_format_direct_reg_list_length; /* length of direct reg list format array */
+	uint32_t save_restore_list_cntl_ucode_ver;
+	uint32_t save_restore_list_cntl_feature_ver;
+	uint32_t save_restore_list_cntl_size_bytes;
+	uint32_t save_restore_list_cntl_offset_bytes;
+	uint32_t save_restore_list_gpm_ucode_ver;
+	uint32_t save_restore_list_gpm_feature_ver;
+	uint32_t save_restore_list_gpm_size_bytes;
+	uint32_t save_restore_list_gpm_offset_bytes;
+	uint32_t save_restore_list_srm_ucode_ver;
+	uint32_t save_restore_list_srm_feature_ver;
+	uint32_t save_restore_list_srm_size_bytes;
+	uint32_t save_restore_list_srm_offset_bytes;
+};
+
+/* version_major=1, version_minor=0 */
+struct sdma_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_feature_version;
+	uint32_t ucode_change_version;
+	uint32_t jt_offset; /* jt location */
+	uint32_t jt_size; /* size of jt */
+};
+
+/* version_major=1, version_minor=1 */
+struct sdma_firmware_header_v1_1 {
+	struct sdma_firmware_header_v1_0 v1_0;
+	uint32_t digest_size;
+};
+
+/* gpu info payload */
+struct gpu_info_firmware_v1_0 {
+	uint32_t gc_num_se;
+	uint32_t gc_num_cu_per_sh;
+	uint32_t gc_num_sh_per_se;
+	uint32_t gc_num_rb_per_se;
+	uint32_t gc_num_tccs;
+	uint32_t gc_num_gprs;
+	uint32_t gc_num_max_gs_thds;
+	uint32_t gc_gs_table_depth;
+	uint32_t gc_gsprim_buff_depth;
+	uint32_t gc_parameter_cache_depth;
+	uint32_t gc_double_offchip_lds_buffer;
+	uint32_t gc_wave_size;
+	uint32_t gc_max_waves_per_simd;
+	uint32_t gc_max_scratch_slots_per_cu;
+	uint32_t gc_lds_size;
+};
+
+/* version_major=1, version_minor=0 */
+struct gpu_info_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint16_t version_major; /* version */
+	uint16_t version_minor; /* version */
+};
+
+/* header is fixed size */
+union amdgpu_firmware_header {
+	struct common_firmware_header common;
+	struct mc_firmware_header_v1_0 mc;
+	struct smc_firmware_header_v1_0 smc;
+	struct psp_firmware_header_v1_0 psp;
+	struct gfx_firmware_header_v1_0 gfx;
+	struct rlc_firmware_header_v1_0 rlc;
+	struct rlc_firmware_header_v2_0 rlc_v2_0;
+	struct rlc_firmware_header_v2_1 rlc_v2_1;
+	struct sdma_firmware_header_v1_0 sdma;
+	struct sdma_firmware_header_v1_1 sdma_v1_1;
+	struct gpu_info_firmware_header_v1_0 gpu_info;
+	uint8_t raw[0x100];
+};
+
+/*
+ * fw loading support
+ */
+enum AMDGPU_UCODE_ID {
+	AMDGPU_UCODE_ID_SDMA0 = 0,
+	AMDGPU_UCODE_ID_SDMA1,
+	AMDGPU_UCODE_ID_CP_CE,
+	AMDGPU_UCODE_ID_CP_PFP,
+	AMDGPU_UCODE_ID_CP_ME,
+	AMDGPU_UCODE_ID_CP_MEC1,
+	AMDGPU_UCODE_ID_CP_MEC1_JT,
+	AMDGPU_UCODE_ID_CP_MEC2,
+	AMDGPU_UCODE_ID_CP_MEC2_JT,
+	AMDGPU_UCODE_ID_RLC_G,
+	AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL,
+	AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM,
+	AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM,
+	AMDGPU_UCODE_ID_STORAGE,
+	AMDGPU_UCODE_ID_SMC,
+	AMDGPU_UCODE_ID_UVD,
+	AMDGPU_UCODE_ID_VCE,
+	AMDGPU_UCODE_ID_VCN,
+	AMDGPU_UCODE_ID_MAXIMUM,
+};
+
+/* engine firmware status */
+enum AMDGPU_UCODE_STATUS {
+	AMDGPU_UCODE_STATUS_INVALID,
+	AMDGPU_UCODE_STATUS_NOT_LOADED,
+	AMDGPU_UCODE_STATUS_LOADED,
+};
+
+/* conform to smu_ucode_xfer_cz.h */
+#define AMDGPU_SDMA0_UCODE_LOADED	0x00000001
+#define AMDGPU_SDMA1_UCODE_LOADED	0x00000002
+#define AMDGPU_CPCE_UCODE_LOADED	0x00000004
+#define AMDGPU_CPPFP_UCODE_LOADED	0x00000008
+#define AMDGPU_CPME_UCODE_LOADED	0x00000010
+#define AMDGPU_CPMEC1_UCODE_LOADED	0x00000020
+#define AMDGPU_CPMEC2_UCODE_LOADED	0x00000040
+#define AMDGPU_CPRLC_UCODE_LOADED	0x00000100
+
+/* amdgpu firmware info */
+struct amdgpu_firmware_info {
+	/* ucode ID */
+	enum AMDGPU_UCODE_ID ucode_id;
+	/* request_firmware */
+	const struct firmware *fw;
+	/* starting mc address */
+	uint64_t mc_addr;
+	/* kernel linear address */
+	void *kaddr;
+	/* ucode_size_bytes */
+	uint32_t ucode_size;
+	/* starting tmr mc address */
+	uint32_t tmr_mc_addr_lo;
+	uint32_t tmr_mc_addr_hi;
+};
+
+void amdgpu_ucode_print_mc_hdr(const struct common_firmware_header *hdr);
+void amdgpu_ucode_print_smc_hdr(const struct common_firmware_header *hdr);
+void amdgpu_ucode_print_gfx_hdr(const struct common_firmware_header *hdr);
+void amdgpu_ucode_print_rlc_hdr(const struct common_firmware_header *hdr);
+void amdgpu_ucode_print_sdma_hdr(const struct common_firmware_header *hdr);
+void amdgpu_ucode_print_gpu_info_hdr(const struct common_firmware_header *hdr);
+int amdgpu_ucode_validate(const struct firmware *fw);
+bool amdgpu_ucode_hdr_version(union amdgpu_firmware_header *hdr,
+				uint16_t hdr_major, uint16_t hdr_minor);
+int amdgpu_ucode_init_bo(struct amdgpu_device *adev);
+int amdgpu_ucode_fini_bo(struct amdgpu_device *adev);
+
+enum amdgpu_firmware_load_type
+amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_uvd.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_uvd.c
new file mode 100644
index 00000000000..ccc22c32535
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_uvd.c
@@ -0,0 +1,1300 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Christian König <deathsimple@vodafone.de>
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <drm/drmP.h>
+#include <drm/drm.h>
+
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_uvd.h"
+#include "cikd.h"
+#include "uvd/uvd_4_2_d.h"
+
+/* 1 second timeout */
+#define UVD_IDLE_TIMEOUT	msecs_to_jiffies(1000)
+
+/* Firmware versions for VI */
+#define FW_1_65_10	((1 << 24) | (65 << 16) | (10 << 8))
+#define FW_1_87_11	((1 << 24) | (87 << 16) | (11 << 8))
+#define FW_1_87_12	((1 << 24) | (87 << 16) | (12 << 8))
+#define FW_1_37_15	((1 << 24) | (37 << 16) | (15 << 8))
+
+/* Polaris10/11 firmware version */
+#define FW_1_66_16	((1 << 24) | (66 << 16) | (16 << 8))
+
+/* Firmware Names */
+#ifdef CONFIG_DRM_AMDGPU_CIK
+#define FIRMWARE_BONAIRE	"amdgpu/bonaire_uvd.bin"
+#define FIRMWARE_KABINI	"amdgpu/kabini_uvd.bin"
+#define FIRMWARE_KAVERI	"amdgpu/kaveri_uvd.bin"
+#define FIRMWARE_HAWAII	"amdgpu/hawaii_uvd.bin"
+#define FIRMWARE_MULLINS	"amdgpu/mullins_uvd.bin"
+#endif
+#define FIRMWARE_TONGA		"amdgpu/tonga_uvd.bin"
+#define FIRMWARE_CARRIZO	"amdgpu/carrizo_uvd.bin"
+#define FIRMWARE_FIJI		"amdgpu/fiji_uvd.bin"
+#define FIRMWARE_STONEY		"amdgpu/stoney_uvd.bin"
+#define FIRMWARE_POLARIS10	"amdgpu/polaris10_uvd.bin"
+#define FIRMWARE_POLARIS11	"amdgpu/polaris11_uvd.bin"
+#define FIRMWARE_POLARIS12	"amdgpu/polaris12_uvd.bin"
+#define FIRMWARE_VEGAM		"amdgpu/vegam_uvd.bin"
+
+#define FIRMWARE_VEGA10		"amdgpu/vega10_uvd.bin"
+#define FIRMWARE_VEGA12		"amdgpu/vega12_uvd.bin"
+#define FIRMWARE_VEGA20		"amdgpu/vega20_uvd.bin"
+
+/* These are common relative offsets for all asics, from uvd_7_0_offset.h,  */
+#define UVD_GPCOM_VCPU_CMD		0x03c3
+#define UVD_GPCOM_VCPU_DATA0	0x03c4
+#define UVD_GPCOM_VCPU_DATA1	0x03c5
+#define UVD_NO_OP				0x03ff
+#define UVD_BASE_SI				0x3800
+
+/**
+ * amdgpu_uvd_cs_ctx - Command submission parser context
+ *
+ * Used for emulating virtual memory support on UVD 4.2.
+ */
+struct amdgpu_uvd_cs_ctx {
+	struct amdgpu_cs_parser *parser;
+	unsigned reg, count;
+	unsigned data0, data1;
+	unsigned idx;
+	unsigned ib_idx;
+
+	/* does the IB has a msg command */
+	bool has_msg_cmd;
+
+	/* minimum buffer sizes */
+	unsigned *buf_sizes;
+};
+
+#ifdef CONFIG_DRM_AMDGPU_CIK
+MODULE_FIRMWARE(FIRMWARE_BONAIRE);
+MODULE_FIRMWARE(FIRMWARE_KABINI);
+MODULE_FIRMWARE(FIRMWARE_KAVERI);
+MODULE_FIRMWARE(FIRMWARE_HAWAII);
+MODULE_FIRMWARE(FIRMWARE_MULLINS);
+#endif
+MODULE_FIRMWARE(FIRMWARE_TONGA);
+MODULE_FIRMWARE(FIRMWARE_CARRIZO);
+MODULE_FIRMWARE(FIRMWARE_FIJI);
+MODULE_FIRMWARE(FIRMWARE_STONEY);
+MODULE_FIRMWARE(FIRMWARE_POLARIS10);
+MODULE_FIRMWARE(FIRMWARE_POLARIS11);
+MODULE_FIRMWARE(FIRMWARE_POLARIS12);
+MODULE_FIRMWARE(FIRMWARE_VEGAM);
+
+MODULE_FIRMWARE(FIRMWARE_VEGA10);
+MODULE_FIRMWARE(FIRMWARE_VEGA12);
+MODULE_FIRMWARE(FIRMWARE_VEGA20);
+
+static void amdgpu_uvd_idle_work_handler(struct work_struct *work);
+
+int amdgpu_uvd_sw_init(struct amdgpu_device *adev)
+{
+	unsigned long bo_size;
+	const char *fw_name;
+	const struct common_firmware_header *hdr;
+	unsigned family_id;
+	int i, j, r;
+
+	INIT_DELAYED_WORK(&adev->uvd.idle_work, amdgpu_uvd_idle_work_handler);
+
+	switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_BONAIRE:
+		fw_name = FIRMWARE_BONAIRE;
+		break;
+	case CHIP_KABINI:
+		fw_name = FIRMWARE_KABINI;
+		break;
+	case CHIP_KAVERI:
+		fw_name = FIRMWARE_KAVERI;
+		break;
+	case CHIP_HAWAII:
+		fw_name = FIRMWARE_HAWAII;
+		break;
+	case CHIP_MULLINS:
+		fw_name = FIRMWARE_MULLINS;
+		break;
+#endif
+	case CHIP_TONGA:
+		fw_name = FIRMWARE_TONGA;
+		break;
+	case CHIP_FIJI:
+		fw_name = FIRMWARE_FIJI;
+		break;
+	case CHIP_CARRIZO:
+		fw_name = FIRMWARE_CARRIZO;
+		break;
+	case CHIP_STONEY:
+		fw_name = FIRMWARE_STONEY;
+		break;
+	case CHIP_POLARIS10:
+		fw_name = FIRMWARE_POLARIS10;
+		break;
+	case CHIP_POLARIS11:
+		fw_name = FIRMWARE_POLARIS11;
+		break;
+	case CHIP_POLARIS12:
+		fw_name = FIRMWARE_POLARIS12;
+		break;
+	case CHIP_VEGA10:
+		fw_name = FIRMWARE_VEGA10;
+		break;
+	case CHIP_VEGA12:
+		fw_name = FIRMWARE_VEGA12;
+		break;
+	case CHIP_VEGAM:
+		fw_name = FIRMWARE_VEGAM;
+		break;
+	case CHIP_VEGA20:
+		fw_name = FIRMWARE_VEGA20;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	r = request_firmware(&adev->uvd.fw, fw_name, adev->dev);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_uvd: Can't load firmware \"%s\"\n",
+			fw_name);
+		return r;
+	}
+
+	r = amdgpu_ucode_validate(adev->uvd.fw);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_uvd: Can't validate firmware \"%s\"\n",
+			fw_name);
+		release_firmware(adev->uvd.fw);
+		adev->uvd.fw = NULL;
+		return r;
+	}
+
+	/* Set the default UVD handles that the firmware can handle */
+	adev->uvd.max_handles = AMDGPU_DEFAULT_UVD_HANDLES;
+
+	hdr = (const struct common_firmware_header *)adev->uvd.fw->data;
+	family_id = le32_to_cpu(hdr->ucode_version) & 0xff;
+
+	if (adev->asic_type < CHIP_VEGA20) {
+		unsigned version_major, version_minor;
+
+		version_major = (le32_to_cpu(hdr->ucode_version) >> 24) & 0xff;
+		version_minor = (le32_to_cpu(hdr->ucode_version) >> 8) & 0xff;
+		DRM_INFO("Found UVD firmware Version: %u.%u Family ID: %u\n",
+			version_major, version_minor, family_id);
+
+		/*
+		 * Limit the number of UVD handles depending on microcode major
+		 * and minor versions. The firmware version which has 40 UVD
+		 * instances support is 1.80. So all subsequent versions should
+		 * also have the same support.
+		 */
+		if ((version_major > 0x01) ||
+		    ((version_major == 0x01) && (version_minor >= 0x50)))
+			adev->uvd.max_handles = AMDGPU_MAX_UVD_HANDLES;
+
+		adev->uvd.fw_version = ((version_major << 24) | (version_minor << 16) |
+					(family_id << 8));
+
+		if ((adev->asic_type == CHIP_POLARIS10 ||
+		     adev->asic_type == CHIP_POLARIS11) &&
+		    (adev->uvd.fw_version < FW_1_66_16))
+			DRM_ERROR("POLARIS10/11 UVD firmware version %hu.%hu is too old.\n",
+				  version_major, version_minor);
+	} else {
+		unsigned int enc_major, enc_minor, dec_minor;
+
+		dec_minor = (le32_to_cpu(hdr->ucode_version) >> 8) & 0xff;
+		enc_minor = (le32_to_cpu(hdr->ucode_version) >> 24) & 0x3f;
+		enc_major = (le32_to_cpu(hdr->ucode_version) >> 30) & 0x3;
+		DRM_INFO("Found UVD firmware ENC: %u.%u DEC: .%u Family ID: %u\n",
+			enc_major, enc_minor, dec_minor, family_id);
+
+		adev->uvd.max_handles = AMDGPU_MAX_UVD_HANDLES;
+
+		adev->uvd.fw_version = le32_to_cpu(hdr->ucode_version);
+	}
+
+	bo_size = AMDGPU_UVD_STACK_SIZE + AMDGPU_UVD_HEAP_SIZE
+		  +  AMDGPU_UVD_SESSION_SIZE * adev->uvd.max_handles;
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		bo_size += AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+	for (j = 0; j < adev->uvd.num_uvd_inst; j++) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		r = amdgpu_bo_create_kernel(adev, bo_size, PAGE_SIZE,
+					    AMDGPU_GEM_DOMAIN_VRAM, &adev->uvd.inst[j].vcpu_bo,
+					    &adev->uvd.inst[j].gpu_addr, &adev->uvd.inst[j].cpu_addr);
+		if (r) {
+			dev_err(adev->dev, "(%d) failed to allocate UVD bo\n", r);
+			return r;
+		}
+	}
+
+	for (i = 0; i < adev->uvd.max_handles; ++i) {
+		atomic_set(&adev->uvd.handles[i], 0);
+		adev->uvd.filp[i] = NULL;
+	}
+
+	/* from uvd v5.0 HW addressing capacity increased to 64 bits */
+	if (!amdgpu_device_ip_block_version_cmp(adev, AMD_IP_BLOCK_TYPE_UVD, 5, 0))
+		adev->uvd.address_64_bit = true;
+
+	switch (adev->asic_type) {
+	case CHIP_TONGA:
+		adev->uvd.use_ctx_buf = adev->uvd.fw_version >= FW_1_65_10;
+		break;
+	case CHIP_CARRIZO:
+		adev->uvd.use_ctx_buf = adev->uvd.fw_version >= FW_1_87_11;
+		break;
+	case CHIP_FIJI:
+		adev->uvd.use_ctx_buf = adev->uvd.fw_version >= FW_1_87_12;
+		break;
+	case CHIP_STONEY:
+		adev->uvd.use_ctx_buf = adev->uvd.fw_version >= FW_1_37_15;
+		break;
+	default:
+		adev->uvd.use_ctx_buf = adev->asic_type >= CHIP_POLARIS10;
+	}
+
+	return 0;
+}
+
+int amdgpu_uvd_sw_fini(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	drm_sched_entity_destroy(&adev->uvd.entity);
+
+	for (j = 0; j < adev->uvd.num_uvd_inst; ++j) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		kvfree(adev->uvd.inst[j].saved_bo);
+
+		amdgpu_bo_free_kernel(&adev->uvd.inst[j].vcpu_bo,
+				      &adev->uvd.inst[j].gpu_addr,
+				      (void **)&adev->uvd.inst[j].cpu_addr);
+
+		amdgpu_ring_fini(&adev->uvd.inst[j].ring);
+
+		for (i = 0; i < AMDGPU_MAX_UVD_ENC_RINGS; ++i)
+			amdgpu_ring_fini(&adev->uvd.inst[j].ring_enc[i]);
+	}
+	release_firmware(adev->uvd.fw);
+
+	return 0;
+}
+
+/**
+ * amdgpu_uvd_entity_init - init entity
+ *
+ * @adev: amdgpu_device pointer
+ *
+ */
+int amdgpu_uvd_entity_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	struct drm_sched_rq *rq;
+	int r;
+
+	ring = &adev->uvd.inst[0].ring;
+	rq = &ring->sched.sched_rq[DRM_SCHED_PRIORITY_NORMAL];
+	r = drm_sched_entity_init(&adev->uvd.entity, &rq, 1, NULL);
+	if (r) {
+		DRM_ERROR("Failed setting up UVD kernel entity.\n");
+		return r;
+	}
+
+	return 0;
+}
+
+int amdgpu_uvd_suspend(struct amdgpu_device *adev)
+{
+	unsigned size;
+	void *ptr;
+	int i, j;
+
+	cancel_delayed_work_sync(&adev->uvd.idle_work);
+
+	/* only valid for physical mode */
+	if (adev->asic_type < CHIP_POLARIS10) {
+		for (i = 0; i < adev->uvd.max_handles; ++i)
+			if (atomic_read(&adev->uvd.handles[i]))
+				break;
+
+		if (i == adev->uvd.max_handles)
+			return 0;
+	}
+
+	for (j = 0; j < adev->uvd.num_uvd_inst; ++j) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		if (adev->uvd.inst[j].vcpu_bo == NULL)
+			continue;
+
+		size = amdgpu_bo_size(adev->uvd.inst[j].vcpu_bo);
+		ptr = adev->uvd.inst[j].cpu_addr;
+
+		adev->uvd.inst[j].saved_bo = kvmalloc(size, GFP_KERNEL);
+		if (!adev->uvd.inst[j].saved_bo)
+			return -ENOMEM;
+
+		memcpy_fromio(adev->uvd.inst[j].saved_bo, ptr, size);
+	}
+	return 0;
+}
+
+int amdgpu_uvd_resume(struct amdgpu_device *adev)
+{
+	unsigned size;
+	void *ptr;
+	int i;
+
+	for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		if (adev->uvd.inst[i].vcpu_bo == NULL)
+			return -EINVAL;
+
+		size = amdgpu_bo_size(adev->uvd.inst[i].vcpu_bo);
+		ptr = adev->uvd.inst[i].cpu_addr;
+
+		if (adev->uvd.inst[i].saved_bo != NULL) {
+			memcpy_toio(ptr, adev->uvd.inst[i].saved_bo, size);
+			kvfree(adev->uvd.inst[i].saved_bo);
+			adev->uvd.inst[i].saved_bo = NULL;
+		} else {
+			const struct common_firmware_header *hdr;
+			unsigned offset;
+
+			hdr = (const struct common_firmware_header *)adev->uvd.fw->data;
+			if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+				offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
+				memcpy_toio(adev->uvd.inst[i].cpu_addr, adev->uvd.fw->data + offset,
+					    le32_to_cpu(hdr->ucode_size_bytes));
+				size -= le32_to_cpu(hdr->ucode_size_bytes);
+				ptr += le32_to_cpu(hdr->ucode_size_bytes);
+			}
+			memset_io(ptr, 0, size);
+			/* to restore uvd fence seq */
+			amdgpu_fence_driver_force_completion(&adev->uvd.inst[i].ring);
+		}
+	}
+	return 0;
+}
+
+void amdgpu_uvd_free_handles(struct amdgpu_device *adev, struct drm_file *filp)
+{
+	struct amdgpu_ring *ring = &adev->uvd.inst[0].ring;
+	int i, r;
+
+	for (i = 0; i < adev->uvd.max_handles; ++i) {
+		uint32_t handle = atomic_read(&adev->uvd.handles[i]);
+
+		if (handle != 0 && adev->uvd.filp[i] == filp) {
+			struct dma_fence *fence;
+
+			r = amdgpu_uvd_get_destroy_msg(ring, handle, false,
+						       &fence);
+			if (r) {
+				DRM_ERROR("Error destroying UVD %d!\n", r);
+				continue;
+			}
+
+			dma_fence_wait(fence, false);
+			dma_fence_put(fence);
+
+			adev->uvd.filp[i] = NULL;
+			atomic_set(&adev->uvd.handles[i], 0);
+		}
+	}
+}
+
+static void amdgpu_uvd_force_into_uvd_segment(struct amdgpu_bo *abo)
+{
+	int i;
+	for (i = 0; i < abo->placement.num_placement; ++i) {
+		abo->placements[i].fpfn = 0 >> PAGE_SHIFT;
+		abo->placements[i].lpfn = (256 * 1024 * 1024) >> PAGE_SHIFT;
+	}
+}
+
+static u64 amdgpu_uvd_get_addr_from_ctx(struct amdgpu_uvd_cs_ctx *ctx)
+{
+	uint32_t lo, hi;
+	uint64_t addr;
+
+	lo = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->data0);
+	hi = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->data1);
+	addr = ((uint64_t)lo) | (((uint64_t)hi) << 32);
+
+	return addr;
+}
+
+/**
+ * amdgpu_uvd_cs_pass1 - first parsing round
+ *
+ * @ctx: UVD parser context
+ *
+ * Make sure UVD message and feedback buffers are in VRAM and
+ * nobody is violating an 256MB boundary.
+ */
+static int amdgpu_uvd_cs_pass1(struct amdgpu_uvd_cs_ctx *ctx)
+{
+	struct ttm_operation_ctx tctx = { false, false };
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_bo *bo;
+	uint32_t cmd;
+	uint64_t addr = amdgpu_uvd_get_addr_from_ctx(ctx);
+	int r = 0;
+
+	r = amdgpu_cs_find_mapping(ctx->parser, addr, &bo, &mapping);
+	if (r) {
+		DRM_ERROR("Can't find BO for addr 0x%08Lx\n", addr);
+		return r;
+	}
+
+	if (!ctx->parser->adev->uvd.address_64_bit) {
+		/* check if it's a message or feedback command */
+		cmd = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->idx) >> 1;
+		if (cmd == 0x0 || cmd == 0x3) {
+			/* yes, force it into VRAM */
+			uint32_t domain = AMDGPU_GEM_DOMAIN_VRAM;
+			amdgpu_bo_placement_from_domain(bo, domain);
+		}
+		amdgpu_uvd_force_into_uvd_segment(bo);
+
+		r = ttm_bo_validate(&bo->tbo, &bo->placement, &tctx);
+	}
+
+	return r;
+}
+
+/**
+ * amdgpu_uvd_cs_msg_decode - handle UVD decode message
+ *
+ * @msg: pointer to message structure
+ * @buf_sizes: returned buffer sizes
+ *
+ * Peek into the decode message and calculate the necessary buffer sizes.
+ */
+static int amdgpu_uvd_cs_msg_decode(struct amdgpu_device *adev, uint32_t *msg,
+	unsigned buf_sizes[])
+{
+	unsigned stream_type = msg[4];
+	unsigned width = msg[6];
+	unsigned height = msg[7];
+	unsigned dpb_size = msg[9];
+	unsigned pitch = msg[28];
+	unsigned level = msg[57];
+
+	unsigned width_in_mb = width / 16;
+	unsigned height_in_mb = roundup2(height / 16, 2);
+	unsigned fs_in_mb = width_in_mb * height_in_mb;
+
+	unsigned image_size, tmp, min_dpb_size, num_dpb_buffer;
+	unsigned min_ctx_size = ~0;
+
+	image_size = width * height;
+	image_size += image_size / 2;
+	image_size = roundup2(image_size, 1024);
+
+	switch (stream_type) {
+	case 0: /* H264 */
+		switch(level) {
+		case 30:
+			num_dpb_buffer = 8100 / fs_in_mb;
+			break;
+		case 31:
+			num_dpb_buffer = 18000 / fs_in_mb;
+			break;
+		case 32:
+			num_dpb_buffer = 20480 / fs_in_mb;
+			break;
+		case 41:
+			num_dpb_buffer = 32768 / fs_in_mb;
+			break;
+		case 42:
+			num_dpb_buffer = 34816 / fs_in_mb;
+			break;
+		case 50:
+			num_dpb_buffer = 110400 / fs_in_mb;
+			break;
+		case 51:
+			num_dpb_buffer = 184320 / fs_in_mb;
+			break;
+		default:
+			num_dpb_buffer = 184320 / fs_in_mb;
+			break;
+		}
+		num_dpb_buffer++;
+		if (num_dpb_buffer > 17)
+			num_dpb_buffer = 17;
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * num_dpb_buffer;
+
+		/* macroblock context buffer */
+		min_dpb_size += width_in_mb * height_in_mb * num_dpb_buffer * 192;
+
+		/* IT surface buffer */
+		min_dpb_size += width_in_mb * height_in_mb * 32;
+		break;
+
+	case 1: /* VC1 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 3;
+
+		/* CONTEXT_BUFFER */
+		min_dpb_size += width_in_mb * height_in_mb * 128;
+
+		/* IT surface buffer */
+		min_dpb_size += width_in_mb * 64;
+
+		/* DB surface buffer */
+		min_dpb_size += width_in_mb * 128;
+
+		/* BP */
+		tmp = max(width_in_mb, height_in_mb);
+		min_dpb_size += roundup2(tmp * 7 * 16, 64);
+		break;
+
+	case 3: /* MPEG2 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 3;
+		break;
+
+	case 4: /* MPEG4 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 3;
+
+		/* CM */
+		min_dpb_size += width_in_mb * height_in_mb * 64;
+
+		/* IT surface buffer */
+		min_dpb_size += roundup2(width_in_mb * height_in_mb * 32, 64);
+		break;
+
+	case 7: /* H264 Perf */
+		switch(level) {
+		case 30:
+			num_dpb_buffer = 8100 / fs_in_mb;
+			break;
+		case 31:
+			num_dpb_buffer = 18000 / fs_in_mb;
+			break;
+		case 32:
+			num_dpb_buffer = 20480 / fs_in_mb;
+			break;
+		case 41:
+			num_dpb_buffer = 32768 / fs_in_mb;
+			break;
+		case 42:
+			num_dpb_buffer = 34816 / fs_in_mb;
+			break;
+		case 50:
+			num_dpb_buffer = 110400 / fs_in_mb;
+			break;
+		case 51:
+			num_dpb_buffer = 184320 / fs_in_mb;
+			break;
+		default:
+			num_dpb_buffer = 184320 / fs_in_mb;
+			break;
+		}
+		num_dpb_buffer++;
+		if (num_dpb_buffer > 17)
+			num_dpb_buffer = 17;
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * num_dpb_buffer;
+
+		if (!adev->uvd.use_ctx_buf){
+			/* macroblock context buffer */
+			min_dpb_size +=
+				width_in_mb * height_in_mb * num_dpb_buffer * 192;
+
+			/* IT surface buffer */
+			min_dpb_size += width_in_mb * height_in_mb * 32;
+		} else {
+			/* macroblock context buffer */
+			min_ctx_size =
+				width_in_mb * height_in_mb * num_dpb_buffer * 192;
+		}
+		break;
+
+	case 8: /* MJPEG */
+		min_dpb_size = 0;
+		break;
+
+	case 16: /* H265 */
+		image_size = (roundup2(width, 16) * roundup2(height, 16) * 3) / 2;
+		image_size = roundup2(image_size, 256);
+
+		num_dpb_buffer = (le32_to_cpu(msg[59]) & 0xff) + 2;
+		min_dpb_size = image_size * num_dpb_buffer;
+		min_ctx_size = ((width + 255) / 16) * ((height + 255) / 16)
+					   * 16 * num_dpb_buffer + 52 * 1024;
+		break;
+
+	default:
+		DRM_ERROR("UVD codec not handled %d!\n", stream_type);
+		return -EINVAL;
+	}
+
+	if (width > pitch) {
+		DRM_ERROR("Invalid UVD decoding target pitch!\n");
+		return -EINVAL;
+	}
+
+	if (dpb_size < min_dpb_size) {
+		DRM_ERROR("Invalid dpb_size in UVD message (%d / %d)!\n",
+			  dpb_size, min_dpb_size);
+		return -EINVAL;
+	}
+
+	buf_sizes[0x1] = dpb_size;
+	buf_sizes[0x2] = image_size;
+	buf_sizes[0x4] = min_ctx_size;
+	return 0;
+}
+
+/**
+ * amdgpu_uvd_cs_msg - handle UVD message
+ *
+ * @ctx: UVD parser context
+ * @bo: buffer object containing the message
+ * @offset: offset into the buffer object
+ *
+ * Peek into the UVD message and extract the session id.
+ * Make sure that we don't open up to many sessions.
+ */
+static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx,
+			     struct amdgpu_bo *bo, unsigned offset)
+{
+	struct amdgpu_device *adev = ctx->parser->adev;
+	int32_t *msg, msg_type, handle;
+	void *ptr;
+	long r;
+	int i;
+
+	if (offset & 0x3F) {
+		DRM_ERROR("UVD messages must be 64 byte aligned!\n");
+		return -EINVAL;
+	}
+
+	r = amdgpu_bo_kmap(bo, &ptr);
+	if (r) {
+		DRM_ERROR("Failed mapping the UVD) message (%ld)!\n", r);
+		return r;
+	}
+
+	msg = ptr + offset;
+
+	msg_type = msg[1];
+	handle = msg[2];
+
+	if (handle == 0) {
+		DRM_ERROR("Invalid UVD handle!\n");
+		return -EINVAL;
+	}
+
+	switch (msg_type) {
+	case 0:
+		/* it's a create msg, calc image size (width * height) */
+		amdgpu_bo_kunmap(bo);
+
+		/* try to alloc a new handle */
+		for (i = 0; i < adev->uvd.max_handles; ++i) {
+			if (atomic_read(&adev->uvd.handles[i]) == handle) {
+				DRM_ERROR(")Handle 0x%x already in use!\n",
+					  handle);
+				return -EINVAL;
+			}
+
+			if (!atomic_cmpxchg(&adev->uvd.handles[i], 0, handle)) {
+				adev->uvd.filp[i] = ctx->parser->filp;
+				return 0;
+			}
+		}
+
+		DRM_ERROR("No more free UVD handles!\n");
+		return -ENOSPC;
+
+	case 1:
+		/* it's a decode msg, calc buffer sizes */
+		r = amdgpu_uvd_cs_msg_decode(adev, msg, ctx->buf_sizes);
+		amdgpu_bo_kunmap(bo);
+		if (r)
+			return r;
+
+		/* validate the handle */
+		for (i = 0; i < adev->uvd.max_handles; ++i) {
+			if (atomic_read(&adev->uvd.handles[i]) == handle) {
+				if (adev->uvd.filp[i] != ctx->parser->filp) {
+					DRM_ERROR("UVD handle collision detected!\n");
+					return -EINVAL;
+				}
+				return 0;
+			}
+		}
+
+		DRM_ERROR("Invalid UVD handle 0x%x!\n", handle);
+		return -ENOENT;
+
+	case 2:
+		/* it's a destroy msg, free the handle */
+		for (i = 0; i < adev->uvd.max_handles; ++i)
+			atomic_cmpxchg(&adev->uvd.handles[i], handle, 0);
+		amdgpu_bo_kunmap(bo);
+		return 0;
+
+	default:
+		DRM_ERROR("Illegal UVD message type (%d)!\n", msg_type);
+		return -EINVAL;
+	}
+	BUG();
+	return -EINVAL;
+}
+
+/**
+ * amdgpu_uvd_cs_pass2 - second parsing round
+ *
+ * @ctx: UVD parser context
+ *
+ * Patch buffer addresses, make sure buffer sizes are correct.
+ */
+static int amdgpu_uvd_cs_pass2(struct amdgpu_uvd_cs_ctx *ctx)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_bo *bo;
+	uint32_t cmd;
+	uint64_t start, end;
+	uint64_t addr = amdgpu_uvd_get_addr_from_ctx(ctx);
+	int r;
+
+	r = amdgpu_cs_find_mapping(ctx->parser, addr, &bo, &mapping);
+	if (r) {
+		DRM_ERROR("Can't find BO for addr 0x%08Lx\n", addr);
+		return r;
+	}
+
+	start = amdgpu_bo_gpu_offset(bo);
+
+	end = (mapping->last + 1 - mapping->start);
+	end = end * AMDGPU_GPU_PAGE_SIZE + start;
+
+	addr -= mapping->start * AMDGPU_GPU_PAGE_SIZE;
+	start += addr;
+
+	amdgpu_set_ib_value(ctx->parser, ctx->ib_idx, ctx->data0,
+			    lower_32_bits(start));
+	amdgpu_set_ib_value(ctx->parser, ctx->ib_idx, ctx->data1,
+			    upper_32_bits(start));
+
+	cmd = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->idx) >> 1;
+	if (cmd < 0x4) {
+		if ((end - start) < ctx->buf_sizes[cmd]) {
+			DRM_ERROR("buffer (%d) to small (%d / %d)!\n", cmd,
+				  (unsigned)(end - start),
+				  ctx->buf_sizes[cmd]);
+			return -EINVAL;
+		}
+
+	} else if (cmd == 0x206) {
+		if ((end - start) < ctx->buf_sizes[4]) {
+			DRM_ERROR("buffer (%d) to small (%d / %d)!\n", cmd,
+					  (unsigned)(end - start),
+					  ctx->buf_sizes[4]);
+			return -EINVAL;
+		}
+	} else if ((cmd != 0x100) && (cmd != 0x204)) {
+		DRM_ERROR("invalid UVD command %X!\n", cmd);
+		return -EINVAL;
+	}
+
+	if (!ctx->parser->adev->uvd.address_64_bit) {
+		if ((start >> 28) != ((end - 1) >> 28)) {
+			DRM_ERROR("reloc %LX-%LX crossing 256MB boundary!\n",
+				  start, end);
+			return -EINVAL;
+		}
+
+		if ((cmd == 0 || cmd == 0x3) &&
+		    (start >> 28) != (ctx->parser->adev->uvd.inst->gpu_addr >> 28)) {
+			DRM_ERROR("msg/fb buffer %LX-%LX out of 256MB segment!\n",
+				  start, end);
+			return -EINVAL;
+		}
+	}
+
+	if (cmd == 0) {
+		ctx->has_msg_cmd = true;
+		r = amdgpu_uvd_cs_msg(ctx, bo, addr);
+		if (r)
+			return r;
+	} else if (!ctx->has_msg_cmd) {
+		DRM_ERROR("Message needed before other commands are send!\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_uvd_cs_reg - parse register writes
+ *
+ * @ctx: UVD parser context
+ * @cb: callback function
+ *
+ * Parse the register writes, call cb on each complete command.
+ */
+static int amdgpu_uvd_cs_reg(struct amdgpu_uvd_cs_ctx *ctx,
+			     int (*cb)(struct amdgpu_uvd_cs_ctx *ctx))
+{
+	struct amdgpu_ib *ib = &ctx->parser->job->ibs[ctx->ib_idx];
+	int i, r;
+
+	ctx->idx++;
+	for (i = 0; i <= ctx->count; ++i) {
+		unsigned reg = ctx->reg + i;
+
+		if (ctx->idx >= ib->length_dw) {
+			DRM_ERROR("Register command after end of CS!\n");
+			return -EINVAL;
+		}
+
+		switch (reg) {
+		case mmUVD_GPCOM_VCPU_DATA0:
+			ctx->data0 = ctx->idx;
+			break;
+		case mmUVD_GPCOM_VCPU_DATA1:
+			ctx->data1 = ctx->idx;
+			break;
+		case mmUVD_GPCOM_VCPU_CMD:
+			r = cb(ctx);
+			if (r)
+				return r;
+			break;
+		case mmUVD_ENGINE_CNTL:
+		case mmUVD_NO_OP:
+			break;
+		default:
+			DRM_ERROR("Invalid reg 0x%X!\n", reg);
+			return -EINVAL;
+		}
+		ctx->idx++;
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_uvd_cs_packets - parse UVD packets
+ *
+ * @ctx: UVD parser context
+ * @cb: callback function
+ *
+ * Parse the command stream packets.
+ */
+static int amdgpu_uvd_cs_packets(struct amdgpu_uvd_cs_ctx *ctx,
+				 int (*cb)(struct amdgpu_uvd_cs_ctx *ctx))
+{
+	struct amdgpu_ib *ib = &ctx->parser->job->ibs[ctx->ib_idx];
+	int r;
+
+	for (ctx->idx = 0 ; ctx->idx < ib->length_dw; ) {
+		uint32_t cmd = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->idx);
+		unsigned type = CP_PACKET_GET_TYPE(cmd);
+		switch (type) {
+		case PACKET_TYPE0:
+			ctx->reg = CP_PACKET0_GET_REG(cmd);
+			ctx->count = CP_PACKET_GET_COUNT(cmd);
+			r = amdgpu_uvd_cs_reg(ctx, cb);
+			if (r)
+				return r;
+			break;
+		case PACKET_TYPE2:
+			++ctx->idx;
+			break;
+		default:
+			DRM_ERROR("Unknown packet type %d !\n", type);
+			return -EINVAL;
+		}
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_uvd_ring_parse_cs - UVD command submission parser
+ *
+ * @parser: Command submission parser context
+ *
+ * Parse the command stream, patch in addresses as necessary.
+ */
+int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx)
+{
+	struct amdgpu_uvd_cs_ctx ctx = {};
+	unsigned buf_sizes[] = {
+		[0x00000000]	=	2048,
+		[0x00000001]	=	0xFFFFFFFF,
+		[0x00000002]	=	0xFFFFFFFF,
+		[0x00000003]	=	2048,
+		[0x00000004]	=	0xFFFFFFFF,
+	};
+	struct amdgpu_ib *ib = &parser->job->ibs[ib_idx];
+	int r;
+
+	parser->job->vm = NULL;
+	ib->gpu_addr = amdgpu_sa_bo_gpu_addr(ib->sa_bo);
+
+	if (ib->length_dw % 16) {
+		DRM_ERROR("UVD IB length (%d) not 16 dwords aligned!\n",
+			  ib->length_dw);
+		return -EINVAL;
+	}
+
+	ctx.parser = parser;
+	ctx.buf_sizes = buf_sizes;
+	ctx.ib_idx = ib_idx;
+
+	/* first round only required on chips without UVD 64 bit address support */
+	if (!parser->adev->uvd.address_64_bit) {
+		/* first round, make sure the buffers are actually in the UVD segment */
+		r = amdgpu_uvd_cs_packets(&ctx, amdgpu_uvd_cs_pass1);
+		if (r)
+			return r;
+	}
+
+	/* second round, patch buffer addresses into the command stream */
+	r = amdgpu_uvd_cs_packets(&ctx, amdgpu_uvd_cs_pass2);
+	if (r)
+		return r;
+
+	if (!ctx.has_msg_cmd) {
+		DRM_ERROR("UVD-IBs need a msg command!\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo,
+			       bool direct, struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct dma_fence *f = NULL;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	uint32_t data[4];
+	uint64_t addr;
+	long r;
+	int i;
+	unsigned offset_idx = 0;
+	unsigned offset[3] = { UVD_BASE_SI, 0, 0 };
+
+	amdgpu_bo_kunmap(bo);
+	amdgpu_bo_unpin(bo);
+
+	if (!ring->adev->uvd.address_64_bit) {
+		struct ttm_operation_ctx ctx = { true, false };
+
+		amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_VRAM);
+		amdgpu_uvd_force_into_uvd_segment(bo);
+		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+		if (r)
+			goto err;
+	}
+
+	r = amdgpu_job_alloc_with_ib(adev, 64, &job);
+	if (r)
+		goto err;
+
+	if (adev->asic_type >= CHIP_VEGA10) {
+		offset_idx = 1 + ring->me;
+		offset[1] = adev->reg_offset[UVD_HWIP][0][1];
+		offset[2] = adev->reg_offset[UVD_HWIP][1][1];
+	}
+
+	data[0] = PACKET0(offset[offset_idx] + UVD_GPCOM_VCPU_DATA0, 0);
+	data[1] = PACKET0(offset[offset_idx] + UVD_GPCOM_VCPU_DATA1, 0);
+	data[2] = PACKET0(offset[offset_idx] + UVD_GPCOM_VCPU_CMD, 0);
+	data[3] = PACKET0(offset[offset_idx] + UVD_NO_OP, 0);
+
+	ib = &job->ibs[0];
+	addr = amdgpu_bo_gpu_offset(bo);
+	ib->ptr[0] = data[0];
+	ib->ptr[1] = addr;
+	ib->ptr[2] = data[1];
+	ib->ptr[3] = addr >> 32;
+	ib->ptr[4] = data[2];
+	ib->ptr[5] = 0;
+	for (i = 6; i < 16; i += 2) {
+		ib->ptr[i] = data[3];
+		ib->ptr[i+1] = 0;
+	}
+	ib->length_dw = 16;
+
+	if (direct) {
+		r = reservation_object_wait_timeout_rcu(bo->tbo.resv,
+							true, false,
+							msecs_to_jiffies(10));
+		if (r == 0)
+			r = -ETIMEDOUT;
+		if (r < 0)
+			goto err_free;
+
+		r = amdgpu_job_submit_direct(job, ring, &f);
+		if (r)
+			goto err_free;
+	} else {
+		r = amdgpu_sync_resv(adev, &job->sync, bo->tbo.resv,
+				     AMDGPU_FENCE_OWNER_UNDEFINED, false);
+		if (r)
+			goto err_free;
+
+		r = amdgpu_job_submit(job, &adev->uvd.entity,
+				      AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+		if (r)
+			goto err_free;
+	}
+
+	amdgpu_bo_fence(bo, f, false);
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&bo);
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+
+	return 0;
+
+err_free:
+	amdgpu_job_free(job);
+
+err:
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&bo);
+	return r;
+}
+
+/* multiple fence commands without any stream commands in between can
+   crash the vcpu so just try to emmit a dummy create/destroy msg to
+   avoid this */
+int amdgpu_uvd_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+			      struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_bo *bo = NULL;
+	uint32_t *msg;
+	int r, i;
+
+	r = amdgpu_bo_create_reserved(adev, 1024, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM,
+				      &bo, NULL, (void **)&msg);
+	if (r)
+		return r;
+
+	/* stitch together an UVD create msg */
+	msg[0] = cpu_to_le32(0x00000de4);
+	msg[1] = cpu_to_le32(0x00000000);
+	msg[2] = cpu_to_le32(handle);
+	msg[3] = cpu_to_le32(0x00000000);
+	msg[4] = cpu_to_le32(0x00000000);
+	msg[5] = cpu_to_le32(0x00000000);
+	msg[6] = cpu_to_le32(0x00000000);
+	msg[7] = cpu_to_le32(0x00000780);
+	msg[8] = cpu_to_le32(0x00000440);
+	msg[9] = cpu_to_le32(0x00000000);
+	msg[10] = cpu_to_le32(0x01b37000);
+	for (i = 11; i < 1024; ++i)
+		msg[i] = cpu_to_le32(0x0);
+
+	return amdgpu_uvd_send_msg(ring, bo, true, fence);
+}
+
+int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+			       bool direct, struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_bo *bo = NULL;
+	uint32_t *msg;
+	int r, i;
+
+	r = amdgpu_bo_create_reserved(adev, 1024, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM,
+				      &bo, NULL, (void **)&msg);
+	if (r)
+		return r;
+
+	/* stitch together an UVD destroy msg */
+	msg[0] = cpu_to_le32(0x00000de4);
+	msg[1] = cpu_to_le32(0x00000002);
+	msg[2] = cpu_to_le32(handle);
+	msg[3] = cpu_to_le32(0x00000000);
+	for (i = 4; i < 1024; ++i)
+		msg[i] = cpu_to_le32(0x0);
+
+	return amdgpu_uvd_send_msg(ring, bo, direct, fence);
+}
+
+static void amdgpu_uvd_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, uvd.idle_work.work);
+	unsigned fences = 0, i, j;
+
+	for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		fences += amdgpu_fence_count_emitted(&adev->uvd.inst[i].ring);
+		for (j = 0; j < adev->uvd.num_enc_rings; ++j) {
+			fences += amdgpu_fence_count_emitted(&adev->uvd.inst[i].ring_enc[j]);
+		}
+	}
+
+	if (fences == 0) {
+		if (adev->pm.dpm_enabled) {
+			amdgpu_dpm_enable_uvd(adev, false);
+		} else {
+			amdgpu_asic_set_uvd_clocks(adev, 0, 0);
+			/* shutdown the UVD block */
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							       AMD_PG_STATE_GATE);
+			amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							       AMD_CG_STATE_GATE);
+		}
+	} else {
+		schedule_delayed_work(&adev->uvd.idle_work, UVD_IDLE_TIMEOUT);
+	}
+}
+
+void amdgpu_uvd_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	bool set_clocks;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	set_clocks = !cancel_delayed_work_sync(&adev->uvd.idle_work);
+	if (set_clocks) {
+		if (adev->pm.dpm_enabled) {
+			amdgpu_dpm_enable_uvd(adev, true);
+		} else {
+			amdgpu_asic_set_uvd_clocks(adev, 53300, 40000);
+			amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							       AMD_CG_STATE_UNGATE);
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							       AMD_PG_STATE_UNGATE);
+		}
+	}
+}
+
+void amdgpu_uvd_ring_end_use(struct amdgpu_ring *ring)
+{
+	if (!amdgpu_sriov_vf(ring->adev))
+		schedule_delayed_work(&ring->adev->uvd.idle_work, UVD_IDLE_TIMEOUT);
+}
+
+/**
+ * amdgpu_uvd_ring_test_ib - test ib execution
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully execute an IB
+ */
+int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct dma_fence *fence;
+	long r;
+	uint32_t ip_instance = ring->me;
+
+	r = amdgpu_uvd_get_create_msg(ring, 1, NULL);
+	if (r) {
+		DRM_ERROR("amdgpu: (%d)failed to get create msg (%ld).\n", ip_instance, r);
+		goto error;
+	}
+
+	r = amdgpu_uvd_get_destroy_msg(ring, 1, true, &fence);
+	if (r) {
+		DRM_ERROR("amdgpu: (%d)failed to get destroy ib (%ld).\n", ip_instance, r);
+		goto error;
+	}
+
+	r = dma_fence_wait_timeout(fence, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: (%d)IB test timed out.\n", ip_instance);
+		r = -ETIMEDOUT;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: (%d)fence wait failed (%ld).\n", ip_instance, r);
+	} else {
+		DRM_DEBUG("ib test on (%d)ring %d succeeded\n", ip_instance, ring->idx);
+		r = 0;
+	}
+
+	dma_fence_put(fence);
+
+error:
+	return r;
+}
+
+/**
+ * amdgpu_uvd_used_handles - returns used UVD handles
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns the number of UVD handles in use
+ */
+uint32_t amdgpu_uvd_used_handles(struct amdgpu_device *adev)
+{
+	unsigned i;
+	uint32_t used_handles = 0;
+
+	for (i = 0; i < adev->uvd.max_handles; ++i) {
+		/*
+		 * Handles can be freed in any order, and not
+		 * necessarily linear. So we need to count
+		 * all non-zero handles.
+		 */
+		if (atomic_read(&adev->uvd.handles[i]))
+			used_handles++;
+	}
+
+	return used_handles;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_uvd.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_uvd.h
new file mode 100644
index 00000000000..a3ab1a41060
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_uvd.h
@@ -0,0 +1,87 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_UVD_H__
+#define __AMDGPU_UVD_H__
+
+#define AMDGPU_DEFAULT_UVD_HANDLES	10
+#define AMDGPU_MAX_UVD_HANDLES		40
+#define AMDGPU_UVD_STACK_SIZE		(200*1024)
+#define AMDGPU_UVD_HEAP_SIZE		(256*1024)
+#define AMDGPU_UVD_SESSION_SIZE		(50*1024)
+#define AMDGPU_UVD_FIRMWARE_OFFSET	256
+
+#define AMDGPU_MAX_UVD_INSTANCES			2
+
+#define AMDGPU_UVD_FIRMWARE_SIZE(adev)    \
+	(AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(((const struct common_firmware_header *)(adev)->uvd.fw->data)->ucode_size_bytes) + \
+			       8) - AMDGPU_UVD_FIRMWARE_OFFSET)
+
+struct amdgpu_uvd_inst {
+	struct amdgpu_bo	*vcpu_bo;
+	void			*cpu_addr;
+	uint64_t		gpu_addr;
+	void			*saved_bo;
+	struct amdgpu_ring	ring;
+	struct amdgpu_ring	ring_enc[AMDGPU_MAX_UVD_ENC_RINGS];
+	struct amdgpu_irq_src	irq;
+	uint32_t                srbm_soft_reset;
+};
+
+#define AMDGPU_UVD_HARVEST_UVD0 (1 << 0)
+#define AMDGPU_UVD_HARVEST_UVD1 (1 << 1)
+
+struct amdgpu_uvd {
+	const struct firmware	*fw;	/* UVD firmware */
+	unsigned		fw_version;
+	unsigned		max_handles;
+	unsigned		num_enc_rings;
+	uint8_t			num_uvd_inst;
+	bool			address_64_bit;
+	bool			use_ctx_buf;
+	struct amdgpu_uvd_inst	inst[AMDGPU_MAX_UVD_INSTANCES];
+	struct drm_file		*filp[AMDGPU_MAX_UVD_HANDLES];
+	atomic_t		handles[AMDGPU_MAX_UVD_HANDLES];
+	struct drm_sched_entity entity;
+	struct delayed_work	idle_work;
+	unsigned		harvest_config;
+};
+
+int amdgpu_uvd_sw_init(struct amdgpu_device *adev);
+int amdgpu_uvd_sw_fini(struct amdgpu_device *adev);
+int amdgpu_uvd_entity_init(struct amdgpu_device *adev);
+int amdgpu_uvd_suspend(struct amdgpu_device *adev);
+int amdgpu_uvd_resume(struct amdgpu_device *adev);
+int amdgpu_uvd_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+			      struct dma_fence **fence);
+int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+			       bool direct, struct dma_fence **fence);
+void amdgpu_uvd_free_handles(struct amdgpu_device *adev,
+			     struct drm_file *filp);
+int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx);
+void amdgpu_uvd_ring_begin_use(struct amdgpu_ring *ring);
+void amdgpu_uvd_ring_end_use(struct amdgpu_ring *ring);
+int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring, long timeout);
+uint32_t amdgpu_uvd_used_handles(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_uvd_v4_2.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_uvd_v4_2.c
new file mode 100644
index 00000000000..9e1b8f03079
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_uvd_v4_2.c
@@ -0,0 +1,793 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_uvd.h"
+#include "cikd.h"
+
+#include "uvd/uvd_4_2_d.h"
+#include "uvd/uvd_4_2_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+#include "bif/bif_4_1_d.h"
+
+#include "smu/smu_7_0_1_d.h"
+#include "smu/smu_7_0_1_sh_mask.h"
+
+static void uvd_v4_2_mc_resume(struct amdgpu_device *adev);
+static void uvd_v4_2_set_ring_funcs(struct amdgpu_device *adev);
+static void uvd_v4_2_set_irq_funcs(struct amdgpu_device *adev);
+static int uvd_v4_2_start(struct amdgpu_device *adev);
+static void uvd_v4_2_stop(struct amdgpu_device *adev);
+static int uvd_v4_2_set_clockgating_state(void *handle,
+				enum amd_clockgating_state state);
+static void uvd_v4_2_set_dcm(struct amdgpu_device *adev,
+			     bool sw_mode);
+/**
+ * uvd_v4_2_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t uvd_v4_2_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_RPTR);
+}
+
+/**
+ * uvd_v4_2_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t uvd_v4_2_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_WPTR);
+}
+
+/**
+ * uvd_v4_2_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void uvd_v4_2_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32(mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+}
+
+static int uvd_v4_2_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	adev->uvd.num_uvd_inst = 1;
+
+	uvd_v4_2_set_ring_funcs(adev);
+	uvd_v4_2_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int uvd_v4_2_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r;
+
+	/* UVD TRAP */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 124, &adev->uvd.inst->irq);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_sw_init(adev);
+	if (r)
+		return r;
+
+	ring = &adev->uvd.inst->ring;
+	snprintf(ring->name, sizeof(ring->name), "uvd");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_entity_init(adev);
+
+	return r;
+}
+
+static int uvd_v4_2_sw_fini(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_suspend(adev);
+	if (r)
+		return r;
+
+	return amdgpu_uvd_sw_fini(adev);
+}
+
+static void uvd_v4_2_enable_mgcg(struct amdgpu_device *adev,
+				 bool enable);
+/**
+ * uvd_v4_2_hw_init - start and test UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int uvd_v4_2_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
+	uint32_t tmp;
+	int r;
+
+	uvd_v4_2_enable_mgcg(adev, true);
+	amdgpu_asic_set_uvd_clocks(adev, 10000, 10000);
+
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		goto done;
+	}
+
+	r = amdgpu_ring_alloc(ring, 10);
+	if (r) {
+		DRM_ERROR("amdgpu: ring failed to lock UVD ring (%d).\n", r);
+		goto done;
+	}
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	/* Clear timeout status bits */
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_TIMEOUT_STATUS, 0));
+	amdgpu_ring_write(ring, 0x8);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_CNTL, 0));
+	amdgpu_ring_write(ring, 3);
+
+	amdgpu_ring_commit(ring);
+
+done:
+	if (!r)
+		DRM_INFO("UVD initialized successfully.\n");
+
+	return r;
+}
+
+/**
+ * uvd_v4_2_hw_fini - stop the hardware block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the UVD block, mark ring as not ready any more
+ */
+static int uvd_v4_2_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
+
+	if (RREG32(mmUVD_STATUS) != 0)
+		uvd_v4_2_stop(adev);
+
+	ring->ready = false;
+
+	return 0;
+}
+
+static int uvd_v4_2_suspend(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = uvd_v4_2_hw_fini(adev);
+	if (r)
+		return r;
+
+	return amdgpu_uvd_suspend(adev);
+}
+
+static int uvd_v4_2_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	return uvd_v4_2_hw_init(adev);
+}
+
+/**
+ * uvd_v4_2_start - start UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the UVD block
+ */
+static int uvd_v4_2_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
+	uint32_t rb_bufsz;
+	int i, j, r;
+	u32 tmp;
+	/* disable byte swapping */
+	u32 lmi_swap_cntl = 0;
+	u32 mp_swap_cntl = 0;
+
+	/* set uvd busy */
+	WREG32_P(mmUVD_STATUS, 1<<2, ~(1<<2));
+
+	uvd_v4_2_set_dcm(adev, true);
+	WREG32(mmUVD_CGC_GATE, 0);
+
+	/* take UVD block out of reset */
+	WREG32_P(mmSRBM_SOFT_RESET, 0, ~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK);
+	mdelay(5);
+
+	/* enable VCPU clock */
+	WREG32(mmUVD_VCPU_CNTL,  1 << 9);
+
+	/* disable interupt */
+	WREG32_P(mmUVD_MASTINT_EN, 0, ~(1 << 1));
+
+#ifdef __BIG_ENDIAN
+	/* swap (8 in 32) RB and IB */
+	lmi_swap_cntl = 0xa;
+	mp_swap_cntl = 0;
+#endif
+	WREG32(mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
+	WREG32(mmUVD_MP_SWAP_CNTL, mp_swap_cntl);
+	/* initialize UVD memory controller */
+	WREG32(mmUVD_LMI_CTRL, 0x203108);
+
+	tmp = RREG32(mmUVD_MPC_CNTL);
+	WREG32(mmUVD_MPC_CNTL, tmp | 0x10);
+
+	WREG32(mmUVD_MPC_SET_MUXA0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXA1, 0x0);
+	WREG32(mmUVD_MPC_SET_MUXB0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXB1, 0x0);
+	WREG32(mmUVD_MPC_SET_ALU, 0);
+	WREG32(mmUVD_MPC_SET_MUX, 0x88);
+
+	uvd_v4_2_mc_resume(adev);
+
+	tmp = RREG32_UVD_CTX(ixUVD_LMI_CACHE_CTRL);
+	WREG32_UVD_CTX(ixUVD_LMI_CACHE_CTRL, tmp & (~0x10));
+
+	/* enable UMC */
+	WREG32_P(mmUVD_LMI_CTRL2, 0, ~(1 << 8));
+
+	WREG32_P(mmUVD_SOFT_RESET, 0, ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK);
+
+	WREG32_P(mmUVD_SOFT_RESET, 0, ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+
+	WREG32_P(mmUVD_SOFT_RESET, 0, ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+
+	mdelay(10);
+
+	for (i = 0; i < 10; ++i) {
+		uint32_t status;
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("UVD not responding, trying to reset the VCPU!!!\n");
+		WREG32_P(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
+				~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		WREG32_P(mmUVD_SOFT_RESET, 0, ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		r = -1;
+	}
+
+	if (r) {
+		DRM_ERROR("UVD not responding, giving up!!!\n");
+		return r;
+	}
+
+	/* enable interupt */
+	WREG32_P(mmUVD_MASTINT_EN, 3<<1, ~(3 << 1));
+
+	WREG32_P(mmUVD_STATUS, 0, ~(1<<2));
+
+	/* force RBC into idle state */
+	WREG32(mmUVD_RBC_RB_CNTL, 0x11010101);
+
+	/* Set the write pointer delay */
+	WREG32(mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* programm the 4GB memory segment for rptr and ring buffer */
+	WREG32(mmUVD_LMI_EXT40_ADDR, upper_32_bits(ring->gpu_addr) |
+				   (0x7 << 16) | (0x1 << 31));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(mmUVD_RBC_RB_RPTR, 0x0);
+
+	ring->wptr = RREG32(mmUVD_RBC_RB_RPTR);
+	WREG32(mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+
+	/* set the ring address */
+	WREG32(mmUVD_RBC_RB_BASE, ring->gpu_addr);
+
+	/* Set ring buffer size */
+	rb_bufsz = order_base_2(ring->ring_size);
+	rb_bufsz = (0x1 << 8) | rb_bufsz;
+	WREG32_P(mmUVD_RBC_RB_CNTL, rb_bufsz, ~0x11f1f);
+
+	return 0;
+}
+
+/**
+ * uvd_v4_2_stop - stop UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the UVD block
+ */
+static void uvd_v4_2_stop(struct amdgpu_device *adev)
+{
+	uint32_t i, j;
+	uint32_t status;
+
+	WREG32(mmUVD_RBC_RB_CNTL, 0x11010101);
+
+	for (i = 0; i < 10; ++i) {
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(1);
+		}
+		if (status & 2)
+			break;
+	}
+
+	for (i = 0; i < 10; ++i) {
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_LMI_STATUS);
+			if (status & 0xf)
+				break;
+			mdelay(1);
+		}
+		if (status & 0xf)
+			break;
+	}
+
+	/* Stall UMC and register bus before resetting VCPU */
+	WREG32_P(mmUVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+
+	for (i = 0; i < 10; ++i) {
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_LMI_STATUS);
+			if (status & 0x240)
+				break;
+			mdelay(1);
+		}
+		if (status & 0x240)
+			break;
+	}
+
+	WREG32_P(0x3D49, 0, ~(1 << 2));
+
+	WREG32_P(mmUVD_VCPU_CNTL, 0, ~(1 << 9));
+
+	/* put LMI, VCPU, RBC etc... into reset */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+
+	WREG32(mmUVD_STATUS, 0);
+
+	uvd_v4_2_set_dcm(adev, false);
+}
+
+/**
+ * uvd_v4_2_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @fence: fence to emit
+ *
+ * Write a fence and a trap command to the ring.
+ */
+static void uvd_v4_2_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				     unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, addr & 0xffffffff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 2);
+}
+
+/**
+ * uvd_v4_2_ring_test_ring - register write test
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully write to the context register
+ */
+static int uvd_v4_2_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmUVD_CONTEXT_ID);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	return r;
+}
+
+/**
+ * uvd_v4_2_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @ib: indirect buffer to execute
+ *
+ * Write ring commands to execute the indirect buffer
+ */
+static void uvd_v4_2_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib,
+				  unsigned vmid, bool ctx_switch)
+{
+	amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_BASE, 0));
+	amdgpu_ring_write(ring, ib->gpu_addr);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_SIZE, 0));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+static void uvd_v4_2_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKET0(mmUVD_NO_OP, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+/**
+ * uvd_v4_2_mc_resume - memory controller programming
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Let the UVD memory controller know it's offsets
+ */
+static void uvd_v4_2_mc_resume(struct amdgpu_device *adev)
+{
+	uint64_t addr;
+	uint32_t size;
+
+	/* programm the VCPU memory controller bits 0-27 */
+	addr = (adev->uvd.inst->gpu_addr + AMDGPU_UVD_FIRMWARE_OFFSET) >> 3;
+	size = AMDGPU_UVD_FIRMWARE_SIZE(adev) >> 3;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET0, addr);
+	WREG32(mmUVD_VCPU_CACHE_SIZE0, size);
+
+	addr += size;
+	size = AMDGPU_UVD_HEAP_SIZE >> 3;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET1, addr);
+	WREG32(mmUVD_VCPU_CACHE_SIZE1, size);
+
+	addr += size;
+	size = (AMDGPU_UVD_STACK_SIZE +
+	       (AMDGPU_UVD_SESSION_SIZE * adev->uvd.max_handles)) >> 3;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET2, addr);
+	WREG32(mmUVD_VCPU_CACHE_SIZE2, size);
+
+	/* bits 28-31 */
+	addr = (adev->uvd.inst->gpu_addr >> 28) & 0xF;
+	WREG32(mmUVD_LMI_ADDR_EXT, (addr << 12) | (addr << 0));
+
+	/* bits 32-39 */
+	addr = (adev->uvd.inst->gpu_addr >> 32) & 0xFF;
+	WREG32(mmUVD_LMI_EXT40_ADDR, addr | (0x9 << 16) | (0x1 << 31));
+
+	WREG32(mmUVD_UDEC_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmUVD_UDEC_DB_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmUVD_UDEC_DBW_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+}
+
+static void uvd_v4_2_enable_mgcg(struct amdgpu_device *adev,
+				 bool enable)
+{
+	u32 orig, data;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_UVD_MGCG)) {
+		data = RREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL);
+		data |= 0xfff;
+		WREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL, data);
+
+		orig = data = RREG32(mmUVD_CGC_CTRL);
+		data |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+		if (orig != data)
+			WREG32(mmUVD_CGC_CTRL, data);
+	} else {
+		data = RREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL);
+		data &= ~0xfff;
+		WREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL, data);
+
+		orig = data = RREG32(mmUVD_CGC_CTRL);
+		data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+		if (orig != data)
+			WREG32(mmUVD_CGC_CTRL, data);
+	}
+}
+
+static void uvd_v4_2_set_dcm(struct amdgpu_device *adev,
+			     bool sw_mode)
+{
+	u32 tmp, tmp2;
+
+	WREG32_FIELD(UVD_CGC_GATE, REGS, 0);
+
+	tmp = RREG32(mmUVD_CGC_CTRL);
+	tmp &= ~(UVD_CGC_CTRL__CLK_OFF_DELAY_MASK | UVD_CGC_CTRL__CLK_GATE_DLY_TIMER_MASK);
+	tmp |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK |
+		(1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT) |
+		(4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT);
+
+	if (sw_mode) {
+		tmp &= ~0x7ffff800;
+		tmp2 = UVD_CGC_CTRL2__DYN_OCLK_RAMP_EN_MASK |
+			UVD_CGC_CTRL2__DYN_RCLK_RAMP_EN_MASK |
+			(7 << UVD_CGC_CTRL2__GATER_DIV_ID__SHIFT);
+	} else {
+		tmp |= 0x7ffff800;
+		tmp2 = 0;
+	}
+
+	WREG32(mmUVD_CGC_CTRL, tmp);
+	WREG32_UVD_CTX(ixUVD_CGC_CTRL2, tmp2);
+}
+
+static bool uvd_v4_2_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return !(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK);
+}
+
+static int uvd_v4_2_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK))
+			return 0;
+	}
+	return -ETIMEDOUT;
+}
+
+static int uvd_v4_2_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	uvd_v4_2_stop(adev);
+
+	WREG32_P(mmSRBM_SOFT_RESET, SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK,
+			~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK);
+	mdelay(5);
+
+	return uvd_v4_2_start(adev);
+}
+
+static int uvd_v4_2_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	// TODO
+	return 0;
+}
+
+static int uvd_v4_2_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: UVD TRAP\n");
+	amdgpu_fence_process(&adev->uvd.inst->ring);
+	return 0;
+}
+
+static int uvd_v4_2_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int uvd_v4_2_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the UVD block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_PG_STATE_GATE) {
+		uvd_v4_2_stop(adev);
+		if (adev->pg_flags & AMD_PG_SUPPORT_UVD && !adev->pm.dpm_enabled) {
+			if (!(RREG32_SMC(ixCURRENT_PG_STATUS) &
+				CURRENT_PG_STATUS__UVD_PG_STATUS_MASK)) {
+				WREG32(mmUVD_PGFSM_CONFIG, (UVD_PGFSM_CONFIG__UVD_PGFSM_FSM_ADDR_MASK   |
+							UVD_PGFSM_CONFIG__UVD_PGFSM_POWER_DOWN_MASK |
+							UVD_PGFSM_CONFIG__UVD_PGFSM_P1_SELECT_MASK));
+				mdelay(20);
+			}
+		}
+		return 0;
+	} else {
+		if (adev->pg_flags & AMD_PG_SUPPORT_UVD && !adev->pm.dpm_enabled) {
+			if (RREG32_SMC(ixCURRENT_PG_STATUS) &
+				CURRENT_PG_STATUS__UVD_PG_STATUS_MASK) {
+				WREG32(mmUVD_PGFSM_CONFIG, (UVD_PGFSM_CONFIG__UVD_PGFSM_FSM_ADDR_MASK   |
+						UVD_PGFSM_CONFIG__UVD_PGFSM_POWER_UP_MASK |
+						UVD_PGFSM_CONFIG__UVD_PGFSM_P1_SELECT_MASK));
+				mdelay(30);
+			}
+		}
+		return uvd_v4_2_start(adev);
+	}
+}
+
+static const struct amd_ip_funcs uvd_v4_2_ip_funcs = {
+	.name = "uvd_v4_2",
+	.early_init = uvd_v4_2_early_init,
+	.late_init = NULL,
+	.sw_init = uvd_v4_2_sw_init,
+	.sw_fini = uvd_v4_2_sw_fini,
+	.hw_init = uvd_v4_2_hw_init,
+	.hw_fini = uvd_v4_2_hw_fini,
+	.suspend = uvd_v4_2_suspend,
+	.resume = uvd_v4_2_resume,
+	.is_idle = uvd_v4_2_is_idle,
+	.wait_for_idle = uvd_v4_2_wait_for_idle,
+	.soft_reset = uvd_v4_2_soft_reset,
+	.set_clockgating_state = uvd_v4_2_set_clockgating_state,
+	.set_powergating_state = uvd_v4_2_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs uvd_v4_2_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_UVD,
+	.align_mask = 0xf,
+	.support_64bit_ptrs = false,
+	.get_rptr = uvd_v4_2_ring_get_rptr,
+	.get_wptr = uvd_v4_2_ring_get_wptr,
+	.set_wptr = uvd_v4_2_ring_set_wptr,
+	.parse_cs = amdgpu_uvd_ring_parse_cs,
+	.emit_frame_size =
+		14, /* uvd_v4_2_ring_emit_fence  x1 no user fence */
+	.emit_ib_size = 4, /* uvd_v4_2_ring_emit_ib */
+	.emit_ib = uvd_v4_2_ring_emit_ib,
+	.emit_fence = uvd_v4_2_ring_emit_fence,
+	.test_ring = uvd_v4_2_ring_test_ring,
+	.test_ib = amdgpu_uvd_ring_test_ib,
+	.insert_nop = uvd_v4_2_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
+};
+
+static void uvd_v4_2_set_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->uvd.inst->ring.funcs = &uvd_v4_2_ring_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs uvd_v4_2_irq_funcs = {
+	.set = uvd_v4_2_set_interrupt_state,
+	.process = uvd_v4_2_process_interrupt,
+};
+
+static void uvd_v4_2_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->uvd.inst->irq.num_types = 1;
+	adev->uvd.inst->irq.funcs = &uvd_v4_2_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version uvd_v4_2_ip_block =
+{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 4,
+		.minor = 2,
+		.rev = 0,
+		.funcs = &uvd_v4_2_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_vce.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vce.c
new file mode 100644
index 00000000000..9f1d89f8967
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vce.c
@@ -0,0 +1,1148 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * Authors: Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <drm/drmP.h>
+#include <drm/drm.h>
+
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_vce.h"
+#include "cikd.h"
+
+/* 1 second timeout */
+#define VCE_IDLE_TIMEOUT	msecs_to_jiffies(1000)
+
+/* Firmware Names */
+#ifdef CONFIG_DRM_AMDGPU_CIK
+#define FIRMWARE_BONAIRE	"amdgpu/bonaire_vce.bin"
+#define FIRMWARE_KABINI	"amdgpu/kabini_vce.bin"
+#define FIRMWARE_KAVERI	"amdgpu/kaveri_vce.bin"
+#define FIRMWARE_HAWAII	"amdgpu/hawaii_vce.bin"
+#define FIRMWARE_MULLINS	"amdgpu/mullins_vce.bin"
+#endif
+#define FIRMWARE_TONGA		"amdgpu/tonga_vce.bin"
+#define FIRMWARE_CARRIZO	"amdgpu/carrizo_vce.bin"
+#define FIRMWARE_FIJI		"amdgpu/fiji_vce.bin"
+#define FIRMWARE_STONEY		"amdgpu/stoney_vce.bin"
+#define FIRMWARE_POLARIS10	"amdgpu/polaris10_vce.bin"
+#define FIRMWARE_POLARIS11	"amdgpu/polaris11_vce.bin"
+#define FIRMWARE_POLARIS12	"amdgpu/polaris12_vce.bin"
+#define FIRMWARE_VEGAM		"amdgpu/vegam_vce.bin"
+
+#define FIRMWARE_VEGA10		"amdgpu/vega10_vce.bin"
+#define FIRMWARE_VEGA12		"amdgpu/vega12_vce.bin"
+#define FIRMWARE_VEGA20		"amdgpu/vega20_vce.bin"
+
+#ifdef CONFIG_DRM_AMDGPU_CIK
+MODULE_FIRMWARE(FIRMWARE_BONAIRE);
+MODULE_FIRMWARE(FIRMWARE_KABINI);
+MODULE_FIRMWARE(FIRMWARE_KAVERI);
+MODULE_FIRMWARE(FIRMWARE_HAWAII);
+MODULE_FIRMWARE(FIRMWARE_MULLINS);
+#endif
+MODULE_FIRMWARE(FIRMWARE_TONGA);
+MODULE_FIRMWARE(FIRMWARE_CARRIZO);
+MODULE_FIRMWARE(FIRMWARE_FIJI);
+MODULE_FIRMWARE(FIRMWARE_STONEY);
+MODULE_FIRMWARE(FIRMWARE_POLARIS10);
+MODULE_FIRMWARE(FIRMWARE_POLARIS11);
+MODULE_FIRMWARE(FIRMWARE_POLARIS12);
+MODULE_FIRMWARE(FIRMWARE_VEGAM);
+
+MODULE_FIRMWARE(FIRMWARE_VEGA10);
+MODULE_FIRMWARE(FIRMWARE_VEGA12);
+MODULE_FIRMWARE(FIRMWARE_VEGA20);
+
+static void amdgpu_vce_idle_work_handler(struct work_struct *work);
+
+/**
+ * amdgpu_vce_init - allocate memory, load vce firmware
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * First step to get VCE online, allocate memory and load the firmware
+ */
+int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
+{
+	const char *fw_name;
+	const struct common_firmware_header *hdr;
+	unsigned ucode_version, version_major, version_minor, binary_id;
+	int i, r;
+
+	switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_BONAIRE:
+		fw_name = FIRMWARE_BONAIRE;
+		break;
+	case CHIP_KAVERI:
+		fw_name = FIRMWARE_KAVERI;
+		break;
+	case CHIP_KABINI:
+		fw_name = FIRMWARE_KABINI;
+		break;
+	case CHIP_HAWAII:
+		fw_name = FIRMWARE_HAWAII;
+		break;
+	case CHIP_MULLINS:
+		fw_name = FIRMWARE_MULLINS;
+		break;
+#endif
+	case CHIP_TONGA:
+		fw_name = FIRMWARE_TONGA;
+		break;
+	case CHIP_CARRIZO:
+		fw_name = FIRMWARE_CARRIZO;
+		break;
+	case CHIP_FIJI:
+		fw_name = FIRMWARE_FIJI;
+		break;
+	case CHIP_STONEY:
+		fw_name = FIRMWARE_STONEY;
+		break;
+	case CHIP_POLARIS10:
+		fw_name = FIRMWARE_POLARIS10;
+		break;
+	case CHIP_POLARIS11:
+		fw_name = FIRMWARE_POLARIS11;
+		break;
+	case CHIP_POLARIS12:
+		fw_name = FIRMWARE_POLARIS12;
+		break;
+	case CHIP_VEGAM:
+		fw_name = FIRMWARE_VEGAM;
+		break;
+	case CHIP_VEGA10:
+		fw_name = FIRMWARE_VEGA10;
+		break;
+	case CHIP_VEGA12:
+		fw_name = FIRMWARE_VEGA12;
+		break;
+	case CHIP_VEGA20:
+		fw_name = FIRMWARE_VEGA20;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	r = request_firmware(&adev->vce.fw, fw_name, adev->dev);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_vce: Can't load firmware \"%s\"\n",
+			fw_name);
+		return r;
+	}
+
+	r = amdgpu_ucode_validate(adev->vce.fw);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_vce: Can't validate firmware \"%s\"\n",
+			fw_name);
+		release_firmware(adev->vce.fw);
+		adev->vce.fw = NULL;
+		return r;
+	}
+
+	hdr = (const struct common_firmware_header *)adev->vce.fw->data;
+
+	ucode_version = le32_to_cpu(hdr->ucode_version);
+	version_major = (ucode_version >> 20) & 0xfff;
+	version_minor = (ucode_version >> 8) & 0xfff;
+	binary_id = ucode_version & 0xff;
+	DRM_INFO("Found VCE firmware Version: %d.%d Binary ID: %d\n",
+		version_major, version_minor, binary_id);
+	adev->vce.fw_version = ((version_major << 24) | (version_minor << 16) |
+				(binary_id << 8));
+
+	r = amdgpu_bo_create_kernel(adev, size, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_VRAM, &adev->vce.vcpu_bo,
+				    &adev->vce.gpu_addr, &adev->vce.cpu_addr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate VCE bo\n", r);
+		return r;
+	}
+
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
+		atomic_set(&adev->vce.handles[i], 0);
+		adev->vce.filp[i] = NULL;
+	}
+
+	INIT_DELAYED_WORK(&adev->vce.idle_work, amdgpu_vce_idle_work_handler);
+	rw_init(&adev->vce.idle_mutex, "vceidle");
+
+	return 0;
+}
+
+/**
+ * amdgpu_vce_fini - free memory
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Last step on VCE teardown, free firmware memory
+ */
+int amdgpu_vce_sw_fini(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	if (adev->vce.vcpu_bo == NULL)
+		return 0;
+
+	drm_sched_entity_destroy(&adev->vce.entity);
+
+	amdgpu_bo_free_kernel(&adev->vce.vcpu_bo, &adev->vce.gpu_addr,
+		(void **)&adev->vce.cpu_addr);
+
+	for (i = 0; i < adev->vce.num_rings; i++)
+		amdgpu_ring_fini(&adev->vce.ring[i]);
+
+	release_firmware(adev->vce.fw);
+	mutex_destroy(&adev->vce.idle_mutex);
+
+	return 0;
+}
+
+/**
+ * amdgpu_vce_entity_init - init entity
+ *
+ * @adev: amdgpu_device pointer
+ *
+ */
+int amdgpu_vce_entity_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	struct drm_sched_rq *rq;
+	int r;
+
+	ring = &adev->vce.ring[0];
+	rq = &ring->sched.sched_rq[DRM_SCHED_PRIORITY_NORMAL];
+	r = drm_sched_entity_init(&adev->vce.entity, &rq, 1, NULL);
+	if (r != 0) {
+		DRM_ERROR("Failed setting up VCE run queue.\n");
+		return r;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_vce_suspend - unpin VCE fw memory
+ *
+ * @adev: amdgpu_device pointer
+ *
+ */
+int amdgpu_vce_suspend(struct amdgpu_device *adev)
+{
+	int i;
+
+	cancel_delayed_work_sync(&adev->vce.idle_work);
+
+	if (adev->vce.vcpu_bo == NULL)
+		return 0;
+
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i)
+		if (atomic_read(&adev->vce.handles[i]))
+			break;
+
+	if (i == AMDGPU_MAX_VCE_HANDLES)
+		return 0;
+
+	/* TODO: suspending running encoding sessions isn't supported */
+	return -EINVAL;
+}
+
+/**
+ * amdgpu_vce_resume - pin VCE fw memory
+ *
+ * @adev: amdgpu_device pointer
+ *
+ */
+int amdgpu_vce_resume(struct amdgpu_device *adev)
+{
+	void *cpu_addr;
+	const struct common_firmware_header *hdr;
+	unsigned offset;
+	int r;
+
+	if (adev->vce.vcpu_bo == NULL)
+		return -EINVAL;
+
+	r = amdgpu_bo_reserve(adev->vce.vcpu_bo, false);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to reserve VCE bo\n", r);
+		return r;
+	}
+
+	r = amdgpu_bo_kmap(adev->vce.vcpu_bo, &cpu_addr);
+	if (r) {
+		amdgpu_bo_unreserve(adev->vce.vcpu_bo);
+		dev_err(adev->dev, "(%d) VCE map failed\n", r);
+		return r;
+	}
+
+	hdr = (const struct common_firmware_header *)adev->vce.fw->data;
+	offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
+	memcpy_toio(cpu_addr, adev->vce.fw->data + offset,
+		    adev->vce.fw->size - offset);
+
+	amdgpu_bo_kunmap(adev->vce.vcpu_bo);
+
+	amdgpu_bo_unreserve(adev->vce.vcpu_bo);
+
+	return 0;
+}
+
+/**
+ * amdgpu_vce_idle_work_handler - power off VCE
+ *
+ * @work: pointer to work structure
+ *
+ * power of VCE when it's not used any more
+ */
+static void amdgpu_vce_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, vce.idle_work.work);
+	unsigned i, count = 0;
+
+	for (i = 0; i < adev->vce.num_rings; i++)
+		count += amdgpu_fence_count_emitted(&adev->vce.ring[i]);
+
+	if (count == 0) {
+		if (adev->pm.dpm_enabled) {
+			amdgpu_dpm_enable_vce(adev, false);
+		} else {
+			amdgpu_asic_set_vce_clocks(adev, 0, 0);
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							       AMD_PG_STATE_GATE);
+			amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							       AMD_CG_STATE_GATE);
+		}
+	} else {
+		schedule_delayed_work(&adev->vce.idle_work, VCE_IDLE_TIMEOUT);
+	}
+}
+
+/**
+ * amdgpu_vce_ring_begin_use - power up VCE
+ *
+ * @ring: amdgpu ring
+ *
+ * Make sure VCE is powerd up when we want to use it
+ */
+void amdgpu_vce_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	bool set_clocks;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	mutex_lock(&adev->vce.idle_mutex);
+	set_clocks = !cancel_delayed_work_sync(&adev->vce.idle_work);
+	if (set_clocks) {
+		if (adev->pm.dpm_enabled) {
+			amdgpu_dpm_enable_vce(adev, true);
+		} else {
+			amdgpu_asic_set_vce_clocks(adev, 53300, 40000);
+			amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							       AMD_CG_STATE_UNGATE);
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							       AMD_PG_STATE_UNGATE);
+
+		}
+	}
+	mutex_unlock(&adev->vce.idle_mutex);
+}
+
+/**
+ * amdgpu_vce_ring_end_use - power VCE down
+ *
+ * @ring: amdgpu ring
+ *
+ * Schedule work to power VCE down again
+ */
+void amdgpu_vce_ring_end_use(struct amdgpu_ring *ring)
+{
+	if (!amdgpu_sriov_vf(ring->adev))
+		schedule_delayed_work(&ring->adev->vce.idle_work, VCE_IDLE_TIMEOUT);
+}
+
+/**
+ * amdgpu_vce_free_handles - free still open VCE handles
+ *
+ * @adev: amdgpu_device pointer
+ * @filp: drm file pointer
+ *
+ * Close all VCE handles still open by this file pointer
+ */
+void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp)
+{
+	struct amdgpu_ring *ring = &adev->vce.ring[0];
+	int i, r;
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
+		uint32_t handle = atomic_read(&adev->vce.handles[i]);
+
+		if (!handle || adev->vce.filp[i] != filp)
+			continue;
+
+		r = amdgpu_vce_get_destroy_msg(ring, handle, false, NULL);
+		if (r)
+			DRM_ERROR("Error destroying VCE handle (%d)!\n", r);
+
+		adev->vce.filp[i] = NULL;
+		atomic_set(&adev->vce.handles[i], 0);
+	}
+}
+
+/**
+ * amdgpu_vce_get_create_msg - generate a VCE create msg
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: ring we should submit the msg to
+ * @handle: VCE session handle to use
+ * @fence: optional fence to return
+ *
+ * Open up a stream for HW test
+ */
+int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+			      struct dma_fence **fence)
+{
+	const unsigned ib_size_dw = 1024;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct dma_fence *f = NULL;
+	uint64_t dummy;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	if (r)
+		return r;
+
+	ib = &job->ibs[0];
+
+	dummy = ib->gpu_addr + 1024;
+
+	/* stitch together an VCE create msg */
+	ib->length_dw = 0;
+	ib->ptr[ib->length_dw++] = 0x0000000c; /* len */
+	ib->ptr[ib->length_dw++] = 0x00000001; /* session cmd */
+	ib->ptr[ib->length_dw++] = handle;
+
+	if ((ring->adev->vce.fw_version >> 24) >= 52)
+		ib->ptr[ib->length_dw++] = 0x00000040; /* len */
+	else
+		ib->ptr[ib->length_dw++] = 0x00000030; /* len */
+	ib->ptr[ib->length_dw++] = 0x01000001; /* create cmd */
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	ib->ptr[ib->length_dw++] = 0x00000042;
+	ib->ptr[ib->length_dw++] = 0x0000000a;
+	ib->ptr[ib->length_dw++] = 0x00000001;
+	ib->ptr[ib->length_dw++] = 0x00000080;
+	ib->ptr[ib->length_dw++] = 0x00000060;
+	ib->ptr[ib->length_dw++] = 0x00000100;
+	ib->ptr[ib->length_dw++] = 0x00000100;
+	ib->ptr[ib->length_dw++] = 0x0000000c;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	if ((ring->adev->vce.fw_version >> 24) >= 52) {
+		ib->ptr[ib->length_dw++] = 0x00000000;
+		ib->ptr[ib->length_dw++] = 0x00000000;
+		ib->ptr[ib->length_dw++] = 0x00000000;
+		ib->ptr[ib->length_dw++] = 0x00000000;
+	}
+
+	ib->ptr[ib->length_dw++] = 0x00000014; /* len */
+	ib->ptr[ib->length_dw++] = 0x05000005; /* feedback buffer */
+	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
+	ib->ptr[ib->length_dw++] = dummy;
+	ib->ptr[ib->length_dw++] = 0x00000001;
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err;
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+	return 0;
+
+err:
+	amdgpu_job_free(job);
+	return r;
+}
+
+/**
+ * amdgpu_vce_get_destroy_msg - generate a VCE destroy msg
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: ring we should submit the msg to
+ * @handle: VCE session handle to use
+ * @fence: optional fence to return
+ *
+ * Close up a stream for HW test or if userspace failed to do so
+ */
+int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+			       bool direct, struct dma_fence **fence)
+{
+	const unsigned ib_size_dw = 1024;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct dma_fence *f = NULL;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	if (r)
+		return r;
+
+	ib = &job->ibs[0];
+
+	/* stitch together an VCE destroy msg */
+	ib->length_dw = 0;
+	ib->ptr[ib->length_dw++] = 0x0000000c; /* len */
+	ib->ptr[ib->length_dw++] = 0x00000001; /* session cmd */
+	ib->ptr[ib->length_dw++] = handle;
+
+	ib->ptr[ib->length_dw++] = 0x00000020; /* len */
+	ib->ptr[ib->length_dw++] = 0x00000002; /* task info */
+	ib->ptr[ib->length_dw++] = 0xffffffff; /* next task info, set to 0xffffffff if no */
+	ib->ptr[ib->length_dw++] = 0x00000001; /* destroy session */
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	ib->ptr[ib->length_dw++] = 0xffffffff; /* feedback is not needed, set to 0xffffffff and firmware will not output feedback */
+	ib->ptr[ib->length_dw++] = 0x00000000;
+
+	ib->ptr[ib->length_dw++] = 0x00000008; /* len */
+	ib->ptr[ib->length_dw++] = 0x02000001; /* destroy cmd */
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+
+	if (direct)
+		r = amdgpu_job_submit_direct(job, ring, &f);
+	else
+		r = amdgpu_job_submit(job, &ring->adev->vce.entity,
+				      AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+	if (r)
+		goto err;
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+	return 0;
+
+err:
+	amdgpu_job_free(job);
+	return r;
+}
+
+/**
+ * amdgpu_vce_cs_validate_bo - make sure not to cross 4GB boundary
+ *
+ * @p: parser context
+ * @lo: address of lower dword
+ * @hi: address of higher dword
+ * @size: minimum size
+ * @index: bs/fb index
+ *
+ * Make sure that no BO cross a 4GB boundary.
+ */
+static int amdgpu_vce_validate_bo(struct amdgpu_cs_parser *p, uint32_t ib_idx,
+				  int lo, int hi, unsigned size, int32_t index)
+{
+	int64_t offset = ((uint64_t)size) * ((int64_t)index);
+	struct ttm_operation_ctx ctx = { false, false };
+	struct amdgpu_bo_va_mapping *mapping;
+	unsigned i, fpfn, lpfn;
+	struct amdgpu_bo *bo;
+	uint64_t addr;
+	int r;
+
+	addr = ((uint64_t)amdgpu_get_ib_value(p, ib_idx, lo)) |
+	       ((uint64_t)amdgpu_get_ib_value(p, ib_idx, hi)) << 32;
+	if (index >= 0) {
+		addr += offset;
+		fpfn = PAGE_ALIGN(offset) >> PAGE_SHIFT;
+		lpfn = 0x100000000ULL >> PAGE_SHIFT;
+	} else {
+		fpfn = 0;
+		lpfn = (0x100000000ULL - PAGE_ALIGN(offset)) >> PAGE_SHIFT;
+	}
+
+	r = amdgpu_cs_find_mapping(p, addr, &bo, &mapping);
+	if (r) {
+		DRM_ERROR("Can't find BO for addr 0x%010Lx %d %d %d %d\n",
+			  addr, lo, hi, size, index);
+		return r;
+	}
+
+	for (i = 0; i < bo->placement.num_placement; ++i) {
+		bo->placements[i].fpfn = max(bo->placements[i].fpfn, fpfn);
+		bo->placements[i].lpfn = bo->placements[i].lpfn ?
+			min(bo->placements[i].lpfn, lpfn) : lpfn;
+	}
+	return ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+}
+
+
+/**
+ * amdgpu_vce_cs_reloc - command submission relocation
+ *
+ * @p: parser context
+ * @lo: address of lower dword
+ * @hi: address of higher dword
+ * @size: minimum size
+ *
+ * Patch relocation inside command stream with real buffer address
+ */
+static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx,
+			       int lo, int hi, unsigned size, uint32_t index)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_bo *bo;
+	uint64_t addr;
+	int r;
+
+	if (index == 0xffffffff)
+		index = 0;
+
+	addr = ((uint64_t)amdgpu_get_ib_value(p, ib_idx, lo)) |
+	       ((uint64_t)amdgpu_get_ib_value(p, ib_idx, hi)) << 32;
+	addr += ((uint64_t)size) * ((uint64_t)index);
+
+	r = amdgpu_cs_find_mapping(p, addr, &bo, &mapping);
+	if (r) {
+		DRM_ERROR("Can't find BO for addr 0x%010Lx %d %d %d %d\n",
+			  addr, lo, hi, size, index);
+		return r;
+	}
+
+	if ((addr + (uint64_t)size) >
+	    (mapping->last + 1) * AMDGPU_GPU_PAGE_SIZE) {
+		DRM_ERROR("BO to small for addr 0x%010Lx %d %d\n",
+			  addr, lo, hi);
+		return -EINVAL;
+	}
+
+	addr -= mapping->start * AMDGPU_GPU_PAGE_SIZE;
+	addr += amdgpu_bo_gpu_offset(bo);
+	addr -= ((uint64_t)size) * ((uint64_t)index);
+
+	amdgpu_set_ib_value(p, ib_idx, lo, lower_32_bits(addr));
+	amdgpu_set_ib_value(p, ib_idx, hi, upper_32_bits(addr));
+
+	return 0;
+}
+
+/**
+ * amdgpu_vce_validate_handle - validate stream handle
+ *
+ * @p: parser context
+ * @handle: handle to validate
+ * @allocated: allocated a new handle?
+ *
+ * Validates the handle and return the found session index or -EINVAL
+ * we we don't have another free session index.
+ */
+static int amdgpu_vce_validate_handle(struct amdgpu_cs_parser *p,
+				      uint32_t handle, uint32_t *allocated)
+{
+	unsigned i;
+
+	/* validate the handle */
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
+		if (atomic_read(&p->adev->vce.handles[i]) == handle) {
+			if (p->adev->vce.filp[i] != p->filp) {
+				DRM_ERROR("VCE handle collision detected!\n");
+				return -EINVAL;
+			}
+			return i;
+		}
+	}
+
+	/* handle not found try to alloc a new one */
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
+		if (!atomic_cmpxchg(&p->adev->vce.handles[i], 0, handle)) {
+			p->adev->vce.filp[i] = p->filp;
+			p->adev->vce.img_size[i] = 0;
+			*allocated |= 1 << i;
+			return i;
+		}
+	}
+
+	DRM_ERROR("No more free VCE handles!\n");
+	return -EINVAL;
+}
+
+/**
+ * amdgpu_vce_cs_parse - parse and validate the command stream
+ *
+ * @p: parser context
+ *
+ */
+int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
+{
+	struct amdgpu_ib *ib = &p->job->ibs[ib_idx];
+	unsigned fb_idx = 0, bs_idx = 0;
+	int session_idx = -1;
+	uint32_t destroyed = 0;
+	uint32_t created = 0;
+	uint32_t allocated = 0;
+	uint32_t tmp, handle = 0;
+	uint32_t *size = &tmp;
+	unsigned idx;
+	int i, r = 0;
+
+	p->job->vm = NULL;
+	ib->gpu_addr = amdgpu_sa_bo_gpu_addr(ib->sa_bo);
+
+	for (idx = 0; idx < ib->length_dw;) {
+		uint32_t len = amdgpu_get_ib_value(p, ib_idx, idx);
+		uint32_t cmd = amdgpu_get_ib_value(p, ib_idx, idx + 1);
+
+		if ((len < 8) || (len & 3)) {
+			DRM_ERROR("invalid VCE command length (%d)!\n", len);
+			r = -EINVAL;
+			goto out;
+		}
+
+		switch (cmd) {
+		case 0x00000002: /* task info */
+			fb_idx = amdgpu_get_ib_value(p, ib_idx, idx + 6);
+			bs_idx = amdgpu_get_ib_value(p, ib_idx, idx + 7);
+			break;
+
+		case 0x03000001: /* encode */
+			r = amdgpu_vce_validate_bo(p, ib_idx, idx + 10,
+						   idx + 9, 0, 0);
+			if (r)
+				goto out;
+
+			r = amdgpu_vce_validate_bo(p, ib_idx, idx + 12,
+						   idx + 11, 0, 0);
+			if (r)
+				goto out;
+			break;
+
+		case 0x05000001: /* context buffer */
+			r = amdgpu_vce_validate_bo(p, ib_idx, idx + 3,
+						   idx + 2, 0, 0);
+			if (r)
+				goto out;
+			break;
+
+		case 0x05000004: /* video bitstream buffer */
+			tmp = amdgpu_get_ib_value(p, ib_idx, idx + 4);
+			r = amdgpu_vce_validate_bo(p, ib_idx, idx + 3, idx + 2,
+						   tmp, bs_idx);
+			if (r)
+				goto out;
+			break;
+
+		case 0x05000005: /* feedback buffer */
+			r = amdgpu_vce_validate_bo(p, ib_idx, idx + 3, idx + 2,
+						   4096, fb_idx);
+			if (r)
+				goto out;
+			break;
+
+		case 0x0500000d: /* MV buffer */
+			r = amdgpu_vce_validate_bo(p, ib_idx, idx + 3,
+							idx + 2, 0, 0);
+			if (r)
+				goto out;
+
+			r = amdgpu_vce_validate_bo(p, ib_idx, idx + 8,
+							idx + 7, 0, 0);
+			if (r)
+				goto out;
+			break;
+		}
+
+		idx += len / 4;
+	}
+
+	for (idx = 0; idx < ib->length_dw;) {
+		uint32_t len = amdgpu_get_ib_value(p, ib_idx, idx);
+		uint32_t cmd = amdgpu_get_ib_value(p, ib_idx, idx + 1);
+
+		switch (cmd) {
+		case 0x00000001: /* session */
+			handle = amdgpu_get_ib_value(p, ib_idx, idx + 2);
+			session_idx = amdgpu_vce_validate_handle(p, handle,
+								 &allocated);
+			if (session_idx < 0) {
+				r = session_idx;
+				goto out;
+			}
+			size = &p->adev->vce.img_size[session_idx];
+			break;
+
+		case 0x00000002: /* task info */
+			fb_idx = amdgpu_get_ib_value(p, ib_idx, idx + 6);
+			bs_idx = amdgpu_get_ib_value(p, ib_idx, idx + 7);
+			break;
+
+		case 0x01000001: /* create */
+			created |= 1 << session_idx;
+			if (destroyed & (1 << session_idx)) {
+				destroyed &= ~(1 << session_idx);
+				allocated |= 1 << session_idx;
+
+			} else if (!(allocated & (1 << session_idx))) {
+				DRM_ERROR("Handle already in use!\n");
+				r = -EINVAL;
+				goto out;
+			}
+
+			*size = amdgpu_get_ib_value(p, ib_idx, idx + 8) *
+				amdgpu_get_ib_value(p, ib_idx, idx + 10) *
+				8 * 3 / 2;
+			break;
+
+		case 0x04000001: /* config extension */
+		case 0x04000002: /* pic control */
+		case 0x04000005: /* rate control */
+		case 0x04000007: /* motion estimation */
+		case 0x04000008: /* rdo */
+		case 0x04000009: /* vui */
+		case 0x05000002: /* auxiliary buffer */
+		case 0x05000009: /* clock table */
+			break;
+
+		case 0x0500000c: /* hw config */
+			switch (p->adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_CIK
+			case CHIP_KAVERI:
+			case CHIP_MULLINS:
+#endif
+			case CHIP_CARRIZO:
+				break;
+			default:
+				r = -EINVAL;
+				goto out;
+			}
+			break;
+
+		case 0x03000001: /* encode */
+			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 10, idx + 9,
+						*size, 0);
+			if (r)
+				goto out;
+
+			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 12, idx + 11,
+						*size / 3, 0);
+			if (r)
+				goto out;
+			break;
+
+		case 0x02000001: /* destroy */
+			destroyed |= 1 << session_idx;
+			break;
+
+		case 0x05000001: /* context buffer */
+			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2,
+						*size * 2, 0);
+			if (r)
+				goto out;
+			break;
+
+		case 0x05000004: /* video bitstream buffer */
+			tmp = amdgpu_get_ib_value(p, ib_idx, idx + 4);
+			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2,
+						tmp, bs_idx);
+			if (r)
+				goto out;
+			break;
+
+		case 0x05000005: /* feedback buffer */
+			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2,
+						4096, fb_idx);
+			if (r)
+				goto out;
+			break;
+
+		case 0x0500000d: /* MV buffer */
+			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3,
+							idx + 2, *size, 0);
+			if (r)
+				goto out;
+
+			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 8,
+							idx + 7, *size / 12, 0);
+			if (r)
+				goto out;
+			break;
+
+		default:
+			DRM_ERROR("invalid VCE command (0x%x)!\n", cmd);
+			r = -EINVAL;
+			goto out;
+		}
+
+		if (session_idx == -1) {
+			DRM_ERROR("no session command at start of IB\n");
+			r = -EINVAL;
+			goto out;
+		}
+
+		idx += len / 4;
+	}
+
+	if (allocated & ~created) {
+		DRM_ERROR("New session without create command!\n");
+		r = -ENOENT;
+	}
+
+out:
+	if (!r) {
+		/* No error, free all destroyed handle slots */
+		tmp = destroyed;
+	} else {
+		/* Error during parsing, free all allocated handle slots */
+		tmp = allocated;
+	}
+
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i)
+		if (tmp & (1 << i))
+			atomic_set(&p->adev->vce.handles[i], 0);
+
+	return r;
+}
+
+/**
+ * amdgpu_vce_cs_parse_vm - parse the command stream in VM mode
+ *
+ * @p: parser context
+ *
+ */
+int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx)
+{
+	struct amdgpu_ib *ib = &p->job->ibs[ib_idx];
+	int session_idx = -1;
+	uint32_t destroyed = 0;
+	uint32_t created = 0;
+	uint32_t allocated = 0;
+	uint32_t tmp, handle = 0;
+	int i, r = 0, idx = 0;
+
+	while (idx < ib->length_dw) {
+		uint32_t len = amdgpu_get_ib_value(p, ib_idx, idx);
+		uint32_t cmd = amdgpu_get_ib_value(p, ib_idx, idx + 1);
+
+		if ((len < 8) || (len & 3)) {
+			DRM_ERROR("invalid VCE command length (%d)!\n", len);
+			r = -EINVAL;
+			goto out;
+		}
+
+		switch (cmd) {
+		case 0x00000001: /* session */
+			handle = amdgpu_get_ib_value(p, ib_idx, idx + 2);
+			session_idx = amdgpu_vce_validate_handle(p, handle,
+								 &allocated);
+			if (session_idx < 0) {
+				r = session_idx;
+				goto out;
+			}
+			break;
+
+		case 0x01000001: /* create */
+			created |= 1 << session_idx;
+			if (destroyed & (1 << session_idx)) {
+				destroyed &= ~(1 << session_idx);
+				allocated |= 1 << session_idx;
+
+			} else if (!(allocated & (1 << session_idx))) {
+				DRM_ERROR("Handle already in use!\n");
+				r = -EINVAL;
+				goto out;
+			}
+
+			break;
+
+		case 0x02000001: /* destroy */
+			destroyed |= 1 << session_idx;
+			break;
+
+		default:
+			break;
+		}
+
+		if (session_idx == -1) {
+			DRM_ERROR("no session command at start of IB\n");
+			r = -EINVAL;
+			goto out;
+		}
+
+		idx += len / 4;
+	}
+
+	if (allocated & ~created) {
+		DRM_ERROR("New session without create command!\n");
+		r = -ENOENT;
+	}
+
+out:
+	if (!r) {
+		/* No error, free all destroyed handle slots */
+		tmp = destroyed;
+		amdgpu_ib_free(p->adev, ib, NULL);
+	} else {
+		/* Error during parsing, free all allocated handle slots */
+		tmp = allocated;
+	}
+
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i)
+		if (tmp & (1 << i))
+			atomic_set(&p->adev->vce.handles[i], 0);
+
+	return r;
+}
+
+/**
+ * amdgpu_vce_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: engine to use
+ * @ib: the IB to execute
+ *
+ */
+void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib,
+			     unsigned vmid, bool ctx_switch)
+{
+	amdgpu_ring_write(ring, VCE_CMD_IB);
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+/**
+ * amdgpu_vce_ring_emit_fence - add a fence command to the ring
+ *
+ * @ring: engine to use
+ * @fence: the fence
+ *
+ */
+void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, VCE_CMD_FENCE);
+	amdgpu_ring_write(ring, addr);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, VCE_CMD_TRAP);
+	amdgpu_ring_write(ring, VCE_CMD_END);
+}
+
+/**
+ * amdgpu_vce_ring_test_ring - test if VCE ring is working
+ *
+ * @ring: the engine to test on
+ *
+ */
+int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t rptr = amdgpu_ring_get_rptr(ring);
+	unsigned i;
+	int r, timeout = adev->usec_timeout;
+
+	/* skip ring test for sriov*/
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	r = amdgpu_ring_alloc(ring, 16);
+	if (r) {
+		DRM_ERROR("amdgpu: vce failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring, VCE_CMD_END);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < timeout; i++) {
+		if (amdgpu_ring_get_rptr(ring) != rptr)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed\n",
+			  ring->idx);
+		r = -ETIMEDOUT;
+	}
+
+	return r;
+}
+
+/**
+ * amdgpu_vce_ring_test_ib - test if VCE IBs are working
+ *
+ * @ring: the engine to test on
+ *
+ */
+int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct dma_fence *fence = NULL;
+	long r;
+
+	/* skip vce ring1/2 ib test for now, since it's not reliable */
+	if (ring != &ring->adev->vce.ring[0])
+		return 0;
+
+	r = amdgpu_vce_get_create_msg(ring, 1, NULL);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
+		goto error;
+	}
+
+	r = amdgpu_vce_get_destroy_msg(ring, 1, true, &fence);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
+		goto error;
+	}
+
+	r = dma_fence_wait_timeout(fence, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out.\n");
+		r = -ETIMEDOUT;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+	} else {
+		DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
+	}
+error:
+	dma_fence_put(fence);
+	return r;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_vce.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vce.h
new file mode 100644
index 00000000000..752dccbacef
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vce.h
@@ -0,0 +1,79 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_VCE_H__
+#define __AMDGPU_VCE_H__
+
+#define AMDGPU_MAX_VCE_HANDLES	16
+#define AMDGPU_VCE_FIRMWARE_OFFSET 256
+
+#define AMDGPU_VCE_HARVEST_VCE0 (1 << 0)
+#define AMDGPU_VCE_HARVEST_VCE1 (1 << 1)
+
+#define AMDGPU_VCE_FW_53_45	((53 << 24) | (45 << 16))
+
+struct amdgpu_vce {
+	struct amdgpu_bo	*vcpu_bo;
+	uint64_t		gpu_addr;
+	void			*cpu_addr;
+	void			*saved_bo;
+	unsigned		fw_version;
+	unsigned		fb_version;
+	atomic_t		handles[AMDGPU_MAX_VCE_HANDLES];
+	struct drm_file		*filp[AMDGPU_MAX_VCE_HANDLES];
+	uint32_t		img_size[AMDGPU_MAX_VCE_HANDLES];
+	struct delayed_work	idle_work;
+	struct rwlock		idle_mutex;
+	const struct firmware	*fw;	/* VCE firmware */
+	struct amdgpu_ring	ring[AMDGPU_MAX_VCE_RINGS];
+	struct amdgpu_irq_src	irq;
+	unsigned		harvest_config;
+	struct drm_sched_entity	entity;
+	uint32_t                srbm_soft_reset;
+	unsigned		num_rings;
+};
+
+int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size);
+int amdgpu_vce_sw_fini(struct amdgpu_device *adev);
+int amdgpu_vce_entity_init(struct amdgpu_device *adev);
+int amdgpu_vce_suspend(struct amdgpu_device *adev);
+int amdgpu_vce_resume(struct amdgpu_device *adev);
+int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+			      struct dma_fence **fence);
+int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+			       bool direct, struct dma_fence **fence);
+void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp);
+int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx);
+int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx);
+void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib,
+			     unsigned vmid, bool ctx_switch);
+void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				unsigned flags);
+int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring);
+int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring, long timeout);
+void amdgpu_vce_ring_begin_use(struct amdgpu_ring *ring);
+void amdgpu_vce_ring_end_use(struct amdgpu_ring *ring);
+unsigned amdgpu_vce_ring_get_emit_ib_size(struct amdgpu_ring *ring);
+unsigned amdgpu_vce_ring_get_dma_frame_size(struct amdgpu_ring *ring);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_vce_v2_0.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vce_v2_0.c
new file mode 100644
index 00000000000..94eeaafa9e5
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vce_v2_0.c
@@ -0,0 +1,656 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * Authors: Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_vce.h"
+#include "cikd.h"
+#include "vce/vce_2_0_d.h"
+#include "vce/vce_2_0_sh_mask.h"
+#include "smu/smu_7_0_1_d.h"
+#include "smu/smu_7_0_1_sh_mask.h"
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+#define VCE_V2_0_FW_SIZE	(256 * 1024)
+#define VCE_V2_0_STACK_SIZE	(64 * 1024)
+#define VCE_V2_0_DATA_SIZE	(23552 * AMDGPU_MAX_VCE_HANDLES)
+#define VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK	0x02
+
+static void vce_v2_0_set_ring_funcs(struct amdgpu_device *adev);
+static void vce_v2_0_set_irq_funcs(struct amdgpu_device *adev);
+
+/**
+ * vce_v2_0_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t vce_v2_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->me == 0)
+		return RREG32(mmVCE_RB_RPTR);
+	else
+		return RREG32(mmVCE_RB_RPTR2);
+}
+
+/**
+ * vce_v2_0_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t vce_v2_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->me == 0)
+		return RREG32(mmVCE_RB_WPTR);
+	else
+		return RREG32(mmVCE_RB_WPTR2);
+}
+
+/**
+ * vce_v2_0_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void vce_v2_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->me == 0)
+		WREG32(mmVCE_RB_WPTR, lower_32_bits(ring->wptr));
+	else
+		WREG32(mmVCE_RB_WPTR2, lower_32_bits(ring->wptr));
+}
+
+static int vce_v2_0_lmi_clean(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	for (i = 0; i < 10; ++i) {
+		for (j = 0; j < 100; ++j) {
+			uint32_t status = RREG32(mmVCE_LMI_STATUS);
+
+			if (status & 0x337f)
+				return 0;
+			mdelay(10);
+		}
+	}
+
+	return -ETIMEDOUT;
+}
+
+static int vce_v2_0_firmware_loaded(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	for (i = 0; i < 10; ++i) {
+		for (j = 0; j < 100; ++j) {
+			uint32_t status = RREG32(mmVCE_STATUS);
+
+			if (status & VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK)
+				return 0;
+			mdelay(10);
+		}
+
+		DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n");
+		WREG32_P(mmVCE_SOFT_RESET,
+			VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK,
+			~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+		mdelay(10);
+		WREG32_P(mmVCE_SOFT_RESET, 0,
+			~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+		mdelay(10);
+	}
+
+	return -ETIMEDOUT;
+}
+
+static void vce_v2_0_disable_cg(struct amdgpu_device *adev)
+{
+	WREG32(mmVCE_CGTT_CLK_OVERRIDE, 7);
+}
+
+static void vce_v2_0_init_cg(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmVCE_CLOCK_GATING_A);
+	tmp &= ~0xfff;
+	tmp |= ((0 << 0) | (4 << 4));
+	tmp |= 0x40000;
+	WREG32(mmVCE_CLOCK_GATING_A, tmp);
+
+	tmp = RREG32(mmVCE_UENC_CLOCK_GATING);
+	tmp &= ~0xfff;
+	tmp |= ((0 << 0) | (4 << 4));
+	WREG32(mmVCE_UENC_CLOCK_GATING, tmp);
+
+	tmp = RREG32(mmVCE_CLOCK_GATING_B);
+	tmp |= 0x10;
+	tmp &= ~0x100000;
+	WREG32(mmVCE_CLOCK_GATING_B, tmp);
+}
+
+static void vce_v2_0_mc_resume(struct amdgpu_device *adev)
+{
+	uint32_t size, offset;
+
+	WREG32_P(mmVCE_CLOCK_GATING_A, 0, ~(1 << 16));
+	WREG32_P(mmVCE_UENC_CLOCK_GATING, 0x1FF000, ~0xFF9FF000);
+	WREG32_P(mmVCE_UENC_REG_CLOCK_GATING, 0x3F, ~0x3F);
+	WREG32(mmVCE_CLOCK_GATING_B, 0xf7);
+
+	WREG32(mmVCE_LMI_CTRL, 0x00398000);
+	WREG32_P(mmVCE_LMI_CACHE_CTRL, 0x0, ~0x1);
+	WREG32(mmVCE_LMI_SWAP_CNTL, 0);
+	WREG32(mmVCE_LMI_SWAP_CNTL1, 0);
+	WREG32(mmVCE_LMI_VM_CTRL, 0);
+
+	WREG32(mmVCE_LMI_VCPU_CACHE_40BIT_BAR, (adev->vce.gpu_addr >> 8));
+
+	offset = AMDGPU_VCE_FIRMWARE_OFFSET;
+	size = VCE_V2_0_FW_SIZE;
+	WREG32(mmVCE_VCPU_CACHE_OFFSET0, offset & 0x7fffffff);
+	WREG32(mmVCE_VCPU_CACHE_SIZE0, size);
+
+	offset += size;
+	size = VCE_V2_0_STACK_SIZE;
+	WREG32(mmVCE_VCPU_CACHE_OFFSET1, offset & 0x7fffffff);
+	WREG32(mmVCE_VCPU_CACHE_SIZE1, size);
+
+	offset += size;
+	size = VCE_V2_0_DATA_SIZE;
+	WREG32(mmVCE_VCPU_CACHE_OFFSET2, offset & 0x7fffffff);
+	WREG32(mmVCE_VCPU_CACHE_SIZE2, size);
+
+	WREG32_P(mmVCE_LMI_CTRL2, 0x0, ~0x100);
+	WREG32_FIELD(VCE_SYS_INT_EN, VCE_SYS_INT_TRAP_INTERRUPT_EN, 1);
+}
+
+static bool vce_v2_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return !(RREG32(mmSRBM_STATUS2) & SRBM_STATUS2__VCE_BUSY_MASK);
+}
+
+static int vce_v2_0_wait_for_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	unsigned i;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (vce_v2_0_is_idle(handle))
+			return 0;
+	}
+	return -ETIMEDOUT;
+}
+
+/**
+ * vce_v2_0_start - start VCE block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the VCE block
+ */
+static int vce_v2_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	int r;
+
+	/* set BUSY flag */
+	WREG32_P(mmVCE_STATUS, 1, ~1);
+
+	vce_v2_0_init_cg(adev);
+	vce_v2_0_disable_cg(adev);
+
+	vce_v2_0_mc_resume(adev);
+
+	ring = &adev->vce.ring[0];
+	WREG32(mmVCE_RB_RPTR, lower_32_bits(ring->wptr));
+	WREG32(mmVCE_RB_WPTR, lower_32_bits(ring->wptr));
+	WREG32(mmVCE_RB_BASE_LO, ring->gpu_addr);
+	WREG32(mmVCE_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32(mmVCE_RB_SIZE, ring->ring_size / 4);
+
+	ring = &adev->vce.ring[1];
+	WREG32(mmVCE_RB_RPTR2, lower_32_bits(ring->wptr));
+	WREG32(mmVCE_RB_WPTR2, lower_32_bits(ring->wptr));
+	WREG32(mmVCE_RB_BASE_LO2, ring->gpu_addr);
+	WREG32(mmVCE_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+	WREG32(mmVCE_RB_SIZE2, ring->ring_size / 4);
+
+	WREG32_FIELD(VCE_VCPU_CNTL, CLK_EN, 1);
+	WREG32_FIELD(VCE_SOFT_RESET, ECPU_SOFT_RESET, 1);
+	mdelay(100);
+	WREG32_FIELD(VCE_SOFT_RESET, ECPU_SOFT_RESET, 0);
+
+	r = vce_v2_0_firmware_loaded(adev);
+
+	/* clear BUSY flag */
+	WREG32_P(mmVCE_STATUS, 0, ~1);
+
+	if (r) {
+		DRM_ERROR("VCE not responding, giving up!!!\n");
+		return r;
+	}
+
+	return 0;
+}
+
+static int vce_v2_0_stop(struct amdgpu_device *adev)
+{
+	int i;
+	int status;
+
+	if (vce_v2_0_lmi_clean(adev)) {
+		DRM_INFO("vce is not idle \n");
+		return 0;
+	}
+
+	if (vce_v2_0_wait_for_idle(adev)) {
+		DRM_INFO("VCE is busy, Can't set clock gateing");
+		return 0;
+	}
+
+	/* Stall UMC and register bus before resetting VCPU */
+	WREG32_P(mmVCE_LMI_CTRL2, 1 << 8, ~(1 << 8));
+
+	for (i = 0; i < 100; ++i) {
+		status = RREG32(mmVCE_LMI_STATUS);
+		if (status & 0x240)
+			break;
+		mdelay(1);
+	}
+
+	WREG32_P(mmVCE_VCPU_CNTL, 0, ~0x80001);
+
+	/* put LMI, VCPU, RBC etc... into reset */
+	WREG32_P(mmVCE_SOFT_RESET, 1, ~0x1);
+
+	WREG32(mmVCE_STATUS, 0);
+
+	return 0;
+}
+
+static void vce_v2_0_set_sw_cg(struct amdgpu_device *adev, bool gated)
+{
+	u32 tmp;
+
+	if (gated) {
+		tmp = RREG32(mmVCE_CLOCK_GATING_B);
+		tmp |= 0xe70000;
+		WREG32(mmVCE_CLOCK_GATING_B, tmp);
+
+		tmp = RREG32(mmVCE_UENC_CLOCK_GATING);
+		tmp |= 0xff000000;
+		WREG32(mmVCE_UENC_CLOCK_GATING, tmp);
+
+		tmp = RREG32(mmVCE_UENC_REG_CLOCK_GATING);
+		tmp &= ~0x3fc;
+		WREG32(mmVCE_UENC_REG_CLOCK_GATING, tmp);
+
+		WREG32(mmVCE_CGTT_CLK_OVERRIDE, 0);
+	} else {
+		tmp = RREG32(mmVCE_CLOCK_GATING_B);
+		tmp |= 0xe7;
+		tmp &= ~0xe70000;
+		WREG32(mmVCE_CLOCK_GATING_B, tmp);
+
+		tmp = RREG32(mmVCE_UENC_CLOCK_GATING);
+		tmp |= 0x1fe000;
+		tmp &= ~0xff000000;
+		WREG32(mmVCE_UENC_CLOCK_GATING, tmp);
+
+		tmp = RREG32(mmVCE_UENC_REG_CLOCK_GATING);
+		tmp |= 0x3fc;
+		WREG32(mmVCE_UENC_REG_CLOCK_GATING, tmp);
+	}
+}
+
+static void vce_v2_0_set_dyn_cg(struct amdgpu_device *adev, bool gated)
+{
+	u32 orig, tmp;
+
+/* LMI_MC/LMI_UMC always set in dynamic,
+ * set {CGC_*_GATE_MODE, CGC_*_SW_GATE} = {0, 0}
+ */
+	tmp = RREG32(mmVCE_CLOCK_GATING_B);
+	tmp &= ~0x00060006;
+
+/* Exception for ECPU, IH, SEM, SYS blocks needs to be turned on/off by SW */
+	if (gated) {
+		tmp |= 0xe10000;
+		WREG32(mmVCE_CLOCK_GATING_B, tmp);
+	} else {
+		tmp |= 0xe1;
+		tmp &= ~0xe10000;
+		WREG32(mmVCE_CLOCK_GATING_B, tmp);
+	}
+
+	orig = tmp = RREG32(mmVCE_UENC_CLOCK_GATING);
+	tmp &= ~0x1fe000;
+	tmp &= ~0xff000000;
+	if (tmp != orig)
+		WREG32(mmVCE_UENC_CLOCK_GATING, tmp);
+
+	orig = tmp = RREG32(mmVCE_UENC_REG_CLOCK_GATING);
+	tmp &= ~0x3fc;
+	if (tmp != orig)
+		WREG32(mmVCE_UENC_REG_CLOCK_GATING, tmp);
+
+	/* set VCE_UENC_REG_CLOCK_GATING always in dynamic mode */
+	WREG32(mmVCE_UENC_REG_CLOCK_GATING, 0x00);
+
+	if(gated)
+		WREG32(mmVCE_CGTT_CLK_OVERRIDE, 0);
+}
+
+static void vce_v2_0_enable_mgcg(struct amdgpu_device *adev, bool enable,
+								bool sw_cg)
+{
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG)) {
+		if (sw_cg)
+			vce_v2_0_set_sw_cg(adev, true);
+		else
+			vce_v2_0_set_dyn_cg(adev, true);
+	} else {
+		vce_v2_0_disable_cg(adev);
+
+		if (sw_cg)
+			vce_v2_0_set_sw_cg(adev, false);
+		else
+			vce_v2_0_set_dyn_cg(adev, false);
+	}
+}
+
+static int vce_v2_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->vce.num_rings = 2;
+
+	vce_v2_0_set_ring_funcs(adev);
+	vce_v2_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int vce_v2_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* VCE */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 167, &adev->vce.irq);
+	if (r)
+		return r;
+
+	r = amdgpu_vce_sw_init(adev, VCE_V2_0_FW_SIZE +
+		VCE_V2_0_STACK_SIZE + VCE_V2_0_DATA_SIZE);
+	if (r)
+		return r;
+
+	r = amdgpu_vce_resume(adev);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->vce.num_rings; i++) {
+		ring = &adev->vce.ring[i];
+		snprintf(ring->name, sizeof(ring->name), "vce%d", i);
+		r = amdgpu_ring_init(adev, ring, 512,
+				     &adev->vce.irq, 0);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_vce_entity_init(adev);
+
+	return r;
+}
+
+static int vce_v2_0_sw_fini(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_vce_suspend(adev);
+	if (r)
+		return r;
+
+	return amdgpu_vce_sw_fini(adev);
+}
+
+static int vce_v2_0_hw_init(void *handle)
+{
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_asic_set_vce_clocks(adev, 10000, 10000);
+	vce_v2_0_enable_mgcg(adev, true, false);
+	for (i = 0; i < adev->vce.num_rings; i++)
+		adev->vce.ring[i].ready = false;
+
+	for (i = 0; i < adev->vce.num_rings; i++) {
+		r = amdgpu_ring_test_ring(&adev->vce.ring[i]);
+		if (r)
+			return r;
+		else
+			adev->vce.ring[i].ready = true;
+	}
+
+	DRM_INFO("VCE initialized successfully.\n");
+
+	return 0;
+}
+
+static int vce_v2_0_hw_fini(void *handle)
+{
+	return 0;
+}
+
+static int vce_v2_0_suspend(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = vce_v2_0_hw_fini(adev);
+	if (r)
+		return r;
+
+	return amdgpu_vce_suspend(adev);
+}
+
+static int vce_v2_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_vce_resume(adev);
+	if (r)
+		return r;
+
+	return vce_v2_0_hw_init(adev);
+}
+
+static int vce_v2_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	WREG32_FIELD(SRBM_SOFT_RESET, SOFT_RESET_VCE, 1);
+	mdelay(5);
+
+	return vce_v2_0_start(adev);
+}
+
+static int vce_v2_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	uint32_t val = 0;
+
+	if (state == AMDGPU_IRQ_STATE_ENABLE)
+		val |= VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK;
+
+	WREG32_P(mmVCE_SYS_INT_EN, val, ~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
+	return 0;
+}
+
+static int vce_v2_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: VCE\n");
+	switch (entry->src_data[0]) {
+	case 0:
+	case 1:
+		amdgpu_fence_process(&adev->vce.ring[entry->src_data[0]]);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int vce_v2_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	bool gate = false;
+	bool sw_cg = false;
+
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_CG_STATE_GATE) {
+		gate = true;
+		sw_cg = true;
+	}
+
+	vce_v2_0_enable_mgcg(adev, gate, sw_cg);
+
+	return 0;
+}
+
+static int vce_v2_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the VCE block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_PG_STATE_GATE)
+		return vce_v2_0_stop(adev);
+	else
+		return vce_v2_0_start(adev);
+}
+
+static const struct amd_ip_funcs vce_v2_0_ip_funcs = {
+	.name = "vce_v2_0",
+	.early_init = vce_v2_0_early_init,
+	.late_init = NULL,
+	.sw_init = vce_v2_0_sw_init,
+	.sw_fini = vce_v2_0_sw_fini,
+	.hw_init = vce_v2_0_hw_init,
+	.hw_fini = vce_v2_0_hw_fini,
+	.suspend = vce_v2_0_suspend,
+	.resume = vce_v2_0_resume,
+	.is_idle = vce_v2_0_is_idle,
+	.wait_for_idle = vce_v2_0_wait_for_idle,
+	.soft_reset = vce_v2_0_soft_reset,
+	.set_clockgating_state = vce_v2_0_set_clockgating_state,
+	.set_powergating_state = vce_v2_0_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs vce_v2_0_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_VCE,
+	.align_mask = 0xf,
+	.nop = VCE_CMD_NO_OP,
+	.support_64bit_ptrs = false,
+	.get_rptr = vce_v2_0_ring_get_rptr,
+	.get_wptr = vce_v2_0_ring_get_wptr,
+	.set_wptr = vce_v2_0_ring_set_wptr,
+	.parse_cs = amdgpu_vce_ring_parse_cs,
+	.emit_frame_size = 6, /* amdgpu_vce_ring_emit_fence  x1 no user fence */
+	.emit_ib_size = 4, /* amdgpu_vce_ring_emit_ib */
+	.emit_ib = amdgpu_vce_ring_emit_ib,
+	.emit_fence = amdgpu_vce_ring_emit_fence,
+	.test_ring = amdgpu_vce_ring_test_ring,
+	.test_ib = amdgpu_vce_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vce_ring_begin_use,
+	.end_use = amdgpu_vce_ring_end_use,
+};
+
+static void vce_v2_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vce.num_rings; i++) {
+		adev->vce.ring[i].funcs = &vce_v2_0_ring_funcs;
+		adev->vce.ring[i].me = i;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs vce_v2_0_irq_funcs = {
+	.set = vce_v2_0_set_interrupt_state,
+	.process = vce_v2_0_process_interrupt,
+};
+
+static void vce_v2_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->vce.irq.num_types = 1;
+	adev->vce.irq.funcs = &vce_v2_0_irq_funcs;
+};
+
+const struct amdgpu_ip_block_version vce_v2_0_ip_block =
+{
+		.type = AMD_IP_BLOCK_TYPE_VCE,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vce_v2_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_vcn.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vcn.c
new file mode 100644
index 00000000000..7ea19217a00
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vcn.c
@@ -0,0 +1,723 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <drm/drmP.h>
+#include <drm/drm.h>
+
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_vcn.h"
+#include "soc15d.h"
+#include "soc15_common.h"
+
+#include "vcn/vcn_1_0_offset.h"
+
+/* 1 second timeout */
+#define VCN_IDLE_TIMEOUT	msecs_to_jiffies(1000)
+
+/* Firmware Names */
+#define FIRMWARE_RAVEN		"amdgpu/raven_vcn.bin"
+
+MODULE_FIRMWARE(FIRMWARE_RAVEN);
+
+static void amdgpu_vcn_idle_work_handler(struct work_struct *work);
+
+int amdgpu_vcn_sw_init(struct amdgpu_device *adev)
+{
+	unsigned long bo_size;
+	const char *fw_name;
+	const struct common_firmware_header *hdr;
+	unsigned char fw_check;
+	int r;
+
+	INIT_DELAYED_WORK(&adev->vcn.idle_work, amdgpu_vcn_idle_work_handler);
+
+	switch (adev->asic_type) {
+	case CHIP_RAVEN:
+		fw_name = FIRMWARE_RAVEN;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	r = request_firmware(&adev->vcn.fw, fw_name, adev->dev);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_vcn: Can't load firmware \"%s\"\n",
+			fw_name);
+		return r;
+	}
+
+	r = amdgpu_ucode_validate(adev->vcn.fw);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_vcn: Can't validate firmware \"%s\"\n",
+			fw_name);
+		release_firmware(adev->vcn.fw);
+		adev->vcn.fw = NULL;
+		return r;
+	}
+
+	hdr = (const struct common_firmware_header *)adev->vcn.fw->data;
+	adev->vcn.fw_version = le32_to_cpu(hdr->ucode_version);
+
+	/* Bit 20-23, it is encode major and non-zero for new naming convention.
+	 * This field is part of version minor and DRM_DISABLED_FLAG in old naming
+	 * convention. Since the l:wq!atest version minor is 0x5B and DRM_DISABLED_FLAG
+	 * is zero in old naming convention, this field is always zero so far.
+	 * These four bits are used to tell which naming convention is present.
+	 */
+	fw_check = (le32_to_cpu(hdr->ucode_version) >> 20) & 0xf;
+	if (fw_check) {
+		unsigned int dec_ver, enc_major, enc_minor, vep, fw_rev;
+
+		fw_rev = le32_to_cpu(hdr->ucode_version) & 0xfff;
+		enc_minor = (le32_to_cpu(hdr->ucode_version) >> 12) & 0xff;
+		enc_major = fw_check;
+		dec_ver = (le32_to_cpu(hdr->ucode_version) >> 24) & 0xf;
+		vep = (le32_to_cpu(hdr->ucode_version) >> 28) & 0xf;
+		DRM_INFO("Found VCN firmware Version ENC: %u.%u DEC: %u VEP: %u Revision: %u\n",
+			enc_major, enc_minor, dec_ver, vep, fw_rev);
+	} else {
+		unsigned int version_major, version_minor, family_id;
+
+		family_id = le32_to_cpu(hdr->ucode_version) & 0xff;
+		version_major = (le32_to_cpu(hdr->ucode_version) >> 24) & 0xff;
+		version_minor = (le32_to_cpu(hdr->ucode_version) >> 8) & 0xff;
+		DRM_INFO("Found VCN firmware Version: %u.%u Family ID: %u\n",
+			version_major, version_minor, family_id);
+	}
+
+	bo_size = AMDGPU_VCN_STACK_SIZE + AMDGPU_VCN_HEAP_SIZE
+		  +  AMDGPU_VCN_SESSION_SIZE * 40;
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		bo_size += AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+	r = amdgpu_bo_create_kernel(adev, bo_size, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_VRAM, &adev->vcn.vcpu_bo,
+				    &adev->vcn.gpu_addr, &adev->vcn.cpu_addr);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate vcn bo\n", r);
+		return r;
+	}
+
+	return 0;
+}
+
+int amdgpu_vcn_sw_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	kvfree(adev->vcn.saved_bo);
+
+	amdgpu_bo_free_kernel(&adev->vcn.vcpu_bo,
+			      &adev->vcn.gpu_addr,
+			      (void **)&adev->vcn.cpu_addr);
+
+	amdgpu_ring_fini(&adev->vcn.ring_dec);
+
+	for (i = 0; i < adev->vcn.num_enc_rings; ++i)
+		amdgpu_ring_fini(&adev->vcn.ring_enc[i]);
+
+	amdgpu_ring_fini(&adev->vcn.ring_jpeg);
+
+	release_firmware(adev->vcn.fw);
+
+	return 0;
+}
+
+int amdgpu_vcn_suspend(struct amdgpu_device *adev)
+{
+	unsigned size;
+	void *ptr;
+
+	cancel_delayed_work_sync(&adev->vcn.idle_work);
+
+	if (adev->vcn.vcpu_bo == NULL)
+		return 0;
+
+	size = amdgpu_bo_size(adev->vcn.vcpu_bo);
+	ptr = adev->vcn.cpu_addr;
+
+	adev->vcn.saved_bo = kvmalloc(size, GFP_KERNEL);
+	if (!adev->vcn.saved_bo)
+		return -ENOMEM;
+
+	memcpy_fromio(adev->vcn.saved_bo, ptr, size);
+
+	return 0;
+}
+
+int amdgpu_vcn_resume(struct amdgpu_device *adev)
+{
+	unsigned size;
+	void *ptr;
+
+	if (adev->vcn.vcpu_bo == NULL)
+		return -EINVAL;
+
+	size = amdgpu_bo_size(adev->vcn.vcpu_bo);
+	ptr = adev->vcn.cpu_addr;
+
+	if (adev->vcn.saved_bo != NULL) {
+		memcpy_toio(ptr, adev->vcn.saved_bo, size);
+		kvfree(adev->vcn.saved_bo);
+		adev->vcn.saved_bo = NULL;
+	} else {
+		const struct common_firmware_header *hdr;
+		unsigned offset;
+
+		hdr = (const struct common_firmware_header *)adev->vcn.fw->data;
+		if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+			offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
+			memcpy_toio(adev->vcn.cpu_addr, adev->vcn.fw->data + offset,
+				    le32_to_cpu(hdr->ucode_size_bytes));
+			size -= le32_to_cpu(hdr->ucode_size_bytes);
+			ptr += le32_to_cpu(hdr->ucode_size_bytes);
+		}
+		memset_io(ptr, 0, size);
+	}
+
+	return 0;
+}
+
+static void amdgpu_vcn_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, vcn.idle_work.work);
+	unsigned fences = amdgpu_fence_count_emitted(&adev->vcn.ring_dec);
+	unsigned i;
+
+	for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
+		fences += amdgpu_fence_count_emitted(&adev->vcn.ring_enc[i]);
+	}
+
+	fences += amdgpu_fence_count_emitted(&adev->vcn.ring_jpeg);
+
+	if (fences == 0) {
+		if (adev->pm.dpm_enabled)
+			amdgpu_dpm_enable_uvd(adev, false);
+		else
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
+							       AMD_PG_STATE_GATE);
+	} else {
+		schedule_delayed_work(&adev->vcn.idle_work, VCN_IDLE_TIMEOUT);
+	}
+}
+
+void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	bool set_clocks = !cancel_delayed_work_sync(&adev->vcn.idle_work);
+
+	if (set_clocks) {
+		if (adev->pm.dpm_enabled)
+			amdgpu_dpm_enable_uvd(adev, true);
+		else
+			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
+							       AMD_PG_STATE_UNGATE);
+	}
+}
+
+void amdgpu_vcn_ring_end_use(struct amdgpu_ring *ring)
+{
+	schedule_delayed_work(&ring->adev->vcn.idle_work, VCN_IDLE_TIMEOUT);
+}
+
+int amdgpu_vcn_dec_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID));
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	return r;
+}
+
+static int amdgpu_vcn_dec_send_msg(struct amdgpu_ring *ring,
+				   struct amdgpu_bo *bo,
+				   struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct dma_fence *f = NULL;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	uint64_t addr;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(adev, 64, &job);
+	if (r)
+		goto err;
+
+	ib = &job->ibs[0];
+	addr = amdgpu_bo_gpu_offset(bo);
+	ib->ptr[0] = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0);
+	ib->ptr[1] = addr;
+	ib->ptr[2] = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0);
+	ib->ptr[3] = addr >> 32;
+	ib->ptr[4] = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0);
+	ib->ptr[5] = 0;
+	for (i = 6; i < 16; i += 2) {
+		ib->ptr[i] = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP), 0);
+		ib->ptr[i+1] = 0;
+	}
+	ib->length_dw = 16;
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err_free;
+
+	amdgpu_bo_fence(bo, f, false);
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&bo);
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+
+	return 0;
+
+err_free:
+	amdgpu_job_free(job);
+
+err:
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&bo);
+	return r;
+}
+
+static int amdgpu_vcn_dec_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+			      struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_bo *bo = NULL;
+	uint32_t *msg;
+	int r, i;
+
+	r = amdgpu_bo_create_reserved(adev, 1024, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM,
+				      &bo, NULL, (void **)&msg);
+	if (r)
+		return r;
+
+	msg[0] = cpu_to_le32(0x00000028);
+	msg[1] = cpu_to_le32(0x00000038);
+	msg[2] = cpu_to_le32(0x00000001);
+	msg[3] = cpu_to_le32(0x00000000);
+	msg[4] = cpu_to_le32(handle);
+	msg[5] = cpu_to_le32(0x00000000);
+	msg[6] = cpu_to_le32(0x00000001);
+	msg[7] = cpu_to_le32(0x00000028);
+	msg[8] = cpu_to_le32(0x00000010);
+	msg[9] = cpu_to_le32(0x00000000);
+	msg[10] = cpu_to_le32(0x00000007);
+	msg[11] = cpu_to_le32(0x00000000);
+	msg[12] = cpu_to_le32(0x00000780);
+	msg[13] = cpu_to_le32(0x00000440);
+	for (i = 14; i < 1024; ++i)
+		msg[i] = cpu_to_le32(0x0);
+
+	return amdgpu_vcn_dec_send_msg(ring, bo, fence);
+}
+
+static int amdgpu_vcn_dec_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+			       struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_bo *bo = NULL;
+	uint32_t *msg;
+	int r, i;
+
+	r = amdgpu_bo_create_reserved(adev, 1024, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM,
+				      &bo, NULL, (void **)&msg);
+	if (r)
+		return r;
+
+	msg[0] = cpu_to_le32(0x00000028);
+	msg[1] = cpu_to_le32(0x00000018);
+	msg[2] = cpu_to_le32(0x00000000);
+	msg[3] = cpu_to_le32(0x00000002);
+	msg[4] = cpu_to_le32(handle);
+	msg[5] = cpu_to_le32(0x00000000);
+	for (i = 6; i < 1024; ++i)
+		msg[i] = cpu_to_le32(0x0);
+
+	return amdgpu_vcn_dec_send_msg(ring, bo, fence);
+}
+
+int amdgpu_vcn_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct dma_fence *fence;
+	long r;
+
+	r = amdgpu_vcn_dec_get_create_msg(ring, 1, NULL);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
+		goto error;
+	}
+
+	r = amdgpu_vcn_dec_get_destroy_msg(ring, 1, &fence);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
+		goto error;
+	}
+
+	r = dma_fence_wait_timeout(fence, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out.\n");
+		r = -ETIMEDOUT;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+	} else {
+		DRM_DEBUG("ib test on ring %d succeeded\n",  ring->idx);
+		r = 0;
+	}
+
+	dma_fence_put(fence);
+
+error:
+	return r;
+}
+
+int amdgpu_vcn_enc_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t rptr = amdgpu_ring_get_rptr(ring);
+	unsigned i;
+	int r;
+
+	r = amdgpu_ring_alloc(ring, 16);
+	if (r) {
+		DRM_ERROR("amdgpu: vcn enc failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring, VCN_ENC_CMD_END);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (amdgpu_ring_get_rptr(ring) != rptr)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed\n",
+			  ring->idx);
+		r = -ETIMEDOUT;
+	}
+
+	return r;
+}
+
+static int amdgpu_vcn_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+			      struct dma_fence **fence)
+{
+	const unsigned ib_size_dw = 16;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct dma_fence *f = NULL;
+	uint64_t dummy;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	if (r)
+		return r;
+
+	ib = &job->ibs[0];
+	dummy = ib->gpu_addr + 1024;
+
+	ib->length_dw = 0;
+	ib->ptr[ib->length_dw++] = 0x00000018;
+	ib->ptr[ib->length_dw++] = 0x00000001; /* session info */
+	ib->ptr[ib->length_dw++] = handle;
+	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
+	ib->ptr[ib->length_dw++] = dummy;
+	ib->ptr[ib->length_dw++] = 0x0000000b;
+
+	ib->ptr[ib->length_dw++] = 0x00000014;
+	ib->ptr[ib->length_dw++] = 0x00000002; /* task info */
+	ib->ptr[ib->length_dw++] = 0x0000001c;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+
+	ib->ptr[ib->length_dw++] = 0x00000008;
+	ib->ptr[ib->length_dw++] = 0x08000001; /* op initialize */
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err;
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+
+	return 0;
+
+err:
+	amdgpu_job_free(job);
+	return r;
+}
+
+static int amdgpu_vcn_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+				struct dma_fence **fence)
+{
+	const unsigned ib_size_dw = 16;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct dma_fence *f = NULL;
+	uint64_t dummy;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	if (r)
+		return r;
+
+	ib = &job->ibs[0];
+	dummy = ib->gpu_addr + 1024;
+
+	ib->length_dw = 0;
+	ib->ptr[ib->length_dw++] = 0x00000018;
+	ib->ptr[ib->length_dw++] = 0x00000001;
+	ib->ptr[ib->length_dw++] = handle;
+	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
+	ib->ptr[ib->length_dw++] = dummy;
+	ib->ptr[ib->length_dw++] = 0x0000000b;
+
+	ib->ptr[ib->length_dw++] = 0x00000014;
+	ib->ptr[ib->length_dw++] = 0x00000002;
+	ib->ptr[ib->length_dw++] = 0x0000001c;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+
+	ib->ptr[ib->length_dw++] = 0x00000008;
+	ib->ptr[ib->length_dw++] = 0x08000002; /* op close session */
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err;
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+
+	return 0;
+
+err:
+	amdgpu_job_free(job);
+	return r;
+}
+
+int amdgpu_vcn_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct dma_fence *fence = NULL;
+	long r;
+
+	r = amdgpu_vcn_enc_get_create_msg(ring, 1, NULL);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
+		goto error;
+	}
+
+	r = amdgpu_vcn_enc_get_destroy_msg(ring, 1, &fence);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
+		goto error;
+	}
+
+	r = dma_fence_wait_timeout(fence, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out.\n");
+		r = -ETIMEDOUT;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+	} else {
+		DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
+	}
+error:
+	dma_fence_put(fence);
+	return r;
+}
+
+int amdgpu_vcn_jpeg_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 3);
+
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+				  ring->idx, r);
+		return r;
+	}
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0, 0, 0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID));
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n",
+				  ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+				  ring->idx, tmp);
+		r = -EINVAL;
+	}
+
+	return r;
+}
+
+static int amdgpu_vcn_jpeg_set_reg(struct amdgpu_ring *ring, uint32_t handle,
+		struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct dma_fence *f = NULL;
+	const unsigned ib_size_dw = 16;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	if (r)
+		return r;
+
+	ib = &job->ibs[0];
+
+	ib->ptr[0] = PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_PITCH), 0, 0, PACKETJ_TYPE0);
+	ib->ptr[1] = 0xDEADBEEF;
+	for (i = 2; i < 16; i += 2) {
+		ib->ptr[i] = PACKETJ(0, 0, 0, PACKETJ_TYPE6);
+		ib->ptr[i+1] = 0;
+	}
+	ib->length_dw = 16;
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err;
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+
+	return 0;
+
+err:
+	amdgpu_job_free(job);
+	return r;
+}
+
+int amdgpu_vcn_jpeg_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	struct dma_fence *fence = NULL;
+	long r = 0;
+
+	r = amdgpu_vcn_jpeg_set_reg(ring, 1, &fence);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to set jpeg register (%ld).\n", r);
+		goto error;
+	}
+
+	r = dma_fence_wait_timeout(fence, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out.\n");
+		r = -ETIMEDOUT;
+		goto error;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto error;
+	} else
+		r = 0;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_PITCH));
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout)
+		DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+	else {
+		DRM_ERROR("ib test failed (0x%08X)\n", tmp);
+		r = -EINVAL;
+	}
+
+	dma_fence_put(fence);
+
+error:
+	return r;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_vcn.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vcn.h
new file mode 100644
index 00000000000..0b0b8638d73
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vcn.h
@@ -0,0 +1,90 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_VCN_H__
+#define __AMDGPU_VCN_H__
+
+#define AMDGPU_VCN_STACK_SIZE		(200*1024)
+#define AMDGPU_VCN_HEAP_SIZE		(256*1024)
+#define AMDGPU_VCN_SESSION_SIZE		(50*1024)
+#define AMDGPU_VCN_FIRMWARE_OFFSET	256
+#define AMDGPU_VCN_MAX_ENC_RINGS	3
+
+#define VCN_DEC_CMD_FENCE		0x00000000
+#define VCN_DEC_CMD_TRAP		0x00000001
+#define VCN_DEC_CMD_WRITE_REG		0x00000004
+#define VCN_DEC_CMD_REG_READ_COND_WAIT	0x00000006
+#define VCN_DEC_CMD_PACKET_START	0x0000000a
+#define VCN_DEC_CMD_PACKET_END		0x0000000b
+
+#define VCN_ENC_CMD_NO_OP		0x00000000
+#define VCN_ENC_CMD_END 		0x00000001
+#define VCN_ENC_CMD_IB			0x00000002
+#define VCN_ENC_CMD_FENCE		0x00000003
+#define VCN_ENC_CMD_TRAP		0x00000004
+#define VCN_ENC_CMD_REG_WRITE		0x0000000b
+#define VCN_ENC_CMD_REG_WAIT		0x0000000c
+
+enum engine_status_constants {
+	UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON = 0x2AAAA0,
+	UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON = 0x00000002,
+	UVD_STATUS__UVD_BUSY = 0x00000004,
+	GB_ADDR_CONFIG_DEFAULT = 0x26010011,
+	UVD_STATUS__IDLE = 0x2,
+	UVD_STATUS__BUSY = 0x5,
+	UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF = 0x1,
+	UVD_STATUS__RBC_BUSY = 0x1,
+};
+
+struct amdgpu_vcn {
+	struct amdgpu_bo	*vcpu_bo;
+	void			*cpu_addr;
+	uint64_t		gpu_addr;
+	unsigned		fw_version;
+	void			*saved_bo;
+	struct delayed_work	idle_work;
+	const struct firmware	*fw;	/* VCN firmware */
+	struct amdgpu_ring	ring_dec;
+	struct amdgpu_ring	ring_enc[AMDGPU_VCN_MAX_ENC_RINGS];
+	struct amdgpu_ring	ring_jpeg;
+	struct amdgpu_irq_src	irq;
+	unsigned		num_enc_rings;
+};
+
+int amdgpu_vcn_sw_init(struct amdgpu_device *adev);
+int amdgpu_vcn_sw_fini(struct amdgpu_device *adev);
+int amdgpu_vcn_suspend(struct amdgpu_device *adev);
+int amdgpu_vcn_resume(struct amdgpu_device *adev);
+void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring);
+void amdgpu_vcn_ring_end_use(struct amdgpu_ring *ring);
+
+int amdgpu_vcn_dec_ring_test_ring(struct amdgpu_ring *ring);
+int amdgpu_vcn_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout);
+
+int amdgpu_vcn_enc_ring_test_ring(struct amdgpu_ring *ring);
+int amdgpu_vcn_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout);
+
+int amdgpu_vcn_jpeg_ring_test_ring(struct amdgpu_ring *ring);
+int amdgpu_vcn_jpeg_ring_test_ib(struct amdgpu_ring *ring, long timeout);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_vf_error.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vf_error.c
new file mode 100644
index 00000000000..7f7097931c6
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vf_error.c
@@ -0,0 +1,86 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_vf_error.h"
+#include "mxgpu_ai.h"
+
+void amdgpu_vf_error_put(struct amdgpu_device *adev,
+			 uint16_t sub_error_code,
+			 uint16_t error_flags,
+			 uint64_t error_data)
+{
+	int index;
+	uint16_t error_code;
+
+	if (!amdgpu_sriov_vf(adev))
+		return;
+
+	error_code = AMDGIM_ERROR_CODE(AMDGIM_ERROR_CATEGORY_VF, sub_error_code);
+
+	mutex_lock(&adev->virt.vf_errors.lock);
+	index = adev->virt.vf_errors.write_count % AMDGPU_VF_ERROR_ENTRY_SIZE;
+	adev->virt.vf_errors.code [index] = error_code;
+	adev->virt.vf_errors.flags [index] = error_flags;
+	adev->virt.vf_errors.data [index] = error_data;
+	adev->virt.vf_errors.write_count ++;
+	mutex_unlock(&adev->virt.vf_errors.lock);
+}
+
+
+void amdgpu_vf_error_trans_all(struct amdgpu_device *adev)
+{
+	/* u32 pf2vf_flags = 0; */
+	u32 data1, data2, data3;
+	int index;
+
+	if ((NULL == adev) || (!amdgpu_sriov_vf(adev)) ||
+	    (!adev->virt.ops) || (!adev->virt.ops->trans_msg)) {
+		return;
+	}
+/*
+ 	TODO: Enable these code when pv2vf_info is merged
+	AMDGPU_FW_VRAM_PF2VF_READ (adev, feature_flags, &pf2vf_flags);
+	if (!(pf2vf_flags & AMDGIM_FEATURE_ERROR_LOG_COLLECT)) {
+		return;
+	}
+*/
+
+	mutex_lock(&adev->virt.vf_errors.lock);
+	/* The errors are overlay of array, correct read_count as full. */
+	if (adev->virt.vf_errors.write_count - adev->virt.vf_errors.read_count > AMDGPU_VF_ERROR_ENTRY_SIZE) {
+		adev->virt.vf_errors.read_count = adev->virt.vf_errors.write_count - AMDGPU_VF_ERROR_ENTRY_SIZE;
+	}
+
+	while (adev->virt.vf_errors.read_count < adev->virt.vf_errors.write_count) {
+		index =adev->virt.vf_errors.read_count % AMDGPU_VF_ERROR_ENTRY_SIZE;
+		data1 = AMDGIM_ERROR_CODE_FLAGS_TO_MAILBOX(adev->virt.vf_errors.code[index],
+							   adev->virt.vf_errors.flags[index]);
+		data2 = adev->virt.vf_errors.data[index] & 0xFFFFFFFF;
+		data3 = (adev->virt.vf_errors.data[index] >> 32) & 0xFFFFFFFF;
+
+		adev->virt.ops->trans_msg(adev, IDH_LOG_VF_ERROR, data1, data2, data3);
+		adev->virt.vf_errors.read_count ++;
+	}
+	mutex_unlock(&adev->virt.vf_errors.lock);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_vf_error.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vf_error.h
new file mode 100644
index 00000000000..6436bd05332
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vf_error.h
@@ -0,0 +1,65 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VF_ERROR_H__
+#define __VF_ERROR_H__
+
+#define AMDGIM_ERROR_CODE_FLAGS_TO_MAILBOX(c,f)    (((c & 0xFFFF) << 16) | (f & 0xFFFF))
+#define AMDGIM_ERROR_CODE(t,c)       (((t&0xF)<<12)|(c&0xFFF))
+
+/* Please keep enum same as AMD GIM driver */
+enum AMDGIM_ERROR_VF {
+	AMDGIM_ERROR_VF_ATOMBIOS_INIT_FAIL = 0,
+	AMDGIM_ERROR_VF_NO_VBIOS,
+	AMDGIM_ERROR_VF_GPU_POST_ERROR,
+	AMDGIM_ERROR_VF_ATOMBIOS_GET_CLOCK_FAIL,
+	AMDGIM_ERROR_VF_FENCE_INIT_FAIL,
+
+	AMDGIM_ERROR_VF_AMDGPU_INIT_FAIL,
+	AMDGIM_ERROR_VF_IB_INIT_FAIL,
+	AMDGIM_ERROR_VF_AMDGPU_LATE_INIT_FAIL,
+	AMDGIM_ERROR_VF_ASIC_RESUME_FAIL,
+	AMDGIM_ERROR_VF_GPU_RESET_FAIL,
+
+	AMDGIM_ERROR_VF_TEST,
+	AMDGIM_ERROR_VF_MAX
+};
+
+enum AMDGIM_ERROR_CATEGORY {
+	AMDGIM_ERROR_CATEGORY_NON_USED = 0,
+	AMDGIM_ERROR_CATEGORY_GIM,
+	AMDGIM_ERROR_CATEGORY_PF,
+	AMDGIM_ERROR_CATEGORY_VF,
+	AMDGIM_ERROR_CATEGORY_VBIOS,
+	AMDGIM_ERROR_CATEGORY_MONITOR,
+
+	AMDGIM_ERROR_CATEGORY_MAX
+};
+
+void amdgpu_vf_error_put(struct amdgpu_device *adev,
+			 uint16_t sub_error_code,
+			 uint16_t error_flags,
+			 uint64_t error_data);
+void amdgpu_vf_error_trans_all (struct amdgpu_device *adev);
+
+#endif /* __VF_ERROR_H__ */
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_virt.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_virt.c
new file mode 100644
index 00000000000..329da412895
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_virt.c
@@ -0,0 +1,439 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#define MAX_KIQ_REG_WAIT	5000 /* in usecs, 5ms */
+#define MAX_KIQ_REG_BAILOUT_INTERVAL	5 /* in msecs, 5ms */
+#define MAX_KIQ_REG_TRY 20
+
+uint64_t amdgpu_csa_vaddr(struct amdgpu_device *adev)
+{
+	uint64_t addr = adev->vm_manager.max_pfn << AMDGPU_GPU_PAGE_SHIFT;
+
+	addr -= AMDGPU_VA_RESERVED_SIZE;
+
+	if (addr >= AMDGPU_VA_HOLE_START)
+		addr |= AMDGPU_VA_HOLE_END;
+
+	return addr;
+}
+
+bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev)
+{
+	/* By now all MMIO pages except mailbox are blocked */
+	/* if blocking is enabled in hypervisor. Choose the */
+	/* SCRATCH_REG0 to test. */
+	return RREG32_NO_KIQ(0xc040) == 0xffffffff;
+}
+
+int amdgpu_allocate_static_csa(struct amdgpu_device *adev)
+{
+	int r;
+	void *ptr;
+
+	r = amdgpu_bo_create_kernel(adev, AMDGPU_CSA_SIZE, PAGE_SIZE,
+				AMDGPU_GEM_DOMAIN_VRAM, &adev->virt.csa_obj,
+				&adev->virt.csa_vmid0_addr, &ptr);
+	if (r)
+		return r;
+
+	memset(ptr, 0, AMDGPU_CSA_SIZE);
+	return 0;
+}
+
+void amdgpu_free_static_csa(struct amdgpu_device *adev) {
+	amdgpu_bo_free_kernel(&adev->virt.csa_obj,
+						&adev->virt.csa_vmid0_addr,
+						NULL);
+}
+
+/*
+ * amdgpu_map_static_csa should be called during amdgpu_vm_init
+ * it maps virtual address amdgpu_csa_vaddr() to this VM, and each command
+ * submission of GFX should use this virtual address within META_DATA init
+ * package to support SRIOV gfx preemption.
+ */
+int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			  struct amdgpu_bo_va **bo_va)
+{
+	uint64_t csa_addr = amdgpu_csa_vaddr(adev) & AMDGPU_VA_HOLE_MASK;
+	struct ww_acquire_ctx ticket;
+	struct list_head list;
+	struct amdgpu_bo_list_entry pd;
+	struct ttm_validate_buffer csa_tv;
+	int r;
+
+	INIT_LIST_HEAD(&list);
+	INIT_LIST_HEAD(&csa_tv.head);
+	csa_tv.bo = &adev->virt.csa_obj->tbo;
+	csa_tv.shared = true;
+
+	list_add(&csa_tv.head, &list);
+	amdgpu_vm_get_pd_bo(vm, &list, &pd);
+
+	r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL);
+	if (r) {
+		DRM_ERROR("failed to reserve CSA,PD BOs: err=%d\n", r);
+		return r;
+	}
+
+	*bo_va = amdgpu_vm_bo_add(adev, vm, adev->virt.csa_obj);
+	if (!*bo_va) {
+		ttm_eu_backoff_reservation(&ticket, &list);
+		DRM_ERROR("failed to create bo_va for static CSA\n");
+		return -ENOMEM;
+	}
+
+	r = amdgpu_vm_alloc_pts(adev, (*bo_va)->base.vm, csa_addr,
+				AMDGPU_CSA_SIZE);
+	if (r) {
+		DRM_ERROR("failed to allocate pts for static CSA, err=%d\n", r);
+		amdgpu_vm_bo_rmv(adev, *bo_va);
+		ttm_eu_backoff_reservation(&ticket, &list);
+		return r;
+	}
+
+	r = amdgpu_vm_bo_map(adev, *bo_va, csa_addr, 0, AMDGPU_CSA_SIZE,
+			     AMDGPU_PTE_READABLE | AMDGPU_PTE_WRITEABLE |
+			     AMDGPU_PTE_EXECUTABLE);
+
+	if (r) {
+		DRM_ERROR("failed to do bo_map on static CSA, err=%d\n", r);
+		amdgpu_vm_bo_rmv(adev, *bo_va);
+		ttm_eu_backoff_reservation(&ticket, &list);
+		return r;
+	}
+
+	ttm_eu_backoff_reservation(&ticket, &list);
+	return 0;
+}
+
+void amdgpu_virt_init_setting(struct amdgpu_device *adev)
+{
+	/* enable virtual display */
+	adev->mode_info.num_crtc = 1;
+	adev->enable_virtual_display = true;
+	adev->cg_flags = 0;
+	adev->pg_flags = 0;
+}
+
+uint32_t amdgpu_virt_kiq_rreg(struct amdgpu_device *adev, uint32_t reg)
+{
+	signed long r, cnt = 0;
+	unsigned long flags;
+	uint32_t seq;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
+	struct amdgpu_ring *ring = &kiq->ring;
+
+	BUG_ON(!ring->funcs->emit_rreg);
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+	amdgpu_ring_alloc(ring, 32);
+	amdgpu_ring_emit_rreg(ring, reg);
+	amdgpu_fence_emit_polling(ring, &seq);
+	amdgpu_ring_commit(ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+
+	/* don't wait anymore for gpu reset case because this way may
+	 * block gpu_recover() routine forever, e.g. this virt_kiq_rreg
+	 * is triggered in TTM and ttm_bo_lock_delayed_workqueue() will
+	 * never return if we keep waiting in virt_kiq_rreg, which cause
+	 * gpu_recover() hang there.
+	 *
+	 * also don't wait anymore for IRQ context
+	 * */
+	if (r < 1 && (adev->in_gpu_reset || in_interrupt()))
+		goto failed_kiq_read;
+
+	if (in_interrupt())
+		might_sleep();
+
+	while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) {
+		drm_msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
+		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+	}
+
+	if (cnt > MAX_KIQ_REG_TRY)
+		goto failed_kiq_read;
+
+	return adev->wb.wb[adev->virt.reg_val_offs];
+
+failed_kiq_read:
+	pr_err("failed to read reg:%x\n", reg);
+	return ~0;
+}
+
+void amdgpu_virt_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v)
+{
+	signed long r, cnt = 0;
+	unsigned long flags;
+	uint32_t seq;
+	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
+	struct amdgpu_ring *ring = &kiq->ring;
+
+	BUG_ON(!ring->funcs->emit_wreg);
+
+	spin_lock_irqsave(&kiq->ring_lock, flags);
+	amdgpu_ring_alloc(ring, 32);
+	amdgpu_ring_emit_wreg(ring, reg, v);
+	amdgpu_fence_emit_polling(ring, &seq);
+	amdgpu_ring_commit(ring);
+	spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+	r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+
+	/* don't wait anymore for gpu reset case because this way may
+	 * block gpu_recover() routine forever, e.g. this virt_kiq_rreg
+	 * is triggered in TTM and ttm_bo_lock_delayed_workqueue() will
+	 * never return if we keep waiting in virt_kiq_rreg, which cause
+	 * gpu_recover() hang there.
+	 *
+	 * also don't wait anymore for IRQ context
+	 * */
+	if (r < 1 && (adev->in_gpu_reset || in_interrupt()))
+		goto failed_kiq_write;
+
+	if (in_interrupt())
+		might_sleep();
+
+	while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) {
+
+		drm_msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
+		r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+	}
+
+	if (cnt > MAX_KIQ_REG_TRY)
+		goto failed_kiq_write;
+
+	return;
+
+failed_kiq_write:
+	pr_err("failed to write reg:%x\n", reg);
+}
+
+/**
+ * amdgpu_virt_request_full_gpu() - request full gpu access
+ * @amdgpu:	amdgpu device.
+ * @init:	is driver init time.
+ * When start to init/fini driver, first need to request full gpu access.
+ * Return: Zero if request success, otherwise will return error.
+ */
+int amdgpu_virt_request_full_gpu(struct amdgpu_device *adev, bool init)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	int r;
+
+	if (virt->ops && virt->ops->req_full_gpu) {
+		r = virt->ops->req_full_gpu(adev, init);
+		if (r)
+			return r;
+
+		adev->virt.caps &= ~AMDGPU_SRIOV_CAPS_RUNTIME;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_virt_release_full_gpu() - release full gpu access
+ * @amdgpu:	amdgpu device.
+ * @init:	is driver init time.
+ * When finishing driver init/fini, need to release full gpu access.
+ * Return: Zero if release success, otherwise will returen error.
+ */
+int amdgpu_virt_release_full_gpu(struct amdgpu_device *adev, bool init)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	int r;
+
+	if (virt->ops && virt->ops->rel_full_gpu) {
+		r = virt->ops->rel_full_gpu(adev, init);
+		if (r)
+			return r;
+
+		adev->virt.caps |= AMDGPU_SRIOV_CAPS_RUNTIME;
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_virt_reset_gpu() - reset gpu
+ * @amdgpu:	amdgpu device.
+ * Send reset command to GPU hypervisor to reset GPU that VM is using
+ * Return: Zero if reset success, otherwise will return error.
+ */
+int amdgpu_virt_reset_gpu(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+	int r;
+
+	if (virt->ops && virt->ops->reset_gpu) {
+		r = virt->ops->reset_gpu(adev);
+		if (r)
+			return r;
+
+		adev->virt.caps &= ~AMDGPU_SRIOV_CAPS_RUNTIME;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_virt_wait_reset() - wait for reset gpu completed
+ * @amdgpu:	amdgpu device.
+ * Wait for GPU reset completed.
+ * Return: Zero if reset success, otherwise will return error.
+ */
+int amdgpu_virt_wait_reset(struct amdgpu_device *adev)
+{
+	struct amdgpu_virt *virt = &adev->virt;
+
+	if (!virt->ops || !virt->ops->wait_reset)
+		return -EINVAL;
+
+	return virt->ops->wait_reset(adev);
+}
+
+/**
+ * amdgpu_virt_alloc_mm_table() - alloc memory for mm table
+ * @amdgpu:	amdgpu device.
+ * MM table is used by UVD and VCE for its initialization
+ * Return: Zero if allocate success.
+ */
+int amdgpu_virt_alloc_mm_table(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!amdgpu_sriov_vf(adev) || adev->virt.mm_table.gpu_addr)
+		return 0;
+
+	r = amdgpu_bo_create_kernel(adev, PAGE_SIZE, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_VRAM,
+				    &adev->virt.mm_table.bo,
+				    &adev->virt.mm_table.gpu_addr,
+				    (void *)&adev->virt.mm_table.cpu_addr);
+	if (r) {
+		DRM_ERROR("failed to alloc mm table and error = %d.\n", r);
+		return r;
+	}
+
+	memset((void *)adev->virt.mm_table.cpu_addr, 0, PAGE_SIZE);
+	DRM_INFO("MM table gpu addr = 0x%llx, cpu addr = %p.\n",
+		 adev->virt.mm_table.gpu_addr,
+		 adev->virt.mm_table.cpu_addr);
+	return 0;
+}
+
+/**
+ * amdgpu_virt_free_mm_table() - free mm table memory
+ * @amdgpu:	amdgpu device.
+ * Free MM table memory
+ */
+void amdgpu_virt_free_mm_table(struct amdgpu_device *adev)
+{
+	if (!amdgpu_sriov_vf(adev) || !adev->virt.mm_table.gpu_addr)
+		return;
+
+	amdgpu_bo_free_kernel(&adev->virt.mm_table.bo,
+			      &adev->virt.mm_table.gpu_addr,
+			      (void *)&adev->virt.mm_table.cpu_addr);
+	adev->virt.mm_table.gpu_addr = 0;
+}
+
+
+int amdgpu_virt_fw_reserve_get_checksum(void *obj,
+					unsigned long obj_size,
+					unsigned int key,
+					unsigned int chksum)
+{
+	unsigned int ret = key;
+	unsigned long i = 0;
+	unsigned char *pos;
+
+	pos = (char *)obj;
+	/* calculate checksum */
+	for (i = 0; i < obj_size; ++i)
+		ret += *(pos + i);
+	/* minus the chksum itself */
+	pos = (char *)&chksum;
+	for (i = 0; i < sizeof(chksum); ++i)
+		ret -= *(pos + i);
+	return ret;
+}
+
+void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev)
+{
+	uint32_t pf2vf_size = 0;
+	uint32_t checksum = 0;
+	uint32_t checkval;
+	char *str;
+
+	adev->virt.fw_reserve.p_pf2vf = NULL;
+	adev->virt.fw_reserve.p_vf2pf = NULL;
+
+	if (adev->fw_vram_usage.va != NULL) {
+		adev->virt.fw_reserve.p_pf2vf =
+			(struct amdgim_pf2vf_info_header *)(
+			adev->fw_vram_usage.va + AMDGIM_DATAEXCHANGE_OFFSET);
+		AMDGPU_FW_VRAM_PF2VF_READ(adev, header.size, &pf2vf_size);
+		AMDGPU_FW_VRAM_PF2VF_READ(adev, checksum, &checksum);
+		AMDGPU_FW_VRAM_PF2VF_READ(adev, feature_flags, &adev->virt.gim_feature);
+
+		/* pf2vf message must be in 4K */
+		if (pf2vf_size > 0 && pf2vf_size < 4096) {
+			checkval = amdgpu_virt_fw_reserve_get_checksum(
+				adev->virt.fw_reserve.p_pf2vf, pf2vf_size,
+				adev->virt.fw_reserve.checksum_key, checksum);
+			if (checkval == checksum) {
+				adev->virt.fw_reserve.p_vf2pf =
+					((void *)adev->virt.fw_reserve.p_pf2vf +
+					pf2vf_size);
+				memset((void *)adev->virt.fw_reserve.p_vf2pf, 0,
+					sizeof(amdgim_vf2pf_info));
+				AMDGPU_FW_VRAM_VF2PF_WRITE(adev, header.version,
+					AMDGPU_FW_VRAM_VF2PF_VER);
+				AMDGPU_FW_VRAM_VF2PF_WRITE(adev, header.size,
+					sizeof(amdgim_vf2pf_info));
+				AMDGPU_FW_VRAM_VF2PF_READ(adev, driver_version,
+					&str);
+#ifdef MODULE
+				if (THIS_MODULE->version != NULL)
+					strcpy(str, THIS_MODULE->version);
+				else
+#endif
+					strlcpy(str, "N/A", 64);
+				AMDGPU_FW_VRAM_VF2PF_WRITE(adev, driver_cert,
+					0);
+				AMDGPU_FW_VRAM_VF2PF_WRITE(adev, checksum,
+					amdgpu_virt_fw_reserve_get_checksum(
+					adev->virt.fw_reserve.p_vf2pf,
+					pf2vf_size,
+					adev->virt.fw_reserve.checksum_key, 0));
+			}
+		}
+	}
+}
+
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_virt.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_virt.h
new file mode 100644
index 00000000000..a4a52484370
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_virt.h
@@ -0,0 +1,302 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: Monk.liu@amd.com
+ */
+#ifndef AMDGPU_VIRT_H
+#define AMDGPU_VIRT_H
+
+#define AMDGPU_SRIOV_CAPS_SRIOV_VBIOS  (1 << 0) /* vBIOS is sr-iov ready */
+#define AMDGPU_SRIOV_CAPS_ENABLE_IOV   (1 << 1) /* sr-iov is enabled on this GPU */
+#define AMDGPU_SRIOV_CAPS_IS_VF        (1 << 2) /* this GPU is a virtual function */
+#define AMDGPU_PASSTHROUGH_MODE        (1 << 3) /* thw whole GPU is pass through for VM */
+#define AMDGPU_SRIOV_CAPS_RUNTIME      (1 << 4) /* is out of full access mode */
+
+struct amdgpu_mm_table {
+	struct amdgpu_bo	*bo;
+	uint32_t		*cpu_addr;
+	uint64_t		gpu_addr;
+};
+
+#define AMDGPU_VF_ERROR_ENTRY_SIZE    16
+
+/* struct error_entry - amdgpu VF error information. */
+struct amdgpu_vf_error_buffer {
+	struct rwlock lock;
+	int read_count;
+	int write_count;
+	uint16_t code[AMDGPU_VF_ERROR_ENTRY_SIZE];
+	uint16_t flags[AMDGPU_VF_ERROR_ENTRY_SIZE];
+	uint64_t data[AMDGPU_VF_ERROR_ENTRY_SIZE];
+};
+
+/**
+ * struct amdgpu_virt_ops - amdgpu device virt operations
+ */
+struct amdgpu_virt_ops {
+	int (*req_full_gpu)(struct amdgpu_device *adev, bool init);
+	int (*rel_full_gpu)(struct amdgpu_device *adev, bool init);
+	int (*reset_gpu)(struct amdgpu_device *adev);
+	int (*wait_reset)(struct amdgpu_device *adev);
+	void (*trans_msg)(struct amdgpu_device *adev, u32 req, u32 data1, u32 data2, u32 data3);
+};
+
+/*
+ * Firmware Reserve Frame buffer
+ */
+struct amdgpu_virt_fw_reserve {
+	struct amdgim_pf2vf_info_header *p_pf2vf;
+	struct amdgim_vf2pf_info_header *p_vf2pf;
+	unsigned int checksum_key;
+};
+/*
+ * Defination between PF and VF
+ * Structures forcibly aligned to 4 to keep the same style as PF.
+ */
+#define AMDGIM_DATAEXCHANGE_OFFSET		(64 * 1024)
+
+#define AMDGIM_GET_STRUCTURE_RESERVED_SIZE(total, u8, u16, u32, u64) \
+		(total - (((u8)+3) / 4 + ((u16)+1) / 2 + (u32) + (u64)*2))
+
+enum AMDGIM_FEATURE_FLAG {
+	/* GIM supports feature of Error log collecting */
+	AMDGIM_FEATURE_ERROR_LOG_COLLECT = 0x1,
+	/* GIM supports feature of loading uCodes */
+	AMDGIM_FEATURE_GIM_LOAD_UCODES   = 0x2,
+	/* VRAM LOST by GIM */
+	AMDGIM_FEATURE_GIM_FLR_VRAMLOST = 0x4,
+};
+
+struct amdgim_pf2vf_info_header {
+	/* the total structure size in byte. */
+	uint32_t size;
+	/* version of this structure, written by the GIM */
+	uint32_t version;
+} __aligned(4);
+struct  amdgim_pf2vf_info_v1 {
+	/* header contains size and version */
+	struct amdgim_pf2vf_info_header header;
+	/* max_width * max_height */
+	unsigned int uvd_enc_max_pixels_count;
+	/* 16x16 pixels/sec, codec independent */
+	unsigned int uvd_enc_max_bandwidth;
+	/* max_width * max_height */
+	unsigned int vce_enc_max_pixels_count;
+	/* 16x16 pixels/sec, codec independent */
+	unsigned int vce_enc_max_bandwidth;
+	/* MEC FW position in kb from the start of visible frame buffer */
+	unsigned int mecfw_kboffset;
+	/* The features flags of the GIM driver supports. */
+	unsigned int feature_flags;
+	/* use private key from mailbox 2 to create chueksum */
+	unsigned int checksum;
+} __aligned(4);
+
+struct  amdgim_pf2vf_info_v2 {
+	/* header contains size and version */
+	struct amdgim_pf2vf_info_header header;
+	/* use private key from mailbox 2 to create chueksum */
+	uint32_t checksum;
+	/* The features flags of the GIM driver supports. */
+	uint32_t feature_flags;
+	/* max_width * max_height */
+	uint32_t uvd_enc_max_pixels_count;
+	/* 16x16 pixels/sec, codec independent */
+	uint32_t uvd_enc_max_bandwidth;
+	/* max_width * max_height */
+	uint32_t vce_enc_max_pixels_count;
+	/* 16x16 pixels/sec, codec independent */
+	uint32_t vce_enc_max_bandwidth;
+	/* MEC FW position in kb from the start of VF visible frame buffer */
+	uint64_t mecfw_kboffset;
+	/* MEC FW size in KB */
+	uint32_t mecfw_ksize;
+	/* UVD FW position in kb from the start of VF visible frame buffer */
+	uint64_t uvdfw_kboffset;
+	/* UVD FW size in KB */
+	uint32_t uvdfw_ksize;
+	/* VCE FW position in kb from the start of VF visible frame buffer */
+	uint64_t vcefw_kboffset;
+	/* VCE FW size in KB */
+	uint32_t vcefw_ksize;
+	uint32_t reserved[AMDGIM_GET_STRUCTURE_RESERVED_SIZE(256, 0, 0, (9 + sizeof(struct amdgim_pf2vf_info_header)/sizeof(uint32_t)), 3)];
+} __aligned(4);
+
+
+struct amdgim_vf2pf_info_header {
+	/* the total structure size in byte. */
+	uint32_t size;
+	/*version of this structure, written by the guest */
+	uint32_t version;
+} __aligned(4);
+
+struct amdgim_vf2pf_info_v1 {
+	/* header contains size and version */
+	struct amdgim_vf2pf_info_header header;
+	/* driver version */
+	char driver_version[64];
+	/* driver certification, 1=WHQL, 0=None */
+	unsigned int driver_cert;
+	/* guest OS type and version: need a define */
+	unsigned int os_info;
+	/* in the unit of 1M */
+	unsigned int fb_usage;
+	/* guest gfx engine usage percentage */
+	unsigned int gfx_usage;
+	/* guest gfx engine health percentage */
+	unsigned int gfx_health;
+	/* guest compute engine usage percentage */
+	unsigned int compute_usage;
+	/* guest compute engine health percentage */
+	unsigned int compute_health;
+	/* guest vce engine usage percentage. 0xffff means N/A. */
+	unsigned int vce_enc_usage;
+	/* guest vce engine health percentage. 0xffff means N/A. */
+	unsigned int vce_enc_health;
+	/* guest uvd engine usage percentage. 0xffff means N/A. */
+	unsigned int uvd_enc_usage;
+	/* guest uvd engine usage percentage. 0xffff means N/A. */
+	unsigned int uvd_enc_health;
+	unsigned int checksum;
+} __aligned(4);
+
+struct amdgim_vf2pf_info_v2 {
+	/* header contains size and version */
+	struct amdgim_vf2pf_info_header header;
+	uint32_t checksum;
+	/* driver version */
+	uint8_t driver_version[64];
+	/* driver certification, 1=WHQL, 0=None */
+	uint32_t driver_cert;
+	/* guest OS type and version: need a define */
+	uint32_t os_info;
+	/* in the unit of 1M */
+	uint32_t fb_usage;
+	/* guest gfx engine usage percentage */
+	uint32_t gfx_usage;
+	/* guest gfx engine health percentage */
+	uint32_t gfx_health;
+	/* guest compute engine usage percentage */
+	uint32_t compute_usage;
+	/* guest compute engine health percentage */
+	uint32_t compute_health;
+	/* guest vce engine usage percentage. 0xffff means N/A. */
+	uint32_t vce_enc_usage;
+	/* guest vce engine health percentage. 0xffff means N/A. */
+	uint32_t vce_enc_health;
+	/* guest uvd engine usage percentage. 0xffff means N/A. */
+	uint32_t uvd_enc_usage;
+	/* guest uvd engine usage percentage. 0xffff means N/A. */
+	uint32_t uvd_enc_health;
+	uint32_t reserved[AMDGIM_GET_STRUCTURE_RESERVED_SIZE(256, 64, 0, (12 + sizeof(struct amdgim_vf2pf_info_header)/sizeof(uint32_t)), 0)];
+} __aligned(4);
+
+#define AMDGPU_FW_VRAM_VF2PF_VER 2
+typedef struct amdgim_vf2pf_info_v2 amdgim_vf2pf_info ;
+
+#define AMDGPU_FW_VRAM_VF2PF_WRITE(adev, field, val) \
+	do { \
+		((amdgim_vf2pf_info *)adev->virt.fw_reserve.p_vf2pf)->field = (val); \
+	} while (0)
+
+#define AMDGPU_FW_VRAM_VF2PF_READ(adev, field, val) \
+	do { \
+		(*val) = ((amdgim_vf2pf_info *)adev->virt.fw_reserve.p_vf2pf)->field; \
+	} while (0)
+
+#define AMDGPU_FW_VRAM_PF2VF_READ(adev, field, val) \
+	do { \
+		if (!adev->virt.fw_reserve.p_pf2vf) \
+			*(val) = 0; \
+		else { \
+			if (adev->virt.fw_reserve.p_pf2vf->version == 1) \
+				*(val) = ((struct amdgim_pf2vf_info_v1 *)adev->virt.fw_reserve.p_pf2vf)->field; \
+			if (adev->virt.fw_reserve.p_pf2vf->version == 2) \
+				*(val) = ((struct amdgim_pf2vf_info_v2 *)adev->virt.fw_reserve.p_pf2vf)->field; \
+		} \
+	} while (0)
+
+/* GPU virtualization */
+struct amdgpu_virt {
+	uint32_t			caps;
+	struct amdgpu_bo		*csa_obj;
+	uint64_t			csa_vmid0_addr;
+	bool chained_ib_support;
+	uint32_t			reg_val_offs;
+	struct amdgpu_irq_src		ack_irq;
+	struct amdgpu_irq_src		rcv_irq;
+	struct work_struct		flr_work;
+	struct amdgpu_mm_table		mm_table;
+	const struct amdgpu_virt_ops	*ops;
+	struct amdgpu_vf_error_buffer   vf_errors;
+	struct amdgpu_virt_fw_reserve	fw_reserve;
+	uint32_t gim_feature;
+};
+
+#define AMDGPU_CSA_SIZE		(8 * 1024)
+
+#define amdgpu_sriov_enabled(adev) \
+((adev)->virt.caps & AMDGPU_SRIOV_CAPS_ENABLE_IOV)
+
+#define amdgpu_sriov_vf(adev) \
+((adev)->virt.caps & AMDGPU_SRIOV_CAPS_IS_VF)
+
+#define amdgpu_sriov_bios(adev) \
+((adev)->virt.caps & AMDGPU_SRIOV_CAPS_SRIOV_VBIOS)
+
+#define amdgpu_sriov_runtime(adev) \
+((adev)->virt.caps & AMDGPU_SRIOV_CAPS_RUNTIME)
+
+#define amdgpu_passthrough(adev) \
+((adev)->virt.caps & AMDGPU_PASSTHROUGH_MODE)
+
+static inline bool is_virtual_machine(void)
+{
+#if defined(__amd64__) || defined(__i386__)
+	return boot_cpu_has(X86_FEATURE_HYPERVISOR);
+#else
+	return false;
+#endif
+}
+
+struct amdgpu_vm;
+
+uint64_t amdgpu_csa_vaddr(struct amdgpu_device *adev);
+bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev);
+int amdgpu_allocate_static_csa(struct amdgpu_device *adev);
+int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			  struct amdgpu_bo_va **bo_va);
+void amdgpu_free_static_csa(struct amdgpu_device *adev);
+void amdgpu_virt_init_setting(struct amdgpu_device *adev);
+uint32_t amdgpu_virt_kiq_rreg(struct amdgpu_device *adev, uint32_t reg);
+void amdgpu_virt_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v);
+int amdgpu_virt_request_full_gpu(struct amdgpu_device *adev, bool init);
+int amdgpu_virt_release_full_gpu(struct amdgpu_device *adev, bool init);
+int amdgpu_virt_reset_gpu(struct amdgpu_device *adev);
+int amdgpu_virt_wait_reset(struct amdgpu_device *adev);
+int amdgpu_virt_alloc_mm_table(struct amdgpu_device *adev);
+void amdgpu_virt_free_mm_table(struct amdgpu_device *adev);
+int amdgpu_virt_fw_reserve_get_checksum(void *obj, unsigned long obj_size,
+					unsigned int key,
+					unsigned int chksum);
+void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_vm.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vm.c
new file mode 100644
index 00000000000..6bec75d65f3
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vm.c
@@ -0,0 +1,3146 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <linux/dma-fence-array.h>
+#include <linux/interval_tree_generic.h>
+#include <linux/idr.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_gmc.h"
+
+/**
+ * DOC: GPUVM
+ *
+ * GPUVM is similar to the legacy gart on older asics, however
+ * rather than there being a single global gart table
+ * for the entire GPU, there are multiple VM page tables active
+ * at any given time.  The VM page tables can contain a mix
+ * vram pages and system memory pages and system memory pages
+ * can be mapped as snooped (cached system pages) or unsnooped
+ * (uncached system pages).
+ * Each VM has an ID associated with it and there is a page table
+ * associated with each VMID.  When execting a command buffer,
+ * the kernel tells the the ring what VMID to use for that command
+ * buffer.  VMIDs are allocated dynamically as commands are submitted.
+ * The userspace drivers maintain their own address space and the kernel
+ * sets up their pages tables accordingly when they submit their
+ * command buffers and a VMID is assigned.
+ * Cayman/Trinity support up to 8 active VMs at any given time;
+ * SI supports 16.
+ */
+
+#define START(node) ((node)->start)
+#define LAST(node) ((node)->last)
+
+#ifdef __linux__
+INTERVAL_TREE_DEFINE(struct amdgpu_bo_va_mapping, rb, uint64_t, __subtree_last,
+		     START, LAST, static, amdgpu_vm_it)
+#else
+static struct amdgpu_bo_va_mapping *
+amdgpu_vm_it_iter_first(struct rb_root_cached *root, uint64_t start,
+    uint64_t last)
+{
+	struct amdgpu_bo_va_mapping *node;
+	struct rb_node *rb;
+
+	for (rb = rb_first(root); rb; rb = rb_next(rb)) {
+		node = rb_entry(rb, typeof(*node), rb);
+		if (LAST(node) >= start && START(node) <= last)
+			return node;
+	}
+	return NULL;
+}
+
+static struct amdgpu_bo_va_mapping *
+amdgpu_vm_it_iter_next(struct amdgpu_bo_va_mapping *node, uint64_t start,
+    uint64_t last)
+{
+	STUB();
+	struct rb_node *rb = &node->rb;
+
+	for (rb = rb_next(rb); rb; rb = rb_next(rb)) {
+		node = rb_entry(rb, typeof(*node), rb);
+		if (LAST(node) >= start && START(node) <= last)
+			return node;
+	}
+	return NULL;
+}
+
+static void
+amdgpu_vm_it_remove(struct amdgpu_bo_va_mapping *node,
+    struct rb_root_cached *root) 
+{
+	rb_erase(&node->rb, root);
+}
+
+static void
+amdgpu_vm_it_insert(struct amdgpu_bo_va_mapping *node,
+    struct rb_root_cached *root)
+{
+	struct rb_node **iter = &root->rb_node;
+	struct rb_node *parent = NULL;
+	struct amdgpu_bo_va_mapping *iter_node;
+
+	while (*iter) {
+		parent = *iter;
+		iter_node = rb_entry(*iter, struct amdgpu_bo_va_mapping, rb);
+
+		if (node->start < iter_node->start)
+			iter = &(*iter)->rb_left;
+		else
+			iter = &(*iter)->rb_right;
+	}
+
+	rb_link_node(&node->rb, parent, iter);
+	rb_insert_color(&node->rb, root);
+}
+#endif
+
+#undef START
+#undef LAST
+
+/**
+ * struct amdgpu_pte_update_params - Local structure
+ *
+ * Encapsulate some VM table update parameters to reduce
+ * the number of function parameters
+ *
+ */
+struct amdgpu_pte_update_params {
+
+	/**
+	 * @adev: amdgpu device we do this update for
+	 */
+	struct amdgpu_device *adev;
+
+	/**
+	 * @vm: optional amdgpu_vm we do this update for
+	 */
+	struct amdgpu_vm *vm;
+
+	/**
+	 * @src: address where to copy page table entries from
+	 */
+	uint64_t src;
+
+	/**
+	 * @ib: indirect buffer to fill with commands
+	 */
+	struct amdgpu_ib *ib;
+
+	/**
+	 * @func: Function which actually does the update
+	 */
+	void (*func)(struct amdgpu_pte_update_params *params,
+		     struct amdgpu_bo *bo, uint64_t pe,
+		     uint64_t addr, unsigned count, uint32_t incr,
+		     uint64_t flags);
+	/**
+	 * @pages_addr:
+	 *
+	 * DMA addresses to use for mapping, used during VM update by CPU
+	 */
+	dma_addr_t *pages_addr;
+
+	/**
+	 * @kptr:
+	 *
+	 * Kernel pointer of PD/PT BO that needs to be updated,
+	 * used during VM update by CPU
+	 */
+	void *kptr;
+};
+
+/**
+ * struct amdgpu_prt_cb - Helper to disable partial resident texture feature from a fence callback
+ */
+struct amdgpu_prt_cb {
+
+	/**
+	 * @adev: amdgpu device
+	 */
+	struct amdgpu_device *adev;
+
+	/**
+	 * @cb: callback
+	 */
+	struct dma_fence_cb cb;
+};
+
+/**
+ * amdgpu_vm_bo_base_init - Adds bo to the list of bos associated with the vm
+ *
+ * @base: base structure for tracking BO usage in a VM
+ * @vm: vm to which bo is to be added
+ * @bo: amdgpu buffer object
+ *
+ * Initialize a bo_va_base structure and add it to the appropriate lists
+ *
+ */
+static void amdgpu_vm_bo_base_init(struct amdgpu_vm_bo_base *base,
+				   struct amdgpu_vm *vm,
+				   struct amdgpu_bo *bo)
+{
+	base->vm = vm;
+	base->bo = bo;
+	INIT_LIST_HEAD(&base->bo_list);
+	INIT_LIST_HEAD(&base->vm_status);
+
+	if (!bo)
+		return;
+	list_add_tail(&base->bo_list, &bo->va);
+
+	if (bo->tbo.type == ttm_bo_type_kernel)
+		list_move(&base->vm_status, &vm->relocated);
+
+	if (bo->tbo.resv != vm->root.base.bo->tbo.resv)
+		return;
+
+	if (bo->preferred_domains &
+	    amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type))
+		return;
+
+	/*
+	 * we checked all the prerequisites, but it looks like this per vm bo
+	 * is currently evicted. add the bo to the evicted list to make sure it
+	 * is validated on next vm use to avoid fault.
+	 * */
+	list_move_tail(&base->vm_status, &vm->evicted);
+	base->moved = true;
+}
+
+/**
+ * amdgpu_vm_level_shift - return the addr shift for each level
+ *
+ * @adev: amdgpu_device pointer
+ * @level: VMPT level
+ *
+ * Returns:
+ * The number of bits the pfn needs to be right shifted for a level.
+ */
+static unsigned amdgpu_vm_level_shift(struct amdgpu_device *adev,
+				      unsigned level)
+{
+	unsigned shift = 0xff;
+
+	switch (level) {
+	case AMDGPU_VM_PDB2:
+	case AMDGPU_VM_PDB1:
+	case AMDGPU_VM_PDB0:
+		shift = 9 * (AMDGPU_VM_PDB0 - level) +
+			adev->vm_manager.block_size;
+		break;
+	case AMDGPU_VM_PTB:
+		shift = 0;
+		break;
+	default:
+		dev_err(adev->dev, "the level%d isn't supported.\n", level);
+	}
+
+	return shift;
+}
+
+/**
+ * amdgpu_vm_num_entries - return the number of entries in a PD/PT
+ *
+ * @adev: amdgpu_device pointer
+ * @level: VMPT level
+ *
+ * Returns:
+ * The number of entries in a page directory or page table.
+ */
+static unsigned amdgpu_vm_num_entries(struct amdgpu_device *adev,
+				      unsigned level)
+{
+	unsigned shift = amdgpu_vm_level_shift(adev,
+					       adev->vm_manager.root_level);
+
+	if (level == adev->vm_manager.root_level)
+		/* For the root directory */
+		return round_up(adev->vm_manager.max_pfn, 1 << shift) >> shift;
+	else if (level != AMDGPU_VM_PTB)
+		/* Everything in between */
+		return 512;
+	else
+		/* For the page tables on the leaves */
+		return AMDGPU_VM_PTE_COUNT(adev);
+}
+
+/**
+ * amdgpu_vm_bo_size - returns the size of the BOs in bytes
+ *
+ * @adev: amdgpu_device pointer
+ * @level: VMPT level
+ *
+ * Returns:
+ * The size of the BO for a page directory or page table in bytes.
+ */
+static unsigned amdgpu_vm_bo_size(struct amdgpu_device *adev, unsigned level)
+{
+	return AMDGPU_GPU_PAGE_ALIGN(amdgpu_vm_num_entries(adev, level) * 8);
+}
+
+/**
+ * amdgpu_vm_get_pd_bo - add the VM PD to a validation list
+ *
+ * @vm: vm providing the BOs
+ * @validated: head of validation list
+ * @entry: entry to add
+ *
+ * Add the page directory to the list of BOs to
+ * validate for command submission.
+ */
+void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
+			 struct list_head *validated,
+			 struct amdgpu_bo_list_entry *entry)
+{
+	entry->robj = vm->root.base.bo;
+	entry->priority = 0;
+	entry->tv.bo = &entry->robj->tbo;
+	entry->tv.shared = true;
+	entry->user_pages = NULL;
+	list_add(&entry->tv.head, validated);
+}
+
+/**
+ * amdgpu_vm_validate_pt_bos - validate the page table BOs
+ *
+ * @adev: amdgpu device pointer
+ * @vm: vm providing the BOs
+ * @validate: callback to do the validation
+ * @param: parameter for the validation callback
+ *
+ * Validate the page table BOs on command submission if neccessary.
+ *
+ * Returns:
+ * Validation result.
+ */
+int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			      int (*validate)(void *p, struct amdgpu_bo *bo),
+			      void *param)
+{
+	struct ttm_bo_global *glob = adev->mman.bdev.glob;
+	struct amdgpu_vm_bo_base *bo_base, *tmp;
+	int r = 0;
+
+	list_for_each_entry_safe(bo_base, tmp, &vm->evicted, vm_status) {
+		struct amdgpu_bo *bo = bo_base->bo;
+
+		if (bo->parent) {
+			r = validate(param, bo);
+			if (r)
+				break;
+
+			spin_lock(&glob->lru_lock);
+			ttm_bo_move_to_lru_tail(&bo->tbo);
+			if (bo->shadow)
+				ttm_bo_move_to_lru_tail(&bo->shadow->tbo);
+			spin_unlock(&glob->lru_lock);
+		}
+
+		if (bo->tbo.type != ttm_bo_type_kernel) {
+			spin_lock(&vm->moved_lock);
+			list_move(&bo_base->vm_status, &vm->moved);
+			spin_unlock(&vm->moved_lock);
+		} else {
+			list_move(&bo_base->vm_status, &vm->relocated);
+		}
+	}
+
+	spin_lock(&glob->lru_lock);
+	list_for_each_entry(bo_base, &vm->idle, vm_status) {
+		struct amdgpu_bo *bo = bo_base->bo;
+
+		if (!bo->parent)
+			continue;
+
+		ttm_bo_move_to_lru_tail(&bo->tbo);
+		if (bo->shadow)
+			ttm_bo_move_to_lru_tail(&bo->shadow->tbo);
+	}
+	spin_unlock(&glob->lru_lock);
+
+	return r;
+}
+
+/**
+ * amdgpu_vm_ready - check VM is ready for updates
+ *
+ * @vm: VM to check
+ *
+ * Check if all VM PDs/PTs are ready for updates
+ *
+ * Returns:
+ * True if eviction list is empty.
+ */
+bool amdgpu_vm_ready(struct amdgpu_vm *vm)
+{
+	return list_empty(&vm->evicted);
+}
+
+/**
+ * amdgpu_vm_clear_bo - initially clear the PDs/PTs
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: VM to clear BO from
+ * @bo: BO to clear
+ * @level: level this BO is at
+ * @pte_support_ats: indicate ATS support from PTE
+ *
+ * Root PD needs to be reserved when calling this.
+ *
+ * Returns:
+ * 0 on success, errno otherwise.
+ */
+static int amdgpu_vm_clear_bo(struct amdgpu_device *adev,
+			      struct amdgpu_vm *vm, struct amdgpu_bo *bo,
+			      unsigned level, bool pte_support_ats)
+{
+	struct ttm_operation_ctx ctx = { true, false };
+	struct dma_fence *fence = NULL;
+	unsigned entries, ats_entries;
+	struct amdgpu_ring *ring;
+	struct amdgpu_job *job;
+	uint64_t addr;
+	int r;
+
+	entries = amdgpu_bo_size(bo) / 8;
+
+	if (pte_support_ats) {
+		if (level == adev->vm_manager.root_level) {
+			ats_entries = amdgpu_vm_level_shift(adev, level);
+			ats_entries += AMDGPU_GPU_PAGE_SHIFT;
+			ats_entries = AMDGPU_VA_HOLE_START >> ats_entries;
+			ats_entries = min(ats_entries, entries);
+			entries -= ats_entries;
+		} else {
+			ats_entries = entries;
+			entries = 0;
+		}
+	} else {
+		ats_entries = 0;
+	}
+
+	ring = container_of(vm->entity.rq->sched, struct amdgpu_ring, sched);
+
+	r = reservation_object_reserve_shared(bo->tbo.resv);
+	if (r)
+		return r;
+
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (r)
+		goto error;
+
+	r = amdgpu_job_alloc_with_ib(adev, 64, &job);
+	if (r)
+		goto error;
+
+	addr = amdgpu_bo_gpu_offset(bo);
+	if (ats_entries) {
+		uint64_t ats_value;
+
+		ats_value = AMDGPU_PTE_DEFAULT_ATC;
+		if (level != AMDGPU_VM_PTB)
+			ats_value |= AMDGPU_PDE_PTE;
+
+		amdgpu_vm_set_pte_pde(adev, &job->ibs[0], addr, 0,
+				      ats_entries, 0, ats_value);
+		addr += ats_entries * 8;
+	}
+
+	if (entries)
+		amdgpu_vm_set_pte_pde(adev, &job->ibs[0], addr, 0,
+				      entries, 0, 0);
+
+	amdgpu_ring_pad_ib(ring, &job->ibs[0]);
+
+	WARN_ON(job->ibs[0].length_dw > 64);
+	r = amdgpu_sync_resv(adev, &job->sync, bo->tbo.resv,
+			     AMDGPU_FENCE_OWNER_UNDEFINED, false);
+	if (r)
+		goto error_free;
+
+	r = amdgpu_job_submit(job, &vm->entity, AMDGPU_FENCE_OWNER_UNDEFINED,
+			      &fence);
+	if (r)
+		goto error_free;
+
+	amdgpu_bo_fence(bo, fence, true);
+	dma_fence_put(fence);
+
+	if (bo->shadow)
+		return amdgpu_vm_clear_bo(adev, vm, bo->shadow,
+					  level, pte_support_ats);
+
+	return 0;
+
+error_free:
+	amdgpu_job_free(job);
+
+error:
+	return r;
+}
+
+/**
+ * amdgpu_vm_alloc_levels - allocate the PD/PT levels
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ * @parent: parent PT
+ * @saddr: start of the address range
+ * @eaddr: end of the address range
+ * @level: VMPT level
+ * @ats: indicate ATS support from PTE
+ *
+ * Make sure the page directories and page tables are allocated
+ *
+ * Returns:
+ * 0 on success, errno otherwise.
+ */
+static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev,
+				  struct amdgpu_vm *vm,
+				  struct amdgpu_vm_pt *parent,
+				  uint64_t saddr, uint64_t eaddr,
+				  unsigned level, bool ats)
+{
+	unsigned shift = amdgpu_vm_level_shift(adev, level);
+	unsigned pt_idx, from, to;
+	u64 flags;
+	int r;
+
+	if (!parent->entries) {
+		unsigned num_entries = amdgpu_vm_num_entries(adev, level);
+
+		parent->entries = kvmalloc_array(num_entries,
+						   sizeof(struct amdgpu_vm_pt),
+						   GFP_KERNEL | __GFP_ZERO);
+		if (!parent->entries)
+			return -ENOMEM;
+		memset(parent->entries, 0 , sizeof(struct amdgpu_vm_pt));
+	}
+
+	from = saddr >> shift;
+	to = eaddr >> shift;
+	if (from >= amdgpu_vm_num_entries(adev, level) ||
+	    to >= amdgpu_vm_num_entries(adev, level))
+		return -EINVAL;
+
+	++level;
+	saddr = saddr & ((1 << shift) - 1);
+	eaddr = eaddr & ((1 << shift) - 1);
+
+	flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	if (vm->root.base.bo->shadow)
+		flags |= AMDGPU_GEM_CREATE_SHADOW;
+	if (vm->use_cpu_for_update)
+		flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+	else
+		flags |= AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
+
+	/* walk over the address space and allocate the page tables */
+	for (pt_idx = from; pt_idx <= to; ++pt_idx) {
+		struct reservation_object *resv = vm->root.base.bo->tbo.resv;
+		struct amdgpu_vm_pt *entry = &parent->entries[pt_idx];
+		struct amdgpu_bo *pt;
+
+		if (!entry->base.bo) {
+			struct amdgpu_bo_param bp;
+
+			memset(&bp, 0, sizeof(bp));
+			bp.size = amdgpu_vm_bo_size(adev, level);
+			bp.byte_align = AMDGPU_GPU_PAGE_SIZE;
+			bp.domain = AMDGPU_GEM_DOMAIN_VRAM;
+			bp.flags = flags;
+			bp.type = ttm_bo_type_kernel;
+			bp.resv = resv;
+			r = amdgpu_bo_create(adev, &bp, &pt);
+			if (r)
+				return r;
+
+			r = amdgpu_vm_clear_bo(adev, vm, pt, level, ats);
+			if (r) {
+				amdgpu_bo_unref(&pt->shadow);
+				amdgpu_bo_unref(&pt);
+				return r;
+			}
+
+			if (vm->use_cpu_for_update) {
+				r = amdgpu_bo_kmap(pt, NULL);
+				if (r) {
+					amdgpu_bo_unref(&pt->shadow);
+					amdgpu_bo_unref(&pt);
+					return r;
+				}
+			}
+
+			/* Keep a reference to the root directory to avoid
+			* freeing them up in the wrong order.
+			*/
+			pt->parent = amdgpu_bo_ref(parent->base.bo);
+
+			amdgpu_vm_bo_base_init(&entry->base, vm, pt);
+		}
+
+		if (level < AMDGPU_VM_PTB) {
+			uint64_t sub_saddr = (pt_idx == from) ? saddr : 0;
+			uint64_t sub_eaddr = (pt_idx == to) ? eaddr :
+				((1 << shift) - 1);
+			r = amdgpu_vm_alloc_levels(adev, vm, entry, sub_saddr,
+						   sub_eaddr, level, ats);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_alloc_pts - Allocate page tables.
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: VM to allocate page tables for
+ * @saddr: Start address which needs to be allocated
+ * @size: Size from start address we need.
+ *
+ * Make sure the page tables are allocated.
+ *
+ * Returns:
+ * 0 on success, errno otherwise.
+ */
+int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
+			struct amdgpu_vm *vm,
+			uint64_t saddr, uint64_t size)
+{
+	uint64_t eaddr;
+	bool ats = false;
+
+	/* validate the parameters */
+	if (saddr & AMDGPU_GPU_PAGE_MASK || size & AMDGPU_GPU_PAGE_MASK)
+		return -EINVAL;
+
+	eaddr = saddr + size - 1;
+
+	if (vm->pte_support_ats)
+		ats = saddr < AMDGPU_VA_HOLE_START;
+
+	saddr /= AMDGPU_GPU_PAGE_SIZE;
+	eaddr /= AMDGPU_GPU_PAGE_SIZE;
+
+	if (eaddr >= adev->vm_manager.max_pfn) {
+		dev_err(adev->dev, "va above limit (0x%08llX >= 0x%08llX)\n",
+			eaddr, adev->vm_manager.max_pfn);
+		return -EINVAL;
+	}
+
+	return amdgpu_vm_alloc_levels(adev, vm, &vm->root, saddr, eaddr,
+				      adev->vm_manager.root_level, ats);
+}
+
+/**
+ * amdgpu_vm_check_compute_bug - check whether asic has compute vm bug
+ *
+ * @adev: amdgpu_device pointer
+ */
+void amdgpu_vm_check_compute_bug(struct amdgpu_device *adev)
+{
+	const struct amdgpu_ip_block *ip_block;
+	bool has_compute_vm_bug;
+	struct amdgpu_ring *ring;
+	int i;
+
+	has_compute_vm_bug = false;
+
+	ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GFX);
+	if (ip_block) {
+		/* Compute has a VM bug for GFX version < 7.
+		   Compute has a VM bug for GFX 8 MEC firmware version < 673.*/
+		if (ip_block->version->major <= 7)
+			has_compute_vm_bug = true;
+		else if (ip_block->version->major == 8)
+			if (adev->gfx.mec_fw_version < 673)
+				has_compute_vm_bug = true;
+	}
+
+	for (i = 0; i < adev->num_rings; i++) {
+		ring = adev->rings[i];
+		if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE)
+			/* only compute rings */
+			ring->has_compute_vm_bug = has_compute_vm_bug;
+		else
+			ring->has_compute_vm_bug = false;
+	}
+}
+
+/**
+ * amdgpu_vm_need_pipeline_sync - Check if pipe sync is needed for job.
+ *
+ * @ring: ring on which the job will be submitted
+ * @job: job to submit
+ *
+ * Returns:
+ * True if sync is needed.
+ */
+bool amdgpu_vm_need_pipeline_sync(struct amdgpu_ring *ring,
+				  struct amdgpu_job *job)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmhub = ring->funcs->vmhub;
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	struct amdgpu_vmid *id;
+	bool gds_switch_needed;
+	bool vm_flush_needed = job->vm_needs_flush || ring->has_compute_vm_bug;
+
+	if (job->vmid == 0)
+		return false;
+	id = &id_mgr->ids[job->vmid];
+	gds_switch_needed = ring->funcs->emit_gds_switch && (
+		id->gds_base != job->gds_base ||
+		id->gds_size != job->gds_size ||
+		id->gws_base != job->gws_base ||
+		id->gws_size != job->gws_size ||
+		id->oa_base != job->oa_base ||
+		id->oa_size != job->oa_size);
+
+	if (amdgpu_vmid_had_gpu_reset(adev, id))
+		return true;
+
+	return vm_flush_needed || gds_switch_needed;
+}
+
+/**
+ * amdgpu_vm_flush - hardware flush the vm
+ *
+ * @ring: ring to use for flush
+ * @job:  related job
+ * @need_pipe_sync: is pipe sync needed
+ *
+ * Emit a VM flush when it is necessary.
+ *
+ * Returns:
+ * 0 on success, errno otherwise.
+ */
+int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, bool need_pipe_sync)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned vmhub = ring->funcs->vmhub;
+	struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub];
+	struct amdgpu_vmid *id = &id_mgr->ids[job->vmid];
+	bool gds_switch_needed = ring->funcs->emit_gds_switch && (
+		id->gds_base != job->gds_base ||
+		id->gds_size != job->gds_size ||
+		id->gws_base != job->gws_base ||
+		id->gws_size != job->gws_size ||
+		id->oa_base != job->oa_base ||
+		id->oa_size != job->oa_size);
+	bool vm_flush_needed = job->vm_needs_flush;
+	bool pasid_mapping_needed = id->pasid != job->pasid ||
+		!id->pasid_mapping ||
+		!dma_fence_is_signaled(id->pasid_mapping);
+	struct dma_fence *fence = NULL;
+	unsigned patch_offset = 0;
+	int r;
+
+	if (amdgpu_vmid_had_gpu_reset(adev, id)) {
+		gds_switch_needed = true;
+		vm_flush_needed = true;
+		pasid_mapping_needed = true;
+	}
+
+	gds_switch_needed &= !!ring->funcs->emit_gds_switch;
+	vm_flush_needed &= !!ring->funcs->emit_vm_flush  &&
+			job->vm_pd_addr != AMDGPU_BO_INVALID_OFFSET;
+	pasid_mapping_needed &= adev->gmc.gmc_funcs->emit_pasid_mapping &&
+		ring->funcs->emit_wreg;
+
+	if (!vm_flush_needed && !gds_switch_needed && !need_pipe_sync)
+		return 0;
+
+	if (ring->funcs->init_cond_exec)
+		patch_offset = amdgpu_ring_init_cond_exec(ring);
+
+	if (need_pipe_sync)
+		amdgpu_ring_emit_pipeline_sync(ring);
+
+	if (vm_flush_needed) {
+		trace_amdgpu_vm_flush(ring, job->vmid, job->vm_pd_addr);
+		amdgpu_ring_emit_vm_flush(ring, job->vmid, job->vm_pd_addr);
+	}
+
+	if (pasid_mapping_needed)
+		amdgpu_gmc_emit_pasid_mapping(ring, job->vmid, job->pasid);
+
+	if (vm_flush_needed || pasid_mapping_needed) {
+		r = amdgpu_fence_emit(ring, &fence, 0);
+		if (r)
+			return r;
+	}
+
+	if (vm_flush_needed) {
+		mutex_lock(&id_mgr->lock);
+		dma_fence_put(id->last_flush);
+		id->last_flush = dma_fence_get(fence);
+		id->current_gpu_reset_count =
+			atomic_read(&adev->gpu_reset_counter);
+		mutex_unlock(&id_mgr->lock);
+	}
+
+	if (pasid_mapping_needed) {
+		id->pasid = job->pasid;
+		dma_fence_put(id->pasid_mapping);
+		id->pasid_mapping = dma_fence_get(fence);
+	}
+	dma_fence_put(fence);
+
+	if (ring->funcs->emit_gds_switch && gds_switch_needed) {
+		id->gds_base = job->gds_base;
+		id->gds_size = job->gds_size;
+		id->gws_base = job->gws_base;
+		id->gws_size = job->gws_size;
+		id->oa_base = job->oa_base;
+		id->oa_size = job->oa_size;
+		amdgpu_ring_emit_gds_switch(ring, job->vmid, job->gds_base,
+					    job->gds_size, job->gws_base,
+					    job->gws_size, job->oa_base,
+					    job->oa_size);
+	}
+
+	if (ring->funcs->patch_cond_exec)
+		amdgpu_ring_patch_cond_exec(ring, patch_offset);
+
+	/* the double SWITCH_BUFFER here *cannot* be skipped by COND_EXEC */
+	if (ring->funcs->emit_switch_buffer) {
+		amdgpu_ring_emit_switch_buffer(ring);
+		amdgpu_ring_emit_switch_buffer(ring);
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_vm_bo_find - find the bo_va for a specific vm & bo
+ *
+ * @vm: requested vm
+ * @bo: requested buffer object
+ *
+ * Find @bo inside the requested vm.
+ * Search inside the @bos vm list for the requested vm
+ * Returns the found bo_va or NULL if none is found
+ *
+ * Object has to be reserved!
+ *
+ * Returns:
+ * Found bo_va or NULL.
+ */
+struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm,
+				       struct amdgpu_bo *bo)
+{
+	struct amdgpu_bo_va *bo_va;
+
+	list_for_each_entry(bo_va, &bo->va, base.bo_list) {
+		if (bo_va->base.vm == vm) {
+			return bo_va;
+		}
+	}
+	return NULL;
+}
+
+/**
+ * amdgpu_vm_do_set_ptes - helper to call the right asic function
+ *
+ * @params: see amdgpu_pte_update_params definition
+ * @bo: PD/PT to update
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: hw access flags
+ *
+ * Traces the parameters and calls the right asic functions
+ * to setup the page table using the DMA.
+ */
+static void amdgpu_vm_do_set_ptes(struct amdgpu_pte_update_params *params,
+				  struct amdgpu_bo *bo,
+				  uint64_t pe, uint64_t addr,
+				  unsigned count, uint32_t incr,
+				  uint64_t flags)
+{
+	pe += amdgpu_bo_gpu_offset(bo);
+	trace_amdgpu_vm_set_ptes(pe, addr, count, incr, flags);
+
+	if (count < 3) {
+		amdgpu_vm_write_pte(params->adev, params->ib, pe,
+				    addr | flags, count, incr);
+
+	} else {
+		amdgpu_vm_set_pte_pde(params->adev, params->ib, pe, addr,
+				      count, incr, flags);
+	}
+}
+
+/**
+ * amdgpu_vm_do_copy_ptes - copy the PTEs from the GART
+ *
+ * @params: see amdgpu_pte_update_params definition
+ * @bo: PD/PT to update
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: hw access flags
+ *
+ * Traces the parameters and calls the DMA function to copy the PTEs.
+ */
+static void amdgpu_vm_do_copy_ptes(struct amdgpu_pte_update_params *params,
+				   struct amdgpu_bo *bo,
+				   uint64_t pe, uint64_t addr,
+				   unsigned count, uint32_t incr,
+				   uint64_t flags)
+{
+	uint64_t src = (params->src + (addr >> 12) * 8);
+
+	pe += amdgpu_bo_gpu_offset(bo);
+	trace_amdgpu_vm_copy_ptes(pe, src, count);
+
+	amdgpu_vm_copy_pte(params->adev, params->ib, pe, src, count);
+}
+
+/**
+ * amdgpu_vm_map_gart - Resolve gart mapping of addr
+ *
+ * @pages_addr: optional DMA address to use for lookup
+ * @addr: the unmapped addr
+ *
+ * Look up the physical address of the page that the pte resolves
+ * to.
+ *
+ * Returns:
+ * The pointer for the page table entry.
+ */
+static uint64_t amdgpu_vm_map_gart(const dma_addr_t *pages_addr, uint64_t addr)
+{
+	uint64_t result;
+
+	/* page table offset */
+	result = pages_addr[addr >> PAGE_SHIFT];
+
+	/* in case cpu page size != gpu page size*/
+	result |= addr & (PAGE_MASK);
+
+	result &= 0xFFFFFFFFFFFFF000ULL;
+
+	return result;
+}
+
+/**
+ * amdgpu_vm_cpu_set_ptes - helper to update page tables via CPU
+ *
+ * @params: see amdgpu_pte_update_params definition
+ * @bo: PD/PT to update
+ * @pe: kmap addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: hw access flags
+ *
+ * Write count number of PT/PD entries directly.
+ */
+static void amdgpu_vm_cpu_set_ptes(struct amdgpu_pte_update_params *params,
+				   struct amdgpu_bo *bo,
+				   uint64_t pe, uint64_t addr,
+				   unsigned count, uint32_t incr,
+				   uint64_t flags)
+{
+	unsigned int i;
+	uint64_t value;
+
+	pe += (unsigned long)amdgpu_bo_kptr(bo);
+
+	trace_amdgpu_vm_set_ptes(pe, addr, count, incr, flags);
+
+	for (i = 0; i < count; i++) {
+		value = params->pages_addr ?
+			amdgpu_vm_map_gart(params->pages_addr, addr) :
+			addr;
+		amdgpu_gmc_set_pte_pde(params->adev, (void *)(uintptr_t)pe,
+				       i, value, flags);
+		addr += incr;
+	}
+}
+
+
+/**
+ * amdgpu_vm_wait_pd - Wait for PT BOs to be free.
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: related vm
+ * @owner: fence owner
+ *
+ * Returns:
+ * 0 on success, errno otherwise.
+ */
+static int amdgpu_vm_wait_pd(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			     void *owner)
+{
+	struct amdgpu_sync sync;
+	int r;
+
+	amdgpu_sync_create(&sync);
+	amdgpu_sync_resv(adev, &sync, vm->root.base.bo->tbo.resv, owner, false);
+	r = amdgpu_sync_wait(&sync, true);
+	amdgpu_sync_free(&sync);
+
+	return r;
+}
+
+/*
+ * amdgpu_vm_update_pde - update a single level in the hierarchy
+ *
+ * @param: parameters for the update
+ * @vm: requested vm
+ * @parent: parent directory
+ * @entry: entry to update
+ *
+ * Makes sure the requested entry in parent is up to date.
+ */
+static void amdgpu_vm_update_pde(struct amdgpu_pte_update_params *params,
+				 struct amdgpu_vm *vm,
+				 struct amdgpu_vm_pt *parent,
+				 struct amdgpu_vm_pt *entry)
+{
+	struct amdgpu_bo *bo = parent->base.bo, *pbo;
+	uint64_t pde, pt, flags;
+	unsigned level;
+
+	/* Don't update huge pages here */
+	if (entry->huge)
+		return;
+
+	for (level = 0, pbo = bo->parent; pbo; ++level)
+		pbo = pbo->parent;
+
+	level += params->adev->vm_manager.root_level;
+	pt = amdgpu_bo_gpu_offset(entry->base.bo);
+	flags = AMDGPU_PTE_VALID;
+	amdgpu_gmc_get_vm_pde(params->adev, level, &pt, &flags);
+	pde = (entry - parent->entries) * 8;
+	if (bo->shadow)
+		params->func(params, bo->shadow, pde, pt, 1, 0, flags);
+	params->func(params, bo, pde, pt, 1, 0, flags);
+}
+
+/*
+ * amdgpu_vm_invalidate_level - mark all PD levels as invalid
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: related vm
+ * @parent: parent PD
+ * @level: VMPT level
+ *
+ * Mark all PD level as invalid after an error.
+ */
+static void amdgpu_vm_invalidate_level(struct amdgpu_device *adev,
+				       struct amdgpu_vm *vm,
+				       struct amdgpu_vm_pt *parent,
+				       unsigned level)
+{
+	unsigned pt_idx, num_entries;
+
+	/*
+	 * Recurse into the subdirectories. This recursion is harmless because
+	 * we only have a maximum of 5 layers.
+	 */
+	num_entries = amdgpu_vm_num_entries(adev, level);
+	for (pt_idx = 0; pt_idx < num_entries; ++pt_idx) {
+		struct amdgpu_vm_pt *entry = &parent->entries[pt_idx];
+
+		if (!entry->base.bo)
+			continue;
+
+		if (!entry->base.moved)
+			list_move(&entry->base.vm_status, &vm->relocated);
+		amdgpu_vm_invalidate_level(adev, vm, entry, level + 1);
+	}
+}
+
+/*
+ * amdgpu_vm_update_directories - make sure that all directories are valid
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ *
+ * Makes sure all directories are up to date.
+ *
+ * Returns:
+ * 0 for success, error for failure.
+ */
+int amdgpu_vm_update_directories(struct amdgpu_device *adev,
+				 struct amdgpu_vm *vm)
+{
+	struct amdgpu_pte_update_params params;
+	struct amdgpu_job *job;
+	unsigned ndw = 0;
+	int r = 0;
+
+	if (list_empty(&vm->relocated))
+		return 0;
+
+restart:
+	memset(&params, 0, sizeof(params));
+	params.adev = adev;
+
+	if (vm->use_cpu_for_update) {
+		struct amdgpu_vm_bo_base *bo_base;
+
+		list_for_each_entry(bo_base, &vm->relocated, vm_status) {
+			r = amdgpu_bo_kmap(bo_base->bo, NULL);
+			if (unlikely(r))
+				return r;
+		}
+
+		r = amdgpu_vm_wait_pd(adev, vm, AMDGPU_FENCE_OWNER_VM);
+		if (unlikely(r))
+			return r;
+
+		params.func = amdgpu_vm_cpu_set_ptes;
+	} else {
+		ndw = 512 * 8;
+		r = amdgpu_job_alloc_with_ib(adev, ndw * 4, &job);
+		if (r)
+			return r;
+
+		params.ib = &job->ibs[0];
+		params.func = amdgpu_vm_do_set_ptes;
+	}
+
+	while (!list_empty(&vm->relocated)) {
+		struct amdgpu_vm_bo_base *bo_base, *parent;
+		struct amdgpu_vm_pt *pt, *entry;
+		struct amdgpu_bo *bo;
+
+		bo_base = list_first_entry(&vm->relocated,
+					   struct amdgpu_vm_bo_base,
+					   vm_status);
+		bo_base->moved = false;
+		list_del_init(&bo_base->vm_status);
+
+		bo = bo_base->bo->parent;
+		if (!bo)
+			continue;
+
+		parent = list_first_entry(&bo->va, struct amdgpu_vm_bo_base,
+					  bo_list);
+		pt = container_of(parent, struct amdgpu_vm_pt, base);
+		entry = container_of(bo_base, struct amdgpu_vm_pt, base);
+
+		amdgpu_vm_update_pde(&params, vm, pt, entry);
+
+		if (!vm->use_cpu_for_update &&
+		    (ndw - params.ib->length_dw) < 32)
+			break;
+	}
+
+	if (vm->use_cpu_for_update) {
+		/* Flush HDP */
+		mb();
+		amdgpu_asic_flush_hdp(adev, NULL);
+	} else if (params.ib->length_dw == 0) {
+		amdgpu_job_free(job);
+	} else {
+		struct amdgpu_bo *root = vm->root.base.bo;
+		struct amdgpu_ring *ring;
+		struct dma_fence *fence;
+
+		ring = container_of(vm->entity.rq->sched, struct amdgpu_ring,
+				    sched);
+
+		amdgpu_ring_pad_ib(ring, params.ib);
+		amdgpu_sync_resv(adev, &job->sync, root->tbo.resv,
+				 AMDGPU_FENCE_OWNER_VM, false);
+		WARN_ON(params.ib->length_dw > ndw);
+		r = amdgpu_job_submit(job, &vm->entity, AMDGPU_FENCE_OWNER_VM,
+				      &fence);
+		if (r)
+			goto error;
+
+		amdgpu_bo_fence(root, fence, true);
+		dma_fence_put(vm->last_update);
+		vm->last_update = fence;
+	}
+
+	if (!list_empty(&vm->relocated))
+		goto restart;
+
+	return 0;
+
+error:
+	amdgpu_vm_invalidate_level(adev, vm, &vm->root,
+				   adev->vm_manager.root_level);
+	amdgpu_job_free(job);
+	return r;
+}
+
+/**
+ * amdgpu_vm_find_entry - find the entry for an address
+ *
+ * @p: see amdgpu_pte_update_params definition
+ * @addr: virtual address in question
+ * @entry: resulting entry or NULL
+ * @parent: parent entry
+ *
+ * Find the vm_pt entry and it's parent for the given address.
+ */
+void amdgpu_vm_get_entry(struct amdgpu_pte_update_params *p, uint64_t addr,
+			 struct amdgpu_vm_pt **entry,
+			 struct amdgpu_vm_pt **parent)
+{
+	unsigned level = p->adev->vm_manager.root_level;
+
+	*parent = NULL;
+	*entry = &p->vm->root;
+	while ((*entry)->entries) {
+		unsigned shift = amdgpu_vm_level_shift(p->adev, level++);
+
+		*parent = *entry;
+		*entry = &(*entry)->entries[addr >> shift];
+		addr &= (1ULL << shift) - 1;
+	}
+
+	if (level != AMDGPU_VM_PTB)
+		*entry = NULL;
+}
+
+/**
+ * amdgpu_vm_handle_huge_pages - handle updating the PD with huge pages
+ *
+ * @p: see amdgpu_pte_update_params definition
+ * @entry: vm_pt entry to check
+ * @parent: parent entry
+ * @nptes: number of PTEs updated with this operation
+ * @dst: destination address where the PTEs should point to
+ * @flags: access flags fro the PTEs
+ *
+ * Check if we can update the PD with a huge page.
+ */
+static void amdgpu_vm_handle_huge_pages(struct amdgpu_pte_update_params *p,
+					struct amdgpu_vm_pt *entry,
+					struct amdgpu_vm_pt *parent,
+					unsigned nptes, uint64_t dst,
+					uint64_t flags)
+{
+	uint64_t pde;
+
+	/* In the case of a mixed PT the PDE must point to it*/
+	if (p->adev->asic_type >= CHIP_VEGA10 && !p->src &&
+	    nptes == AMDGPU_VM_PTE_COUNT(p->adev)) {
+		/* Set the huge page flag to stop scanning at this PDE */
+		flags |= AMDGPU_PDE_PTE;
+	}
+
+	if (!(flags & AMDGPU_PDE_PTE)) {
+		if (entry->huge) {
+			/* Add the entry to the relocated list to update it. */
+			entry->huge = false;
+			list_move(&entry->base.vm_status, &p->vm->relocated);
+		}
+		return;
+	}
+
+	entry->huge = true;
+	amdgpu_gmc_get_vm_pde(p->adev, AMDGPU_VM_PDB0, &dst, &flags);
+
+	pde = (entry - parent->entries) * 8;
+	if (parent->base.bo->shadow)
+		p->func(p, parent->base.bo->shadow, pde, dst, 1, 0, flags);
+	p->func(p, parent->base.bo, pde, dst, 1, 0, flags);
+}
+
+/**
+ * amdgpu_vm_update_ptes - make sure that page tables are valid
+ *
+ * @params: see amdgpu_pte_update_params definition
+ * @start: start of GPU address range
+ * @end: end of GPU address range
+ * @dst: destination address to map to, the next dst inside the function
+ * @flags: mapping flags
+ *
+ * Update the page tables in the range @start - @end.
+ *
+ * Returns:
+ * 0 for success, -EINVAL for failure.
+ */
+static int amdgpu_vm_update_ptes(struct amdgpu_pte_update_params *params,
+				  uint64_t start, uint64_t end,
+				  uint64_t dst, uint64_t flags)
+{
+	struct amdgpu_device *adev = params->adev;
+	const uint64_t mask = AMDGPU_VM_PTE_COUNT(adev) - 1;
+
+	uint64_t addr, pe_start;
+	struct amdgpu_bo *pt;
+	unsigned nptes;
+
+	/* walk over the address space and update the page tables */
+	for (addr = start; addr < end; addr += nptes,
+	     dst += nptes * AMDGPU_GPU_PAGE_SIZE) {
+		struct amdgpu_vm_pt *entry, *parent;
+
+		amdgpu_vm_get_entry(params, addr, &entry, &parent);
+		if (!entry)
+			return -ENOENT;
+
+		if ((addr & ~mask) == (end & ~mask))
+			nptes = end - addr;
+		else
+			nptes = AMDGPU_VM_PTE_COUNT(adev) - (addr & mask);
+
+		amdgpu_vm_handle_huge_pages(params, entry, parent,
+					    nptes, dst, flags);
+		/* We don't need to update PTEs for huge pages */
+		if (entry->huge)
+			continue;
+
+		pt = entry->base.bo;
+		pe_start = (addr & mask) * 8;
+		if (pt->shadow)
+			params->func(params, pt->shadow, pe_start, dst, nptes,
+				     AMDGPU_GPU_PAGE_SIZE, flags);
+		params->func(params, pt, pe_start, dst, nptes,
+			     AMDGPU_GPU_PAGE_SIZE, flags);
+	}
+
+	return 0;
+}
+
+/*
+ * amdgpu_vm_frag_ptes - add fragment information to PTEs
+ *
+ * @params: see amdgpu_pte_update_params definition
+ * @vm: requested vm
+ * @start: first PTE to handle
+ * @end: last PTE to handle
+ * @dst: addr those PTEs should point to
+ * @flags: hw mapping flags
+ *
+ * Returns:
+ * 0 for success, -EINVAL for failure.
+ */
+static int amdgpu_vm_frag_ptes(struct amdgpu_pte_update_params	*params,
+				uint64_t start, uint64_t end,
+				uint64_t dst, uint64_t flags)
+{
+	/**
+	 * The MC L1 TLB supports variable sized pages, based on a fragment
+	 * field in the PTE. When this field is set to a non-zero value, page
+	 * granularity is increased from 4KB to (1 << (12 + frag)). The PTE
+	 * flags are considered valid for all PTEs within the fragment range
+	 * and corresponding mappings are assumed to be physically contiguous.
+	 *
+	 * The L1 TLB can store a single PTE for the whole fragment,
+	 * significantly increasing the space available for translation
+	 * caching. This leads to large improvements in throughput when the
+	 * TLB is under pressure.
+	 *
+	 * The L2 TLB distributes small and large fragments into two
+	 * asymmetric partitions. The large fragment cache is significantly
+	 * larger. Thus, we try to use large fragments wherever possible.
+	 * Userspace can support this by aligning virtual base address and
+	 * allocation size to the fragment size.
+	 */
+	unsigned max_frag = params->adev->vm_manager.fragment_size;
+	int r;
+
+	/* system pages are non continuously */
+	if (params->src || !(flags & AMDGPU_PTE_VALID))
+		return amdgpu_vm_update_ptes(params, start, end, dst, flags);
+
+	while (start != end) {
+		uint64_t frag_flags, frag_end;
+		unsigned frag;
+
+		/* This intentionally wraps around if no bit is set */
+		frag = min((unsigned)ffs(start) - 1,
+			   (unsigned)fls64(end - start) - 1);
+		if (frag >= max_frag) {
+			frag_flags = AMDGPU_PTE_FRAG(max_frag);
+			frag_end = end & ~((1ULL << max_frag) - 1);
+		} else {
+			frag_flags = AMDGPU_PTE_FRAG(frag);
+			frag_end = start + (1 << frag);
+		}
+
+		r = amdgpu_vm_update_ptes(params, start, frag_end, dst,
+					  flags | frag_flags);
+		if (r)
+			return r;
+
+		dst += (frag_end - start) * AMDGPU_GPU_PAGE_SIZE;
+		start = frag_end;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_bo_update_mapping - update a mapping in the vm page table
+ *
+ * @adev: amdgpu_device pointer
+ * @exclusive: fence we need to sync to
+ * @pages_addr: DMA addresses to use for mapping
+ * @vm: requested vm
+ * @start: start of mapped range
+ * @last: last mapped entry
+ * @flags: flags for the entries
+ * @addr: addr to set the area to
+ * @fence: optional resulting fence
+ *
+ * Fill in the page table entries between @start and @last.
+ *
+ * Returns:
+ * 0 for success, -EINVAL for failure.
+ */
+static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
+				       struct dma_fence *exclusive,
+				       dma_addr_t *pages_addr,
+				       struct amdgpu_vm *vm,
+				       uint64_t start, uint64_t last,
+				       uint64_t flags, uint64_t addr,
+				       struct dma_fence **fence)
+{
+	struct amdgpu_ring *ring;
+	void *owner = AMDGPU_FENCE_OWNER_VM;
+	unsigned nptes, ncmds, ndw;
+	struct amdgpu_job *job;
+	struct amdgpu_pte_update_params params;
+	struct dma_fence *f = NULL;
+	int r;
+
+	memset(&params, 0, sizeof(params));
+	params.adev = adev;
+	params.vm = vm;
+
+	/* sync to everything on unmapping */
+	if (!(flags & AMDGPU_PTE_VALID))
+		owner = AMDGPU_FENCE_OWNER_UNDEFINED;
+
+	if (vm->use_cpu_for_update) {
+		/* params.src is used as flag to indicate system Memory */
+		if (pages_addr)
+			params.src = ~0;
+
+		/* Wait for PT BOs to be free. PTs share the same resv. object
+		 * as the root PD BO
+		 */
+		r = amdgpu_vm_wait_pd(adev, vm, owner);
+		if (unlikely(r))
+			return r;
+
+		params.func = amdgpu_vm_cpu_set_ptes;
+		params.pages_addr = pages_addr;
+		return amdgpu_vm_frag_ptes(&params, start, last + 1,
+					   addr, flags);
+	}
+
+	ring = container_of(vm->entity.rq->sched, struct amdgpu_ring, sched);
+
+	nptes = last - start + 1;
+
+	/*
+	 * reserve space for two commands every (1 << BLOCK_SIZE)
+	 *  entries or 2k dwords (whatever is smaller)
+         *
+         * The second command is for the shadow pagetables.
+	 */
+	if (vm->root.base.bo->shadow)
+		ncmds = ((nptes >> min(adev->vm_manager.block_size, 11u)) + 1) * 2;
+	else
+		ncmds = ((nptes >> min(adev->vm_manager.block_size, 11u)) + 1);
+
+	/* padding, etc. */
+	ndw = 64;
+
+	if (pages_addr) {
+		/* copy commands needed */
+		ndw += ncmds * adev->vm_manager.vm_pte_funcs->copy_pte_num_dw;
+
+		/* and also PTEs */
+		ndw += nptes * 2;
+
+		params.func = amdgpu_vm_do_copy_ptes;
+
+	} else {
+		/* set page commands needed */
+		ndw += ncmds * 10;
+
+		/* extra commands for begin/end fragments */
+		if (vm->root.base.bo->shadow)
+		        ndw += 2 * 10 * adev->vm_manager.fragment_size * 2;
+		else
+		        ndw += 2 * 10 * adev->vm_manager.fragment_size;
+
+		params.func = amdgpu_vm_do_set_ptes;
+	}
+
+	r = amdgpu_job_alloc_with_ib(adev, ndw * 4, &job);
+	if (r)
+		return r;
+
+	params.ib = &job->ibs[0];
+
+	if (pages_addr) {
+		uint64_t *pte;
+		unsigned i;
+
+		/* Put the PTEs at the end of the IB. */
+		i = ndw - nptes * 2;
+		pte= (uint64_t *)&(job->ibs->ptr[i]);
+		params.src = job->ibs->gpu_addr + i * 4;
+
+		for (i = 0; i < nptes; ++i) {
+			pte[i] = amdgpu_vm_map_gart(pages_addr, addr + i *
+						    AMDGPU_GPU_PAGE_SIZE);
+			pte[i] |= flags;
+		}
+		addr = 0;
+	}
+
+	r = amdgpu_sync_fence(adev, &job->sync, exclusive, false);
+	if (r)
+		goto error_free;
+
+	r = amdgpu_sync_resv(adev, &job->sync, vm->root.base.bo->tbo.resv,
+			     owner, false);
+	if (r)
+		goto error_free;
+
+	r = reservation_object_reserve_shared(vm->root.base.bo->tbo.resv);
+	if (r)
+		goto error_free;
+
+	r = amdgpu_vm_frag_ptes(&params, start, last + 1, addr, flags);
+	if (r)
+		goto error_free;
+
+	amdgpu_ring_pad_ib(ring, params.ib);
+	WARN_ON(params.ib->length_dw > ndw);
+	r = amdgpu_job_submit(job, &vm->entity, AMDGPU_FENCE_OWNER_VM, &f);
+	if (r)
+		goto error_free;
+
+	amdgpu_bo_fence(vm->root.base.bo, f, true);
+	dma_fence_put(*fence);
+	*fence = f;
+	return 0;
+
+error_free:
+	amdgpu_job_free(job);
+	return r;
+}
+
+/**
+ * amdgpu_vm_bo_split_mapping - split a mapping into smaller chunks
+ *
+ * @adev: amdgpu_device pointer
+ * @exclusive: fence we need to sync to
+ * @pages_addr: DMA addresses to use for mapping
+ * @vm: requested vm
+ * @mapping: mapped range and flags to use for the update
+ * @flags: HW flags for the mapping
+ * @nodes: array of drm_mm_nodes with the MC addresses
+ * @fence: optional resulting fence
+ *
+ * Split the mapping into smaller chunks so that each update fits
+ * into a SDMA IB.
+ *
+ * Returns:
+ * 0 for success, -EINVAL for failure.
+ */
+static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev,
+				      struct dma_fence *exclusive,
+				      dma_addr_t *pages_addr,
+				      struct amdgpu_vm *vm,
+				      struct amdgpu_bo_va_mapping *mapping,
+				      uint64_t flags,
+				      struct drm_mm_node *nodes,
+				      struct dma_fence **fence)
+{
+	unsigned min_linear_pages = 1 << adev->vm_manager.fragment_size;
+	uint64_t pfn, start = mapping->start;
+	int r;
+
+	/* normally,bo_va->flags only contians READABLE and WIRTEABLE bit go here
+	 * but in case of something, we filter the flags in first place
+	 */
+	if (!(mapping->flags & AMDGPU_PTE_READABLE))
+		flags &= ~AMDGPU_PTE_READABLE;
+	if (!(mapping->flags & AMDGPU_PTE_WRITEABLE))
+		flags &= ~AMDGPU_PTE_WRITEABLE;
+
+	flags &= ~AMDGPU_PTE_EXECUTABLE;
+	flags |= mapping->flags & AMDGPU_PTE_EXECUTABLE;
+
+	flags &= ~AMDGPU_PTE_MTYPE_MASK;
+	flags |= (mapping->flags & AMDGPU_PTE_MTYPE_MASK);
+
+	if ((mapping->flags & AMDGPU_PTE_PRT) &&
+	    (adev->asic_type >= CHIP_VEGA10)) {
+		flags |= AMDGPU_PTE_PRT;
+		flags &= ~AMDGPU_PTE_VALID;
+	}
+
+	trace_amdgpu_vm_bo_update(mapping);
+
+	pfn = mapping->offset >> PAGE_SHIFT;
+	if (nodes) {
+		while (pfn >= nodes->size) {
+			pfn -= nodes->size;
+			++nodes;
+		}
+	}
+
+	do {
+		dma_addr_t *dma_addr = NULL;
+		uint64_t max_entries;
+		uint64_t addr, last;
+
+		if (nodes) {
+			addr = nodes->start << PAGE_SHIFT;
+			max_entries = (nodes->size - pfn) *
+				AMDGPU_GPU_PAGES_IN_CPU_PAGE;
+		} else {
+			addr = 0;
+			max_entries = S64_MAX;
+		}
+
+		if (pages_addr) {
+			uint64_t count;
+
+			max_entries = min(max_entries, 16ull * 1024ull);
+			for (count = 1;
+			     count < max_entries / AMDGPU_GPU_PAGES_IN_CPU_PAGE;
+			     ++count) {
+				uint64_t idx = pfn + count;
+
+				if (pages_addr[idx] !=
+				    (pages_addr[idx - 1] + PAGE_SIZE))
+					break;
+			}
+
+			if (count < min_linear_pages) {
+				addr = pfn << PAGE_SHIFT;
+				dma_addr = pages_addr;
+			} else {
+				addr = pages_addr[pfn];
+				max_entries = count * AMDGPU_GPU_PAGES_IN_CPU_PAGE;
+			}
+
+		} else if (flags & AMDGPU_PTE_VALID) {
+			addr += adev->vm_manager.vram_base_offset;
+			addr += pfn << PAGE_SHIFT;
+		}
+
+		last = min((uint64_t)mapping->last, start + max_entries - 1);
+		r = amdgpu_vm_bo_update_mapping(adev, exclusive, dma_addr, vm,
+						start, last, flags, addr,
+						fence);
+		if (r)
+			return r;
+
+		pfn += (last - start + 1) / AMDGPU_GPU_PAGES_IN_CPU_PAGE;
+		if (nodes && nodes->size == pfn) {
+			pfn = 0;
+			++nodes;
+		}
+		start = last + 1;
+
+	} while (unlikely(start != mapping->last + 1));
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_bo_update - update all BO mappings in the vm page table
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: requested BO and VM object
+ * @clear: if true clear the entries
+ *
+ * Fill in the page table entries for @bo_va.
+ *
+ * Returns:
+ * 0 for success, -EINVAL for failure.
+ */
+int amdgpu_vm_bo_update(struct amdgpu_device *adev,
+			struct amdgpu_bo_va *bo_va,
+			bool clear)
+{
+	struct amdgpu_bo *bo = bo_va->base.bo;
+	struct amdgpu_vm *vm = bo_va->base.vm;
+	struct amdgpu_bo_va_mapping *mapping;
+	dma_addr_t *pages_addr = NULL;
+	struct ttm_mem_reg *mem;
+	struct drm_mm_node *nodes;
+	struct dma_fence *exclusive, **last_update;
+	uint64_t flags;
+	int r;
+
+	if (clear || !bo) {
+		mem = NULL;
+		nodes = NULL;
+		exclusive = NULL;
+	} else {
+		struct ttm_dma_tt *ttm;
+
+		mem = &bo->tbo.mem;
+		nodes = mem->mm_node;
+		if (mem->mem_type == TTM_PL_TT) {
+			ttm = container_of(bo->tbo.ttm, struct ttm_dma_tt, ttm);
+			pages_addr = ttm->dma_address;
+		}
+		exclusive = reservation_object_get_excl(bo->tbo.resv);
+	}
+
+	if (bo)
+		flags = amdgpu_ttm_tt_pte_flags(adev, bo->tbo.ttm, mem);
+	else
+		flags = 0x0;
+
+	if (clear || (bo && bo->tbo.resv == vm->root.base.bo->tbo.resv))
+		last_update = &vm->last_update;
+	else
+		last_update = &bo_va->last_pt_update;
+
+	if (!clear && bo_va->base.moved) {
+		bo_va->base.moved = false;
+		list_splice_init(&bo_va->valids, &bo_va->invalids);
+
+	} else if (bo_va->cleared != clear) {
+		list_splice_init(&bo_va->valids, &bo_va->invalids);
+	}
+
+	list_for_each_entry(mapping, &bo_va->invalids, list) {
+		r = amdgpu_vm_bo_split_mapping(adev, exclusive, pages_addr, vm,
+					       mapping, flags, nodes,
+					       last_update);
+		if (r)
+			return r;
+	}
+
+	if (vm->use_cpu_for_update) {
+		/* Flush HDP */
+		mb();
+		amdgpu_asic_flush_hdp(adev, NULL);
+	}
+
+	spin_lock(&vm->moved_lock);
+	list_del_init(&bo_va->base.vm_status);
+	spin_unlock(&vm->moved_lock);
+
+	/* If the BO is not in its preferred location add it back to
+	 * the evicted list so that it gets validated again on the
+	 * next command submission.
+	 */
+	if (bo && bo->tbo.resv == vm->root.base.bo->tbo.resv) {
+		uint32_t mem_type = bo->tbo.mem.mem_type;
+
+		if (!(bo->preferred_domains & amdgpu_mem_type_to_domain(mem_type)))
+			list_add_tail(&bo_va->base.vm_status, &vm->evicted);
+		else
+			list_add(&bo_va->base.vm_status, &vm->idle);
+	}
+
+	list_splice_init(&bo_va->invalids, &bo_va->valids);
+	bo_va->cleared = clear;
+
+	if (trace_amdgpu_vm_bo_mapping_enabled()) {
+		list_for_each_entry(mapping, &bo_va->valids, list)
+			trace_amdgpu_vm_bo_mapping(mapping);
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_update_prt_state - update the global PRT state
+ *
+ * @adev: amdgpu_device pointer
+ */
+static void amdgpu_vm_update_prt_state(struct amdgpu_device *adev)
+{
+	unsigned long flags;
+	bool enable;
+
+	spin_lock_irqsave(&adev->vm_manager.prt_lock, flags);
+	enable = !!atomic_read(&adev->vm_manager.num_prt_users);
+	adev->gmc.gmc_funcs->set_prt(adev, enable);
+	spin_unlock_irqrestore(&adev->vm_manager.prt_lock, flags);
+}
+
+/**
+ * amdgpu_vm_prt_get - add a PRT user
+ *
+ * @adev: amdgpu_device pointer
+ */
+static void amdgpu_vm_prt_get(struct amdgpu_device *adev)
+{
+	if (!adev->gmc.gmc_funcs->set_prt)
+		return;
+
+	if (atomic_inc_return(&adev->vm_manager.num_prt_users) == 1)
+		amdgpu_vm_update_prt_state(adev);
+}
+
+/**
+ * amdgpu_vm_prt_put - drop a PRT user
+ *
+ * @adev: amdgpu_device pointer
+ */
+static void amdgpu_vm_prt_put(struct amdgpu_device *adev)
+{
+	if (atomic_dec_return(&adev->vm_manager.num_prt_users) == 0)
+		amdgpu_vm_update_prt_state(adev);
+}
+
+/**
+ * amdgpu_vm_prt_cb - callback for updating the PRT status
+ *
+ * @fence: fence for the callback
+ * @_cb: the callback function
+ */
+static void amdgpu_vm_prt_cb(struct dma_fence *fence, struct dma_fence_cb *_cb)
+{
+	struct amdgpu_prt_cb *cb = container_of(_cb, struct amdgpu_prt_cb, cb);
+
+	amdgpu_vm_prt_put(cb->adev);
+	kfree(cb);
+}
+
+/**
+ * amdgpu_vm_add_prt_cb - add callback for updating the PRT status
+ *
+ * @adev: amdgpu_device pointer
+ * @fence: fence for the callback
+ */
+static void amdgpu_vm_add_prt_cb(struct amdgpu_device *adev,
+				 struct dma_fence *fence)
+{
+	struct amdgpu_prt_cb *cb;
+
+	if (!adev->gmc.gmc_funcs->set_prt)
+		return;
+
+	cb = kmalloc(sizeof(struct amdgpu_prt_cb), GFP_KERNEL);
+	if (!cb) {
+		/* Last resort when we are OOM */
+		if (fence)
+			dma_fence_wait(fence, false);
+
+		amdgpu_vm_prt_put(adev);
+	} else {
+		cb->adev = adev;
+		if (!fence || dma_fence_add_callback(fence, &cb->cb,
+						     amdgpu_vm_prt_cb))
+			amdgpu_vm_prt_cb(fence, &cb->cb);
+	}
+}
+
+/**
+ * amdgpu_vm_free_mapping - free a mapping
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ * @mapping: mapping to be freed
+ * @fence: fence of the unmap operation
+ *
+ * Free a mapping and make sure we decrease the PRT usage count if applicable.
+ */
+static void amdgpu_vm_free_mapping(struct amdgpu_device *adev,
+				   struct amdgpu_vm *vm,
+				   struct amdgpu_bo_va_mapping *mapping,
+				   struct dma_fence *fence)
+{
+	if (mapping->flags & AMDGPU_PTE_PRT)
+		amdgpu_vm_add_prt_cb(adev, fence);
+	kfree(mapping);
+}
+
+/**
+ * amdgpu_vm_prt_fini - finish all prt mappings
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ *
+ * Register a cleanup callback to disable PRT support after VM dies.
+ */
+static void amdgpu_vm_prt_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
+{
+	struct reservation_object *resv = vm->root.base.bo->tbo.resv;
+	struct dma_fence *excl, **shared;
+	unsigned i, shared_count;
+	int r;
+
+	r = reservation_object_get_fences_rcu(resv, &excl,
+					      &shared_count, &shared);
+	if (r) {
+		/* Not enough memory to grab the fence list, as last resort
+		 * block for all the fences to complete.
+		 */
+		reservation_object_wait_timeout_rcu(resv, true, false,
+						    MAX_SCHEDULE_TIMEOUT);
+		return;
+	}
+
+	/* Add a callback for each fence in the reservation object */
+	amdgpu_vm_prt_get(adev);
+	amdgpu_vm_add_prt_cb(adev, excl);
+
+	for (i = 0; i < shared_count; ++i) {
+		amdgpu_vm_prt_get(adev);
+		amdgpu_vm_add_prt_cb(adev, shared[i]);
+	}
+
+	kfree(shared);
+}
+
+/**
+ * amdgpu_vm_clear_freed - clear freed BOs in the PT
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ * @fence: optional resulting fence (unchanged if no work needed to be done
+ * or if an error occurred)
+ *
+ * Make sure all freed BOs are cleared in the PT.
+ * PTs have to be reserved and mutex must be locked!
+ *
+ * Returns:
+ * 0 for success.
+ *
+ */
+int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
+			  struct amdgpu_vm *vm,
+			  struct dma_fence **fence)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	uint64_t init_pte_value = 0;
+	struct dma_fence *f = NULL;
+	int r;
+
+	while (!list_empty(&vm->freed)) {
+		mapping = list_first_entry(&vm->freed,
+			struct amdgpu_bo_va_mapping, list);
+		list_del(&mapping->list);
+
+		if (vm->pte_support_ats && mapping->start < AMDGPU_VA_HOLE_START)
+			init_pte_value = AMDGPU_PTE_DEFAULT_ATC;
+
+		r = amdgpu_vm_bo_update_mapping(adev, NULL, NULL, vm,
+						mapping->start, mapping->last,
+						init_pte_value, 0, &f);
+		amdgpu_vm_free_mapping(adev, vm, mapping, f);
+		if (r) {
+			dma_fence_put(f);
+			return r;
+		}
+	}
+
+	if (fence && f) {
+		dma_fence_put(*fence);
+		*fence = f;
+	} else {
+		dma_fence_put(f);
+	}
+
+	return 0;
+
+}
+
+/**
+ * amdgpu_vm_handle_moved - handle moved BOs in the PT
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ *
+ * Make sure all BOs which are moved are updated in the PTs.
+ *
+ * Returns:
+ * 0 for success.
+ *
+ * PTs have to be reserved!
+ */
+int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
+			   struct amdgpu_vm *vm)
+{
+	struct amdgpu_bo_va *bo_va, *tmp;
+	struct list_head moved;
+	bool clear;
+	int r;
+
+	INIT_LIST_HEAD(&moved);
+	spin_lock(&vm->moved_lock);
+	list_splice_init(&vm->moved, &moved);
+	spin_unlock(&vm->moved_lock);
+
+	list_for_each_entry_safe(bo_va, tmp, &moved, base.vm_status) {
+		struct reservation_object *resv = bo_va->base.bo->tbo.resv;
+
+		/* Per VM BOs never need to bo cleared in the page tables */
+		if (resv == vm->root.base.bo->tbo.resv)
+			clear = false;
+		/* Try to reserve the BO to avoid clearing its ptes */
+		else if (!amdgpu_vm_debug && reservation_object_trylock(resv))
+			clear = false;
+		/* Somebody else is using the BO right now */
+		else
+			clear = true;
+
+		r = amdgpu_vm_bo_update(adev, bo_va, clear);
+		if (r) {
+			spin_lock(&vm->moved_lock);
+			list_splice(&moved, &vm->moved);
+			spin_unlock(&vm->moved_lock);
+			return r;
+		}
+
+		if (!clear && resv != vm->root.base.bo->tbo.resv)
+			reservation_object_unlock(resv);
+
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_bo_add - add a bo to a specific vm
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ * @bo: amdgpu buffer object
+ *
+ * Add @bo into the requested vm.
+ * Add @bo to the list of bos associated with the vm
+ *
+ * Returns:
+ * Newly added bo_va or NULL for failure
+ *
+ * Object has to be reserved!
+ */
+struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
+				      struct amdgpu_vm *vm,
+				      struct amdgpu_bo *bo)
+{
+	struct amdgpu_bo_va *bo_va;
+
+	bo_va = kzalloc(sizeof(struct amdgpu_bo_va), GFP_KERNEL);
+	if (bo_va == NULL) {
+		return NULL;
+	}
+	amdgpu_vm_bo_base_init(&bo_va->base, vm, bo);
+
+	bo_va->ref_count = 1;
+	INIT_LIST_HEAD(&bo_va->valids);
+	INIT_LIST_HEAD(&bo_va->invalids);
+
+	return bo_va;
+}
+
+
+/**
+ * amdgpu_vm_bo_insert_mapping - insert a new mapping
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: bo_va to store the address
+ * @mapping: the mapping to insert
+ *
+ * Insert a new mapping into all structures.
+ */
+static void amdgpu_vm_bo_insert_map(struct amdgpu_device *adev,
+				    struct amdgpu_bo_va *bo_va,
+				    struct amdgpu_bo_va_mapping *mapping)
+{
+	struct amdgpu_vm *vm = bo_va->base.vm;
+	struct amdgpu_bo *bo = bo_va->base.bo;
+
+	mapping->bo_va = bo_va;
+	list_add(&mapping->list, &bo_va->invalids);
+	amdgpu_vm_it_insert(mapping, &vm->va);
+
+	if (mapping->flags & AMDGPU_PTE_PRT)
+		amdgpu_vm_prt_get(adev);
+
+	if (bo && bo->tbo.resv == vm->root.base.bo->tbo.resv &&
+	    !bo_va->base.moved) {
+		spin_lock(&vm->moved_lock);
+		list_move(&bo_va->base.vm_status, &vm->moved);
+		spin_unlock(&vm->moved_lock);
+	}
+	trace_amdgpu_vm_bo_map(bo_va, mapping);
+}
+
+/**
+ * amdgpu_vm_bo_map - map bo inside a vm
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: bo_va to store the address
+ * @saddr: where to map the BO
+ * @offset: requested offset in the BO
+ * @size: BO size in bytes
+ * @flags: attributes of pages (read/write/valid/etc.)
+ *
+ * Add a mapping of the BO at the specefied addr into the VM.
+ *
+ * Returns:
+ * 0 for success, error for failure.
+ *
+ * Object has to be reserved and unreserved outside!
+ */
+int amdgpu_vm_bo_map(struct amdgpu_device *adev,
+		     struct amdgpu_bo_va *bo_va,
+		     uint64_t saddr, uint64_t offset,
+		     uint64_t size, uint64_t flags)
+{
+	struct amdgpu_bo_va_mapping *mapping, *tmp;
+	struct amdgpu_bo *bo = bo_va->base.bo;
+	struct amdgpu_vm *vm = bo_va->base.vm;
+	uint64_t eaddr;
+
+	/* validate the parameters */
+	if (saddr & AMDGPU_GPU_PAGE_MASK || offset & AMDGPU_GPU_PAGE_MASK ||
+	    size == 0 || size & AMDGPU_GPU_PAGE_MASK)
+		return -EINVAL;
+
+	/* make sure object fit at this offset */
+	eaddr = saddr + size - 1;
+	if (saddr >= eaddr ||
+	    (bo && offset + size > amdgpu_bo_size(bo)))
+		return -EINVAL;
+
+	saddr /= AMDGPU_GPU_PAGE_SIZE;
+	eaddr /= AMDGPU_GPU_PAGE_SIZE;
+
+	tmp = amdgpu_vm_it_iter_first(&vm->va, saddr, eaddr);
+	if (tmp) {
+		/* bo and tmp overlap, invalid addr */
+		dev_err(adev->dev, "bo %p va 0x%010Lx-0x%010Lx conflict with "
+			"0x%010Lx-0x%010Lx\n", bo, saddr, eaddr,
+			tmp->start, tmp->last + 1);
+		return -EINVAL;
+	}
+
+	mapping = kmalloc(sizeof(*mapping), GFP_KERNEL);
+	if (!mapping)
+		return -ENOMEM;
+
+	mapping->start = saddr;
+	mapping->last = eaddr;
+	mapping->offset = offset;
+	mapping->flags = flags;
+
+	amdgpu_vm_bo_insert_map(adev, bo_va, mapping);
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_bo_replace_map - map bo inside a vm, replacing existing mappings
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: bo_va to store the address
+ * @saddr: where to map the BO
+ * @offset: requested offset in the BO
+ * @size: BO size in bytes
+ * @flags: attributes of pages (read/write/valid/etc.)
+ *
+ * Add a mapping of the BO at the specefied addr into the VM. Replace existing
+ * mappings as we do so.
+ *
+ * Returns:
+ * 0 for success, error for failure.
+ *
+ * Object has to be reserved and unreserved outside!
+ */
+int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
+			     struct amdgpu_bo_va *bo_va,
+			     uint64_t saddr, uint64_t offset,
+			     uint64_t size, uint64_t flags)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_bo *bo = bo_va->base.bo;
+	uint64_t eaddr;
+	int r;
+
+	/* validate the parameters */
+	if (saddr & AMDGPU_GPU_PAGE_MASK || offset & AMDGPU_GPU_PAGE_MASK ||
+	    size == 0 || size & AMDGPU_GPU_PAGE_MASK)
+		return -EINVAL;
+
+	/* make sure object fit at this offset */
+	eaddr = saddr + size - 1;
+	if (saddr >= eaddr ||
+	    (bo && offset + size > amdgpu_bo_size(bo)))
+		return -EINVAL;
+
+	/* Allocate all the needed memory */
+	mapping = kmalloc(sizeof(*mapping), GFP_KERNEL);
+	if (!mapping)
+		return -ENOMEM;
+
+	r = amdgpu_vm_bo_clear_mappings(adev, bo_va->base.vm, saddr, size);
+	if (r) {
+		kfree(mapping);
+		return r;
+	}
+
+	saddr /= AMDGPU_GPU_PAGE_SIZE;
+	eaddr /= AMDGPU_GPU_PAGE_SIZE;
+
+	mapping->start = saddr;
+	mapping->last = eaddr;
+	mapping->offset = offset;
+	mapping->flags = flags;
+
+	amdgpu_vm_bo_insert_map(adev, bo_va, mapping);
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_bo_unmap - remove bo mapping from vm
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: bo_va to remove the address from
+ * @saddr: where to the BO is mapped
+ *
+ * Remove a mapping of the BO at the specefied addr from the VM.
+ *
+ * Returns:
+ * 0 for success, error for failure.
+ *
+ * Object has to be reserved and unreserved outside!
+ */
+int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
+		       struct amdgpu_bo_va *bo_va,
+		       uint64_t saddr)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_vm *vm = bo_va->base.vm;
+	bool valid = true;
+
+	saddr /= AMDGPU_GPU_PAGE_SIZE;
+
+	list_for_each_entry(mapping, &bo_va->valids, list) {
+		if (mapping->start == saddr)
+			break;
+	}
+
+	if (&mapping->list == &bo_va->valids) {
+		valid = false;
+
+		list_for_each_entry(mapping, &bo_va->invalids, list) {
+			if (mapping->start == saddr)
+				break;
+		}
+
+		if (&mapping->list == &bo_va->invalids)
+			return -ENOENT;
+	}
+
+	list_del(&mapping->list);
+	amdgpu_vm_it_remove(mapping, &vm->va);
+	mapping->bo_va = NULL;
+	trace_amdgpu_vm_bo_unmap(bo_va, mapping);
+
+	if (valid)
+		list_add(&mapping->list, &vm->freed);
+	else
+		amdgpu_vm_free_mapping(adev, vm, mapping,
+				       bo_va->last_pt_update);
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_bo_clear_mappings - remove all mappings in a specific range
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: VM structure to use
+ * @saddr: start of the range
+ * @size: size of the range
+ *
+ * Remove all mappings in a range, split them as appropriate.
+ *
+ * Returns:
+ * 0 for success, error for failure.
+ */
+int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm,
+				uint64_t saddr, uint64_t size)
+{
+	struct amdgpu_bo_va_mapping *before, *after, *tmp, *next;
+	DRM_LIST_HEAD(removed);
+	uint64_t eaddr;
+
+	eaddr = saddr + size - 1;
+	saddr /= AMDGPU_GPU_PAGE_SIZE;
+	eaddr /= AMDGPU_GPU_PAGE_SIZE;
+
+	/* Allocate all the needed memory */
+	before = kzalloc(sizeof(*before), GFP_KERNEL);
+	if (!before)
+		return -ENOMEM;
+	INIT_LIST_HEAD(&before->list);
+
+	after = kzalloc(sizeof(*after), GFP_KERNEL);
+	if (!after) {
+		kfree(before);
+		return -ENOMEM;
+	}
+	INIT_LIST_HEAD(&after->list);
+
+	/* Now gather all removed mappings */
+	tmp = amdgpu_vm_it_iter_first(&vm->va, saddr, eaddr);
+	while (tmp) {
+		/* Remember mapping split at the start */
+		if (tmp->start < saddr) {
+			before->start = tmp->start;
+			before->last = saddr - 1;
+			before->offset = tmp->offset;
+			before->flags = tmp->flags;
+			before->bo_va = tmp->bo_va;
+			list_add(&before->list, &tmp->bo_va->invalids);
+		}
+
+		/* Remember mapping split at the end */
+		if (tmp->last > eaddr) {
+			after->start = eaddr + 1;
+			after->last = tmp->last;
+			after->offset = tmp->offset;
+			after->offset += after->start - tmp->start;
+			after->flags = tmp->flags;
+			after->bo_va = tmp->bo_va;
+			list_add(&after->list, &tmp->bo_va->invalids);
+		}
+
+		list_del(&tmp->list);
+		list_add(&tmp->list, &removed);
+
+		tmp = amdgpu_vm_it_iter_next(tmp, saddr, eaddr);
+	}
+
+	/* And free them up */
+	list_for_each_entry_safe(tmp, next, &removed, list) {
+		amdgpu_vm_it_remove(tmp, &vm->va);
+		list_del(&tmp->list);
+
+		if (tmp->start < saddr)
+		    tmp->start = saddr;
+		if (tmp->last > eaddr)
+		    tmp->last = eaddr;
+
+		tmp->bo_va = NULL;
+		list_add(&tmp->list, &vm->freed);
+		trace_amdgpu_vm_bo_unmap(NULL, tmp);
+	}
+
+	/* Insert partial mapping before the range */
+	if (!list_empty(&before->list)) {
+		amdgpu_vm_it_insert(before, &vm->va);
+		if (before->flags & AMDGPU_PTE_PRT)
+			amdgpu_vm_prt_get(adev);
+	} else {
+		kfree(before);
+	}
+
+	/* Insert partial mapping after the range */
+	if (!list_empty(&after->list)) {
+		amdgpu_vm_it_insert(after, &vm->va);
+		if (after->flags & AMDGPU_PTE_PRT)
+			amdgpu_vm_prt_get(adev);
+	} else {
+		kfree(after);
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_bo_lookup_mapping - find mapping by address
+ *
+ * @vm: the requested VM
+ * @addr: the address
+ *
+ * Find a mapping by it's address.
+ *
+ * Returns:
+ * The amdgpu_bo_va_mapping matching for addr or NULL
+ *
+ */
+struct amdgpu_bo_va_mapping *amdgpu_vm_bo_lookup_mapping(struct amdgpu_vm *vm,
+							 uint64_t addr)
+{
+	return amdgpu_vm_it_iter_first(&vm->va, addr, addr);
+}
+
+/**
+ * amdgpu_vm_bo_trace_cs - trace all reserved mappings
+ *
+ * @vm: the requested vm
+ * @ticket: CS ticket
+ *
+ * Trace all mappings of BOs reserved during a command submission.
+ */
+void amdgpu_vm_bo_trace_cs(struct amdgpu_vm *vm, struct ww_acquire_ctx *ticket)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+
+	if (!trace_amdgpu_vm_bo_cs_enabled())
+		return;
+
+	for (mapping = amdgpu_vm_it_iter_first(&vm->va, 0, U64_MAX); mapping;
+	     mapping = amdgpu_vm_it_iter_next(mapping, 0, U64_MAX)) {
+		if (mapping->bo_va && mapping->bo_va->base.bo) {
+			struct amdgpu_bo *bo;
+
+			bo = mapping->bo_va->base.bo;
+			if (READ_ONCE(bo->tbo.resv->lock.ctx) != ticket)
+				continue;
+		}
+
+		trace_amdgpu_vm_bo_cs(mapping);
+	}
+}
+
+/**
+ * amdgpu_vm_bo_rmv - remove a bo to a specific vm
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: requested bo_va
+ *
+ * Remove @bo_va->bo from the requested vm.
+ *
+ * Object have to be reserved!
+ */
+void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
+		      struct amdgpu_bo_va *bo_va)
+{
+	struct amdgpu_bo_va_mapping *mapping, *next;
+	struct amdgpu_vm *vm = bo_va->base.vm;
+
+	list_del(&bo_va->base.bo_list);
+
+	spin_lock(&vm->moved_lock);
+	list_del(&bo_va->base.vm_status);
+	spin_unlock(&vm->moved_lock);
+
+	list_for_each_entry_safe(mapping, next, &bo_va->valids, list) {
+		list_del(&mapping->list);
+		amdgpu_vm_it_remove(mapping, &vm->va);
+		mapping->bo_va = NULL;
+		trace_amdgpu_vm_bo_unmap(bo_va, mapping);
+		list_add(&mapping->list, &vm->freed);
+	}
+	list_for_each_entry_safe(mapping, next, &bo_va->invalids, list) {
+		list_del(&mapping->list);
+		amdgpu_vm_it_remove(mapping, &vm->va);
+		amdgpu_vm_free_mapping(adev, vm, mapping,
+				       bo_va->last_pt_update);
+	}
+
+	dma_fence_put(bo_va->last_pt_update);
+	kfree(bo_va);
+}
+
+/**
+ * amdgpu_vm_bo_invalidate - mark the bo as invalid
+ *
+ * @adev: amdgpu_device pointer
+ * @bo: amdgpu buffer object
+ * @evicted: is the BO evicted
+ *
+ * Mark @bo as invalid.
+ */
+void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev,
+			     struct amdgpu_bo *bo, bool evicted)
+{
+	struct amdgpu_vm_bo_base *bo_base;
+
+	/* shadow bo doesn't have bo base, its validation needs its parent */
+	if (bo->parent && bo->parent->shadow == bo)
+		bo = bo->parent;
+
+	list_for_each_entry(bo_base, &bo->va, bo_list) {
+		struct amdgpu_vm *vm = bo_base->vm;
+		bool was_moved = bo_base->moved;
+
+		bo_base->moved = true;
+		if (evicted && bo->tbo.resv == vm->root.base.bo->tbo.resv) {
+			if (bo->tbo.type == ttm_bo_type_kernel)
+				list_move(&bo_base->vm_status, &vm->evicted);
+			else
+				list_move_tail(&bo_base->vm_status,
+					       &vm->evicted);
+			continue;
+		}
+
+		if (was_moved)
+			continue;
+
+		if (bo->tbo.type == ttm_bo_type_kernel) {
+			list_move(&bo_base->vm_status, &vm->relocated);
+		} else {
+			spin_lock(&bo_base->vm->moved_lock);
+			list_move(&bo_base->vm_status, &vm->moved);
+			spin_unlock(&bo_base->vm->moved_lock);
+		}
+	}
+}
+
+/**
+ * amdgpu_vm_get_block_size - calculate VM page table size as power of two
+ *
+ * @vm_size: VM size
+ *
+ * Returns:
+ * VM page table as power of two
+ */
+static uint32_t amdgpu_vm_get_block_size(uint64_t vm_size)
+{
+	/* Total bits covered by PD + PTs */
+	unsigned bits = ilog2(vm_size) + 18;
+
+	/* Make sure the PD is 4K in size up to 8GB address space.
+	   Above that split equal between PD and PTs */
+	if (vm_size <= 8)
+		return (bits - 9);
+	else
+		return ((bits + 3) / 2);
+}
+
+/**
+ * amdgpu_vm_adjust_size - adjust vm size, block size and fragment size
+ *
+ * @adev: amdgpu_device pointer
+ * @min_vm_size: the minimum vm size in GB if it's set auto
+ * @fragment_size_default: Default PTE fragment size
+ * @max_level: max VMPT level
+ * @max_bits: max address space size in bits
+ *
+ */
+void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t min_vm_size,
+			   uint32_t fragment_size_default, unsigned max_level,
+			   unsigned max_bits)
+{
+	unsigned int max_size = 1 << (max_bits - 30);
+	unsigned int vm_size;
+	uint64_t tmp;
+
+	/* adjust vm size first */
+	if (amdgpu_vm_size != -1) {
+		vm_size = amdgpu_vm_size;
+		if (vm_size > max_size) {
+			dev_warn(adev->dev, "VM size (%d) too large, max is %u GB\n",
+				 amdgpu_vm_size, max_size);
+			vm_size = max_size;
+		}
+	} else {
+#ifdef __linux__
+		struct sysinfo si;
+#endif
+		unsigned int phys_ram_gb;
+
+		/* Optimal VM size depends on the amount of physical
+		 * RAM available. Underlying requirements and
+		 * assumptions:
+		 *
+		 *  - Need to map system memory and VRAM from all GPUs
+		 *     - VRAM from other GPUs not known here
+		 *     - Assume VRAM <= system memory
+		 *  - On GFX8 and older, VM space can be segmented for
+		 *    different MTYPEs
+		 *  - Need to allow room for fragmentation, guard pages etc.
+		 *
+		 * This adds up to a rough guess of system memory x3.
+		 * Round up to power of two to maximize the available
+		 * VM size with the given page table size.
+		 */
+#ifdef __linux__
+		si_meminfo(&si);
+		phys_ram_gb = ((uint64_t)si.totalram * si.mem_unit +
+			       (1 << 30) - 1) >> 30;
+#else
+		phys_ram_gb = ((uint64_t)ptoa(physmem) +
+			       (1 << 30) - 1) >> 30;
+#endif
+		vm_size = roundup_pow_of_two(
+			min(max(phys_ram_gb * 3, min_vm_size), max_size));
+	}
+
+	adev->vm_manager.max_pfn = (uint64_t)vm_size << 18;
+
+	tmp = roundup_pow_of_two(adev->vm_manager.max_pfn);
+	if (amdgpu_vm_block_size != -1)
+		tmp >>= amdgpu_vm_block_size - 9;
+	tmp = DIV_ROUND_UP(fls64(tmp) - 1, 9) - 1;
+	adev->vm_manager.num_level = min(max_level, (unsigned)tmp);
+	switch (adev->vm_manager.num_level) {
+	case 3:
+		adev->vm_manager.root_level = AMDGPU_VM_PDB2;
+		break;
+	case 2:
+		adev->vm_manager.root_level = AMDGPU_VM_PDB1;
+		break;
+	case 1:
+		adev->vm_manager.root_level = AMDGPU_VM_PDB0;
+		break;
+	default:
+		dev_err(adev->dev, "VMPT only supports 2~4+1 levels\n");
+	}
+	/* block size depends on vm size and hw setup*/
+	if (amdgpu_vm_block_size != -1)
+		adev->vm_manager.block_size =
+			min((unsigned)amdgpu_vm_block_size, max_bits
+			    - AMDGPU_GPU_PAGE_SHIFT
+			    - 9 * adev->vm_manager.num_level);
+	else if (adev->vm_manager.num_level > 1)
+		adev->vm_manager.block_size = 9;
+	else
+		adev->vm_manager.block_size = amdgpu_vm_get_block_size(tmp);
+
+	if (amdgpu_vm_fragment_size == -1)
+		adev->vm_manager.fragment_size = fragment_size_default;
+	else
+		adev->vm_manager.fragment_size = amdgpu_vm_fragment_size;
+
+	DRM_INFO("vm size is %u GB, %u levels, block size is %u-bit, fragment size is %u-bit\n",
+		 vm_size, adev->vm_manager.num_level + 1,
+		 adev->vm_manager.block_size,
+		 adev->vm_manager.fragment_size);
+}
+
+/**
+ * amdgpu_vm_init - initialize a vm instance
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ * @vm_context: Indicates if it GFX or Compute context
+ * @pasid: Process address space identifier
+ *
+ * Init @vm fields.
+ *
+ * Returns:
+ * 0 for success, error for failure.
+ */
+int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		   int vm_context, unsigned int pasid)
+{
+	struct amdgpu_bo_param bp;
+	struct amdgpu_bo *root;
+	const unsigned align = min(AMDGPU_VM_PTB_ALIGN_SIZE,
+		AMDGPU_VM_PTE_COUNT(adev) * 8);
+	unsigned ring_instance;
+	struct amdgpu_ring *ring;
+	struct drm_sched_rq *rq;
+	unsigned long size;
+	uint64_t flags;
+	int r, i;
+
+	vm->va = RB_ROOT_CACHED;
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++)
+		vm->reserved_vmid[i] = NULL;
+	INIT_LIST_HEAD(&vm->evicted);
+	INIT_LIST_HEAD(&vm->relocated);
+	mtx_init(&vm->moved_lock, IPL_TTY);
+	INIT_LIST_HEAD(&vm->moved);
+	INIT_LIST_HEAD(&vm->idle);
+	INIT_LIST_HEAD(&vm->freed);
+
+	/* create scheduler entity for page table updates */
+
+	ring_instance = atomic_inc_return(&adev->vm_manager.vm_pte_next_ring);
+	ring_instance %= adev->vm_manager.vm_pte_num_rings;
+	ring = adev->vm_manager.vm_pte_rings[ring_instance];
+	rq = &ring->sched.sched_rq[DRM_SCHED_PRIORITY_KERNEL];
+	r = drm_sched_entity_init(&vm->entity, &rq, 1, NULL);
+	if (r)
+		return r;
+
+	vm->pte_support_ats = false;
+
+	if (vm_context == AMDGPU_VM_CONTEXT_COMPUTE) {
+		vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
+						AMDGPU_VM_USE_CPU_FOR_COMPUTE);
+
+		if (adev->asic_type == CHIP_RAVEN)
+			vm->pte_support_ats = true;
+	} else {
+		vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
+						AMDGPU_VM_USE_CPU_FOR_GFX);
+	}
+	DRM_DEBUG_DRIVER("VM update mode is %s\n",
+			 vm->use_cpu_for_update ? "CPU" : "SDMA");
+	WARN_ONCE((vm->use_cpu_for_update & !amdgpu_gmc_vram_full_visible(&adev->gmc)),
+		  "CPU update of VM recommended only for large BAR system\n");
+	vm->last_update = NULL;
+
+	flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	if (vm->use_cpu_for_update)
+		flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+	else if (vm_context != AMDGPU_VM_CONTEXT_COMPUTE)
+		flags |= AMDGPU_GEM_CREATE_SHADOW;
+
+	size = amdgpu_vm_bo_size(adev, adev->vm_manager.root_level);
+	memset(&bp, 0, sizeof(bp));
+	bp.size = size;
+	bp.byte_align = align;
+	bp.domain = AMDGPU_GEM_DOMAIN_VRAM;
+	bp.flags = flags;
+	bp.type = ttm_bo_type_kernel;
+	bp.resv = NULL;
+	r = amdgpu_bo_create(adev, &bp, &root);
+	if (r)
+		goto error_free_sched_entity;
+
+	r = amdgpu_bo_reserve(root, true);
+	if (r)
+		goto error_free_root;
+
+	r = amdgpu_vm_clear_bo(adev, vm, root,
+			       adev->vm_manager.root_level,
+			       vm->pte_support_ats);
+	if (r)
+		goto error_unreserve;
+
+	amdgpu_vm_bo_base_init(&vm->root.base, vm, root);
+	amdgpu_bo_unreserve(vm->root.base.bo);
+
+	if (pasid) {
+		unsigned long flags;
+
+		spin_lock_irqsave(&adev->vm_manager.pasid_lock, flags);
+		r = idr_alloc(&adev->vm_manager.pasid_idr, vm, pasid, pasid + 1,
+			      GFP_ATOMIC);
+		spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, flags);
+		if (r < 0)
+			goto error_free_root;
+
+		vm->pasid = pasid;
+	}
+
+#ifdef __linux__
+	INIT_KFIFO(vm->faults);
+#else
+	SIMPLEQ_INIT(&vm->faults);
+#endif
+	vm->fault_credit = 16;
+
+	return 0;
+
+error_unreserve:
+	amdgpu_bo_unreserve(vm->root.base.bo);
+
+error_free_root:
+	amdgpu_bo_unref(&vm->root.base.bo->shadow);
+	amdgpu_bo_unref(&vm->root.base.bo);
+	vm->root.base.bo = NULL;
+
+error_free_sched_entity:
+	drm_sched_entity_destroy(&vm->entity);
+
+	return r;
+}
+
+/**
+ * amdgpu_vm_make_compute - Turn a GFX VM into a compute VM
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ *
+ * This only works on GFX VMs that don't have any BOs added and no
+ * page tables allocated yet.
+ *
+ * Changes the following VM parameters:
+ * - use_cpu_for_update
+ * - pte_supports_ats
+ * - pasid (old PASID is released, because compute manages its own PASIDs)
+ *
+ * Reinitializes the page directory to reflect the changed ATS
+ * setting.
+ *
+ * Returns:
+ * 0 for success, -errno for errors.
+ */
+int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm)
+{
+	bool pte_support_ats = (adev->asic_type == CHIP_RAVEN);
+	int r;
+
+	r = amdgpu_bo_reserve(vm->root.base.bo, true);
+	if (r)
+		return r;
+
+	/* Sanity checks */
+#ifdef __linux__
+	if (!RB_EMPTY_ROOT(&vm->va.rb_root) || vm->root.entries) {
+#else
+	if (!RB_EMPTY_ROOT(&vm->va) || vm->root.entries) {
+#endif
+		r = -EINVAL;
+		goto error;
+	}
+
+	/* Check if PD needs to be reinitialized and do it before
+	 * changing any other state, in case it fails.
+	 */
+	if (pte_support_ats != vm->pte_support_ats) {
+		r = amdgpu_vm_clear_bo(adev, vm, vm->root.base.bo,
+			       adev->vm_manager.root_level,
+			       pte_support_ats);
+		if (r)
+			goto error;
+	}
+
+	/* Update VM state */
+	vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
+				    AMDGPU_VM_USE_CPU_FOR_COMPUTE);
+	vm->pte_support_ats = pte_support_ats;
+	DRM_DEBUG_DRIVER("VM update mode is %s\n",
+			 vm->use_cpu_for_update ? "CPU" : "SDMA");
+	WARN_ONCE((vm->use_cpu_for_update & !amdgpu_gmc_vram_full_visible(&adev->gmc)),
+		  "CPU update of VM recommended only for large BAR system\n");
+
+	if (vm->pasid) {
+		unsigned long flags;
+
+		spin_lock_irqsave(&adev->vm_manager.pasid_lock, flags);
+		idr_remove(&adev->vm_manager.pasid_idr, vm->pasid);
+		spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, flags);
+
+		vm->pasid = 0;
+	}
+
+	/* Free the shadow bo for compute VM */
+	amdgpu_bo_unref(&vm->root.base.bo->shadow);
+
+error:
+	amdgpu_bo_unreserve(vm->root.base.bo);
+	return r;
+}
+
+/**
+ * amdgpu_vm_free_levels - free PD/PT levels
+ *
+ * @adev: amdgpu device structure
+ * @parent: PD/PT starting level to free
+ * @level: level of parent structure
+ *
+ * Free the page directory or page table level and all sub levels.
+ */
+static void amdgpu_vm_free_levels(struct amdgpu_device *adev,
+				  struct amdgpu_vm_pt *parent,
+				  unsigned level)
+{
+	unsigned i, num_entries = amdgpu_vm_num_entries(adev, level);
+
+	if (parent->base.bo) {
+		list_del(&parent->base.bo_list);
+		list_del(&parent->base.vm_status);
+		amdgpu_bo_unref(&parent->base.bo->shadow);
+		amdgpu_bo_unref(&parent->base.bo);
+	}
+
+	if (parent->entries)
+		for (i = 0; i < num_entries; i++)
+			amdgpu_vm_free_levels(adev, &parent->entries[i],
+					      level + 1);
+
+	kvfree(parent->entries);
+}
+
+/**
+ * amdgpu_vm_fini - tear down a vm instance
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ *
+ * Tear down @vm.
+ * Unbind the VM and remove all bos from the vm bo list
+ */
+void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
+{
+	struct amdgpu_bo_va_mapping *mapping, *tmp;
+	bool prt_fini_needed = !!adev->gmc.gmc_funcs->set_prt;
+	struct amdgpu_bo *root;
+	u64 fault;
+	int i, r;
+	struct amdgpu_vm_fault *vmf;
+
+	amdgpu_amdkfd_gpuvm_destroy_cb(adev, vm);
+
+	/* Clear pending page faults from IH when the VM is destroyed */
+#ifdef __linux__
+	while (kfifo_get(&vm->faults, &fault))
+		amdgpu_ih_clear_fault(adev, fault);
+#else
+	while (!SIMPLEQ_EMPTY(&vm->faults)) {
+		vmf = SIMPLEQ_FIRST(&vm->faults);
+		fault = vmf->val;
+		SIMPLEQ_REMOVE_HEAD(&vm->faults, vm_fault_entry);
+		free(vmf, M_DRM, sizeof(*vmf));
+		amdgpu_ih_clear_fault(adev, fault);
+	}
+#endif
+
+	if (vm->pasid) {
+		unsigned long flags;
+
+		spin_lock_irqsave(&adev->vm_manager.pasid_lock, flags);
+		idr_remove(&adev->vm_manager.pasid_idr, vm->pasid);
+		spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, flags);
+	}
+
+	drm_sched_entity_destroy(&vm->entity);
+
+#ifdef __linux__
+	if (!RB_EMPTY_ROOT(&vm->va.rb_root)) {
+		dev_err(adev->dev, "still active bo inside vm\n");
+	}
+	rbtree_postorder_for_each_entry_safe(mapping, tmp,
+					     &vm->va.rb_root, rb) {
+		list_del(&mapping->list);
+		amdgpu_vm_it_remove(mapping, &vm->va);
+		kfree(mapping);
+	}
+#else
+	if (!RB_EMPTY_ROOT(&vm->va)) {
+		dev_err(adev->dev, "still active bo inside vm\n");
+	}
+	rbtree_postorder_for_each_entry_safe(mapping, tmp,
+					     &vm->va, rb) {
+		list_del(&mapping->list);
+		amdgpu_vm_it_remove(mapping, &vm->va);
+		kfree(mapping);
+	}
+#endif
+	list_for_each_entry_safe(mapping, tmp, &vm->freed, list) {
+		if (mapping->flags & AMDGPU_PTE_PRT && prt_fini_needed) {
+			amdgpu_vm_prt_fini(adev, vm);
+			prt_fini_needed = false;
+		}
+
+		list_del(&mapping->list);
+		amdgpu_vm_free_mapping(adev, vm, mapping, NULL);
+	}
+
+	root = amdgpu_bo_ref(vm->root.base.bo);
+	r = amdgpu_bo_reserve(root, true);
+	if (r) {
+		dev_err(adev->dev, "Leaking page tables because BO reservation failed\n");
+	} else {
+		amdgpu_vm_free_levels(adev, &vm->root,
+				      adev->vm_manager.root_level);
+		amdgpu_bo_unreserve(root);
+	}
+	amdgpu_bo_unref(&root);
+	dma_fence_put(vm->last_update);
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++)
+		amdgpu_vmid_free_reserved(adev, vm, i);
+}
+
+/**
+ * amdgpu_vm_pasid_fault_credit - Check fault credit for given PASID
+ *
+ * @adev: amdgpu_device pointer
+ * @pasid: PASID do identify the VM
+ *
+ * This function is expected to be called in interrupt context.
+ *
+ * Returns:
+ * True if there was fault credit, false otherwise
+ */
+bool amdgpu_vm_pasid_fault_credit(struct amdgpu_device *adev,
+				  unsigned int pasid)
+{
+	struct amdgpu_vm *vm;
+
+	spin_lock(&adev->vm_manager.pasid_lock);
+	vm = idr_find(&adev->vm_manager.pasid_idr, pasid);
+	if (!vm) {
+		/* VM not found, can't track fault credit */
+		spin_unlock(&adev->vm_manager.pasid_lock);
+		return true;
+	}
+
+	/* No lock needed. only accessed by IRQ handler */
+	if (!vm->fault_credit) {
+		/* Too many faults in this VM */
+		spin_unlock(&adev->vm_manager.pasid_lock);
+		return false;
+	}
+
+	vm->fault_credit--;
+	spin_unlock(&adev->vm_manager.pasid_lock);
+	return true;
+}
+
+/**
+ * amdgpu_vm_manager_init - init the VM manager
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the VM manager structures
+ */
+void amdgpu_vm_manager_init(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	amdgpu_vmid_mgr_init(adev);
+
+	adev->vm_manager.fence_context =
+		dma_fence_context_alloc(AMDGPU_MAX_RINGS);
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
+		adev->vm_manager.seqno[i] = 0;
+
+	atomic_set(&adev->vm_manager.vm_pte_next_ring, 0);
+	mtx_init(&adev->vm_manager.prt_lock, IPL_TTY);
+	atomic_set(&adev->vm_manager.num_prt_users, 0);
+
+	/* If not overridden by the user, by default, only in large BAR systems
+	 * Compute VM tables will be updated by CPU
+	 */
+#ifdef __amd64__
+	if (amdgpu_vm_update_mode == -1) {
+		if (amdgpu_gmc_vram_full_visible(&adev->gmc))
+			adev->vm_manager.vm_update_mode =
+				AMDGPU_VM_USE_CPU_FOR_COMPUTE;
+		else
+			adev->vm_manager.vm_update_mode = 0;
+	} else
+		adev->vm_manager.vm_update_mode = amdgpu_vm_update_mode;
+#else
+	adev->vm_manager.vm_update_mode = 0;
+#endif
+
+	idr_init(&adev->vm_manager.pasid_idr);
+	mtx_init(&adev->vm_manager.pasid_lock, IPL_TTY);
+}
+
+/**
+ * amdgpu_vm_manager_fini - cleanup VM manager
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Cleanup the VM manager and free resources.
+ */
+void amdgpu_vm_manager_fini(struct amdgpu_device *adev)
+{
+	WARN_ON(!idr_is_empty(&adev->vm_manager.pasid_idr));
+	idr_destroy(&adev->vm_manager.pasid_idr);
+
+	amdgpu_vmid_mgr_fini(adev);
+}
+
+/**
+ * amdgpu_vm_ioctl - Manages VMID reservation for vm hubs.
+ *
+ * @dev: drm device pointer
+ * @data: drm_amdgpu_vm
+ * @filp: drm file pointer
+ *
+ * Returns:
+ * 0 for success, -errno for errors.
+ */
+int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
+{
+	union drm_amdgpu_vm *args = data;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	int r;
+
+	switch (args->in.op) {
+	case AMDGPU_VM_OP_RESERVE_VMID:
+		/* current, we only have requirement to reserve vmid from gfxhub */
+		r = amdgpu_vmid_alloc_reserved(adev, &fpriv->vm, AMDGPU_GFXHUB);
+		if (r)
+			return r;
+		break;
+	case AMDGPU_VM_OP_UNRESERVE_VMID:
+		amdgpu_vmid_free_reserved(adev, &fpriv->vm, AMDGPU_GFXHUB);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_get_task_info - Extracts task info for a PASID.
+ *
+ * @dev: drm device pointer
+ * @pasid: PASID identifier for VM
+ * @task_info: task_info to fill.
+ */
+void amdgpu_vm_get_task_info(struct amdgpu_device *adev, unsigned int pasid,
+			 struct amdgpu_task_info *task_info)
+{
+	struct amdgpu_vm *vm;
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->vm_manager.pasid_lock, flags);
+
+	vm = idr_find(&adev->vm_manager.pasid_idr, pasid);
+	if (vm)
+		*task_info = vm->task_info;
+
+	spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, flags);
+}
+
+/**
+ * amdgpu_vm_set_task_info - Sets VMs task info.
+ *
+ * @vm: vm for which to set the info
+ */
+void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
+{
+	if (!vm->task_info.pid) {
+#ifdef __linux__
+		vm->task_info.pid = current->pid;
+		get_task_comm(vm->task_info.task_name, current);
+
+		if (current->group_leader->mm == current->mm) {
+			vm->task_info.tgid = current->group_leader->pid;
+			get_task_comm(vm->task_info.process_name, current->group_leader);
+		}
+#else
+		vm->task_info.pid = curproc->p_p->ps_pid;
+		strlcpy(vm->task_info.task_name, curproc->p_p->ps_comm,
+		    sizeof(vm->task_info.task_name));
+#endif
+	}
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_vm.h b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vm.h
new file mode 100644
index 00000000000..d74b59729b3
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vm.h
@@ -0,0 +1,347 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König
+ */
+#ifndef __AMDGPU_VM_H__
+#define __AMDGPU_VM_H__
+
+#include <linux/idr.h>
+#include <linux/kfifo.h>
+#include <linux/rbtree.h>
+#include <drm/gpu_scheduler.h>
+#include <drm/drm_file.h>
+
+#include "amdgpu_sync.h"
+#include "amdgpu_ring.h"
+#include "amdgpu_ids.h"
+
+struct amdgpu_bo_va;
+struct amdgpu_job;
+struct amdgpu_bo_list_entry;
+
+/*
+ * GPUVM handling
+ */
+
+/* Maximum number of PTEs the hardware can write with one command */
+#define AMDGPU_VM_MAX_UPDATE_SIZE	0x3FFFF
+
+/* number of entries in page table */
+#define AMDGPU_VM_PTE_COUNT(adev) (1 << (adev)->vm_manager.block_size)
+
+/* PTBs (Page Table Blocks) need to be aligned to 32K */
+#define AMDGPU_VM_PTB_ALIGN_SIZE   32768
+
+#define AMDGPU_PTE_VALID	(1ULL << 0)
+#define AMDGPU_PTE_SYSTEM	(1ULL << 1)
+#define AMDGPU_PTE_SNOOPED	(1ULL << 2)
+
+/* VI only */
+#define AMDGPU_PTE_EXECUTABLE	(1ULL << 4)
+
+#define AMDGPU_PTE_READABLE	(1ULL << 5)
+#define AMDGPU_PTE_WRITEABLE	(1ULL << 6)
+
+#define AMDGPU_PTE_FRAG(x)	((x & 0x1fULL) << 7)
+
+/* TILED for VEGA10, reserved for older ASICs  */
+#define AMDGPU_PTE_PRT		(1ULL << 51)
+
+/* PDE is handled as PTE for VEGA10 */
+#define AMDGPU_PDE_PTE		(1ULL << 54)
+
+/* PTE is handled as PDE for VEGA10 (Translate Further) */
+#define AMDGPU_PTE_TF		(1ULL << 56)
+
+/* PDE Block Fragment Size for VEGA10 */
+#define AMDGPU_PDE_BFS(a)	((uint64_t)a << 59)
+
+
+/* For GFX9 */
+#define AMDGPU_PTE_MTYPE(a)    ((uint64_t)a << 57)
+#define AMDGPU_PTE_MTYPE_MASK	AMDGPU_PTE_MTYPE(3ULL)
+
+#define AMDGPU_MTYPE_NC 0
+#define AMDGPU_MTYPE_CC 2
+
+#define AMDGPU_PTE_DEFAULT_ATC  (AMDGPU_PTE_SYSTEM      \
+                                | AMDGPU_PTE_SNOOPED    \
+                                | AMDGPU_PTE_EXECUTABLE \
+                                | AMDGPU_PTE_READABLE   \
+                                | AMDGPU_PTE_WRITEABLE  \
+                                | AMDGPU_PTE_MTYPE(AMDGPU_MTYPE_CC))
+
+/* How to programm VM fault handling */
+#define AMDGPU_VM_FAULT_STOP_NEVER	0
+#define AMDGPU_VM_FAULT_STOP_FIRST	1
+#define AMDGPU_VM_FAULT_STOP_ALWAYS	2
+
+/* max number of VMHUB */
+#define AMDGPU_MAX_VMHUBS			2
+#define AMDGPU_GFXHUB				0
+#define AMDGPU_MMHUB				1
+
+/* hardcode that limit for now */
+#define AMDGPU_VA_RESERVED_SIZE			(1ULL << 20)
+
+/* VA hole for 48bit addresses on Vega10 */
+#define AMDGPU_VA_HOLE_START			0x0000800000000000ULL
+#define AMDGPU_VA_HOLE_END			0xffff800000000000ULL
+
+/*
+ * Hardware is programmed as if the hole doesn't exists with start and end
+ * address values.
+ *
+ * This mask is used to remove the upper 16bits of the VA and so come up with
+ * the linear addr value.
+ */
+#define AMDGPU_VA_HOLE_MASK			0x0000ffffffffffffULL
+
+/* max vmids dedicated for process */
+#define AMDGPU_VM_MAX_RESERVED_VMID	1
+
+#define AMDGPU_VM_CONTEXT_GFX 0
+#define AMDGPU_VM_CONTEXT_COMPUTE 1
+
+/* See vm_update_mode */
+#define AMDGPU_VM_USE_CPU_FOR_GFX (1 << 0)
+#define AMDGPU_VM_USE_CPU_FOR_COMPUTE (1 << 1)
+
+/* VMPT level enumerate, and the hiberachy is:
+ * PDB2->PDB1->PDB0->PTB
+ */
+enum amdgpu_vm_level {
+	AMDGPU_VM_PDB2,
+	AMDGPU_VM_PDB1,
+	AMDGPU_VM_PDB0,
+	AMDGPU_VM_PTB
+};
+
+/* base structure for tracking BO usage in a VM */
+struct amdgpu_vm_bo_base {
+	/* constant after initialization */
+	struct amdgpu_vm		*vm;
+	struct amdgpu_bo		*bo;
+
+	/* protected by bo being reserved */
+	struct list_head		bo_list;
+
+	/* protected by spinlock */
+	struct list_head		vm_status;
+
+	/* protected by the BO being reserved */
+	bool				moved;
+};
+
+struct amdgpu_vm_pt {
+	struct amdgpu_vm_bo_base	base;
+	bool				huge;
+
+	/* array of page tables, one for each directory entry */
+	struct amdgpu_vm_pt		*entries;
+};
+
+#define AMDGPU_VM_FAULT(pasid, addr) (((u64)(pasid) << 48) | (addr))
+#define AMDGPU_VM_FAULT_PASID(fault) ((u64)(fault) >> 48)
+#define AMDGPU_VM_FAULT_ADDR(fault)  ((u64)(fault) & 0xfffffffff000ULL)
+
+
+struct amdgpu_task_info {
+	char	process_name[TASK_COMM_LEN];
+	char	task_name[TASK_COMM_LEN];
+	pid_t	pid;
+	pid_t	tgid;
+};
+
+struct amdgpu_vm_fault {
+	SIMPLEQ_ENTRY(amdgpu_vm_fault)	vm_fault_entry;
+	uint64_t			val;
+};
+SIMPLEQ_HEAD(amdgpu_vm_faults, amdgpu_vm_fault);
+
+struct amdgpu_vm {
+	/* tree of virtual addresses mapped */
+	struct rb_root_cached	va;
+
+	/* BOs who needs a validation */
+	struct list_head	evicted;
+
+	/* PT BOs which relocated and their parent need an update */
+	struct list_head	relocated;
+
+	/* BOs moved, but not yet updated in the PT */
+	struct list_head	moved;
+	spinlock_t		moved_lock;
+
+	/* All BOs of this VM not currently in the state machine */
+	struct list_head	idle;
+
+	/* BO mappings freed, but not yet updated in the PT */
+	struct list_head	freed;
+
+	/* contains the page directory */
+	struct amdgpu_vm_pt     root;
+	struct dma_fence	*last_update;
+
+	/* Scheduler entity for page table updates */
+	struct drm_sched_entity	entity;
+
+	unsigned int		pasid;
+	/* dedicated to vm */
+	struct amdgpu_vmid	*reserved_vmid[AMDGPU_MAX_VMHUBS];
+
+	/* Flag to indicate if VM tables are updated by CPU or GPU (SDMA) */
+	bool                    use_cpu_for_update;
+
+	/* Flag to indicate ATS support from PTE for GFX9 */
+	bool			pte_support_ats;
+
+#ifdef __linux__
+	/* Up to 128 pending retry page faults */
+	DECLARE_KFIFO(faults, u64, 128);
+#else
+	struct amdgpu_vm_faults	faults;
+#endif
+
+	/* Limit non-retry fault storms */
+	unsigned int		fault_credit;
+
+	/* Points to the KFD process VM info */
+	struct amdkfd_process_info *process_info;
+
+	/* List node in amdkfd_process_info.vm_list_head */
+	struct list_head	vm_list_node;
+
+	/* Valid while the PD is reserved or fenced */
+	uint64_t		pd_phys_addr;
+
+	/* Some basic info about the task */
+	struct amdgpu_task_info task_info;
+};
+
+struct amdgpu_vm_manager {
+	/* Handling of VMIDs */
+	struct amdgpu_vmid_mgr			id_mgr[AMDGPU_MAX_VMHUBS];
+
+	/* Handling of VM fences */
+	u64					fence_context;
+	unsigned				seqno[AMDGPU_MAX_RINGS];
+
+	uint64_t				max_pfn;
+	uint32_t				num_level;
+	uint32_t				block_size;
+	uint32_t				fragment_size;
+	enum amdgpu_vm_level			root_level;
+	/* vram base address for page table entry  */
+	u64					vram_base_offset;
+	/* vm pte handling */
+	const struct amdgpu_vm_pte_funcs        *vm_pte_funcs;
+	struct amdgpu_ring                      *vm_pte_rings[AMDGPU_MAX_RINGS];
+	unsigned				vm_pte_num_rings;
+	atomic_t				vm_pte_next_ring;
+
+	/* partial resident texture handling */
+	spinlock_t				prt_lock;
+	atomic_t				num_prt_users;
+
+	/* controls how VM page tables are updated for Graphics and Compute.
+	 * BIT0[= 0] Graphics updated by SDMA [= 1] by CPU
+	 * BIT1[= 0] Compute updated by SDMA [= 1] by CPU
+	 */
+	int					vm_update_mode;
+
+	/* PASID to VM mapping, will be used in interrupt context to
+	 * look up VM of a page fault
+	 */
+	struct idr				pasid_idr;
+	spinlock_t				pasid_lock;
+};
+
+void amdgpu_vm_manager_init(struct amdgpu_device *adev);
+void amdgpu_vm_manager_fini(struct amdgpu_device *adev);
+int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		   int vm_context, unsigned int pasid);
+int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm);
+void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm);
+bool amdgpu_vm_pasid_fault_credit(struct amdgpu_device *adev,
+				  unsigned int pasid);
+void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
+			 struct list_head *validated,
+			 struct amdgpu_bo_list_entry *entry);
+bool amdgpu_vm_ready(struct amdgpu_vm *vm);
+int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			      int (*callback)(void *p, struct amdgpu_bo *bo),
+			      void *param);
+int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
+			struct amdgpu_vm *vm,
+			uint64_t saddr, uint64_t size);
+int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, bool need_pipe_sync);
+int amdgpu_vm_update_directories(struct amdgpu_device *adev,
+				 struct amdgpu_vm *vm);
+int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
+			  struct amdgpu_vm *vm,
+			  struct dma_fence **fence);
+int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
+			   struct amdgpu_vm *vm);
+int amdgpu_vm_bo_update(struct amdgpu_device *adev,
+			struct amdgpu_bo_va *bo_va,
+			bool clear);
+void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev,
+			     struct amdgpu_bo *bo, bool evicted);
+struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm,
+				       struct amdgpu_bo *bo);
+struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
+				      struct amdgpu_vm *vm,
+				      struct amdgpu_bo *bo);
+int amdgpu_vm_bo_map(struct amdgpu_device *adev,
+		     struct amdgpu_bo_va *bo_va,
+		     uint64_t addr, uint64_t offset,
+		     uint64_t size, uint64_t flags);
+int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
+			     struct amdgpu_bo_va *bo_va,
+			     uint64_t addr, uint64_t offset,
+			     uint64_t size, uint64_t flags);
+int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
+		       struct amdgpu_bo_va *bo_va,
+		       uint64_t addr);
+int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm,
+				uint64_t saddr, uint64_t size);
+struct amdgpu_bo_va_mapping *amdgpu_vm_bo_lookup_mapping(struct amdgpu_vm *vm,
+							 uint64_t addr);
+void amdgpu_vm_bo_trace_cs(struct amdgpu_vm *vm, struct ww_acquire_ctx *ticket);
+void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
+		      struct amdgpu_bo_va *bo_va);
+void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t min_vm_size,
+			   uint32_t fragment_size_default, unsigned max_level,
+			   unsigned max_bits);
+int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp);
+bool amdgpu_vm_need_pipeline_sync(struct amdgpu_ring *ring,
+				  struct amdgpu_job *job);
+void amdgpu_vm_check_compute_bug(struct amdgpu_device *adev);
+
+void amdgpu_vm_get_task_info(struct amdgpu_device *adev, unsigned int pasid,
+			 struct amdgpu_task_info *task_info);
+
+void amdgpu_vm_set_task_info(struct amdgpu_vm *vm);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/amdgpu_vram_mgr.c b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vram_mgr.c
new file mode 100644
index 00000000000..aeb90503373
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/amdgpu_vram_mgr.c
@@ -0,0 +1,321 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+
+struct amdgpu_vram_mgr {
+	struct drm_mm mm;
+	spinlock_t lock;
+	atomic64_t usage;
+	atomic64_t vis_usage;
+};
+
+/**
+ * amdgpu_vram_mgr_init - init VRAM manager and DRM MM
+ *
+ * @man: TTM memory type manager
+ * @p_size: maximum size of VRAM
+ *
+ * Allocate and initialize the VRAM manager.
+ */
+static int amdgpu_vram_mgr_init(struct ttm_mem_type_manager *man,
+				unsigned long p_size)
+{
+	struct amdgpu_vram_mgr *mgr;
+
+	mgr = kzalloc(sizeof(*mgr), GFP_KERNEL);
+	if (!mgr)
+		return -ENOMEM;
+
+	drm_mm_init(&mgr->mm, 0, p_size);
+	mtx_init(&mgr->lock, IPL_TTY);
+	man->priv = mgr;
+	return 0;
+}
+
+/**
+ * amdgpu_vram_mgr_fini - free and destroy VRAM manager
+ *
+ * @man: TTM memory type manager
+ *
+ * Destroy and free the VRAM manager, returns -EBUSY if ranges are still
+ * allocated inside it.
+ */
+static int amdgpu_vram_mgr_fini(struct ttm_mem_type_manager *man)
+{
+	struct amdgpu_vram_mgr *mgr = man->priv;
+
+	spin_lock(&mgr->lock);
+	drm_mm_takedown(&mgr->mm);
+	spin_unlock(&mgr->lock);
+	kfree(mgr);
+	man->priv = NULL;
+	return 0;
+}
+
+/**
+ * amdgpu_vram_mgr_vis_size - Calculate visible node size
+ *
+ * @adev: amdgpu device structure
+ * @node: MM node structure
+ *
+ * Calculate how many bytes of the MM node are inside visible VRAM
+ */
+static u64 amdgpu_vram_mgr_vis_size(struct amdgpu_device *adev,
+				    struct drm_mm_node *node)
+{
+	uint64_t start = node->start << PAGE_SHIFT;
+	uint64_t end = (node->size + node->start) << PAGE_SHIFT;
+
+	if (start >= adev->gmc.visible_vram_size)
+		return 0;
+
+	return (end > adev->gmc.visible_vram_size ?
+		adev->gmc.visible_vram_size : end) - start;
+}
+
+/**
+ * amdgpu_vram_mgr_bo_visible_size - CPU visible BO size
+ *
+ * @bo: &amdgpu_bo buffer object (must be in VRAM)
+ *
+ * Returns:
+ * How much of the given &amdgpu_bo buffer object lies in CPU visible VRAM.
+ */
+u64 amdgpu_vram_mgr_bo_visible_size(struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct ttm_mem_reg *mem = &bo->tbo.mem;
+	struct drm_mm_node *nodes = mem->mm_node;
+	unsigned pages = mem->num_pages;
+	u64 usage;
+
+	if (amdgpu_gmc_vram_full_visible(&adev->gmc))
+		return amdgpu_bo_size(bo);
+
+	if (mem->start >= adev->gmc.visible_vram_size >> PAGE_SHIFT)
+		return 0;
+
+	for (usage = 0; nodes && pages; pages -= nodes->size, nodes++)
+		usage += amdgpu_vram_mgr_vis_size(adev, nodes);
+
+	return usage;
+}
+
+/**
+ * amdgpu_vram_mgr_new - allocate new ranges
+ *
+ * @man: TTM memory type manager
+ * @tbo: TTM BO we need this range for
+ * @place: placement flags and restrictions
+ * @mem: the resulting mem object
+ *
+ * Allocate VRAM for the given BO.
+ */
+static int amdgpu_vram_mgr_new(struct ttm_mem_type_manager *man,
+			       struct ttm_buffer_object *tbo,
+			       const struct ttm_place *place,
+			       struct ttm_mem_reg *mem)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
+	struct amdgpu_vram_mgr *mgr = man->priv;
+	struct drm_mm *mm = &mgr->mm;
+	struct drm_mm_node *nodes;
+	enum drm_mm_insert_mode mode;
+	unsigned long lpfn, num_nodes, pages_per_node, pages_left;
+	uint64_t usage = 0, vis_usage = 0;
+	unsigned i;
+	int r;
+
+	lpfn = place->lpfn;
+	if (!lpfn)
+		lpfn = man->size;
+
+	if (place->flags & TTM_PL_FLAG_CONTIGUOUS ||
+	    amdgpu_vram_page_split == -1) {
+		pages_per_node = ~0ul;
+		num_nodes = 1;
+	} else {
+		pages_per_node = max((uint32_t)amdgpu_vram_page_split,
+				     mem->page_alignment);
+		num_nodes = DIV_ROUND_UP(mem->num_pages, pages_per_node);
+	}
+
+	nodes = kvmalloc_array(num_nodes, sizeof(*nodes),
+			       GFP_KERNEL | __GFP_ZERO);
+	if (!nodes)
+		return -ENOMEM;
+
+	mode = DRM_MM_INSERT_BEST;
+	if (place->flags & TTM_PL_FLAG_TOPDOWN)
+		mode = DRM_MM_INSERT_HIGH;
+
+	mem->start = 0;
+	pages_left = mem->num_pages;
+
+	spin_lock(&mgr->lock);
+	for (i = 0; i < num_nodes; ++i) {
+		unsigned long pages = min(pages_left, pages_per_node);
+		uint32_t alignment = mem->page_alignment;
+		unsigned long start;
+
+		if (pages == pages_per_node)
+			alignment = pages_per_node;
+
+		r = drm_mm_insert_node_in_range(mm, &nodes[i],
+						pages, alignment, 0,
+						place->fpfn, lpfn,
+						mode);
+		if (unlikely(r))
+			goto error;
+
+		usage += nodes[i].size << PAGE_SHIFT;
+		vis_usage += amdgpu_vram_mgr_vis_size(adev, &nodes[i]);
+
+		/* Calculate a virtual BO start address to easily check if
+		 * everything is CPU accessible.
+		 */
+		start = nodes[i].start + nodes[i].size;
+		if (start > mem->num_pages)
+			start -= mem->num_pages;
+		else
+			start = 0;
+		mem->start = max(mem->start, start);
+		pages_left -= pages;
+	}
+	spin_unlock(&mgr->lock);
+
+	atomic64_add(usage, &mgr->usage);
+	atomic64_add(vis_usage, &mgr->vis_usage);
+
+	mem->mm_node = nodes;
+
+	return 0;
+
+error:
+	while (i--)
+		drm_mm_remove_node(&nodes[i]);
+	spin_unlock(&mgr->lock);
+
+	kvfree(nodes);
+	return r == -ENOSPC ? 0 : r;
+}
+
+/**
+ * amdgpu_vram_mgr_del - free ranges
+ *
+ * @man: TTM memory type manager
+ * @tbo: TTM BO we need this range for
+ * @place: placement flags and restrictions
+ * @mem: TTM memory object
+ *
+ * Free the allocated VRAM again.
+ */
+static void amdgpu_vram_mgr_del(struct ttm_mem_type_manager *man,
+				struct ttm_mem_reg *mem)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
+	struct amdgpu_vram_mgr *mgr = man->priv;
+	struct drm_mm_node *nodes = mem->mm_node;
+	uint64_t usage = 0, vis_usage = 0;
+	unsigned pages = mem->num_pages;
+
+	if (!mem->mm_node)
+		return;
+
+	spin_lock(&mgr->lock);
+	while (pages) {
+		pages -= nodes->size;
+		drm_mm_remove_node(nodes);
+		usage += nodes->size << PAGE_SHIFT;
+		vis_usage += amdgpu_vram_mgr_vis_size(adev, nodes);
+		++nodes;
+	}
+	spin_unlock(&mgr->lock);
+
+	atomic64_sub(usage, &mgr->usage);
+	atomic64_sub(vis_usage, &mgr->vis_usage);
+
+	kvfree(mem->mm_node);
+	mem->mm_node = NULL;
+}
+
+/**
+ * amdgpu_vram_mgr_usage - how many bytes are used in this domain
+ *
+ * @man: TTM memory type manager
+ *
+ * Returns how many bytes are used in this domain.
+ */
+uint64_t amdgpu_vram_mgr_usage(struct ttm_mem_type_manager *man)
+{
+	struct amdgpu_vram_mgr *mgr = man->priv;
+
+	return atomic64_read(&mgr->usage);
+}
+
+/**
+ * amdgpu_vram_mgr_vis_usage - how many bytes are used in the visible part
+ *
+ * @man: TTM memory type manager
+ *
+ * Returns how many bytes are used in the visible part of VRAM
+ */
+uint64_t amdgpu_vram_mgr_vis_usage(struct ttm_mem_type_manager *man)
+{
+	struct amdgpu_vram_mgr *mgr = man->priv;
+
+	return atomic64_read(&mgr->vis_usage);
+}
+
+/**
+ * amdgpu_vram_mgr_debug - dump VRAM table
+ *
+ * @man: TTM memory type manager
+ * @printer: DRM printer to use
+ *
+ * Dump the table content using printk.
+ */
+static void amdgpu_vram_mgr_debug(struct ttm_mem_type_manager *man,
+				  struct drm_printer *printer)
+{
+	struct amdgpu_vram_mgr *mgr = man->priv;
+
+	spin_lock(&mgr->lock);
+	drm_mm_print(&mgr->mm, printer);
+	spin_unlock(&mgr->lock);
+
+	drm_printf(printer, "man size:%llu pages, ram usage:%lluMB, vis usage:%lluMB\n",
+		   man->size, amdgpu_vram_mgr_usage(man) >> 20,
+		   amdgpu_vram_mgr_vis_usage(man) >> 20);
+}
+
+const struct ttm_mem_type_manager_func amdgpu_vram_mgr_func = {
+	.init		= amdgpu_vram_mgr_init,
+	.takedown	= amdgpu_vram_mgr_fini,
+	.get_node	= amdgpu_vram_mgr_new,
+	.put_node	= amdgpu_vram_mgr_del,
+	.debug		= amdgpu_vram_mgr_debug
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/atom.h b/sys/dev/pci/drm/amd/amdgpu/atom.h
new file mode 100644
index 00000000000..1d61a4d6a5e
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/atom.h
@@ -0,0 +1,160 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: Stanislaw Skowronek
+ */
+
+#ifndef ATOM_H
+#define ATOM_H
+
+#include <linux/types.h>
+#include <drm/drmP.h>
+
+#define ATOM_BIOS_MAGIC		0xAA55
+#define ATOM_ATI_MAGIC_PTR	0x30
+#define ATOM_ATI_MAGIC		" 761295520"
+#define ATOM_ROM_TABLE_PTR	0x48
+#define ATOM_ROM_PART_NUMBER_PTR	0x6E
+
+#define ATOM_ROM_MAGIC		"ATOM"
+#define ATOM_ROM_MAGIC_PTR	4
+
+#define ATOM_ROM_MSG_PTR	0x10
+#define ATOM_ROM_CMD_PTR	0x1E
+#define ATOM_ROM_DATA_PTR	0x20
+
+#define ATOM_CMD_INIT		0
+#define ATOM_CMD_SETSCLK	0x0A
+#define ATOM_CMD_SETMCLK	0x0B
+#define ATOM_CMD_SETPCLK	0x0C
+#define ATOM_CMD_SPDFANCNTL	0x39
+
+#define ATOM_DATA_FWI_PTR	0xC
+#define ATOM_DATA_IIO_PTR	0x32
+
+#define ATOM_FWI_DEFSCLK_PTR	8
+#define ATOM_FWI_DEFMCLK_PTR	0xC
+#define ATOM_FWI_MAXSCLK_PTR	0x24
+#define ATOM_FWI_MAXMCLK_PTR	0x28
+
+#define ATOM_CT_SIZE_PTR	0
+#define ATOM_CT_WS_PTR		4
+#define ATOM_CT_PS_PTR		5
+#define ATOM_CT_PS_MASK		0x7F
+#define ATOM_CT_CODE_PTR	6
+
+#define ATOM_OP_CNT		127
+#define ATOM_OP_EOT		91
+
+#define ATOM_CASE_MAGIC		0x63
+#define ATOM_CASE_END		0x5A5A
+
+#define ATOM_ARG_REG		0
+#define ATOM_ARG_PS		1
+#define ATOM_ARG_WS		2
+#define ATOM_ARG_FB		3
+#define ATOM_ARG_ID		4
+#define ATOM_ARG_IMM		5
+#define ATOM_ARG_PLL		6
+#define ATOM_ARG_MC		7
+
+#define ATOM_SRC_DWORD		0
+#define ATOM_SRC_WORD0		1
+#define ATOM_SRC_WORD8		2
+#define ATOM_SRC_WORD16		3
+#define ATOM_SRC_BYTE0		4
+#define ATOM_SRC_BYTE8		5
+#define ATOM_SRC_BYTE16		6
+#define ATOM_SRC_BYTE24		7
+
+#define ATOM_WS_QUOTIENT	0x40
+#define ATOM_WS_REMAINDER	0x41
+#define ATOM_WS_DATAPTR		0x42
+#define ATOM_WS_SHIFT		0x43
+#define ATOM_WS_OR_MASK		0x44
+#define ATOM_WS_AND_MASK	0x45
+#define ATOM_WS_FB_WINDOW	0x46
+#define ATOM_WS_ATTRIBUTES	0x47
+#define ATOM_WS_REGPTR		0x48
+
+#define ATOM_IIO_NOP		0
+#define ATOM_IIO_START		1
+#define ATOM_IIO_READ		2
+#define ATOM_IIO_WRITE		3
+#define ATOM_IIO_CLEAR		4
+#define ATOM_IIO_SET		5
+#define ATOM_IIO_MOVE_INDEX	6
+#define ATOM_IIO_MOVE_ATTR	7
+#define ATOM_IIO_MOVE_DATA	8
+#define ATOM_IIO_END		9
+
+#define ATOM_IO_MM		0
+#define ATOM_IO_PCI		1
+#define ATOM_IO_SYSIO		2
+#define ATOM_IO_IIO		0x80
+
+struct card_info {
+	struct drm_device *dev;
+	void (* reg_write)(struct card_info *, uint32_t, uint32_t);   /*  filled by driver */
+	uint32_t (* reg_read)(struct card_info *, uint32_t);          /*  filled by driver */
+	void (* ioreg_write)(struct card_info *, uint32_t, uint32_t);   /*  filled by driver */
+	uint32_t (* ioreg_read)(struct card_info *, uint32_t);          /*  filled by driver */
+	void (* mc_write)(struct card_info *, uint32_t, uint32_t);   /*  filled by driver */
+	uint32_t (* mc_read)(struct card_info *, uint32_t);          /*  filled by driver */
+	void (* pll_write)(struct card_info *, uint32_t, uint32_t);   /*  filled by driver */
+	uint32_t (* pll_read)(struct card_info *, uint32_t);          /*  filled by driver */
+};
+
+struct atom_context {
+	struct card_info *card;
+	struct rwlock mutex;
+	void *bios;
+	uint32_t cmd_table, data_table;
+	uint16_t *iio;
+
+	uint16_t data_block;
+	uint32_t fb_base;
+	uint32_t divmul[2];
+	uint16_t io_attr;
+	uint16_t reg_block;
+	uint8_t shift;
+	int cs_equal, cs_above;
+	int io_mode;
+	uint32_t *scratch;
+	int scratch_size_bytes;
+	char vbios_version[20];
+};
+
+extern int amdgpu_atom_debug;
+
+struct atom_context *amdgpu_atom_parse(struct card_info *, void *);
+int amdgpu_atom_execute_table(struct atom_context *, int, uint32_t *);
+int amdgpu_atom_asic_init(struct atom_context *);
+void amdgpu_atom_destroy(struct atom_context *);
+bool amdgpu_atom_parse_data_header(struct atom_context *ctx, int index, uint16_t *size,
+			    uint8_t *frev, uint8_t *crev, uint16_t *data_start);
+bool amdgpu_atom_parse_cmd_header(struct atom_context *ctx, int index,
+			   uint8_t *frev, uint8_t *crev);
+#include "atom-types.h"
+#include "atombios.h"
+#include "ObjectID.h"
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/atombios_crtc.h b/sys/dev/pci/drm/amd/amdgpu/atombios_crtc.h
new file mode 100644
index 00000000000..0eeda8e3bf5
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/atombios_crtc.h
@@ -0,0 +1,60 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ATOMBIOS_CRTC_H__
+#define __ATOMBIOS_CRTC_H__
+
+void amdgpu_atombios_crtc_overscan_setup(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode);
+void amdgpu_atombios_crtc_scaler_setup(struct drm_crtc *crtc);
+void amdgpu_atombios_crtc_lock(struct drm_crtc *crtc, int lock);
+void amdgpu_atombios_crtc_enable(struct drm_crtc *crtc, int state);
+void amdgpu_atombios_crtc_blank(struct drm_crtc *crtc, int state);
+void amdgpu_atombios_crtc_powergate(struct drm_crtc *crtc, int state);
+void amdgpu_atombios_crtc_powergate_init(struct amdgpu_device *adev);
+void amdgpu_atombios_crtc_set_dtd_timing(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode);
+void amdgpu_atombios_crtc_set_disp_eng_pll(struct amdgpu_device *adev,
+				    u32 dispclk);
+u32 amdgpu_atombios_crtc_set_dce_clock(struct amdgpu_device *adev,
+				       u32 freq, u8 clk_type, u8 clk_src);
+void amdgpu_atombios_crtc_program_pll(struct drm_crtc *crtc,
+			       u32 crtc_id,
+			       int pll_id,
+			       u32 encoder_mode,
+			       u32 encoder_id,
+			       u32 clock,
+			       u32 ref_div,
+			       u32 fb_div,
+			       u32 frac_fb_div,
+			       u32 post_div,
+			       int bpc,
+			       bool ss_enabled,
+			       struct amdgpu_atom_ss *ss);
+int amdgpu_atombios_crtc_prepare_pll(struct drm_crtc *crtc,
+			      struct drm_display_mode *mode);
+void amdgpu_atombios_crtc_set_pll(struct drm_crtc *crtc,
+			   struct drm_display_mode *mode);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/atombios_dp.h b/sys/dev/pci/drm/amd/amdgpu/atombios_dp.h
new file mode 100644
index 00000000000..f59d85eaddf
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/atombios_dp.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ATOMBIOS_DP_H__
+#define __ATOMBIOS_DP_H__
+
+void amdgpu_atombios_dp_aux_init(struct amdgpu_connector *amdgpu_connector);
+u8 amdgpu_atombios_dp_get_sinktype(struct amdgpu_connector *amdgpu_connector);
+int amdgpu_atombios_dp_get_dpcd(struct amdgpu_connector *amdgpu_connector);
+int amdgpu_atombios_dp_get_panel_mode(struct drm_encoder *encoder,
+			       struct drm_connector *connector);
+void amdgpu_atombios_dp_set_link_config(struct drm_connector *connector,
+				 const struct drm_display_mode *mode);
+int amdgpu_atombios_dp_mode_valid_helper(struct drm_connector *connector,
+				  struct drm_display_mode *mode);
+bool amdgpu_atombios_dp_needs_link_train(struct amdgpu_connector *amdgpu_connector);
+void amdgpu_atombios_dp_set_rx_power_state(struct drm_connector *connector,
+				    u8 power_state);
+void amdgpu_atombios_dp_link_train(struct drm_encoder *encoder,
+			    struct drm_connector *connector);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/atombios_encoders.h b/sys/dev/pci/drm/amd/amdgpu/atombios_encoders.h
new file mode 100644
index 00000000000..f77cbdef679
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/atombios_encoders.h
@@ -0,0 +1,78 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ATOMBIOS_ENCODER_H__
+#define __ATOMBIOS_ENCODER_H__
+
+u8
+amdgpu_atombios_encoder_get_backlight_level_from_reg(struct amdgpu_device *adev);
+void
+amdgpu_atombios_encoder_set_backlight_level_to_reg(struct amdgpu_device *adev,
+						   u8 backlight_level);
+u8
+amdgpu_atombios_encoder_get_backlight_level(struct amdgpu_encoder *amdgpu_encoder);
+void
+amdgpu_atombios_encoder_set_backlight_level(struct amdgpu_encoder *amdgpu_encoder,
+				     u8 level);
+void amdgpu_atombios_encoder_init_backlight(struct amdgpu_encoder *amdgpu_encoder,
+				     struct drm_connector *drm_connector);
+void
+amdgpu_atombios_encoder_fini_backlight(struct amdgpu_encoder *amdgpu_encoder);
+bool amdgpu_atombios_encoder_is_digital(struct drm_encoder *encoder);
+bool amdgpu_atombios_encoder_mode_fixup(struct drm_encoder *encoder,
+				 const struct drm_display_mode *mode,
+				 struct drm_display_mode *adjusted_mode);
+int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder);
+void
+amdgpu_atombios_encoder_setup_dig_encoder(struct drm_encoder *encoder,
+				   int action, int panel_mode);
+void
+amdgpu_atombios_encoder_setup_dig_transmitter(struct drm_encoder *encoder, int action,
+				       uint8_t lane_num, uint8_t lane_set);
+bool
+amdgpu_atombios_encoder_set_edp_panel_power(struct drm_connector *connector,
+				     int action);
+void
+amdgpu_atombios_encoder_dpms(struct drm_encoder *encoder, int mode);
+void
+amdgpu_atombios_encoder_set_crtc_source(struct drm_encoder *encoder);
+void
+amdgpu_atombios_encoder_init_dig(struct amdgpu_device *adev);
+enum drm_connector_status
+amdgpu_atombios_encoder_dac_detect(struct drm_encoder *encoder,
+			    struct drm_connector *connector);
+enum drm_connector_status
+amdgpu_atombios_encoder_dig_detect(struct drm_encoder *encoder,
+			    struct drm_connector *connector);
+void
+amdgpu_atombios_encoder_setup_ext_encoder_ddc(struct drm_encoder *encoder);
+void
+amdgpu_atombios_encoder_set_bios_scratch_regs(struct drm_connector *connector,
+				       struct drm_encoder *encoder,
+				       bool connected);
+struct amdgpu_encoder_atom_dig *
+amdgpu_atombios_encoder_get_lcd_info(struct amdgpu_encoder *encoder);
+struct amdgpu_encoder_atom_dig *
+amdgpu_atombios_encoder_get_dig_info(struct amdgpu_encoder *amdgpu_encoder);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/atombios_i2c.h b/sys/dev/pci/drm/amd/amdgpu/atombios_i2c.h
new file mode 100644
index 00000000000..251aaf41f65
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/atombios_i2c.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ATOMBIOS_I2C_H__
+#define __ATOMBIOS_I2C_H__
+
+int amdgpu_atombios_i2c_xfer(struct i2c_adapter *i2c_adap,
+		      struct i2c_msg *msgs, int num);
+u32 amdgpu_atombios_i2c_func(struct i2c_adapter *adap);
+void amdgpu_atombios_i2c_channel_trans(struct amdgpu_device* adev,
+		u8 slave_addr, u8 line_number, u8 offset, u8 data);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/ci_dpm.h b/sys/dev/pci/drm/amd/amdgpu/ci_dpm.h
new file mode 100644
index 00000000000..91be2996ae7
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/ci_dpm.h
@@ -0,0 +1,349 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __CI_DPM_H__
+#define __CI_DPM_H__
+
+#include "amdgpu_atombios.h"
+#include "ppsmc.h"
+
+#define SMU__NUM_SCLK_DPM_STATE  8
+#define SMU__NUM_MCLK_DPM_LEVELS 6
+#define SMU__NUM_LCLK_DPM_LEVELS 8
+#define SMU__NUM_PCIE_DPM_LEVELS 8
+#include "smu7_discrete.h"
+
+#define CISLANDS_MAX_HARDWARE_POWERLEVELS 2
+
+#define CISLANDS_UNUSED_GPIO_PIN 0x7F
+
+struct ci_pl {
+	u32 mclk;
+	u32 sclk;
+	enum amdgpu_pcie_gen pcie_gen;
+	u16 pcie_lane;
+};
+
+struct ci_ps {
+	u16 performance_level_count;
+	bool dc_compatible;
+	u32 sclk_t;
+	struct ci_pl performance_levels[CISLANDS_MAX_HARDWARE_POWERLEVELS];
+};
+
+struct ci_dpm_level {
+	bool enabled;
+	u32 value;
+	u32 param1;
+};
+
+#define CISLAND_MAX_DEEPSLEEP_DIVIDER_ID 5
+#define MAX_REGULAR_DPM_NUMBER 8
+#define CISLAND_MINIMUM_ENGINE_CLOCK 800
+
+struct ci_single_dpm_table {
+	u32 count;
+	struct ci_dpm_level dpm_levels[MAX_REGULAR_DPM_NUMBER];
+};
+
+struct ci_dpm_table {
+	struct ci_single_dpm_table sclk_table;
+	struct ci_single_dpm_table mclk_table;
+	struct ci_single_dpm_table pcie_speed_table;
+	struct ci_single_dpm_table vddc_table;
+	struct ci_single_dpm_table vddci_table;
+	struct ci_single_dpm_table mvdd_table;
+};
+
+struct ci_mc_reg_entry {
+	u32 mclk_max;
+	u32 mc_data[SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE];
+};
+
+struct ci_mc_reg_table {
+	u8 last;
+	u8 num_entries;
+	u16 valid_flag;
+	struct ci_mc_reg_entry mc_reg_table_entry[MAX_AC_TIMING_ENTRIES];
+	SMU7_Discrete_MCRegisterAddress mc_reg_address[SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE];
+};
+
+struct ci_ulv_parm
+{
+	bool supported;
+	u32 cg_ulv_parameter;
+	u32 volt_change_delay;
+	struct ci_pl pl;
+};
+
+#define CISLANDS_MAX_LEAKAGE_COUNT  8
+
+struct ci_leakage_voltage {
+	u16 count;
+	u16 leakage_id[CISLANDS_MAX_LEAKAGE_COUNT];
+	u16 actual_voltage[CISLANDS_MAX_LEAKAGE_COUNT];
+};
+
+struct ci_dpm_level_enable_mask {
+	u32 uvd_dpm_enable_mask;
+	u32 vce_dpm_enable_mask;
+	u32 acp_dpm_enable_mask;
+	u32 samu_dpm_enable_mask;
+	u32 sclk_dpm_enable_mask;
+	u32 mclk_dpm_enable_mask;
+	u32 pcie_dpm_enable_mask;
+};
+
+struct ci_vbios_boot_state
+{
+	u16 mvdd_bootup_value;
+	u16 vddc_bootup_value;
+	u16 vddci_bootup_value;
+	u32 sclk_bootup_value;
+	u32 mclk_bootup_value;
+	u16 pcie_gen_bootup_value;
+	u16 pcie_lane_bootup_value;
+};
+
+struct ci_clock_registers {
+	u32 cg_spll_func_cntl;
+	u32 cg_spll_func_cntl_2;
+	u32 cg_spll_func_cntl_3;
+	u32 cg_spll_func_cntl_4;
+	u32 cg_spll_spread_spectrum;
+	u32 cg_spll_spread_spectrum_2;
+	u32 dll_cntl;
+	u32 mclk_pwrmgt_cntl;
+	u32 mpll_ad_func_cntl;
+	u32 mpll_dq_func_cntl;
+	u32 mpll_func_cntl;
+	u32 mpll_func_cntl_1;
+	u32 mpll_func_cntl_2;
+	u32 mpll_ss1;
+	u32 mpll_ss2;
+};
+
+struct ci_thermal_temperature_setting {
+	s32 temperature_low;
+	s32 temperature_high;
+	s32 temperature_shutdown;
+};
+
+struct ci_pcie_perf_range {
+	u16 max;
+	u16 min;
+};
+
+enum ci_pt_config_reg_type {
+	CISLANDS_CONFIGREG_MMR = 0,
+	CISLANDS_CONFIGREG_SMC_IND,
+	CISLANDS_CONFIGREG_DIDT_IND,
+	CISLANDS_CONFIGREG_CACHE,
+	CISLANDS_CONFIGREG_MAX
+};
+
+#define POWERCONTAINMENT_FEATURE_BAPM            0x00000001
+#define POWERCONTAINMENT_FEATURE_TDCLimit        0x00000002
+#define POWERCONTAINMENT_FEATURE_PkgPwrLimit     0x00000004
+
+struct ci_pt_config_reg {
+	u32 offset;
+	u32 mask;
+	u32 shift;
+	u32 value;
+	enum ci_pt_config_reg_type type;
+};
+
+struct ci_pt_defaults {
+	u8 svi_load_line_en;
+	u8 svi_load_line_vddc;
+	u8 tdc_vddc_throttle_release_limit_perc;
+	u8 tdc_mawt;
+	u8 tdc_waterfall_ctl;
+	u8 dte_ambient_temp_base;
+	u32 display_cac;
+	u32 bapm_temp_gradient;
+	u16 bapmti_r[SMU7_DTE_ITERATIONS * SMU7_DTE_SOURCES * SMU7_DTE_SINKS];
+	u16 bapmti_rc[SMU7_DTE_ITERATIONS * SMU7_DTE_SOURCES * SMU7_DTE_SINKS];
+};
+
+#define DPMTABLE_OD_UPDATE_SCLK     0x00000001
+#define DPMTABLE_OD_UPDATE_MCLK     0x00000002
+#define DPMTABLE_UPDATE_SCLK        0x00000004
+#define DPMTABLE_UPDATE_MCLK        0x00000008
+
+struct ci_power_info {
+	struct ci_dpm_table dpm_table;
+	struct ci_dpm_table golden_dpm_table;
+	u32 voltage_control;
+	u32 mvdd_control;
+	u32 vddci_control;
+	u32 active_auto_throttle_sources;
+	struct ci_clock_registers clock_registers;
+	u16 acpi_vddc;
+	u16 acpi_vddci;
+	enum amdgpu_pcie_gen force_pcie_gen;
+	enum amdgpu_pcie_gen acpi_pcie_gen;
+	struct ci_leakage_voltage vddc_leakage;
+	struct ci_leakage_voltage vddci_leakage;
+	u16 max_vddc_in_pp_table;
+	u16 min_vddc_in_pp_table;
+	u16 max_vddci_in_pp_table;
+	u16 min_vddci_in_pp_table;
+	u32 mclk_strobe_mode_threshold;
+	u32 mclk_stutter_mode_threshold;
+	u32 mclk_edc_enable_threshold;
+	u32 mclk_edc_wr_enable_threshold;
+	struct ci_vbios_boot_state vbios_boot_state;
+	/* smc offsets */
+	u32 sram_end;
+	u32 dpm_table_start;
+	u32 soft_regs_start;
+	u32 mc_reg_table_start;
+	u32 fan_table_start;
+	u32 arb_table_start;
+	/* smc tables */
+	SMU7_Discrete_DpmTable smc_state_table;
+	SMU7_Discrete_MCRegisters smc_mc_reg_table;
+	SMU7_Discrete_PmFuses smc_powertune_table;
+	/* other stuff */
+	struct ci_mc_reg_table mc_reg_table;
+	struct atom_voltage_table vddc_voltage_table;
+	struct atom_voltage_table vddci_voltage_table;
+	struct atom_voltage_table mvdd_voltage_table;
+	struct ci_ulv_parm ulv;
+	u32 power_containment_features;
+	const struct ci_pt_defaults *powertune_defaults;
+	u32 dte_tj_offset;
+	bool vddc_phase_shed_control;
+	struct ci_thermal_temperature_setting thermal_temp_setting;
+	struct ci_dpm_level_enable_mask dpm_level_enable_mask;
+	u32 need_update_smu7_dpm_table;
+	u32 sclk_dpm_key_disabled;
+	u32 mclk_dpm_key_disabled;
+	u32 pcie_dpm_key_disabled;
+	u32 thermal_sclk_dpm_enabled;
+	struct ci_pcie_perf_range pcie_gen_performance;
+	struct ci_pcie_perf_range pcie_lane_performance;
+	struct ci_pcie_perf_range pcie_gen_powersaving;
+	struct ci_pcie_perf_range pcie_lane_powersaving;
+	u32 activity_target[SMU7_MAX_LEVELS_GRAPHICS];
+	u32 mclk_activity_target;
+	u32 low_sclk_interrupt_t;
+	u32 last_mclk_dpm_enable_mask;
+	u32 sys_pcie_mask;
+	/* caps */
+	bool caps_power_containment;
+	bool caps_cac;
+	bool caps_sq_ramping;
+	bool caps_db_ramping;
+	bool caps_td_ramping;
+	bool caps_tcp_ramping;
+	bool caps_fps;
+	bool caps_sclk_ds;
+	bool caps_sclk_ss_support;
+	bool caps_mclk_ss_support;
+	bool caps_uvd_dpm;
+	bool caps_vce_dpm;
+	bool caps_samu_dpm;
+	bool caps_acp_dpm;
+	bool caps_automatic_dc_transition;
+	bool caps_sclk_throttle_low_notification;
+	bool caps_dynamic_ac_timing;
+	bool caps_od_fuzzy_fan_control_support;
+	/* flags */
+	bool thermal_protection;
+	bool pcie_performance_request;
+	bool dynamic_ss;
+	bool dll_default_on;
+	bool cac_enabled;
+	bool uvd_enabled;
+	bool battery_state;
+	bool pspp_notify_required;
+	bool enable_bapm_feature;
+	bool enable_tdc_limit_feature;
+	bool enable_pkg_pwr_tracking_feature;
+	bool use_pcie_performance_levels;
+	bool use_pcie_powersaving_levels;
+	bool uvd_power_gated;
+	/* driver states */
+	struct amdgpu_ps current_rps;
+	struct ci_ps current_ps;
+	struct amdgpu_ps requested_rps;
+	struct ci_ps requested_ps;
+	/* fan control */
+	bool fan_ctrl_is_in_default_mode;
+	bool fan_is_controlled_by_smc;
+	u32 t_min;
+	u32 fan_ctrl_default_mode;
+};
+
+#define CISLANDS_VOLTAGE_CONTROL_NONE                   0x0
+#define CISLANDS_VOLTAGE_CONTROL_BY_GPIO                0x1
+#define CISLANDS_VOLTAGE_CONTROL_BY_SVID2               0x2
+
+#define CISLANDS_Q88_FORMAT_CONVERSION_UNIT             256
+
+#define CISLANDS_VRC_DFLT0                              0x3FFFC000
+#define CISLANDS_VRC_DFLT1                              0x000400
+#define CISLANDS_VRC_DFLT2                              0xC00080
+#define CISLANDS_VRC_DFLT3                              0xC00200
+#define CISLANDS_VRC_DFLT4                              0xC01680
+#define CISLANDS_VRC_DFLT5                              0xC00033
+#define CISLANDS_VRC_DFLT6                              0xC00033
+#define CISLANDS_VRC_DFLT7                              0x3FFFC000
+
+#define CISLANDS_CGULVPARAMETER_DFLT                    0x00040035
+#define CISLAND_TARGETACTIVITY_DFLT                     30
+#define CISLAND_MCLK_TARGETACTIVITY_DFLT                10
+
+#define PCIE_PERF_REQ_REMOVE_REGISTRY   0
+#define PCIE_PERF_REQ_FORCE_LOWPOWER    1
+#define PCIE_PERF_REQ_PECI_GEN1         2
+#define PCIE_PERF_REQ_PECI_GEN2         3
+#define PCIE_PERF_REQ_PECI_GEN3         4
+
+#define CISLANDS_SSTU_DFLT                               0
+#define CISLANDS_SST_DFLT                                0x00C8
+
+/* XXX are these ok? */
+#define CISLANDS_TEMP_RANGE_MIN (90 * 1000)
+#define CISLANDS_TEMP_RANGE_MAX (120 * 1000)
+
+int amdgpu_ci_copy_bytes_to_smc(struct amdgpu_device *adev,
+			 u32 smc_start_address,
+			 const u8 *src, u32 byte_count, u32 limit);
+void amdgpu_ci_start_smc(struct amdgpu_device *adev);
+void amdgpu_ci_reset_smc(struct amdgpu_device *adev);
+int amdgpu_ci_program_jump_on_start(struct amdgpu_device *adev);
+void amdgpu_ci_stop_smc_clock(struct amdgpu_device *adev);
+void amdgpu_ci_start_smc_clock(struct amdgpu_device *adev);
+bool amdgpu_ci_is_smc_running(struct amdgpu_device *adev);
+PPSMC_Result amdgpu_ci_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg);
+PPSMC_Result amdgpu_ci_wait_for_smc_inactive(struct amdgpu_device *adev);
+int amdgpu_ci_load_smc_ucode(struct amdgpu_device *adev, u32 limit);
+int amdgpu_ci_read_smc_sram_dword(struct amdgpu_device *adev,
+			   u32 smc_address, u32 *value, u32 limit);
+int amdgpu_ci_write_smc_sram_dword(struct amdgpu_device *adev,
+			    u32 smc_address, u32 value, u32 limit);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/cik.h b/sys/dev/pci/drm/amd/amdgpu/cik.h
new file mode 100644
index 00000000000..e49c6f15a0a
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/cik.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __CIK_H__
+#define __CIK_H__
+
+#define CIK_FLUSH_GPU_TLB_NUM_WREG	3
+
+void cik_srbm_select(struct amdgpu_device *adev,
+		     u32 me, u32 pipe, u32 queue, u32 vmid);
+int cik_set_ip_blocks(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/cik_dpm.h b/sys/dev/pci/drm/amd/amdgpu/cik_dpm.h
new file mode 100644
index 00000000000..2a086610f74
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/cik_dpm.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __CIK_DPM_H__
+#define __CIK_DPM_H__
+
+extern const struct amdgpu_ip_block_version ci_smu_ip_block;
+extern const struct amdgpu_ip_block_version kv_smu_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/cik_ih.c b/sys/dev/pci/drm/amd/amdgpu/cik_ih.c
new file mode 100644
index 00000000000..44d10c2172f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/cik_ih.c
@@ -0,0 +1,482 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "cikd.h"
+
+#include "bif/bif_4_1_d.h"
+#include "bif/bif_4_1_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+/*
+ * Interrupts
+ * Starting with r6xx, interrupts are handled via a ring buffer.
+ * Ring buffers are areas of GPU accessible memory that the GPU
+ * writes interrupt vectors into and the host reads vectors out of.
+ * There is a rptr (read pointer) that determines where the
+ * host is currently reading, and a wptr (write pointer)
+ * which determines where the GPU has written.  When the
+ * pointers are equal, the ring is idle.  When the GPU
+ * writes vectors to the ring buffer, it increments the
+ * wptr.  When there is an interrupt, the host then starts
+ * fetching commands and processing them until the pointers are
+ * equal again at which point it updates the rptr.
+ */
+
+static void cik_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * cik_ih_enable_interrupts - Enable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enable the interrupt ring buffer (CIK).
+ */
+static void cik_ih_enable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_cntl = RREG32(mmIH_CNTL);
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+
+	ih_cntl |= IH_CNTL__ENABLE_INTR_MASK;
+	ih_rb_cntl |= IH_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(mmIH_CNTL, ih_cntl);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	adev->irq.ih.enabled = true;
+}
+
+/**
+ * cik_ih_disable_interrupts - Disable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable the interrupt ring buffer (CIK).
+ */
+static void cik_ih_disable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+	u32 ih_cntl = RREG32(mmIH_CNTL);
+
+	ih_rb_cntl &= ~IH_RB_CNTL__RB_ENABLE_MASK;
+	ih_cntl &= ~IH_CNTL__ENABLE_INTR_MASK;
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	WREG32(mmIH_CNTL, ih_cntl);
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+	adev->irq.ih.enabled = false;
+	adev->irq.ih.rptr = 0;
+}
+
+/**
+ * cik_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (CIK).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int cik_ih_irq_init(struct amdgpu_device *adev)
+{
+	int rb_bufsz;
+	u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
+	u64 wptr_off;
+
+	/* disable irqs */
+	cik_ih_disable_interrupts(adev);
+
+	/* setup interrupt control */
+	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
+	interrupt_cntl = RREG32(mmINTERRUPT_CNTL);
+	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl &= ~INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK;
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl &= ~INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK;
+	WREG32(mmINTERRUPT_CNTL, interrupt_cntl);
+
+	WREG32(mmIH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
+	rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
+
+	ih_rb_cntl = (IH_RB_CNTL__WPTR_OVERFLOW_ENABLE_MASK |
+		      IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK |
+		      (rb_bufsz << 1));
+
+	ih_rb_cntl |= IH_RB_CNTL__WPTR_WRITEBACK_ENABLE_MASK;
+
+	/* set the writeback address whether it's enabled or not */
+	wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
+	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
+	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+
+	/* Default settings for IH_CNTL (disabled at first) */
+	ih_cntl = (0x10 << IH_CNTL__MC_WRREQ_CREDIT__SHIFT) |
+		(0x10 << IH_CNTL__MC_WR_CLEAN_CNT__SHIFT) |
+		(0 << IH_CNTL__MC_VMID__SHIFT);
+	/* IH_CNTL__RPTR_REARM_MASK only works if msi's are enabled */
+	if (adev->irq.msi_enabled)
+		ih_cntl |= IH_CNTL__RPTR_REARM_MASK;
+	WREG32(mmIH_CNTL, ih_cntl);
+
+	pci_set_master(adev->pdev);
+
+	/* enable irqs */
+	cik_ih_enable_interrupts(adev);
+
+	return 0;
+}
+
+/**
+ * cik_ih_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw (CIK).
+ */
+static void cik_ih_irq_disable(struct amdgpu_device *adev)
+{
+	cik_ih_disable_interrupts(adev);
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * cik_ih_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer (CIK).  Also check for
+ * ring buffer overflow and deal with it.
+ * Used by cik_irq_process().
+ * Returns the value of the wptr.
+ */
+static u32 cik_ih_get_wptr(struct amdgpu_device *adev)
+{
+	u32 wptr, tmp;
+
+	wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
+
+	if (wptr & IH_RB_WPTR__RB_OVERFLOW_MASK) {
+		wptr &= ~IH_RB_WPTR__RB_OVERFLOW_MASK;
+		/* When a ring buffer overflow happen start parsing interrupt
+		 * from the last not overwritten vector (wptr + 16). Hopefully
+		 * this should allow us to catchup.
+		 */
+		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
+		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
+		tmp = RREG32(mmIH_RB_CNTL);
+		tmp |= IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK;
+		WREG32(mmIH_RB_CNTL, tmp);
+	}
+	return (wptr & adev->irq.ih.ptr_mask);
+}
+
+/*        CIK IV Ring
+ * Each IV ring entry is 128 bits:
+ * [7:0]    - interrupt source id
+ * [31:8]   - reserved
+ * [59:32]  - interrupt source data
+ * [63:60]  - reserved
+ * [71:64]  - RINGID
+ *            CP:
+ *            ME_ID [1:0], PIPE_ID[1:0], QUEUE_ID[2:0]
+ *            QUEUE_ID - for compute, which of the 8 queues owned by the dispatcher
+ *                     - for gfx, hw shader state (0=PS...5=LS, 6=CS)
+ *            ME_ID - 0 = gfx, 1 = first 4 CS pipes, 2 = second 4 CS pipes
+ *            PIPE_ID - ME0 0=3D
+ *                    - ME1&2 compute dispatcher (4 pipes each)
+ *            SDMA:
+ *            INSTANCE_ID [1:0], QUEUE_ID[1:0]
+ *            INSTANCE_ID - 0 = sdma0, 1 = sdma1
+ *            QUEUE_ID - 0 = gfx, 1 = rlc0, 2 = rlc1
+ * [79:72]  - VMID
+ * [95:80]  - PASID
+ * [127:96] - reserved
+ */
+
+/**
+ * cik_ih_prescreen_iv - prescreen an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns true if the interrupt vector should be further processed.
+ */
+static bool cik_ih_prescreen_iv(struct amdgpu_device *adev)
+{
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	u16 pasid;
+
+	switch (le32_to_cpu(adev->irq.ih.ring[ring_index]) & 0xff) {
+	case 146:
+	case 147:
+		pasid = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]) >> 16;
+		if (!pasid || amdgpu_vm_pasid_fault_credit(adev, pasid))
+			return true;
+		break;
+	default:
+		/* Not a VM fault */
+		return true;
+	}
+
+	adev->irq.ih.rptr += 16;
+	return false;
+}
+
+ /**
+ * cik_ih_decode_iv - decode an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Decodes the interrupt vector at the current rptr
+ * position and also advance the position.
+ */
+static void cik_ih_decode_iv(struct amdgpu_device *adev,
+			     struct amdgpu_iv_entry *entry)
+{
+	/* wptr/rptr are in bytes! */
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	uint32_t dw[4];
+
+	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+
+	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->src_id = dw[0] & 0xff;
+	entry->src_data[0] = dw[1] & 0xfffffff;
+	entry->ring_id = dw[2] & 0xff;
+	entry->vmid = (dw[2] >> 8) & 0xff;
+	entry->pasid = (dw[2] >> 16) & 0xffff;
+
+	/* wptr/rptr are in bytes! */
+	adev->irq.ih.rptr += 16;
+}
+
+/**
+ * cik_ih_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void cik_ih_set_rptr(struct amdgpu_device *adev)
+{
+	WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
+}
+
+static int cik_ih_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret;
+
+	ret = amdgpu_irq_add_domain(adev);
+	if (ret)
+		return ret;
+
+	cik_ih_set_interrupt_funcs(adev);
+
+	return 0;
+}
+
+static int cik_ih_sw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int cik_ih_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_fini(adev);
+	amdgpu_ih_ring_fini(adev);
+	amdgpu_irq_remove_domain(adev);
+
+	return 0;
+}
+
+static int cik_ih_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = cik_ih_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int cik_ih_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	cik_ih_irq_disable(adev);
+
+	return 0;
+}
+
+static int cik_ih_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cik_ih_hw_fini(adev);
+}
+
+static int cik_ih_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cik_ih_hw_init(adev);
+}
+
+static bool cik_ih_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		return false;
+
+	return true;
+}
+
+static int cik_ih_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS) & SRBM_STATUS__IH_BUSY_MASK;
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int cik_ih_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	u32 srbm_soft_reset = 0;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_IH_MASK;
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int cik_ih_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int cik_ih_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs cik_ih_ip_funcs = {
+	.name = "cik_ih",
+	.early_init = cik_ih_early_init,
+	.late_init = NULL,
+	.sw_init = cik_ih_sw_init,
+	.sw_fini = cik_ih_sw_fini,
+	.hw_init = cik_ih_hw_init,
+	.hw_fini = cik_ih_hw_fini,
+	.suspend = cik_ih_suspend,
+	.resume = cik_ih_resume,
+	.is_idle = cik_ih_is_idle,
+	.wait_for_idle = cik_ih_wait_for_idle,
+	.soft_reset = cik_ih_soft_reset,
+	.set_clockgating_state = cik_ih_set_clockgating_state,
+	.set_powergating_state = cik_ih_set_powergating_state,
+};
+
+static const struct amdgpu_ih_funcs cik_ih_funcs = {
+	.get_wptr = cik_ih_get_wptr,
+	.prescreen_iv = cik_ih_prescreen_iv,
+	.decode_iv = cik_ih_decode_iv,
+	.set_rptr = cik_ih_set_rptr
+};
+
+static void cik_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih_funcs == NULL)
+		adev->irq.ih_funcs = &cik_ih_funcs;
+}
+
+const struct amdgpu_ip_block_version cik_ih_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_IH,
+	.major = 2,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &cik_ih_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/cik_ih.h b/sys/dev/pci/drm/amd/amdgpu/cik_ih.h
new file mode 100644
index 00000000000..1d9ddee2868
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/cik_ih.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __CIK_IH_H__
+#define __CIK_IH_H__
+
+extern const struct amdgpu_ip_block_version cik_ih_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/cik_sdma.h b/sys/dev/pci/drm/amd/amdgpu/cik_sdma.h
new file mode 100644
index 00000000000..a4a8fe01410
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/cik_sdma.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __CIK_SDMA_H__
+#define __CIK_SDMA_H__
+
+extern const struct amdgpu_ip_block_version cik_sdma_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/cikd.h b/sys/dev/pci/drm/amd/amdgpu/cikd.h
new file mode 100644
index 00000000000..cee6e8a3ad9
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/cikd.h
@@ -0,0 +1,609 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+#ifndef CIK_H
+#define CIK_H
+
+#define MC_SEQ_MISC0__MT__MASK	0xf0000000
+#define MC_SEQ_MISC0__MT__GDDR1  0x10000000
+#define MC_SEQ_MISC0__MT__DDR2   0x20000000
+#define MC_SEQ_MISC0__MT__GDDR3  0x30000000
+#define MC_SEQ_MISC0__MT__GDDR4  0x40000000
+#define MC_SEQ_MISC0__MT__GDDR5  0x50000000
+#define MC_SEQ_MISC0__MT__HBM    0x60000000
+#define MC_SEQ_MISC0__MT__DDR3   0xB0000000
+
+#define CP_ME_TABLE_SIZE    96
+
+/* display controller offsets used for crtc/cur/lut/grph/viewport/etc. */
+#define CRTC0_REGISTER_OFFSET                 (0x1b7c - 0x1b7c)
+#define CRTC1_REGISTER_OFFSET                 (0x1e7c - 0x1b7c)
+#define CRTC2_REGISTER_OFFSET                 (0x417c - 0x1b7c)
+#define CRTC3_REGISTER_OFFSET                 (0x447c - 0x1b7c)
+#define CRTC4_REGISTER_OFFSET                 (0x477c - 0x1b7c)
+#define CRTC5_REGISTER_OFFSET                 (0x4a7c - 0x1b7c)
+
+/* hpd instance offsets */
+#define HPD0_REGISTER_OFFSET                 (0x1807 - 0x1807)
+#define HPD1_REGISTER_OFFSET                 (0x180a - 0x1807)
+#define HPD2_REGISTER_OFFSET                 (0x180d - 0x1807)
+#define HPD3_REGISTER_OFFSET                 (0x1810 - 0x1807)
+#define HPD4_REGISTER_OFFSET                 (0x1813 - 0x1807)
+#define HPD5_REGISTER_OFFSET                 (0x1816 - 0x1807)
+
+#define BONAIRE_GB_ADDR_CONFIG_GOLDEN        0x12010001
+#define HAWAII_GB_ADDR_CONFIG_GOLDEN         0x12011003
+
+#define AMDGPU_NUM_OF_VMIDS	8
+
+#define		PIPEID(x)					((x) << 0)
+#define		MEID(x)						((x) << 2)
+#define		VMID(x)						((x) << 4)
+#define		QUEUEID(x)					((x) << 8)
+
+#define mmCC_DRM_ID_STRAPS				0x1559
+#define CC_DRM_ID_STRAPS__ATI_REV_ID_MASK		0xf0000000
+
+#define mmCHUB_CONTROL					0x619
+#define		BYPASS_VM					(1 << 0)
+
+#define		SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU	(0 << 5)
+
+#define mmGRPH_LUT_10BIT_BYPASS_CONTROL			0x1a02
+#define		LUT_10BIT_BYPASS_EN			(1 << 8)
+
+#       define CURSOR_MONO                    0
+#       define CURSOR_24_1                    1
+#       define CURSOR_24_8_PRE_MULT           2
+#       define CURSOR_24_8_UNPRE_MULT         3
+#       define CURSOR_URGENT_ALWAYS           0
+#       define CURSOR_URGENT_1_8              1
+#       define CURSOR_URGENT_1_4              2
+#       define CURSOR_URGENT_3_8              3
+#       define CURSOR_URGENT_1_2              4
+
+#       define GRPH_DEPTH_8BPP                0
+#       define GRPH_DEPTH_16BPP               1
+#       define GRPH_DEPTH_32BPP               2
+/* 8 BPP */
+#       define GRPH_FORMAT_INDEXED            0
+/* 16 BPP */
+#       define GRPH_FORMAT_ARGB1555           0
+#       define GRPH_FORMAT_ARGB565            1
+#       define GRPH_FORMAT_ARGB4444           2
+#       define GRPH_FORMAT_AI88               3
+#       define GRPH_FORMAT_MONO16             4
+#       define GRPH_FORMAT_BGRA5551           5
+/* 32 BPP */
+#       define GRPH_FORMAT_ARGB8888           0
+#       define GRPH_FORMAT_ARGB2101010        1
+#       define GRPH_FORMAT_32BPP_DIG          2
+#       define GRPH_FORMAT_8B_ARGB2101010     3
+#       define GRPH_FORMAT_BGRA1010102        4
+#       define GRPH_FORMAT_8B_BGRA1010102     5
+#       define GRPH_FORMAT_RGB111110          6
+#       define GRPH_FORMAT_BGR101111          7
+#       define ADDR_SURF_MACRO_TILE_ASPECT_1  0
+#       define ADDR_SURF_MACRO_TILE_ASPECT_2  1
+#       define ADDR_SURF_MACRO_TILE_ASPECT_4  2
+#       define ADDR_SURF_MACRO_TILE_ASPECT_8  3
+#       define GRPH_ARRAY_LINEAR_GENERAL      0
+#       define GRPH_ARRAY_LINEAR_ALIGNED      1
+#       define GRPH_ARRAY_1D_TILED_THIN1      2
+#       define GRPH_ARRAY_2D_TILED_THIN1      4
+#       define DISPLAY_MICRO_TILING          0
+#       define THIN_MICRO_TILING             1
+#       define DEPTH_MICRO_TILING            2
+#       define ROTATED_MICRO_TILING          4
+#       define GRPH_ENDIAN_NONE               0
+#       define GRPH_ENDIAN_8IN16              1
+#       define GRPH_ENDIAN_8IN32              2
+#       define GRPH_ENDIAN_8IN64              3
+#       define GRPH_RED_SEL_R                 0
+#       define GRPH_RED_SEL_G                 1
+#       define GRPH_RED_SEL_B                 2
+#       define GRPH_RED_SEL_A                 3
+#       define GRPH_GREEN_SEL_G               0
+#       define GRPH_GREEN_SEL_B               1
+#       define GRPH_GREEN_SEL_A               2
+#       define GRPH_GREEN_SEL_R               3
+#       define GRPH_BLUE_SEL_B                0
+#       define GRPH_BLUE_SEL_A                1
+#       define GRPH_BLUE_SEL_R                2
+#       define GRPH_BLUE_SEL_G                3
+#       define GRPH_ALPHA_SEL_A               0
+#       define GRPH_ALPHA_SEL_R               1
+#       define GRPH_ALPHA_SEL_G               2
+#       define GRPH_ALPHA_SEL_B               3
+#       define INPUT_GAMMA_USE_LUT                  0
+#       define INPUT_GAMMA_BYPASS                   1
+#       define INPUT_GAMMA_SRGB_24                  2
+#       define INPUT_GAMMA_XVYCC_222                3
+
+#       define INPUT_CSC_BYPASS                     0
+#       define INPUT_CSC_PROG_COEFF                 1
+#       define INPUT_CSC_PROG_SHARED_MATRIXA        2
+
+#       define OUTPUT_CSC_BYPASS                    0
+#       define OUTPUT_CSC_TV_RGB                    1
+#       define OUTPUT_CSC_YCBCR_601                 2
+#       define OUTPUT_CSC_YCBCR_709                 3
+#       define OUTPUT_CSC_PROG_COEFF                4
+#       define OUTPUT_CSC_PROG_SHARED_MATRIXB       5
+
+#       define DEGAMMA_BYPASS                       0
+#       define DEGAMMA_SRGB_24                      1
+#       define DEGAMMA_XVYCC_222                    2
+#       define GAMUT_REMAP_BYPASS                   0
+#       define GAMUT_REMAP_PROG_COEFF               1
+#       define GAMUT_REMAP_PROG_SHARED_MATRIXA      2
+#       define GAMUT_REMAP_PROG_SHARED_MATRIXB      3
+
+#       define REGAMMA_BYPASS                       0
+#       define REGAMMA_SRGB_24                      1
+#       define REGAMMA_XVYCC_222                    2
+#       define REGAMMA_PROG_A                       3
+#       define REGAMMA_PROG_B                       4
+
+#       define FMT_CLAMP_6BPC                0
+#       define FMT_CLAMP_8BPC                1
+#       define FMT_CLAMP_10BPC               2
+
+#       define HDMI_24BIT_DEEP_COLOR         0
+#       define HDMI_30BIT_DEEP_COLOR         1
+#       define HDMI_36BIT_DEEP_COLOR         2
+#       define HDMI_ACR_HW                   0
+#       define HDMI_ACR_32                   1
+#       define HDMI_ACR_44                   2
+#       define HDMI_ACR_48                   3
+#       define HDMI_ACR_X1                   1
+#       define HDMI_ACR_X2                   2
+#       define HDMI_ACR_X4                   4
+#       define AFMT_AVI_INFO_Y_RGB           0
+#       define AFMT_AVI_INFO_Y_YCBCR422      1
+#       define AFMT_AVI_INFO_Y_YCBCR444      2
+
+#define			NO_AUTO						0
+#define			ES_AUTO						1
+#define			GS_AUTO						2
+#define			ES_AND_GS_AUTO					3
+
+#       define ARRAY_MODE(x)					((x) << 2)
+#       define PIPE_CONFIG(x)					((x) << 6)
+#       define TILE_SPLIT(x)					((x) << 11)
+#       define MICRO_TILE_MODE_NEW(x)				((x) << 22)
+#       define SAMPLE_SPLIT(x)					((x) << 25)
+#       define BANK_WIDTH(x)					((x) << 0)
+#       define BANK_HEIGHT(x)					((x) << 2)
+#       define MACRO_TILE_ASPECT(x)				((x) << 4)
+#       define NUM_BANKS(x)					((x) << 6)
+
+#define		MSG_ENTER_RLC_SAFE_MODE			1
+#define		MSG_EXIT_RLC_SAFE_MODE			0
+
+/*
+ * PM4
+ */
+#define	PACKET_TYPE0	0
+#define	PACKET_TYPE1	1
+#define	PACKET_TYPE2	2
+#define	PACKET_TYPE3	3
+
+#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3)
+#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF)
+#define CP_PACKET0_GET_REG(h) ((h) & 0xFFFF)
+#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF)
+#define PACKET0(reg, n)	((PACKET_TYPE0 << 30) |				\
+			 ((reg) & 0xFFFF) |			\
+			 ((n) & 0x3FFF) << 16)
+#define CP_PACKET2			0x80000000
+#define		PACKET2_PAD_SHIFT		0
+#define		PACKET2_PAD_MASK		(0x3fffffff << 0)
+
+#define PACKET2(v)	(CP_PACKET2 | REG_SET(PACKET2_PAD, (v)))
+
+#define PACKET3(op, n)	((PACKET_TYPE3 << 30) |				\
+			 (((op) & 0xFF) << 8) |				\
+			 ((n) & 0x3FFF) << 16)
+
+#define PACKET3_COMPUTE(op, n) (PACKET3(op, n) | 1 << 1)
+
+/* Packet 3 types */
+#define	PACKET3_NOP					0x10
+#define	PACKET3_SET_BASE				0x11
+#define		PACKET3_BASE_INDEX(x)                  ((x) << 0)
+#define			CE_PARTITION_BASE		3
+#define	PACKET3_CLEAR_STATE				0x12
+#define	PACKET3_INDEX_BUFFER_SIZE			0x13
+#define	PACKET3_DISPATCH_DIRECT				0x15
+#define	PACKET3_DISPATCH_INDIRECT			0x16
+#define	PACKET3_ATOMIC_GDS				0x1D
+#define	PACKET3_ATOMIC_MEM				0x1E
+#define	PACKET3_OCCLUSION_QUERY				0x1F
+#define	PACKET3_SET_PREDICATION				0x20
+#define	PACKET3_REG_RMW					0x21
+#define	PACKET3_COND_EXEC				0x22
+#define	PACKET3_PRED_EXEC				0x23
+#define	PACKET3_DRAW_INDIRECT				0x24
+#define	PACKET3_DRAW_INDEX_INDIRECT			0x25
+#define	PACKET3_INDEX_BASE				0x26
+#define	PACKET3_DRAW_INDEX_2				0x27
+#define	PACKET3_CONTEXT_CONTROL				0x28
+#define	PACKET3_INDEX_TYPE				0x2A
+#define	PACKET3_DRAW_INDIRECT_MULTI			0x2C
+#define	PACKET3_DRAW_INDEX_AUTO				0x2D
+#define	PACKET3_NUM_INSTANCES				0x2F
+#define	PACKET3_DRAW_INDEX_MULTI_AUTO			0x30
+#define	PACKET3_INDIRECT_BUFFER_CONST			0x33
+#define	PACKET3_STRMOUT_BUFFER_UPDATE			0x34
+#define	PACKET3_DRAW_INDEX_OFFSET_2			0x35
+#define	PACKET3_DRAW_PREAMBLE				0x36
+#define	PACKET3_WRITE_DATA				0x37
+#define		WRITE_DATA_DST_SEL(x)                   ((x) << 8)
+		/* 0 - register
+		 * 1 - memory (sync - via GRBM)
+		 * 2 - gl2
+		 * 3 - gds
+		 * 4 - reserved
+		 * 5 - memory (async - direct)
+		 */
+#define		WR_ONE_ADDR                             (1 << 16)
+#define		WR_CONFIRM                              (1 << 20)
+#define		WRITE_DATA_CACHE_POLICY(x)              ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 */
+#define		WRITE_DATA_ENGINE_SEL(x)                ((x) << 30)
+		/* 0 - me
+		 * 1 - pfp
+		 * 2 - ce
+		 */
+#define	PACKET3_DRAW_INDEX_INDIRECT_MULTI		0x38
+#define	PACKET3_MEM_SEMAPHORE				0x39
+#              define PACKET3_SEM_USE_MAILBOX       (0x1 << 16)
+#              define PACKET3_SEM_SEL_SIGNAL_TYPE   (0x1 << 20) /* 0 = increment, 1 = write 1 */
+#              define PACKET3_SEM_CLIENT_CODE	    ((x) << 24) /* 0 = CP, 1 = CB, 2 = DB */
+#              define PACKET3_SEM_SEL_SIGNAL	    (0x6 << 29)
+#              define PACKET3_SEM_SEL_WAIT	    (0x7 << 29)
+#define	PACKET3_COPY_DW					0x3B
+#define	PACKET3_WAIT_REG_MEM				0x3C
+#define		WAIT_REG_MEM_FUNCTION(x)                ((x) << 0)
+		/* 0 - always
+		 * 1 - <
+		 * 2 - <=
+		 * 3 - ==
+		 * 4 - !=
+		 * 5 - >=
+		 * 6 - >
+		 */
+#define		WAIT_REG_MEM_MEM_SPACE(x)               ((x) << 4)
+		/* 0 - reg
+		 * 1 - mem
+		 */
+#define		WAIT_REG_MEM_OPERATION(x)               ((x) << 6)
+		/* 0 - wait_reg_mem
+		 * 1 - wr_wait_wr_reg
+		 */
+#define		WAIT_REG_MEM_ENGINE(x)                  ((x) << 8)
+		/* 0 - me
+		 * 1 - pfp
+		 */
+#define	PACKET3_INDIRECT_BUFFER				0x3F
+#define		INDIRECT_BUFFER_TCL2_VOLATILE           (1 << 22)
+#define		INDIRECT_BUFFER_VALID                   (1 << 23)
+#define		INDIRECT_BUFFER_CACHE_POLICY(x)         ((x) << 28)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#define	PACKET3_COPY_DATA				0x40
+#define	PACKET3_PFP_SYNC_ME				0x42
+#define	PACKET3_SURFACE_SYNC				0x43
+#              define PACKET3_DEST_BASE_0_ENA      (1 << 0)
+#              define PACKET3_DEST_BASE_1_ENA      (1 << 1)
+#              define PACKET3_CB0_DEST_BASE_ENA    (1 << 6)
+#              define PACKET3_CB1_DEST_BASE_ENA    (1 << 7)
+#              define PACKET3_CB2_DEST_BASE_ENA    (1 << 8)
+#              define PACKET3_CB3_DEST_BASE_ENA    (1 << 9)
+#              define PACKET3_CB4_DEST_BASE_ENA    (1 << 10)
+#              define PACKET3_CB5_DEST_BASE_ENA    (1 << 11)
+#              define PACKET3_CB6_DEST_BASE_ENA    (1 << 12)
+#              define PACKET3_CB7_DEST_BASE_ENA    (1 << 13)
+#              define PACKET3_DB_DEST_BASE_ENA     (1 << 14)
+#              define PACKET3_TCL1_VOL_ACTION_ENA  (1 << 15)
+#              define PACKET3_TC_VOL_ACTION_ENA    (1 << 16) /* L2 */
+#              define PACKET3_TC_WB_ACTION_ENA     (1 << 18) /* L2 */
+#              define PACKET3_DEST_BASE_2_ENA      (1 << 19)
+#              define PACKET3_DEST_BASE_3_ENA      (1 << 21)
+#              define PACKET3_TCL1_ACTION_ENA      (1 << 22)
+#              define PACKET3_TC_ACTION_ENA        (1 << 23) /* L2 */
+#              define PACKET3_CB_ACTION_ENA        (1 << 25)
+#              define PACKET3_DB_ACTION_ENA        (1 << 26)
+#              define PACKET3_SH_KCACHE_ACTION_ENA (1 << 27)
+#              define PACKET3_SH_KCACHE_VOL_ACTION_ENA (1 << 28)
+#              define PACKET3_SH_ICACHE_ACTION_ENA (1 << 29)
+#define	PACKET3_COND_WRITE				0x45
+#define	PACKET3_EVENT_WRITE				0x46
+#define		EVENT_TYPE(x)                           ((x) << 0)
+#define		EVENT_INDEX(x)                          ((x) << 8)
+		/* 0 - any non-TS event
+		 * 1 - ZPASS_DONE, PIXEL_PIPE_STAT_*
+		 * 2 - SAMPLE_PIPELINESTAT
+		 * 3 - SAMPLE_STREAMOUTSTAT*
+		 * 4 - *S_PARTIAL_FLUSH
+		 * 5 - EOP events
+		 * 6 - EOS events
+		 */
+#define	PACKET3_EVENT_WRITE_EOP				0x47
+#define		EOP_TCL1_VOL_ACTION_EN                  (1 << 12)
+#define		EOP_TC_VOL_ACTION_EN                    (1 << 13) /* L2 */
+#define		EOP_TC_WB_ACTION_EN                     (1 << 15) /* L2 */
+#define		EOP_TCL1_ACTION_EN                      (1 << 16)
+#define		EOP_TC_ACTION_EN                        (1 << 17) /* L2 */
+#define		EOP_TCL2_VOLATILE                       (1 << 24)
+#define		EOP_CACHE_POLICY(x)                     ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#define		DATA_SEL(x)                             ((x) << 29)
+		/* 0 - discard
+		 * 1 - send low 32bit data
+		 * 2 - send 64bit data
+		 * 3 - send 64bit GPU counter value
+		 * 4 - send 64bit sys counter value
+		 */
+#define		INT_SEL(x)                              ((x) << 24)
+		/* 0 - none
+		 * 1 - interrupt only (DATA_SEL = 0)
+		 * 2 - interrupt when data write is confirmed
+		 */
+#define		DST_SEL(x)                              ((x) << 16)
+		/* 0 - MC
+		 * 1 - TC/L2
+		 */
+#define	PACKET3_EVENT_WRITE_EOS				0x48
+#define	PACKET3_RELEASE_MEM				0x49
+#define	PACKET3_PREAMBLE_CNTL				0x4A
+#              define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE     (2 << 28)
+#              define PACKET3_PREAMBLE_END_CLEAR_STATE       (3 << 28)
+#define	PACKET3_DMA_DATA				0x50
+/* 1. header
+ * 2. CONTROL
+ * 3. SRC_ADDR_LO or DATA [31:0]
+ * 4. SRC_ADDR_HI [31:0]
+ * 5. DST_ADDR_LO [31:0]
+ * 6. DST_ADDR_HI [7:0]
+ * 7. COMMAND [30:21] | BYTE_COUNT [20:0]
+ */
+/* CONTROL */
+#              define PACKET3_DMA_DATA_ENGINE(x)     ((x) << 0)
+		/* 0 - ME
+		 * 1 - PFP
+		 */
+#              define PACKET3_DMA_DATA_SRC_CACHE_POLICY(x) ((x) << 13)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#              define PACKET3_DMA_DATA_SRC_VOLATILE (1 << 15)
+#              define PACKET3_DMA_DATA_DST_SEL(x)  ((x) << 20)
+		/* 0 - DST_ADDR using DAS
+		 * 1 - GDS
+		 * 3 - DST_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_DST_CACHE_POLICY(x) ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#              define PACKET3_DMA_DATA_DST_VOLATILE (1 << 27)
+#              define PACKET3_DMA_DATA_SRC_SEL(x)  ((x) << 29)
+		/* 0 - SRC_ADDR using SAS
+		 * 1 - GDS
+		 * 2 - DATA
+		 * 3 - SRC_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_CP_SYNC     (1 << 31)
+/* COMMAND */
+#              define PACKET3_DMA_DATA_DIS_WC      (1 << 21)
+#              define PACKET3_DMA_DATA_CMD_SRC_SWAP(x) ((x) << 22)
+		/* 0 - none
+		 * 1 - 8 in 16
+		 * 2 - 8 in 32
+		 * 3 - 8 in 64
+		 */
+#              define PACKET3_DMA_DATA_CMD_DST_SWAP(x) ((x) << 24)
+		/* 0 - none
+		 * 1 - 8 in 16
+		 * 2 - 8 in 32
+		 * 3 - 8 in 64
+		 */
+#              define PACKET3_DMA_DATA_CMD_SAS     (1 << 26)
+		/* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_DAS     (1 << 27)
+		/* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_SAIC    (1 << 28)
+#              define PACKET3_DMA_DATA_CMD_DAIC    (1 << 29)
+#              define PACKET3_DMA_DATA_CMD_RAW_WAIT  (1 << 30)
+#define	PACKET3_AQUIRE_MEM				0x58
+#define	PACKET3_REWIND					0x59
+#define	PACKET3_LOAD_UCONFIG_REG			0x5E
+#define	PACKET3_LOAD_SH_REG				0x5F
+#define	PACKET3_LOAD_CONFIG_REG				0x60
+#define	PACKET3_LOAD_CONTEXT_REG			0x61
+#define	PACKET3_SET_CONFIG_REG				0x68
+#define		PACKET3_SET_CONFIG_REG_START			0x00002000
+#define		PACKET3_SET_CONFIG_REG_END			0x00002c00
+#define	PACKET3_SET_CONTEXT_REG				0x69
+#define		PACKET3_SET_CONTEXT_REG_START			0x0000a000
+#define		PACKET3_SET_CONTEXT_REG_END			0x0000a400
+#define	PACKET3_SET_CONTEXT_REG_INDIRECT		0x73
+#define	PACKET3_SET_SH_REG				0x76
+#define		PACKET3_SET_SH_REG_START			0x00002c00
+#define		PACKET3_SET_SH_REG_END				0x00003000
+#define	PACKET3_SET_SH_REG_OFFSET			0x77
+#define	PACKET3_SET_QUEUE_REG				0x78
+#define	PACKET3_SET_UCONFIG_REG				0x79
+#define		PACKET3_SET_UCONFIG_REG_START			0x0000c000
+#define		PACKET3_SET_UCONFIG_REG_END			0x0000c400
+#define	PACKET3_SCRATCH_RAM_WRITE			0x7D
+#define	PACKET3_SCRATCH_RAM_READ			0x7E
+#define	PACKET3_LOAD_CONST_RAM				0x80
+#define	PACKET3_WRITE_CONST_RAM				0x81
+#define	PACKET3_DUMP_CONST_RAM				0x83
+#define	PACKET3_INCREMENT_CE_COUNTER			0x84
+#define	PACKET3_INCREMENT_DE_COUNTER			0x85
+#define	PACKET3_WAIT_ON_CE_COUNTER			0x86
+#define	PACKET3_WAIT_ON_DE_COUNTER_DIFF			0x88
+#define	PACKET3_SWITCH_BUFFER				0x8B
+
+/* SDMA - first instance at 0xd000, second at 0xd800 */
+#define SDMA0_REGISTER_OFFSET                             0x0 /* not a register */
+#define SDMA1_REGISTER_OFFSET                             0x200 /* not a register */
+#define SDMA_MAX_INSTANCE 2
+
+#define SDMA_PACKET(op, sub_op, e)	((((e) & 0xFFFF) << 16) |	\
+					 (((sub_op) & 0xFF) << 8) |	\
+					 (((op) & 0xFF) << 0))
+/* sDMA opcodes */
+#define	SDMA_OPCODE_NOP					  0
+#	define SDMA_NOP_COUNT(x)			  (((x) & 0x3FFF) << 16)
+#define	SDMA_OPCODE_COPY				  1
+#       define SDMA_COPY_SUB_OPCODE_LINEAR                0
+#       define SDMA_COPY_SUB_OPCODE_TILED                 1
+#       define SDMA_COPY_SUB_OPCODE_SOA                   3
+#       define SDMA_COPY_SUB_OPCODE_LINEAR_SUB_WINDOW     4
+#       define SDMA_COPY_SUB_OPCODE_TILED_SUB_WINDOW      5
+#       define SDMA_COPY_SUB_OPCODE_T2T_SUB_WINDOW        6
+#define	SDMA_OPCODE_WRITE				  2
+#       define SDMA_WRITE_SUB_OPCODE_LINEAR               0
+#       define SDMA_WRITE_SUB_OPCODE_TILED                1
+#define	SDMA_OPCODE_INDIRECT_BUFFER			  4
+#define	SDMA_OPCODE_FENCE				  5
+#define	SDMA_OPCODE_TRAP				  6
+#define	SDMA_OPCODE_SEMAPHORE				  7
+#       define SDMA_SEMAPHORE_EXTRA_O                     (1 << 13)
+		/* 0 - increment
+		 * 1 - write 1
+		 */
+#       define SDMA_SEMAPHORE_EXTRA_S                     (1 << 14)
+		/* 0 - wait
+		 * 1 - signal
+		 */
+#       define SDMA_SEMAPHORE_EXTRA_M                     (1 << 15)
+		/* mailbox */
+#define	SDMA_OPCODE_POLL_REG_MEM			  8
+#       define SDMA_POLL_REG_MEM_EXTRA_OP(x)              ((x) << 10)
+		/* 0 - wait_reg_mem
+		 * 1 - wr_wait_wr_reg
+		 */
+#       define SDMA_POLL_REG_MEM_EXTRA_FUNC(x)            ((x) << 12)
+		/* 0 - always
+		 * 1 - <
+		 * 2 - <=
+		 * 3 - ==
+		 * 4 - !=
+		 * 5 - >=
+		 * 6 - >
+		 */
+#       define SDMA_POLL_REG_MEM_EXTRA_M                  (1 << 15)
+		/* 0 = register
+		 * 1 = memory
+		 */
+#define	SDMA_OPCODE_COND_EXEC				  9
+#define	SDMA_OPCODE_CONSTANT_FILL			  11
+#       define SDMA_CONSTANT_FILL_EXTRA_SIZE(x)           ((x) << 14)
+		/* 0 = byte fill
+		 * 2 = DW fill
+		 */
+#define	SDMA_OPCODE_GENERATE_PTE_PDE			  12
+#define	SDMA_OPCODE_TIMESTAMP				  13
+#       define SDMA_TIMESTAMP_SUB_OPCODE_SET_LOCAL        0
+#       define SDMA_TIMESTAMP_SUB_OPCODE_GET_LOCAL        1
+#       define SDMA_TIMESTAMP_SUB_OPCODE_GET_GLOBAL       2
+#define	SDMA_OPCODE_SRBM_WRITE				  14
+#       define SDMA_SRBM_WRITE_EXTRA_BYTE_ENABLE(x)       ((x) << 12)
+		/* byte mask */
+
+#define VCE_CMD_NO_OP		0x00000000
+#define VCE_CMD_END		0x00000001
+#define VCE_CMD_IB		0x00000002
+#define VCE_CMD_FENCE		0x00000003
+#define VCE_CMD_TRAP		0x00000004
+#define VCE_CMD_IB_AUTO		0x00000005
+#define VCE_CMD_SEMAPHORE	0x00000006
+
+/* if PTR32, these are the bases for scratch and lds */
+#define	PRIVATE_BASE(x)	((x) << 0) /* scratch */
+#define	SHARED_BASE(x)	((x) << 16) /* LDS */
+
+#define KFD_CIK_SDMA_QUEUE_OFFSET (mmSDMA0_RLC1_RB_CNTL - mmSDMA0_RLC0_RB_CNTL)
+
+/* valid for both DEFAULT_MTYPE and APE1_MTYPE */
+enum {
+	MTYPE_CACHED = 0,
+	MTYPE_NONCACHED = 3
+};
+
+/* mmPA_SC_RASTER_CONFIG mask */
+#define RB_MAP_PKR0(x)				((x) << 0)
+#define RB_MAP_PKR0_MASK			(0x3 << 0)
+#define RB_MAP_PKR1(x)				((x) << 2)
+#define RB_MAP_PKR1_MASK			(0x3 << 2)
+#define RB_XSEL2(x)				((x) << 4)
+#define RB_XSEL2_MASK				(0x3 << 4)
+#define RB_XSEL					(1 << 6)
+#define RB_YSEL					(1 << 7)
+#define PKR_MAP(x)				((x) << 8)
+#define PKR_MAP_MASK				(0x3 << 8)
+#define PKR_XSEL(x)				((x) << 10)
+#define PKR_XSEL_MASK				(0x3 << 10)
+#define PKR_YSEL(x)				((x) << 12)
+#define PKR_YSEL_MASK				(0x3 << 12)
+#define SC_MAP(x)				((x) << 16)
+#define SC_MAP_MASK				(0x3 << 16)
+#define SC_XSEL(x)				((x) << 18)
+#define SC_XSEL_MASK				(0x3 << 18)
+#define SC_YSEL(x)				((x) << 20)
+#define SC_YSEL_MASK				(0x3 << 20)
+#define SE_MAP(x)				((x) << 24)
+#define SE_MAP_MASK				(0x3 << 24)
+#define SE_XSEL(x)				((x) << 26)
+#define SE_XSEL_MASK				(0x3 << 26)
+#define SE_YSEL(x)				((x) << 28)
+#define SE_YSEL_MASK				(0x3 << 28)
+
+/* mmPA_SC_RASTER_CONFIG_1 mask */
+#define SE_PAIR_MAP(x)				((x) << 0)
+#define SE_PAIR_MAP_MASK			(0x3 << 0)
+#define SE_PAIR_XSEL(x)				((x) << 2)
+#define SE_PAIR_XSEL_MASK			(0x3 << 2)
+#define SE_PAIR_YSEL(x)				((x) << 4)
+#define SE_PAIR_YSEL_MASK			(0x3 << 4)
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/clearstate_ci.h b/sys/dev/pci/drm/amd/amdgpu/clearstate_ci.h
new file mode 100644
index 00000000000..c3982f9475f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/clearstate_ci.h
@@ -0,0 +1,944 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+static const unsigned int ci_SECT_CONTEXT_def_1[] =
+{
+    0x00000000, // DB_RENDER_CONTROL
+    0x00000000, // DB_COUNT_CONTROL
+    0x00000000, // DB_DEPTH_VIEW
+    0x00000000, // DB_RENDER_OVERRIDE
+    0x00000000, // DB_RENDER_OVERRIDE2
+    0x00000000, // DB_HTILE_DATA_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // DB_DEPTH_BOUNDS_MIN
+    0x00000000, // DB_DEPTH_BOUNDS_MAX
+    0x00000000, // DB_STENCIL_CLEAR
+    0x00000000, // DB_DEPTH_CLEAR
+    0x00000000, // PA_SC_SCREEN_SCISSOR_TL
+    0x40004000, // PA_SC_SCREEN_SCISSOR_BR
+    0, // HOLE
+    0x00000000, // DB_DEPTH_INFO
+    0x00000000, // DB_Z_INFO
+    0x00000000, // DB_STENCIL_INFO
+    0x00000000, // DB_Z_READ_BASE
+    0x00000000, // DB_STENCIL_READ_BASE
+    0x00000000, // DB_Z_WRITE_BASE
+    0x00000000, // DB_STENCIL_WRITE_BASE
+    0x00000000, // DB_DEPTH_SIZE
+    0x00000000, // DB_DEPTH_SLICE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // TA_BC_BASE_ADDR
+    0x00000000, // TA_BC_BASE_ADDR_HI
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // COHER_DEST_BASE_HI_0
+    0x00000000, // COHER_DEST_BASE_HI_1
+    0x00000000, // COHER_DEST_BASE_HI_2
+    0x00000000, // COHER_DEST_BASE_HI_3
+    0x00000000, // COHER_DEST_BASE_2
+    0x00000000, // COHER_DEST_BASE_3
+    0x00000000, // PA_SC_WINDOW_OFFSET
+    0x80000000, // PA_SC_WINDOW_SCISSOR_TL
+    0x40004000, // PA_SC_WINDOW_SCISSOR_BR
+    0x0000ffff, // PA_SC_CLIPRECT_RULE
+    0x00000000, // PA_SC_CLIPRECT_0_TL
+    0x40004000, // PA_SC_CLIPRECT_0_BR
+    0x00000000, // PA_SC_CLIPRECT_1_TL
+    0x40004000, // PA_SC_CLIPRECT_1_BR
+    0x00000000, // PA_SC_CLIPRECT_2_TL
+    0x40004000, // PA_SC_CLIPRECT_2_BR
+    0x00000000, // PA_SC_CLIPRECT_3_TL
+    0x40004000, // PA_SC_CLIPRECT_3_BR
+    0xaa99aaaa, // PA_SC_EDGERULE
+    0x00000000, // PA_SU_HARDWARE_SCREEN_OFFSET
+    0xffffffff, // CB_TARGET_MASK
+    0xffffffff, // CB_SHADER_MASK
+    0x80000000, // PA_SC_GENERIC_SCISSOR_TL
+    0x40004000, // PA_SC_GENERIC_SCISSOR_BR
+    0x00000000, // COHER_DEST_BASE_0
+    0x00000000, // COHER_DEST_BASE_1
+    0x80000000, // PA_SC_VPORT_SCISSOR_0_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_0_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_1_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_1_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_2_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_2_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_3_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_3_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_4_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_4_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_5_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_5_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_6_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_6_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_7_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_7_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_8_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_8_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_9_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_9_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_10_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_10_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_11_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_11_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_12_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_12_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_13_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_13_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_14_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_14_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_15_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_15_BR
+    0x00000000, // PA_SC_VPORT_ZMIN_0
+    0x3f800000, // PA_SC_VPORT_ZMAX_0
+    0x00000000, // PA_SC_VPORT_ZMIN_1
+    0x3f800000, // PA_SC_VPORT_ZMAX_1
+    0x00000000, // PA_SC_VPORT_ZMIN_2
+    0x3f800000, // PA_SC_VPORT_ZMAX_2
+    0x00000000, // PA_SC_VPORT_ZMIN_3
+    0x3f800000, // PA_SC_VPORT_ZMAX_3
+    0x00000000, // PA_SC_VPORT_ZMIN_4
+    0x3f800000, // PA_SC_VPORT_ZMAX_4
+    0x00000000, // PA_SC_VPORT_ZMIN_5
+    0x3f800000, // PA_SC_VPORT_ZMAX_5
+    0x00000000, // PA_SC_VPORT_ZMIN_6
+    0x3f800000, // PA_SC_VPORT_ZMAX_6
+    0x00000000, // PA_SC_VPORT_ZMIN_7
+    0x3f800000, // PA_SC_VPORT_ZMAX_7
+    0x00000000, // PA_SC_VPORT_ZMIN_8
+    0x3f800000, // PA_SC_VPORT_ZMAX_8
+    0x00000000, // PA_SC_VPORT_ZMIN_9
+    0x3f800000, // PA_SC_VPORT_ZMAX_9
+    0x00000000, // PA_SC_VPORT_ZMIN_10
+    0x3f800000, // PA_SC_VPORT_ZMAX_10
+    0x00000000, // PA_SC_VPORT_ZMIN_11
+    0x3f800000, // PA_SC_VPORT_ZMAX_11
+    0x00000000, // PA_SC_VPORT_ZMIN_12
+    0x3f800000, // PA_SC_VPORT_ZMAX_12
+    0x00000000, // PA_SC_VPORT_ZMIN_13
+    0x3f800000, // PA_SC_VPORT_ZMAX_13
+    0x00000000, // PA_SC_VPORT_ZMIN_14
+    0x3f800000, // PA_SC_VPORT_ZMAX_14
+    0x00000000, // PA_SC_VPORT_ZMIN_15
+    0x3f800000, // PA_SC_VPORT_ZMAX_15
+};
+static const unsigned int ci_SECT_CONTEXT_def_2[] =
+{
+    0x00000000, // PA_SC_SCREEN_EXTENT_CONTROL
+    0, // HOLE
+    0x00000000, // CP_PERFMON_CNTX_CNTL
+    0x00000000, // CP_RINGID
+    0x00000000, // CP_VMID
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0xffffffff, // VGT_MAX_VTX_INDX
+    0x00000000, // VGT_MIN_VTX_INDX
+    0x00000000, // VGT_INDX_OFFSET
+    0x00000000, // VGT_MULTI_PRIM_IB_RESET_INDX
+    0, // HOLE
+    0x00000000, // CB_BLEND_RED
+    0x00000000, // CB_BLEND_GREEN
+    0x00000000, // CB_BLEND_BLUE
+    0x00000000, // CB_BLEND_ALPHA
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // DB_STENCIL_CONTROL
+    0x00000000, // DB_STENCILREFMASK
+    0x00000000, // DB_STENCILREFMASK_BF
+    0, // HOLE
+    0x00000000, // PA_CL_VPORT_XSCALE
+    0x00000000, // PA_CL_VPORT_XOFFSET
+    0x00000000, // PA_CL_VPORT_YSCALE
+    0x00000000, // PA_CL_VPORT_YOFFSET
+    0x00000000, // PA_CL_VPORT_ZSCALE
+    0x00000000, // PA_CL_VPORT_ZOFFSET
+    0x00000000, // PA_CL_VPORT_XSCALE_1
+    0x00000000, // PA_CL_VPORT_XOFFSET_1
+    0x00000000, // PA_CL_VPORT_YSCALE_1
+    0x00000000, // PA_CL_VPORT_YOFFSET_1
+    0x00000000, // PA_CL_VPORT_ZSCALE_1
+    0x00000000, // PA_CL_VPORT_ZOFFSET_1
+    0x00000000, // PA_CL_VPORT_XSCALE_2
+    0x00000000, // PA_CL_VPORT_XOFFSET_2
+    0x00000000, // PA_CL_VPORT_YSCALE_2
+    0x00000000, // PA_CL_VPORT_YOFFSET_2
+    0x00000000, // PA_CL_VPORT_ZSCALE_2
+    0x00000000, // PA_CL_VPORT_ZOFFSET_2
+    0x00000000, // PA_CL_VPORT_XSCALE_3
+    0x00000000, // PA_CL_VPORT_XOFFSET_3
+    0x00000000, // PA_CL_VPORT_YSCALE_3
+    0x00000000, // PA_CL_VPORT_YOFFSET_3
+    0x00000000, // PA_CL_VPORT_ZSCALE_3
+    0x00000000, // PA_CL_VPORT_ZOFFSET_3
+    0x00000000, // PA_CL_VPORT_XSCALE_4
+    0x00000000, // PA_CL_VPORT_XOFFSET_4
+    0x00000000, // PA_CL_VPORT_YSCALE_4
+    0x00000000, // PA_CL_VPORT_YOFFSET_4
+    0x00000000, // PA_CL_VPORT_ZSCALE_4
+    0x00000000, // PA_CL_VPORT_ZOFFSET_4
+    0x00000000, // PA_CL_VPORT_XSCALE_5
+    0x00000000, // PA_CL_VPORT_XOFFSET_5
+    0x00000000, // PA_CL_VPORT_YSCALE_5
+    0x00000000, // PA_CL_VPORT_YOFFSET_5
+    0x00000000, // PA_CL_VPORT_ZSCALE_5
+    0x00000000, // PA_CL_VPORT_ZOFFSET_5
+    0x00000000, // PA_CL_VPORT_XSCALE_6
+    0x00000000, // PA_CL_VPORT_XOFFSET_6
+    0x00000000, // PA_CL_VPORT_YSCALE_6
+    0x00000000, // PA_CL_VPORT_YOFFSET_6
+    0x00000000, // PA_CL_VPORT_ZSCALE_6
+    0x00000000, // PA_CL_VPORT_ZOFFSET_6
+    0x00000000, // PA_CL_VPORT_XSCALE_7
+    0x00000000, // PA_CL_VPORT_XOFFSET_7
+    0x00000000, // PA_CL_VPORT_YSCALE_7
+    0x00000000, // PA_CL_VPORT_YOFFSET_7
+    0x00000000, // PA_CL_VPORT_ZSCALE_7
+    0x00000000, // PA_CL_VPORT_ZOFFSET_7
+    0x00000000, // PA_CL_VPORT_XSCALE_8
+    0x00000000, // PA_CL_VPORT_XOFFSET_8
+    0x00000000, // PA_CL_VPORT_YSCALE_8
+    0x00000000, // PA_CL_VPORT_YOFFSET_8
+    0x00000000, // PA_CL_VPORT_ZSCALE_8
+    0x00000000, // PA_CL_VPORT_ZOFFSET_8
+    0x00000000, // PA_CL_VPORT_XSCALE_9
+    0x00000000, // PA_CL_VPORT_XOFFSET_9
+    0x00000000, // PA_CL_VPORT_YSCALE_9
+    0x00000000, // PA_CL_VPORT_YOFFSET_9
+    0x00000000, // PA_CL_VPORT_ZSCALE_9
+    0x00000000, // PA_CL_VPORT_ZOFFSET_9
+    0x00000000, // PA_CL_VPORT_XSCALE_10
+    0x00000000, // PA_CL_VPORT_XOFFSET_10
+    0x00000000, // PA_CL_VPORT_YSCALE_10
+    0x00000000, // PA_CL_VPORT_YOFFSET_10
+    0x00000000, // PA_CL_VPORT_ZSCALE_10
+    0x00000000, // PA_CL_VPORT_ZOFFSET_10
+    0x00000000, // PA_CL_VPORT_XSCALE_11
+    0x00000000, // PA_CL_VPORT_XOFFSET_11
+    0x00000000, // PA_CL_VPORT_YSCALE_11
+    0x00000000, // PA_CL_VPORT_YOFFSET_11
+    0x00000000, // PA_CL_VPORT_ZSCALE_11
+    0x00000000, // PA_CL_VPORT_ZOFFSET_11
+    0x00000000, // PA_CL_VPORT_XSCALE_12
+    0x00000000, // PA_CL_VPORT_XOFFSET_12
+    0x00000000, // PA_CL_VPORT_YSCALE_12
+    0x00000000, // PA_CL_VPORT_YOFFSET_12
+    0x00000000, // PA_CL_VPORT_ZSCALE_12
+    0x00000000, // PA_CL_VPORT_ZOFFSET_12
+    0x00000000, // PA_CL_VPORT_XSCALE_13
+    0x00000000, // PA_CL_VPORT_XOFFSET_13
+    0x00000000, // PA_CL_VPORT_YSCALE_13
+    0x00000000, // PA_CL_VPORT_YOFFSET_13
+    0x00000000, // PA_CL_VPORT_ZSCALE_13
+    0x00000000, // PA_CL_VPORT_ZOFFSET_13
+    0x00000000, // PA_CL_VPORT_XSCALE_14
+    0x00000000, // PA_CL_VPORT_XOFFSET_14
+    0x00000000, // PA_CL_VPORT_YSCALE_14
+    0x00000000, // PA_CL_VPORT_YOFFSET_14
+    0x00000000, // PA_CL_VPORT_ZSCALE_14
+    0x00000000, // PA_CL_VPORT_ZOFFSET_14
+    0x00000000, // PA_CL_VPORT_XSCALE_15
+    0x00000000, // PA_CL_VPORT_XOFFSET_15
+    0x00000000, // PA_CL_VPORT_YSCALE_15
+    0x00000000, // PA_CL_VPORT_YOFFSET_15
+    0x00000000, // PA_CL_VPORT_ZSCALE_15
+    0x00000000, // PA_CL_VPORT_ZOFFSET_15
+    0x00000000, // PA_CL_UCP_0_X
+    0x00000000, // PA_CL_UCP_0_Y
+    0x00000000, // PA_CL_UCP_0_Z
+    0x00000000, // PA_CL_UCP_0_W
+    0x00000000, // PA_CL_UCP_1_X
+    0x00000000, // PA_CL_UCP_1_Y
+    0x00000000, // PA_CL_UCP_1_Z
+    0x00000000, // PA_CL_UCP_1_W
+    0x00000000, // PA_CL_UCP_2_X
+    0x00000000, // PA_CL_UCP_2_Y
+    0x00000000, // PA_CL_UCP_2_Z
+    0x00000000, // PA_CL_UCP_2_W
+    0x00000000, // PA_CL_UCP_3_X
+    0x00000000, // PA_CL_UCP_3_Y
+    0x00000000, // PA_CL_UCP_3_Z
+    0x00000000, // PA_CL_UCP_3_W
+    0x00000000, // PA_CL_UCP_4_X
+    0x00000000, // PA_CL_UCP_4_Y
+    0x00000000, // PA_CL_UCP_4_Z
+    0x00000000, // PA_CL_UCP_4_W
+    0x00000000, // PA_CL_UCP_5_X
+    0x00000000, // PA_CL_UCP_5_Y
+    0x00000000, // PA_CL_UCP_5_Z
+    0x00000000, // PA_CL_UCP_5_W
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_CNTL_0
+    0x00000000, // SPI_PS_INPUT_CNTL_1
+    0x00000000, // SPI_PS_INPUT_CNTL_2
+    0x00000000, // SPI_PS_INPUT_CNTL_3
+    0x00000000, // SPI_PS_INPUT_CNTL_4
+    0x00000000, // SPI_PS_INPUT_CNTL_5
+    0x00000000, // SPI_PS_INPUT_CNTL_6
+    0x00000000, // SPI_PS_INPUT_CNTL_7
+    0x00000000, // SPI_PS_INPUT_CNTL_8
+    0x00000000, // SPI_PS_INPUT_CNTL_9
+    0x00000000, // SPI_PS_INPUT_CNTL_10
+    0x00000000, // SPI_PS_INPUT_CNTL_11
+    0x00000000, // SPI_PS_INPUT_CNTL_12
+    0x00000000, // SPI_PS_INPUT_CNTL_13
+    0x00000000, // SPI_PS_INPUT_CNTL_14
+    0x00000000, // SPI_PS_INPUT_CNTL_15
+    0x00000000, // SPI_PS_INPUT_CNTL_16
+    0x00000000, // SPI_PS_INPUT_CNTL_17
+    0x00000000, // SPI_PS_INPUT_CNTL_18
+    0x00000000, // SPI_PS_INPUT_CNTL_19
+    0x00000000, // SPI_PS_INPUT_CNTL_20
+    0x00000000, // SPI_PS_INPUT_CNTL_21
+    0x00000000, // SPI_PS_INPUT_CNTL_22
+    0x00000000, // SPI_PS_INPUT_CNTL_23
+    0x00000000, // SPI_PS_INPUT_CNTL_24
+    0x00000000, // SPI_PS_INPUT_CNTL_25
+    0x00000000, // SPI_PS_INPUT_CNTL_26
+    0x00000000, // SPI_PS_INPUT_CNTL_27
+    0x00000000, // SPI_PS_INPUT_CNTL_28
+    0x00000000, // SPI_PS_INPUT_CNTL_29
+    0x00000000, // SPI_PS_INPUT_CNTL_30
+    0x00000000, // SPI_PS_INPUT_CNTL_31
+    0x00000000, // SPI_VS_OUT_CONFIG
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_ENA
+    0x00000000, // SPI_PS_INPUT_ADDR
+    0x00000000, // SPI_INTERP_CONTROL_0
+    0x00000002, // SPI_PS_IN_CONTROL
+    0, // HOLE
+    0x00000000, // SPI_BARYC_CNTL
+    0, // HOLE
+    0x00000000, // SPI_TMPRING_SIZE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_SHADER_POS_FORMAT
+    0x00000000, // SPI_SHADER_Z_FORMAT
+    0x00000000, // SPI_SHADER_COL_FORMAT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_BLEND0_CONTROL
+    0x00000000, // CB_BLEND1_CONTROL
+    0x00000000, // CB_BLEND2_CONTROL
+    0x00000000, // CB_BLEND3_CONTROL
+    0x00000000, // CB_BLEND4_CONTROL
+    0x00000000, // CB_BLEND5_CONTROL
+    0x00000000, // CB_BLEND6_CONTROL
+    0x00000000, // CB_BLEND7_CONTROL
+};
+static const unsigned int ci_SECT_CONTEXT_def_3[] =
+{
+    0x00000000, // PA_CL_POINT_X_RAD
+    0x00000000, // PA_CL_POINT_Y_RAD
+    0x00000000, // PA_CL_POINT_SIZE
+    0x00000000, // PA_CL_POINT_CULL_RAD
+    0x00000000, // VGT_DMA_BASE_HI
+    0x00000000, // VGT_DMA_BASE
+};
+static const unsigned int ci_SECT_CONTEXT_def_4[] =
+{
+    0x00000000, // DB_DEPTH_CONTROL
+    0x00000000, // DB_EQAA
+    0x00000000, // CB_COLOR_CONTROL
+    0x00000000, // DB_SHADER_CONTROL
+    0x00090000, // PA_CL_CLIP_CNTL
+    0x00000004, // PA_SU_SC_MODE_CNTL
+    0x00000000, // PA_CL_VTE_CNTL
+    0x00000000, // PA_CL_VS_OUT_CNTL
+    0x00000000, // PA_CL_NANINF_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_SCALE
+    0x00000000, // PA_SU_PRIM_FILTER_CNTL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SU_POINT_SIZE
+    0x00000000, // PA_SU_POINT_MINMAX
+    0x00000000, // PA_SU_LINE_CNTL
+    0x00000000, // PA_SC_LINE_STIPPLE
+    0x00000000, // VGT_OUTPUT_PATH_CNTL
+    0x00000000, // VGT_HOS_CNTL
+    0x00000000, // VGT_HOS_MAX_TESS_LEVEL
+    0x00000000, // VGT_HOS_MIN_TESS_LEVEL
+    0x00000000, // VGT_HOS_REUSE_DEPTH
+    0x00000000, // VGT_GROUP_PRIM_TYPE
+    0x00000000, // VGT_GROUP_FIRST_DECR
+    0x00000000, // VGT_GROUP_DECR
+    0x00000000, // VGT_GROUP_VECT_0_CNTL
+    0x00000000, // VGT_GROUP_VECT_1_CNTL
+    0x00000000, // VGT_GROUP_VECT_0_FMT_CNTL
+    0x00000000, // VGT_GROUP_VECT_1_FMT_CNTL
+    0x00000000, // VGT_GS_MODE
+    0x00000000, // VGT_GS_ONCHIP_CNTL
+    0x00000000, // PA_SC_MODE_CNTL_0
+    0x00000000, // PA_SC_MODE_CNTL_1
+    0x00000000, // VGT_ENHANCE
+    0x00000100, // VGT_GS_PER_ES
+    0x00000080, // VGT_ES_PER_GS
+    0x00000002, // VGT_GS_PER_VS
+    0x00000000, // VGT_GSVS_RING_OFFSET_1
+    0x00000000, // VGT_GSVS_RING_OFFSET_2
+    0x00000000, // VGT_GSVS_RING_OFFSET_3
+    0x00000000, // VGT_GS_OUT_PRIM_TYPE
+    0x00000000, // IA_ENHANCE
+};
+static const unsigned int ci_SECT_CONTEXT_def_5[] =
+{
+    0x00000000, // WD_ENHANCE
+    0x00000000, // VGT_PRIMITIVEID_EN
+};
+static const unsigned int ci_SECT_CONTEXT_def_6[] =
+{
+    0x00000000, // VGT_PRIMITIVEID_RESET
+};
+static const unsigned int ci_SECT_CONTEXT_def_7[] =
+{
+    0x00000000, // VGT_MULTI_PRIM_IB_RESET_EN
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_INSTANCE_STEP_RATE_0
+    0x00000000, // VGT_INSTANCE_STEP_RATE_1
+    0x000000ff, // IA_MULTI_VGT_PARAM
+    0x00000000, // VGT_ESGS_RING_ITEMSIZE
+    0x00000000, // VGT_GSVS_RING_ITEMSIZE
+    0x00000000, // VGT_REUSE_OFF
+    0x00000000, // VGT_VTX_CNT_EN
+    0x00000000, // DB_HTILE_SURFACE
+    0x00000000, // DB_SRESULTS_COMPARE_STATE0
+    0x00000000, // DB_SRESULTS_COMPARE_STATE1
+    0x00000000, // DB_PRELOAD_CONTROL
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_0
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_0
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_0
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_1
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_1
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_1
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_2
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_2
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_2
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_3
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_3
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_3
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_OFFSET
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_BUFFER_FILLED_SIZE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_VERTEX_STRIDE
+    0, // HOLE
+    0x00000000, // VGT_GS_MAX_VERT_OUT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_SHADER_STAGES_EN
+    0x00000000, // VGT_LS_HS_CONFIG
+    0x00000000, // VGT_GS_VERT_ITEMSIZE
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_1
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_2
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_3
+    0x00000000, // VGT_TF_PARAM
+    0x00000000, // DB_ALPHA_TO_MASK
+    0, // HOLE
+    0x00000000, // PA_SU_POLY_OFFSET_DB_FMT_CNTL
+    0x00000000, // PA_SU_POLY_OFFSET_CLAMP
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_OFFSET
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_OFFSET
+    0x00000000, // VGT_GS_INSTANCE_CNT
+    0x00000000, // VGT_STRMOUT_CONFIG
+    0x00000000, // VGT_STRMOUT_BUFFER_CONFIG
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SC_CENTROID_PRIORITY_0
+    0x00000000, // PA_SC_CENTROID_PRIORITY_1
+    0x00001000, // PA_SC_LINE_CNTL
+    0x00000000, // PA_SC_AA_CONFIG
+    0x00000005, // PA_SU_VTX_CNTL
+    0x3f800000, // PA_CL_GB_VERT_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_VERT_DISC_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_DISC_ADJ
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_3
+    0xffffffff, // PA_SC_AA_MASK_X0Y0_X1Y0
+    0xffffffff, // PA_SC_AA_MASK_X0Y1_X1Y1
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x0000000e, // VGT_VERTEX_REUSE_BLOCK_CNTL
+    0x00000010, // VGT_OUT_DEALLOC_CNTL
+    0x00000000, // CB_COLOR0_BASE
+    0x00000000, // CB_COLOR0_PITCH
+    0x00000000, // CB_COLOR0_SLICE
+    0x00000000, // CB_COLOR0_VIEW
+    0x00000000, // CB_COLOR0_INFO
+    0x00000000, // CB_COLOR0_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR0_CMASK
+    0x00000000, // CB_COLOR0_CMASK_SLICE
+    0x00000000, // CB_COLOR0_FMASK
+    0x00000000, // CB_COLOR0_FMASK_SLICE
+    0x00000000, // CB_COLOR0_CLEAR_WORD0
+    0x00000000, // CB_COLOR0_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR1_BASE
+    0x00000000, // CB_COLOR1_PITCH
+    0x00000000, // CB_COLOR1_SLICE
+    0x00000000, // CB_COLOR1_VIEW
+    0x00000000, // CB_COLOR1_INFO
+    0x00000000, // CB_COLOR1_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR1_CMASK
+    0x00000000, // CB_COLOR1_CMASK_SLICE
+    0x00000000, // CB_COLOR1_FMASK
+    0x00000000, // CB_COLOR1_FMASK_SLICE
+    0x00000000, // CB_COLOR1_CLEAR_WORD0
+    0x00000000, // CB_COLOR1_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR2_BASE
+    0x00000000, // CB_COLOR2_PITCH
+    0x00000000, // CB_COLOR2_SLICE
+    0x00000000, // CB_COLOR2_VIEW
+    0x00000000, // CB_COLOR2_INFO
+    0x00000000, // CB_COLOR2_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR2_CMASK
+    0x00000000, // CB_COLOR2_CMASK_SLICE
+    0x00000000, // CB_COLOR2_FMASK
+    0x00000000, // CB_COLOR2_FMASK_SLICE
+    0x00000000, // CB_COLOR2_CLEAR_WORD0
+    0x00000000, // CB_COLOR2_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR3_BASE
+    0x00000000, // CB_COLOR3_PITCH
+    0x00000000, // CB_COLOR3_SLICE
+    0x00000000, // CB_COLOR3_VIEW
+    0x00000000, // CB_COLOR3_INFO
+    0x00000000, // CB_COLOR3_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR3_CMASK
+    0x00000000, // CB_COLOR3_CMASK_SLICE
+    0x00000000, // CB_COLOR3_FMASK
+    0x00000000, // CB_COLOR3_FMASK_SLICE
+    0x00000000, // CB_COLOR3_CLEAR_WORD0
+    0x00000000, // CB_COLOR3_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR4_BASE
+    0x00000000, // CB_COLOR4_PITCH
+    0x00000000, // CB_COLOR4_SLICE
+    0x00000000, // CB_COLOR4_VIEW
+    0x00000000, // CB_COLOR4_INFO
+    0x00000000, // CB_COLOR4_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR4_CMASK
+    0x00000000, // CB_COLOR4_CMASK_SLICE
+    0x00000000, // CB_COLOR4_FMASK
+    0x00000000, // CB_COLOR4_FMASK_SLICE
+    0x00000000, // CB_COLOR4_CLEAR_WORD0
+    0x00000000, // CB_COLOR4_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR5_BASE
+    0x00000000, // CB_COLOR5_PITCH
+    0x00000000, // CB_COLOR5_SLICE
+    0x00000000, // CB_COLOR5_VIEW
+    0x00000000, // CB_COLOR5_INFO
+    0x00000000, // CB_COLOR5_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR5_CMASK
+    0x00000000, // CB_COLOR5_CMASK_SLICE
+    0x00000000, // CB_COLOR5_FMASK
+    0x00000000, // CB_COLOR5_FMASK_SLICE
+    0x00000000, // CB_COLOR5_CLEAR_WORD0
+    0x00000000, // CB_COLOR5_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR6_BASE
+    0x00000000, // CB_COLOR6_PITCH
+    0x00000000, // CB_COLOR6_SLICE
+    0x00000000, // CB_COLOR6_VIEW
+    0x00000000, // CB_COLOR6_INFO
+    0x00000000, // CB_COLOR6_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR6_CMASK
+    0x00000000, // CB_COLOR6_CMASK_SLICE
+    0x00000000, // CB_COLOR6_FMASK
+    0x00000000, // CB_COLOR6_FMASK_SLICE
+    0x00000000, // CB_COLOR6_CLEAR_WORD0
+    0x00000000, // CB_COLOR6_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR7_BASE
+    0x00000000, // CB_COLOR7_PITCH
+    0x00000000, // CB_COLOR7_SLICE
+    0x00000000, // CB_COLOR7_VIEW
+    0x00000000, // CB_COLOR7_INFO
+    0x00000000, // CB_COLOR7_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR7_CMASK
+    0x00000000, // CB_COLOR7_CMASK_SLICE
+    0x00000000, // CB_COLOR7_FMASK
+    0x00000000, // CB_COLOR7_FMASK_SLICE
+    0x00000000, // CB_COLOR7_CLEAR_WORD0
+    0x00000000, // CB_COLOR7_CLEAR_WORD1
+};
+static const struct cs_extent_def ci_SECT_CONTEXT_defs[] =
+{
+    {ci_SECT_CONTEXT_def_1, 0x0000a000, 212 },
+    {ci_SECT_CONTEXT_def_2, 0x0000a0d6, 274 },
+    {ci_SECT_CONTEXT_def_3, 0x0000a1f5, 6 },
+    {ci_SECT_CONTEXT_def_4, 0x0000a200, 157 },
+    {ci_SECT_CONTEXT_def_5, 0x0000a2a0, 2 },
+    {ci_SECT_CONTEXT_def_6, 0x0000a2a3, 1 },
+    {ci_SECT_CONTEXT_def_7, 0x0000a2a5, 233 },
+    { 0, 0, 0 }
+};
+static const struct cs_section_def ci_cs_data[] = {
+    { ci_SECT_CONTEXT_defs, SECT_CONTEXT },
+    { 0, SECT_NONE }
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/clearstate_defs.h b/sys/dev/pci/drm/amd/amdgpu/clearstate_defs.h
new file mode 100644
index 00000000000..3eda707d738
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/clearstate_defs.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef CLEARSTATE_DEFS_H
+#define CLEARSTATE_DEFS_H
+
+enum section_id {
+    SECT_NONE,
+    SECT_CONTEXT,
+    SECT_CLEAR,
+    SECT_CTRLCONST
+};
+
+struct cs_extent_def {
+    const unsigned int *extent;
+    const unsigned int reg_index;
+    const unsigned int reg_count;
+};
+
+struct cs_section_def {
+    const struct cs_extent_def *section;
+    const enum section_id id;
+};
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/clearstate_gfx9.h b/sys/dev/pci/drm/amd/amdgpu/clearstate_gfx9.h
new file mode 100644
index 00000000000..567a904804b
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/clearstate_gfx9.h
@@ -0,0 +1,942 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+static const unsigned int gfx9_SECT_CONTEXT_def_1[] =
+{
+    0x00000000, // DB_RENDER_CONTROL
+    0x00000000, // DB_COUNT_CONTROL
+    0x00000000, // DB_DEPTH_VIEW
+    0x00000000, // DB_RENDER_OVERRIDE
+    0x00000000, // DB_RENDER_OVERRIDE2
+    0x00000000, // DB_HTILE_DATA_BASE
+    0x00000000, // DB_HTILE_DATA_BASE_HI
+    0x00000000, // DB_DEPTH_SIZE
+    0x00000000, // DB_DEPTH_BOUNDS_MIN
+    0x00000000, // DB_DEPTH_BOUNDS_MAX
+    0x00000000, // DB_STENCIL_CLEAR
+    0x00000000, // DB_DEPTH_CLEAR
+    0x00000000, // PA_SC_SCREEN_SCISSOR_TL
+    0x40004000, // PA_SC_SCREEN_SCISSOR_BR
+    0x00000000, // DB_Z_INFO
+    0x00000000, // DB_STENCIL_INFO
+    0x00000000, // DB_Z_READ_BASE
+    0x00000000, // DB_Z_READ_BASE_HI
+    0x00000000, // DB_STENCIL_READ_BASE
+    0x00000000, // DB_STENCIL_READ_BASE_HI
+    0x00000000, // DB_Z_WRITE_BASE
+    0x00000000, // DB_Z_WRITE_BASE_HI
+    0x00000000, // DB_STENCIL_WRITE_BASE
+    0x00000000, // DB_STENCIL_WRITE_BASE_HI
+    0x00000000, // DB_DFSM_CONTROL
+    0, // HOLE
+    0x00000000, // DB_Z_INFO2
+    0x00000000, // DB_STENCIL_INFO2
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // TA_BC_BASE_ADDR
+    0x00000000, // TA_BC_BASE_ADDR_HI
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // COHER_DEST_BASE_HI_0
+    0x00000000, // COHER_DEST_BASE_HI_1
+    0x00000000, // COHER_DEST_BASE_HI_2
+    0x00000000, // COHER_DEST_BASE_HI_3
+    0x00000000, // COHER_DEST_BASE_2
+    0x00000000, // COHER_DEST_BASE_3
+    0x00000000, // PA_SC_WINDOW_OFFSET
+    0x80000000, // PA_SC_WINDOW_SCISSOR_TL
+    0x40004000, // PA_SC_WINDOW_SCISSOR_BR
+    0x0000ffff, // PA_SC_CLIPRECT_RULE
+    0x00000000, // PA_SC_CLIPRECT_0_TL
+    0x40004000, // PA_SC_CLIPRECT_0_BR
+    0x00000000, // PA_SC_CLIPRECT_1_TL
+    0x40004000, // PA_SC_CLIPRECT_1_BR
+    0x00000000, // PA_SC_CLIPRECT_2_TL
+    0x40004000, // PA_SC_CLIPRECT_2_BR
+    0x00000000, // PA_SC_CLIPRECT_3_TL
+    0x40004000, // PA_SC_CLIPRECT_3_BR
+    0xaa99aaaa, // PA_SC_EDGERULE
+    0x00000000, // PA_SU_HARDWARE_SCREEN_OFFSET
+    0xffffffff, // CB_TARGET_MASK
+    0xffffffff, // CB_SHADER_MASK
+    0x80000000, // PA_SC_GENERIC_SCISSOR_TL
+    0x40004000, // PA_SC_GENERIC_SCISSOR_BR
+    0x00000000, // COHER_DEST_BASE_0
+    0x00000000, // COHER_DEST_BASE_1
+    0x80000000, // PA_SC_VPORT_SCISSOR_0_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_0_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_1_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_1_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_2_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_2_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_3_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_3_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_4_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_4_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_5_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_5_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_6_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_6_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_7_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_7_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_8_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_8_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_9_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_9_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_10_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_10_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_11_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_11_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_12_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_12_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_13_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_13_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_14_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_14_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_15_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_15_BR
+    0x00000000, // PA_SC_VPORT_ZMIN_0
+    0x3f800000, // PA_SC_VPORT_ZMAX_0
+    0x00000000, // PA_SC_VPORT_ZMIN_1
+    0x3f800000, // PA_SC_VPORT_ZMAX_1
+    0x00000000, // PA_SC_VPORT_ZMIN_2
+    0x3f800000, // PA_SC_VPORT_ZMAX_2
+    0x00000000, // PA_SC_VPORT_ZMIN_3
+    0x3f800000, // PA_SC_VPORT_ZMAX_3
+    0x00000000, // PA_SC_VPORT_ZMIN_4
+    0x3f800000, // PA_SC_VPORT_ZMAX_4
+    0x00000000, // PA_SC_VPORT_ZMIN_5
+    0x3f800000, // PA_SC_VPORT_ZMAX_5
+    0x00000000, // PA_SC_VPORT_ZMIN_6
+    0x3f800000, // PA_SC_VPORT_ZMAX_6
+    0x00000000, // PA_SC_VPORT_ZMIN_7
+    0x3f800000, // PA_SC_VPORT_ZMAX_7
+    0x00000000, // PA_SC_VPORT_ZMIN_8
+    0x3f800000, // PA_SC_VPORT_ZMAX_8
+    0x00000000, // PA_SC_VPORT_ZMIN_9
+    0x3f800000, // PA_SC_VPORT_ZMAX_9
+    0x00000000, // PA_SC_VPORT_ZMIN_10
+    0x3f800000, // PA_SC_VPORT_ZMAX_10
+    0x00000000, // PA_SC_VPORT_ZMIN_11
+    0x3f800000, // PA_SC_VPORT_ZMAX_11
+    0x00000000, // PA_SC_VPORT_ZMIN_12
+    0x3f800000, // PA_SC_VPORT_ZMAX_12
+    0x00000000, // PA_SC_VPORT_ZMIN_13
+    0x3f800000, // PA_SC_VPORT_ZMAX_13
+    0x00000000, // PA_SC_VPORT_ZMIN_14
+    0x3f800000, // PA_SC_VPORT_ZMAX_14
+    0x00000000, // PA_SC_VPORT_ZMIN_15
+    0x3f800000, // PA_SC_VPORT_ZMAX_15
+};
+static const unsigned int gfx9_SECT_CONTEXT_def_2[] =
+{
+    0x00000000, // PA_SC_SCREEN_EXTENT_CONTROL
+    0x00000000, // PA_SC_TILE_STEERING_OVERRIDE
+    0x00000000, // CP_PERFMON_CNTX_CNTL
+    0x00000000, // CP_RINGID
+    0x00000000, // CP_VMID
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SC_RIGHT_VERT_GRID
+    0x00000000, // PA_SC_LEFT_VERT_GRID
+    0x00000000, // PA_SC_HORIZ_GRID
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_MULTI_PRIM_IB_RESET_INDX
+    0, // HOLE
+    0x00000000, // CB_BLEND_RED
+    0x00000000, // CB_BLEND_GREEN
+    0x00000000, // CB_BLEND_BLUE
+    0x00000000, // CB_BLEND_ALPHA
+    0x00000000, // CB_DCC_CONTROL
+    0, // HOLE
+    0x00000000, // DB_STENCIL_CONTROL
+    0x01000000, // DB_STENCILREFMASK
+    0x01000000, // DB_STENCILREFMASK_BF
+    0, // HOLE
+    0x00000000, // PA_CL_VPORT_XSCALE
+    0x00000000, // PA_CL_VPORT_XOFFSET
+    0x00000000, // PA_CL_VPORT_YSCALE
+    0x00000000, // PA_CL_VPORT_YOFFSET
+    0x00000000, // PA_CL_VPORT_ZSCALE
+    0x00000000, // PA_CL_VPORT_ZOFFSET
+    0x00000000, // PA_CL_VPORT_XSCALE_1
+    0x00000000, // PA_CL_VPORT_XOFFSET_1
+    0x00000000, // PA_CL_VPORT_YSCALE_1
+    0x00000000, // PA_CL_VPORT_YOFFSET_1
+    0x00000000, // PA_CL_VPORT_ZSCALE_1
+    0x00000000, // PA_CL_VPORT_ZOFFSET_1
+    0x00000000, // PA_CL_VPORT_XSCALE_2
+    0x00000000, // PA_CL_VPORT_XOFFSET_2
+    0x00000000, // PA_CL_VPORT_YSCALE_2
+    0x00000000, // PA_CL_VPORT_YOFFSET_2
+    0x00000000, // PA_CL_VPORT_ZSCALE_2
+    0x00000000, // PA_CL_VPORT_ZOFFSET_2
+    0x00000000, // PA_CL_VPORT_XSCALE_3
+    0x00000000, // PA_CL_VPORT_XOFFSET_3
+    0x00000000, // PA_CL_VPORT_YSCALE_3
+    0x00000000, // PA_CL_VPORT_YOFFSET_3
+    0x00000000, // PA_CL_VPORT_ZSCALE_3
+    0x00000000, // PA_CL_VPORT_ZOFFSET_3
+    0x00000000, // PA_CL_VPORT_XSCALE_4
+    0x00000000, // PA_CL_VPORT_XOFFSET_4
+    0x00000000, // PA_CL_VPORT_YSCALE_4
+    0x00000000, // PA_CL_VPORT_YOFFSET_4
+    0x00000000, // PA_CL_VPORT_ZSCALE_4
+    0x00000000, // PA_CL_VPORT_ZOFFSET_4
+    0x00000000, // PA_CL_VPORT_XSCALE_5
+    0x00000000, // PA_CL_VPORT_XOFFSET_5
+    0x00000000, // PA_CL_VPORT_YSCALE_5
+    0x00000000, // PA_CL_VPORT_YOFFSET_5
+    0x00000000, // PA_CL_VPORT_ZSCALE_5
+    0x00000000, // PA_CL_VPORT_ZOFFSET_5
+    0x00000000, // PA_CL_VPORT_XSCALE_6
+    0x00000000, // PA_CL_VPORT_XOFFSET_6
+    0x00000000, // PA_CL_VPORT_YSCALE_6
+    0x00000000, // PA_CL_VPORT_YOFFSET_6
+    0x00000000, // PA_CL_VPORT_ZSCALE_6
+    0x00000000, // PA_CL_VPORT_ZOFFSET_6
+    0x00000000, // PA_CL_VPORT_XSCALE_7
+    0x00000000, // PA_CL_VPORT_XOFFSET_7
+    0x00000000, // PA_CL_VPORT_YSCALE_7
+    0x00000000, // PA_CL_VPORT_YOFFSET_7
+    0x00000000, // PA_CL_VPORT_ZSCALE_7
+    0x00000000, // PA_CL_VPORT_ZOFFSET_7
+    0x00000000, // PA_CL_VPORT_XSCALE_8
+    0x00000000, // PA_CL_VPORT_XOFFSET_8
+    0x00000000, // PA_CL_VPORT_YSCALE_8
+    0x00000000, // PA_CL_VPORT_YOFFSET_8
+    0x00000000, // PA_CL_VPORT_ZSCALE_8
+    0x00000000, // PA_CL_VPORT_ZOFFSET_8
+    0x00000000, // PA_CL_VPORT_XSCALE_9
+    0x00000000, // PA_CL_VPORT_XOFFSET_9
+    0x00000000, // PA_CL_VPORT_YSCALE_9
+    0x00000000, // PA_CL_VPORT_YOFFSET_9
+    0x00000000, // PA_CL_VPORT_ZSCALE_9
+    0x00000000, // PA_CL_VPORT_ZOFFSET_9
+    0x00000000, // PA_CL_VPORT_XSCALE_10
+    0x00000000, // PA_CL_VPORT_XOFFSET_10
+    0x00000000, // PA_CL_VPORT_YSCALE_10
+    0x00000000, // PA_CL_VPORT_YOFFSET_10
+    0x00000000, // PA_CL_VPORT_ZSCALE_10
+    0x00000000, // PA_CL_VPORT_ZOFFSET_10
+    0x00000000, // PA_CL_VPORT_XSCALE_11
+    0x00000000, // PA_CL_VPORT_XOFFSET_11
+    0x00000000, // PA_CL_VPORT_YSCALE_11
+    0x00000000, // PA_CL_VPORT_YOFFSET_11
+    0x00000000, // PA_CL_VPORT_ZSCALE_11
+    0x00000000, // PA_CL_VPORT_ZOFFSET_11
+    0x00000000, // PA_CL_VPORT_XSCALE_12
+    0x00000000, // PA_CL_VPORT_XOFFSET_12
+    0x00000000, // PA_CL_VPORT_YSCALE_12
+    0x00000000, // PA_CL_VPORT_YOFFSET_12
+    0x00000000, // PA_CL_VPORT_ZSCALE_12
+    0x00000000, // PA_CL_VPORT_ZOFFSET_12
+    0x00000000, // PA_CL_VPORT_XSCALE_13
+    0x00000000, // PA_CL_VPORT_XOFFSET_13
+    0x00000000, // PA_CL_VPORT_YSCALE_13
+    0x00000000, // PA_CL_VPORT_YOFFSET_13
+    0x00000000, // PA_CL_VPORT_ZSCALE_13
+    0x00000000, // PA_CL_VPORT_ZOFFSET_13
+    0x00000000, // PA_CL_VPORT_XSCALE_14
+    0x00000000, // PA_CL_VPORT_XOFFSET_14
+    0x00000000, // PA_CL_VPORT_YSCALE_14
+    0x00000000, // PA_CL_VPORT_YOFFSET_14
+    0x00000000, // PA_CL_VPORT_ZSCALE_14
+    0x00000000, // PA_CL_VPORT_ZOFFSET_14
+    0x00000000, // PA_CL_VPORT_XSCALE_15
+    0x00000000, // PA_CL_VPORT_XOFFSET_15
+    0x00000000, // PA_CL_VPORT_YSCALE_15
+    0x00000000, // PA_CL_VPORT_YOFFSET_15
+    0x00000000, // PA_CL_VPORT_ZSCALE_15
+    0x00000000, // PA_CL_VPORT_ZOFFSET_15
+    0x00000000, // PA_CL_UCP_0_X
+    0x00000000, // PA_CL_UCP_0_Y
+    0x00000000, // PA_CL_UCP_0_Z
+    0x00000000, // PA_CL_UCP_0_W
+    0x00000000, // PA_CL_UCP_1_X
+    0x00000000, // PA_CL_UCP_1_Y
+    0x00000000, // PA_CL_UCP_1_Z
+    0x00000000, // PA_CL_UCP_1_W
+    0x00000000, // PA_CL_UCP_2_X
+    0x00000000, // PA_CL_UCP_2_Y
+    0x00000000, // PA_CL_UCP_2_Z
+    0x00000000, // PA_CL_UCP_2_W
+    0x00000000, // PA_CL_UCP_3_X
+    0x00000000, // PA_CL_UCP_3_Y
+    0x00000000, // PA_CL_UCP_3_Z
+    0x00000000, // PA_CL_UCP_3_W
+    0x00000000, // PA_CL_UCP_4_X
+    0x00000000, // PA_CL_UCP_4_Y
+    0x00000000, // PA_CL_UCP_4_Z
+    0x00000000, // PA_CL_UCP_4_W
+    0x00000000, // PA_CL_UCP_5_X
+    0x00000000, // PA_CL_UCP_5_Y
+    0x00000000, // PA_CL_UCP_5_Z
+    0x00000000, // PA_CL_UCP_5_W
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_CNTL_0
+    0x00000000, // SPI_PS_INPUT_CNTL_1
+    0x00000000, // SPI_PS_INPUT_CNTL_2
+    0x00000000, // SPI_PS_INPUT_CNTL_3
+    0x00000000, // SPI_PS_INPUT_CNTL_4
+    0x00000000, // SPI_PS_INPUT_CNTL_5
+    0x00000000, // SPI_PS_INPUT_CNTL_6
+    0x00000000, // SPI_PS_INPUT_CNTL_7
+    0x00000000, // SPI_PS_INPUT_CNTL_8
+    0x00000000, // SPI_PS_INPUT_CNTL_9
+    0x00000000, // SPI_PS_INPUT_CNTL_10
+    0x00000000, // SPI_PS_INPUT_CNTL_11
+    0x00000000, // SPI_PS_INPUT_CNTL_12
+    0x00000000, // SPI_PS_INPUT_CNTL_13
+    0x00000000, // SPI_PS_INPUT_CNTL_14
+    0x00000000, // SPI_PS_INPUT_CNTL_15
+    0x00000000, // SPI_PS_INPUT_CNTL_16
+    0x00000000, // SPI_PS_INPUT_CNTL_17
+    0x00000000, // SPI_PS_INPUT_CNTL_18
+    0x00000000, // SPI_PS_INPUT_CNTL_19
+    0x00000000, // SPI_PS_INPUT_CNTL_20
+    0x00000000, // SPI_PS_INPUT_CNTL_21
+    0x00000000, // SPI_PS_INPUT_CNTL_22
+    0x00000000, // SPI_PS_INPUT_CNTL_23
+    0x00000000, // SPI_PS_INPUT_CNTL_24
+    0x00000000, // SPI_PS_INPUT_CNTL_25
+    0x00000000, // SPI_PS_INPUT_CNTL_26
+    0x00000000, // SPI_PS_INPUT_CNTL_27
+    0x00000000, // SPI_PS_INPUT_CNTL_28
+    0x00000000, // SPI_PS_INPUT_CNTL_29
+    0x00000000, // SPI_PS_INPUT_CNTL_30
+    0x00000000, // SPI_PS_INPUT_CNTL_31
+    0x00000000, // SPI_VS_OUT_CONFIG
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_ENA
+    0x00000000, // SPI_PS_INPUT_ADDR
+    0x00000000, // SPI_INTERP_CONTROL_0
+    0x00000002, // SPI_PS_IN_CONTROL
+    0, // HOLE
+    0x00000000, // SPI_BARYC_CNTL
+    0, // HOLE
+    0x00000000, // SPI_TMPRING_SIZE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_SHADER_POS_FORMAT
+    0x00000000, // SPI_SHADER_Z_FORMAT
+    0x00000000, // SPI_SHADER_COL_FORMAT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SX_PS_DOWNCONVERT
+    0x00000000, // SX_BLEND_OPT_EPSILON
+    0x00000000, // SX_BLEND_OPT_CONTROL
+    0x00000000, // SX_MRT0_BLEND_OPT
+    0x00000000, // SX_MRT1_BLEND_OPT
+    0x00000000, // SX_MRT2_BLEND_OPT
+    0x00000000, // SX_MRT3_BLEND_OPT
+    0x00000000, // SX_MRT4_BLEND_OPT
+    0x00000000, // SX_MRT5_BLEND_OPT
+    0x00000000, // SX_MRT6_BLEND_OPT
+    0x00000000, // SX_MRT7_BLEND_OPT
+    0x00000000, // CB_BLEND0_CONTROL
+    0x00000000, // CB_BLEND1_CONTROL
+    0x00000000, // CB_BLEND2_CONTROL
+    0x00000000, // CB_BLEND3_CONTROL
+    0x00000000, // CB_BLEND4_CONTROL
+    0x00000000, // CB_BLEND5_CONTROL
+    0x00000000, // CB_BLEND6_CONTROL
+    0x00000000, // CB_BLEND7_CONTROL
+    0x00000000, // CB_MRT0_EPITCH
+    0x00000000, // CB_MRT1_EPITCH
+    0x00000000, // CB_MRT2_EPITCH
+    0x00000000, // CB_MRT3_EPITCH
+    0x00000000, // CB_MRT4_EPITCH
+    0x00000000, // CB_MRT5_EPITCH
+    0x00000000, // CB_MRT6_EPITCH
+    0x00000000, // CB_MRT7_EPITCH
+};
+static const unsigned int gfx9_SECT_CONTEXT_def_3[] =
+{
+    0x00000000, // PA_CL_POINT_X_RAD
+    0x00000000, // PA_CL_POINT_Y_RAD
+    0x00000000, // PA_CL_POINT_SIZE
+    0x00000000, // PA_CL_POINT_CULL_RAD
+};
+static const unsigned int gfx9_SECT_CONTEXT_def_4[] =
+{
+    0x00000000, // DB_DEPTH_CONTROL
+    0x00000000, // DB_EQAA
+    0x00000000, // CB_COLOR_CONTROL
+    0x00000000, // DB_SHADER_CONTROL
+    0x00090000, // PA_CL_CLIP_CNTL
+    0x00000004, // PA_SU_SC_MODE_CNTL
+    0x00000000, // PA_CL_VTE_CNTL
+    0x00000000, // PA_CL_VS_OUT_CNTL
+    0x00000000, // PA_CL_NANINF_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_SCALE
+    0x00000000, // PA_SU_PRIM_FILTER_CNTL
+    0x00000000, // PA_SU_SMALL_PRIM_FILTER_CNTL
+    0x00000000, // PA_CL_OBJPRIM_ID_CNTL
+    0x00000000, // PA_CL_NGG_CNTL
+    0x00000000, // PA_SU_OVER_RASTERIZATION_CNTL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SU_POINT_SIZE
+    0x00000000, // PA_SU_POINT_MINMAX
+    0x00000000, // PA_SU_LINE_CNTL
+    0x00000000, // PA_SC_LINE_STIPPLE
+    0x00000000, // VGT_OUTPUT_PATH_CNTL
+    0x00000000, // VGT_HOS_CNTL
+    0x00000000, // VGT_HOS_MAX_TESS_LEVEL
+    0x00000000, // VGT_HOS_MIN_TESS_LEVEL
+    0x00000000, // VGT_HOS_REUSE_DEPTH
+    0x00000000, // VGT_GROUP_PRIM_TYPE
+    0x00000000, // VGT_GROUP_FIRST_DECR
+    0x00000000, // VGT_GROUP_DECR
+    0x00000000, // VGT_GROUP_VECT_0_CNTL
+    0x00000000, // VGT_GROUP_VECT_1_CNTL
+    0x00000000, // VGT_GROUP_VECT_0_FMT_CNTL
+    0x00000000, // VGT_GROUP_VECT_1_FMT_CNTL
+    0x00000000, // VGT_GS_MODE
+    0x00000000, // VGT_GS_ONCHIP_CNTL
+    0x00000000, // PA_SC_MODE_CNTL_0
+    0x00000000, // PA_SC_MODE_CNTL_1
+    0x00000000, // VGT_ENHANCE
+    0x00000100, // VGT_GS_PER_ES
+    0x00000080, // VGT_ES_PER_GS
+    0x00000002, // VGT_GS_PER_VS
+    0x00000000, // VGT_GSVS_RING_OFFSET_1
+    0x00000000, // VGT_GSVS_RING_OFFSET_2
+    0x00000000, // VGT_GSVS_RING_OFFSET_3
+    0x00000000, // VGT_GS_OUT_PRIM_TYPE
+    0x00000000, // IA_ENHANCE
+};
+static const unsigned int gfx9_SECT_CONTEXT_def_5[] =
+{
+    0x00000000, // WD_ENHANCE
+    0x00000000, // VGT_PRIMITIVEID_EN
+};
+static const unsigned int gfx9_SECT_CONTEXT_def_6[] =
+{
+    0x00000000, // VGT_PRIMITIVEID_RESET
+};
+static const unsigned int gfx9_SECT_CONTEXT_def_7[] =
+{
+    0x00000000, // VGT_GS_MAX_PRIMS_PER_SUBGROUP
+    0x00000000, // VGT_DRAW_PAYLOAD_CNTL
+    0, // HOLE
+    0x00000000, // VGT_INSTANCE_STEP_RATE_0
+    0x00000000, // VGT_INSTANCE_STEP_RATE_1
+    0, // HOLE
+    0x00000000, // VGT_ESGS_RING_ITEMSIZE
+    0x00000000, // VGT_GSVS_RING_ITEMSIZE
+    0x00000000, // VGT_REUSE_OFF
+    0x00000000, // VGT_VTX_CNT_EN
+    0x00000000, // DB_HTILE_SURFACE
+    0x00000000, // DB_SRESULTS_COMPARE_STATE0
+    0x00000000, // DB_SRESULTS_COMPARE_STATE1
+    0x00000000, // DB_PRELOAD_CONTROL
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_0
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_0
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_0
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_1
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_1
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_1
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_2
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_2
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_2
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_3
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_3
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_3
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_OFFSET
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_BUFFER_FILLED_SIZE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_VERTEX_STRIDE
+    0, // HOLE
+    0x00000000, // VGT_GS_MAX_VERT_OUT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_TESS_DISTRIBUTION
+    0x00000000, // VGT_SHADER_STAGES_EN
+    0x00000000, // VGT_LS_HS_CONFIG
+    0x00000000, // VGT_GS_VERT_ITEMSIZE
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_1
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_2
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_3
+    0x00000000, // VGT_TF_PARAM
+    0x00000000, // DB_ALPHA_TO_MASK
+    0x00000000, // VGT_DISPATCH_DRAW_INDEX
+    0x00000000, // PA_SU_POLY_OFFSET_DB_FMT_CNTL
+    0x00000000, // PA_SU_POLY_OFFSET_CLAMP
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_OFFSET
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_OFFSET
+    0x00000000, // VGT_GS_INSTANCE_CNT
+    0x00000000, // VGT_STRMOUT_CONFIG
+    0x00000000, // VGT_STRMOUT_BUFFER_CONFIG
+};
+static const unsigned int gfx9_SECT_CONTEXT_def_8[] =
+{
+    0x00000000, // PA_SC_CENTROID_PRIORITY_0
+    0x00000000, // PA_SC_CENTROID_PRIORITY_1
+    0x00001000, // PA_SC_LINE_CNTL
+    0x00000000, // PA_SC_AA_CONFIG
+    0x00000005, // PA_SU_VTX_CNTL
+    0x3f800000, // PA_CL_GB_VERT_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_VERT_DISC_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_DISC_ADJ
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_3
+    0xffffffff, // PA_SC_AA_MASK_X0Y0_X1Y0
+    0xffffffff, // PA_SC_AA_MASK_X0Y1_X1Y1
+    0x00000000, // PA_SC_SHADER_CONTROL
+    0x00000003, // PA_SC_BINNER_CNTL_0
+    0x00000000, // PA_SC_BINNER_CNTL_1
+    0x00000000, // PA_SC_CONSERVATIVE_RASTERIZATION_CNTL
+    0x00000000, // PA_SC_NGG_MODE_CNTL
+    0, // HOLE
+    0x0000001e, // VGT_VERTEX_REUSE_BLOCK_CNTL
+    0x00000020, // VGT_OUT_DEALLOC_CNTL
+    0x00000000, // CB_COLOR0_BASE
+    0x00000000, // CB_COLOR0_BASE_EXT
+    0x00000000, // CB_COLOR0_ATTRIB2
+    0x00000000, // CB_COLOR0_VIEW
+    0x00000000, // CB_COLOR0_INFO
+    0x00000000, // CB_COLOR0_ATTRIB
+    0x00000000, // CB_COLOR0_DCC_CONTROL
+    0x00000000, // CB_COLOR0_CMASK
+    0x00000000, // CB_COLOR0_CMASK_BASE_EXT
+    0x00000000, // CB_COLOR0_FMASK
+    0x00000000, // CB_COLOR0_FMASK_BASE_EXT
+    0x00000000, // CB_COLOR0_CLEAR_WORD0
+    0x00000000, // CB_COLOR0_CLEAR_WORD1
+    0x00000000, // CB_COLOR0_DCC_BASE
+    0x00000000, // CB_COLOR0_DCC_BASE_EXT
+    0x00000000, // CB_COLOR1_BASE
+    0x00000000, // CB_COLOR1_BASE_EXT
+    0x00000000, // CB_COLOR1_ATTRIB2
+    0x00000000, // CB_COLOR1_VIEW
+    0x00000000, // CB_COLOR1_INFO
+    0x00000000, // CB_COLOR1_ATTRIB
+    0x00000000, // CB_COLOR1_DCC_CONTROL
+    0x00000000, // CB_COLOR1_CMASK
+    0x00000000, // CB_COLOR1_CMASK_BASE_EXT
+    0x00000000, // CB_COLOR1_FMASK
+    0x00000000, // CB_COLOR1_FMASK_BASE_EXT
+    0x00000000, // CB_COLOR1_CLEAR_WORD0
+    0x00000000, // CB_COLOR1_CLEAR_WORD1
+    0x00000000, // CB_COLOR1_DCC_BASE
+    0x00000000, // CB_COLOR1_DCC_BASE_EXT
+    0x00000000, // CB_COLOR2_BASE
+    0x00000000, // CB_COLOR2_BASE_EXT
+    0x00000000, // CB_COLOR2_ATTRIB2
+    0x00000000, // CB_COLOR2_VIEW
+    0x00000000, // CB_COLOR2_INFO
+    0x00000000, // CB_COLOR2_ATTRIB
+    0x00000000, // CB_COLOR2_DCC_CONTROL
+    0x00000000, // CB_COLOR2_CMASK
+    0x00000000, // CB_COLOR2_CMASK_BASE_EXT
+    0x00000000, // CB_COLOR2_FMASK
+    0x00000000, // CB_COLOR2_FMASK_BASE_EXT
+    0x00000000, // CB_COLOR2_CLEAR_WORD0
+    0x00000000, // CB_COLOR2_CLEAR_WORD1
+    0x00000000, // CB_COLOR2_DCC_BASE
+    0x00000000, // CB_COLOR2_DCC_BASE_EXT
+    0x00000000, // CB_COLOR3_BASE
+    0x00000000, // CB_COLOR3_BASE_EXT
+    0x00000000, // CB_COLOR3_ATTRIB2
+    0x00000000, // CB_COLOR3_VIEW
+    0x00000000, // CB_COLOR3_INFO
+    0x00000000, // CB_COLOR3_ATTRIB
+    0x00000000, // CB_COLOR3_DCC_CONTROL
+    0x00000000, // CB_COLOR3_CMASK
+    0x00000000, // CB_COLOR3_CMASK_BASE_EXT
+    0x00000000, // CB_COLOR3_FMASK
+    0x00000000, // CB_COLOR3_FMASK_BASE_EXT
+    0x00000000, // CB_COLOR3_CLEAR_WORD0
+    0x00000000, // CB_COLOR3_CLEAR_WORD1
+    0x00000000, // CB_COLOR3_DCC_BASE
+    0x00000000, // CB_COLOR3_DCC_BASE_EXT
+    0x00000000, // CB_COLOR4_BASE
+    0x00000000, // CB_COLOR4_BASE_EXT
+    0x00000000, // CB_COLOR4_ATTRIB2
+    0x00000000, // CB_COLOR4_VIEW
+    0x00000000, // CB_COLOR4_INFO
+    0x00000000, // CB_COLOR4_ATTRIB
+    0x00000000, // CB_COLOR4_DCC_CONTROL
+    0x00000000, // CB_COLOR4_CMASK
+    0x00000000, // CB_COLOR4_CMASK_BASE_EXT
+    0x00000000, // CB_COLOR4_FMASK
+    0x00000000, // CB_COLOR4_FMASK_BASE_EXT
+    0x00000000, // CB_COLOR4_CLEAR_WORD0
+    0x00000000, // CB_COLOR4_CLEAR_WORD1
+    0x00000000, // CB_COLOR4_DCC_BASE
+    0x00000000, // CB_COLOR4_DCC_BASE_EXT
+    0x00000000, // CB_COLOR5_BASE
+    0x00000000, // CB_COLOR5_BASE_EXT
+    0x00000000, // CB_COLOR5_ATTRIB2
+    0x00000000, // CB_COLOR5_VIEW
+    0x00000000, // CB_COLOR5_INFO
+    0x00000000, // CB_COLOR5_ATTRIB
+    0x00000000, // CB_COLOR5_DCC_CONTROL
+    0x00000000, // CB_COLOR5_CMASK
+    0x00000000, // CB_COLOR5_CMASK_BASE_EXT
+    0x00000000, // CB_COLOR5_FMASK
+    0x00000000, // CB_COLOR5_FMASK_BASE_EXT
+    0x00000000, // CB_COLOR5_CLEAR_WORD0
+    0x00000000, // CB_COLOR5_CLEAR_WORD1
+    0x00000000, // CB_COLOR5_DCC_BASE
+    0x00000000, // CB_COLOR5_DCC_BASE_EXT
+    0x00000000, // CB_COLOR6_BASE
+    0x00000000, // CB_COLOR6_BASE_EXT
+    0x00000000, // CB_COLOR6_ATTRIB2
+    0x00000000, // CB_COLOR6_VIEW
+    0x00000000, // CB_COLOR6_INFO
+    0x00000000, // CB_COLOR6_ATTRIB
+    0x00000000, // CB_COLOR6_DCC_CONTROL
+    0x00000000, // CB_COLOR6_CMASK
+    0x00000000, // CB_COLOR6_CMASK_BASE_EXT
+    0x00000000, // CB_COLOR6_FMASK
+    0x00000000, // CB_COLOR6_FMASK_BASE_EXT
+    0x00000000, // CB_COLOR6_CLEAR_WORD0
+    0x00000000, // CB_COLOR6_CLEAR_WORD1
+    0x00000000, // CB_COLOR6_DCC_BASE
+    0x00000000, // CB_COLOR6_DCC_BASE_EXT
+    0x00000000, // CB_COLOR7_BASE
+    0x00000000, // CB_COLOR7_BASE_EXT
+    0x00000000, // CB_COLOR7_ATTRIB2
+    0x00000000, // CB_COLOR7_VIEW
+    0x00000000, // CB_COLOR7_INFO
+    0x00000000, // CB_COLOR7_ATTRIB
+    0x00000000, // CB_COLOR7_DCC_CONTROL
+    0x00000000, // CB_COLOR7_CMASK
+    0x00000000, // CB_COLOR7_CMASK_BASE_EXT
+    0x00000000, // CB_COLOR7_FMASK
+    0x00000000, // CB_COLOR7_FMASK_BASE_EXT
+    0x00000000, // CB_COLOR7_CLEAR_WORD0
+    0x00000000, // CB_COLOR7_CLEAR_WORD1
+    0x00000000, // CB_COLOR7_DCC_BASE
+    0x00000000, // CB_COLOR7_DCC_BASE_EXT
+};
+static const struct cs_extent_def gfx9_SECT_CONTEXT_defs[] =
+{
+    {gfx9_SECT_CONTEXT_def_1, 0x0000a000, 212 },
+    {gfx9_SECT_CONTEXT_def_2, 0x0000a0d6, 282 },
+    {gfx9_SECT_CONTEXT_def_3, 0x0000a1f5, 4 },
+    {gfx9_SECT_CONTEXT_def_4, 0x0000a200, 157 },
+    {gfx9_SECT_CONTEXT_def_5, 0x0000a2a0, 2 },
+    {gfx9_SECT_CONTEXT_def_6, 0x0000a2a3, 1 },
+    {gfx9_SECT_CONTEXT_def_7, 0x0000a2a5, 66 },
+    {gfx9_SECT_CONTEXT_def_8, 0x0000a2f5, 155 },
+    { 0, 0, 0 }
+};
+static const struct cs_section_def gfx9_cs_data[] = {
+    { gfx9_SECT_CONTEXT_defs, SECT_CONTEXT },
+    { 0, SECT_NONE }
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/clearstate_si.h b/sys/dev/pci/drm/amd/amdgpu/clearstate_si.h
new file mode 100644
index 00000000000..66e39cdb5cb
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/clearstate_si.h
@@ -0,0 +1,941 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+static const u32 si_SECT_CONTEXT_def_1[] =
+{
+    0x00000000, // DB_RENDER_CONTROL
+    0x00000000, // DB_COUNT_CONTROL
+    0x00000000, // DB_DEPTH_VIEW
+    0x00000000, // DB_RENDER_OVERRIDE
+    0x00000000, // DB_RENDER_OVERRIDE2
+    0x00000000, // DB_HTILE_DATA_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // DB_DEPTH_BOUNDS_MIN
+    0x00000000, // DB_DEPTH_BOUNDS_MAX
+    0x00000000, // DB_STENCIL_CLEAR
+    0x00000000, // DB_DEPTH_CLEAR
+    0x00000000, // PA_SC_SCREEN_SCISSOR_TL
+    0x40004000, // PA_SC_SCREEN_SCISSOR_BR
+    0, // HOLE
+    0x00000000, // DB_DEPTH_INFO
+    0x00000000, // DB_Z_INFO
+    0x00000000, // DB_STENCIL_INFO
+    0x00000000, // DB_Z_READ_BASE
+    0x00000000, // DB_STENCIL_READ_BASE
+    0x00000000, // DB_Z_WRITE_BASE
+    0x00000000, // DB_STENCIL_WRITE_BASE
+    0x00000000, // DB_DEPTH_SIZE
+    0x00000000, // DB_DEPTH_SLICE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // TA_BC_BASE_ADDR
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // COHER_DEST_BASE_2
+    0x00000000, // COHER_DEST_BASE_3
+    0x00000000, // PA_SC_WINDOW_OFFSET
+    0x80000000, // PA_SC_WINDOW_SCISSOR_TL
+    0x40004000, // PA_SC_WINDOW_SCISSOR_BR
+    0x0000ffff, // PA_SC_CLIPRECT_RULE
+    0x00000000, // PA_SC_CLIPRECT_0_TL
+    0x40004000, // PA_SC_CLIPRECT_0_BR
+    0x00000000, // PA_SC_CLIPRECT_1_TL
+    0x40004000, // PA_SC_CLIPRECT_1_BR
+    0x00000000, // PA_SC_CLIPRECT_2_TL
+    0x40004000, // PA_SC_CLIPRECT_2_BR
+    0x00000000, // PA_SC_CLIPRECT_3_TL
+    0x40004000, // PA_SC_CLIPRECT_3_BR
+    0xaa99aaaa, // PA_SC_EDGERULE
+    0x00000000, // PA_SU_HARDWARE_SCREEN_OFFSET
+    0xffffffff, // CB_TARGET_MASK
+    0xffffffff, // CB_SHADER_MASK
+    0x80000000, // PA_SC_GENERIC_SCISSOR_TL
+    0x40004000, // PA_SC_GENERIC_SCISSOR_BR
+    0x00000000, // COHER_DEST_BASE_0
+    0x00000000, // COHER_DEST_BASE_1
+    0x80000000, // PA_SC_VPORT_SCISSOR_0_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_0_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_1_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_1_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_2_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_2_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_3_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_3_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_4_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_4_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_5_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_5_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_6_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_6_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_7_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_7_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_8_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_8_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_9_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_9_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_10_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_10_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_11_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_11_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_12_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_12_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_13_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_13_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_14_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_14_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_15_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_15_BR
+    0x00000000, // PA_SC_VPORT_ZMIN_0
+    0x3f800000, // PA_SC_VPORT_ZMAX_0
+    0x00000000, // PA_SC_VPORT_ZMIN_1
+    0x3f800000, // PA_SC_VPORT_ZMAX_1
+    0x00000000, // PA_SC_VPORT_ZMIN_2
+    0x3f800000, // PA_SC_VPORT_ZMAX_2
+    0x00000000, // PA_SC_VPORT_ZMIN_3
+    0x3f800000, // PA_SC_VPORT_ZMAX_3
+    0x00000000, // PA_SC_VPORT_ZMIN_4
+    0x3f800000, // PA_SC_VPORT_ZMAX_4
+    0x00000000, // PA_SC_VPORT_ZMIN_5
+    0x3f800000, // PA_SC_VPORT_ZMAX_5
+    0x00000000, // PA_SC_VPORT_ZMIN_6
+    0x3f800000, // PA_SC_VPORT_ZMAX_6
+    0x00000000, // PA_SC_VPORT_ZMIN_7
+    0x3f800000, // PA_SC_VPORT_ZMAX_7
+    0x00000000, // PA_SC_VPORT_ZMIN_8
+    0x3f800000, // PA_SC_VPORT_ZMAX_8
+    0x00000000, // PA_SC_VPORT_ZMIN_9
+    0x3f800000, // PA_SC_VPORT_ZMAX_9
+    0x00000000, // PA_SC_VPORT_ZMIN_10
+    0x3f800000, // PA_SC_VPORT_ZMAX_10
+    0x00000000, // PA_SC_VPORT_ZMIN_11
+    0x3f800000, // PA_SC_VPORT_ZMAX_11
+    0x00000000, // PA_SC_VPORT_ZMIN_12
+    0x3f800000, // PA_SC_VPORT_ZMAX_12
+    0x00000000, // PA_SC_VPORT_ZMIN_13
+    0x3f800000, // PA_SC_VPORT_ZMAX_13
+    0x00000000, // PA_SC_VPORT_ZMIN_14
+    0x3f800000, // PA_SC_VPORT_ZMAX_14
+    0x00000000, // PA_SC_VPORT_ZMIN_15
+    0x3f800000, // PA_SC_VPORT_ZMAX_15
+};
+static const u32 si_SECT_CONTEXT_def_2[] =
+{
+    0x00000000, // CP_PERFMON_CNTX_CNTL
+    0x00000000, // CP_RINGID
+    0x00000000, // CP_VMID
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0xffffffff, // VGT_MAX_VTX_INDX
+    0x00000000, // VGT_MIN_VTX_INDX
+    0x00000000, // VGT_INDX_OFFSET
+    0x00000000, // VGT_MULTI_PRIM_IB_RESET_INDX
+    0, // HOLE
+    0x00000000, // CB_BLEND_RED
+    0x00000000, // CB_BLEND_GREEN
+    0x00000000, // CB_BLEND_BLUE
+    0x00000000, // CB_BLEND_ALPHA
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // DB_STENCIL_CONTROL
+    0x00000000, // DB_STENCILREFMASK
+    0x00000000, // DB_STENCILREFMASK_BF
+    0, // HOLE
+    0x00000000, // PA_CL_VPORT_XSCALE
+    0x00000000, // PA_CL_VPORT_XOFFSET
+    0x00000000, // PA_CL_VPORT_YSCALE
+    0x00000000, // PA_CL_VPORT_YOFFSET
+    0x00000000, // PA_CL_VPORT_ZSCALE
+    0x00000000, // PA_CL_VPORT_ZOFFSET
+    0x00000000, // PA_CL_VPORT_XSCALE_1
+    0x00000000, // PA_CL_VPORT_XOFFSET_1
+    0x00000000, // PA_CL_VPORT_YSCALE_1
+    0x00000000, // PA_CL_VPORT_YOFFSET_1
+    0x00000000, // PA_CL_VPORT_ZSCALE_1
+    0x00000000, // PA_CL_VPORT_ZOFFSET_1
+    0x00000000, // PA_CL_VPORT_XSCALE_2
+    0x00000000, // PA_CL_VPORT_XOFFSET_2
+    0x00000000, // PA_CL_VPORT_YSCALE_2
+    0x00000000, // PA_CL_VPORT_YOFFSET_2
+    0x00000000, // PA_CL_VPORT_ZSCALE_2
+    0x00000000, // PA_CL_VPORT_ZOFFSET_2
+    0x00000000, // PA_CL_VPORT_XSCALE_3
+    0x00000000, // PA_CL_VPORT_XOFFSET_3
+    0x00000000, // PA_CL_VPORT_YSCALE_3
+    0x00000000, // PA_CL_VPORT_YOFFSET_3
+    0x00000000, // PA_CL_VPORT_ZSCALE_3
+    0x00000000, // PA_CL_VPORT_ZOFFSET_3
+    0x00000000, // PA_CL_VPORT_XSCALE_4
+    0x00000000, // PA_CL_VPORT_XOFFSET_4
+    0x00000000, // PA_CL_VPORT_YSCALE_4
+    0x00000000, // PA_CL_VPORT_YOFFSET_4
+    0x00000000, // PA_CL_VPORT_ZSCALE_4
+    0x00000000, // PA_CL_VPORT_ZOFFSET_4
+    0x00000000, // PA_CL_VPORT_XSCALE_5
+    0x00000000, // PA_CL_VPORT_XOFFSET_5
+    0x00000000, // PA_CL_VPORT_YSCALE_5
+    0x00000000, // PA_CL_VPORT_YOFFSET_5
+    0x00000000, // PA_CL_VPORT_ZSCALE_5
+    0x00000000, // PA_CL_VPORT_ZOFFSET_5
+    0x00000000, // PA_CL_VPORT_XSCALE_6
+    0x00000000, // PA_CL_VPORT_XOFFSET_6
+    0x00000000, // PA_CL_VPORT_YSCALE_6
+    0x00000000, // PA_CL_VPORT_YOFFSET_6
+    0x00000000, // PA_CL_VPORT_ZSCALE_6
+    0x00000000, // PA_CL_VPORT_ZOFFSET_6
+    0x00000000, // PA_CL_VPORT_XSCALE_7
+    0x00000000, // PA_CL_VPORT_XOFFSET_7
+    0x00000000, // PA_CL_VPORT_YSCALE_7
+    0x00000000, // PA_CL_VPORT_YOFFSET_7
+    0x00000000, // PA_CL_VPORT_ZSCALE_7
+    0x00000000, // PA_CL_VPORT_ZOFFSET_7
+    0x00000000, // PA_CL_VPORT_XSCALE_8
+    0x00000000, // PA_CL_VPORT_XOFFSET_8
+    0x00000000, // PA_CL_VPORT_YSCALE_8
+    0x00000000, // PA_CL_VPORT_YOFFSET_8
+    0x00000000, // PA_CL_VPORT_ZSCALE_8
+    0x00000000, // PA_CL_VPORT_ZOFFSET_8
+    0x00000000, // PA_CL_VPORT_XSCALE_9
+    0x00000000, // PA_CL_VPORT_XOFFSET_9
+    0x00000000, // PA_CL_VPORT_YSCALE_9
+    0x00000000, // PA_CL_VPORT_YOFFSET_9
+    0x00000000, // PA_CL_VPORT_ZSCALE_9
+    0x00000000, // PA_CL_VPORT_ZOFFSET_9
+    0x00000000, // PA_CL_VPORT_XSCALE_10
+    0x00000000, // PA_CL_VPORT_XOFFSET_10
+    0x00000000, // PA_CL_VPORT_YSCALE_10
+    0x00000000, // PA_CL_VPORT_YOFFSET_10
+    0x00000000, // PA_CL_VPORT_ZSCALE_10
+    0x00000000, // PA_CL_VPORT_ZOFFSET_10
+    0x00000000, // PA_CL_VPORT_XSCALE_11
+    0x00000000, // PA_CL_VPORT_XOFFSET_11
+    0x00000000, // PA_CL_VPORT_YSCALE_11
+    0x00000000, // PA_CL_VPORT_YOFFSET_11
+    0x00000000, // PA_CL_VPORT_ZSCALE_11
+    0x00000000, // PA_CL_VPORT_ZOFFSET_11
+    0x00000000, // PA_CL_VPORT_XSCALE_12
+    0x00000000, // PA_CL_VPORT_XOFFSET_12
+    0x00000000, // PA_CL_VPORT_YSCALE_12
+    0x00000000, // PA_CL_VPORT_YOFFSET_12
+    0x00000000, // PA_CL_VPORT_ZSCALE_12
+    0x00000000, // PA_CL_VPORT_ZOFFSET_12
+    0x00000000, // PA_CL_VPORT_XSCALE_13
+    0x00000000, // PA_CL_VPORT_XOFFSET_13
+    0x00000000, // PA_CL_VPORT_YSCALE_13
+    0x00000000, // PA_CL_VPORT_YOFFSET_13
+    0x00000000, // PA_CL_VPORT_ZSCALE_13
+    0x00000000, // PA_CL_VPORT_ZOFFSET_13
+    0x00000000, // PA_CL_VPORT_XSCALE_14
+    0x00000000, // PA_CL_VPORT_XOFFSET_14
+    0x00000000, // PA_CL_VPORT_YSCALE_14
+    0x00000000, // PA_CL_VPORT_YOFFSET_14
+    0x00000000, // PA_CL_VPORT_ZSCALE_14
+    0x00000000, // PA_CL_VPORT_ZOFFSET_14
+    0x00000000, // PA_CL_VPORT_XSCALE_15
+    0x00000000, // PA_CL_VPORT_XOFFSET_15
+    0x00000000, // PA_CL_VPORT_YSCALE_15
+    0x00000000, // PA_CL_VPORT_YOFFSET_15
+    0x00000000, // PA_CL_VPORT_ZSCALE_15
+    0x00000000, // PA_CL_VPORT_ZOFFSET_15
+    0x00000000, // PA_CL_UCP_0_X
+    0x00000000, // PA_CL_UCP_0_Y
+    0x00000000, // PA_CL_UCP_0_Z
+    0x00000000, // PA_CL_UCP_0_W
+    0x00000000, // PA_CL_UCP_1_X
+    0x00000000, // PA_CL_UCP_1_Y
+    0x00000000, // PA_CL_UCP_1_Z
+    0x00000000, // PA_CL_UCP_1_W
+    0x00000000, // PA_CL_UCP_2_X
+    0x00000000, // PA_CL_UCP_2_Y
+    0x00000000, // PA_CL_UCP_2_Z
+    0x00000000, // PA_CL_UCP_2_W
+    0x00000000, // PA_CL_UCP_3_X
+    0x00000000, // PA_CL_UCP_3_Y
+    0x00000000, // PA_CL_UCP_3_Z
+    0x00000000, // PA_CL_UCP_3_W
+    0x00000000, // PA_CL_UCP_4_X
+    0x00000000, // PA_CL_UCP_4_Y
+    0x00000000, // PA_CL_UCP_4_Z
+    0x00000000, // PA_CL_UCP_4_W
+    0x00000000, // PA_CL_UCP_5_X
+    0x00000000, // PA_CL_UCP_5_Y
+    0x00000000, // PA_CL_UCP_5_Z
+    0x00000000, // PA_CL_UCP_5_W
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_CNTL_0
+    0x00000000, // SPI_PS_INPUT_CNTL_1
+    0x00000000, // SPI_PS_INPUT_CNTL_2
+    0x00000000, // SPI_PS_INPUT_CNTL_3
+    0x00000000, // SPI_PS_INPUT_CNTL_4
+    0x00000000, // SPI_PS_INPUT_CNTL_5
+    0x00000000, // SPI_PS_INPUT_CNTL_6
+    0x00000000, // SPI_PS_INPUT_CNTL_7
+    0x00000000, // SPI_PS_INPUT_CNTL_8
+    0x00000000, // SPI_PS_INPUT_CNTL_9
+    0x00000000, // SPI_PS_INPUT_CNTL_10
+    0x00000000, // SPI_PS_INPUT_CNTL_11
+    0x00000000, // SPI_PS_INPUT_CNTL_12
+    0x00000000, // SPI_PS_INPUT_CNTL_13
+    0x00000000, // SPI_PS_INPUT_CNTL_14
+    0x00000000, // SPI_PS_INPUT_CNTL_15
+    0x00000000, // SPI_PS_INPUT_CNTL_16
+    0x00000000, // SPI_PS_INPUT_CNTL_17
+    0x00000000, // SPI_PS_INPUT_CNTL_18
+    0x00000000, // SPI_PS_INPUT_CNTL_19
+    0x00000000, // SPI_PS_INPUT_CNTL_20
+    0x00000000, // SPI_PS_INPUT_CNTL_21
+    0x00000000, // SPI_PS_INPUT_CNTL_22
+    0x00000000, // SPI_PS_INPUT_CNTL_23
+    0x00000000, // SPI_PS_INPUT_CNTL_24
+    0x00000000, // SPI_PS_INPUT_CNTL_25
+    0x00000000, // SPI_PS_INPUT_CNTL_26
+    0x00000000, // SPI_PS_INPUT_CNTL_27
+    0x00000000, // SPI_PS_INPUT_CNTL_28
+    0x00000000, // SPI_PS_INPUT_CNTL_29
+    0x00000000, // SPI_PS_INPUT_CNTL_30
+    0x00000000, // SPI_PS_INPUT_CNTL_31
+    0x00000000, // SPI_VS_OUT_CONFIG
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_ENA
+    0x00000000, // SPI_PS_INPUT_ADDR
+    0x00000000, // SPI_INTERP_CONTROL_0
+    0x00000002, // SPI_PS_IN_CONTROL
+    0, // HOLE
+    0x00000000, // SPI_BARYC_CNTL
+    0, // HOLE
+    0x00000000, // SPI_TMPRING_SIZE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_WAVE_MGMT_1
+    0x00000000, // SPI_WAVE_MGMT_2
+    0x00000000, // SPI_SHADER_POS_FORMAT
+    0x00000000, // SPI_SHADER_Z_FORMAT
+    0x00000000, // SPI_SHADER_COL_FORMAT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_BLEND0_CONTROL
+    0x00000000, // CB_BLEND1_CONTROL
+    0x00000000, // CB_BLEND2_CONTROL
+    0x00000000, // CB_BLEND3_CONTROL
+    0x00000000, // CB_BLEND4_CONTROL
+    0x00000000, // CB_BLEND5_CONTROL
+    0x00000000, // CB_BLEND6_CONTROL
+    0x00000000, // CB_BLEND7_CONTROL
+};
+static const u32 si_SECT_CONTEXT_def_3[] =
+{
+    0x00000000, // PA_CL_POINT_X_RAD
+    0x00000000, // PA_CL_POINT_Y_RAD
+    0x00000000, // PA_CL_POINT_SIZE
+    0x00000000, // PA_CL_POINT_CULL_RAD
+    0x00000000, // VGT_DMA_BASE_HI
+    0x00000000, // VGT_DMA_BASE
+};
+static const u32 si_SECT_CONTEXT_def_4[] =
+{
+    0x00000000, // DB_DEPTH_CONTROL
+    0x00000000, // DB_EQAA
+    0x00000000, // CB_COLOR_CONTROL
+    0x00000000, // DB_SHADER_CONTROL
+    0x00090000, // PA_CL_CLIP_CNTL
+    0x00000004, // PA_SU_SC_MODE_CNTL
+    0x00000000, // PA_CL_VTE_CNTL
+    0x00000000, // PA_CL_VS_OUT_CNTL
+    0x00000000, // PA_CL_NANINF_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_SCALE
+    0x00000000, // PA_SU_PRIM_FILTER_CNTL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SU_POINT_SIZE
+    0x00000000, // PA_SU_POINT_MINMAX
+    0x00000000, // PA_SU_LINE_CNTL
+    0x00000000, // PA_SC_LINE_STIPPLE
+    0x00000000, // VGT_OUTPUT_PATH_CNTL
+    0x00000000, // VGT_HOS_CNTL
+    0x00000000, // VGT_HOS_MAX_TESS_LEVEL
+    0x00000000, // VGT_HOS_MIN_TESS_LEVEL
+    0x00000000, // VGT_HOS_REUSE_DEPTH
+    0x00000000, // VGT_GROUP_PRIM_TYPE
+    0x00000000, // VGT_GROUP_FIRST_DECR
+    0x00000000, // VGT_GROUP_DECR
+    0x00000000, // VGT_GROUP_VECT_0_CNTL
+    0x00000000, // VGT_GROUP_VECT_1_CNTL
+    0x00000000, // VGT_GROUP_VECT_0_FMT_CNTL
+    0x00000000, // VGT_GROUP_VECT_1_FMT_CNTL
+    0x00000000, // VGT_GS_MODE
+    0, // HOLE
+    0x00000000, // PA_SC_MODE_CNTL_0
+    0x00000000, // PA_SC_MODE_CNTL_1
+    0x00000000, // VGT_ENHANCE
+    0x00000100, // VGT_GS_PER_ES
+    0x00000080, // VGT_ES_PER_GS
+    0x00000002, // VGT_GS_PER_VS
+    0x00000000, // VGT_GSVS_RING_OFFSET_1
+    0x00000000, // VGT_GSVS_RING_OFFSET_2
+    0x00000000, // VGT_GSVS_RING_OFFSET_3
+    0x00000000, // VGT_GS_OUT_PRIM_TYPE
+    0x00000000, // IA_ENHANCE
+};
+static const u32 si_SECT_CONTEXT_def_5[] =
+{
+    0x00000000, // VGT_PRIMITIVEID_EN
+};
+static const u32 si_SECT_CONTEXT_def_6[] =
+{
+    0x00000000, // VGT_PRIMITIVEID_RESET
+};
+static const u32 si_SECT_CONTEXT_def_7[] =
+{
+    0x00000000, // VGT_MULTI_PRIM_IB_RESET_EN
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_INSTANCE_STEP_RATE_0
+    0x00000000, // VGT_INSTANCE_STEP_RATE_1
+    0x000000ff, // IA_MULTI_VGT_PARAM
+    0x00000000, // VGT_ESGS_RING_ITEMSIZE
+    0x00000000, // VGT_GSVS_RING_ITEMSIZE
+    0x00000000, // VGT_REUSE_OFF
+    0x00000000, // VGT_VTX_CNT_EN
+    0x00000000, // DB_HTILE_SURFACE
+    0x00000000, // DB_SRESULTS_COMPARE_STATE0
+    0x00000000, // DB_SRESULTS_COMPARE_STATE1
+    0x00000000, // DB_PRELOAD_CONTROL
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_0
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_0
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_0
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_1
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_1
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_1
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_2
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_2
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_2
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_3
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_3
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_3
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_OFFSET
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_BUFFER_FILLED_SIZE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_VERTEX_STRIDE
+    0, // HOLE
+    0x00000000, // VGT_GS_MAX_VERT_OUT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_SHADER_STAGES_EN
+    0x00000000, // VGT_LS_HS_CONFIG
+    0x00000000, // VGT_GS_VERT_ITEMSIZE
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_1
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_2
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_3
+    0x00000000, // VGT_TF_PARAM
+    0x00000000, // DB_ALPHA_TO_MASK
+    0, // HOLE
+    0x00000000, // PA_SU_POLY_OFFSET_DB_FMT_CNTL
+    0x00000000, // PA_SU_POLY_OFFSET_CLAMP
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_OFFSET
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_OFFSET
+    0x00000000, // VGT_GS_INSTANCE_CNT
+    0x00000000, // VGT_STRMOUT_CONFIG
+    0x00000000, // VGT_STRMOUT_BUFFER_CONFIG
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SC_CENTROID_PRIORITY_0
+    0x00000000, // PA_SC_CENTROID_PRIORITY_1
+    0x00001000, // PA_SC_LINE_CNTL
+    0x00000000, // PA_SC_AA_CONFIG
+    0x00000005, // PA_SU_VTX_CNTL
+    0x3f800000, // PA_CL_GB_VERT_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_VERT_DISC_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_DISC_ADJ
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_3
+    0xffffffff, // PA_SC_AA_MASK_X0Y0_X1Y0
+    0xffffffff, // PA_SC_AA_MASK_X0Y1_X1Y1
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x0000000e, // VGT_VERTEX_REUSE_BLOCK_CNTL
+    0x00000010, // VGT_OUT_DEALLOC_CNTL
+    0x00000000, // CB_COLOR0_BASE
+    0x00000000, // CB_COLOR0_PITCH
+    0x00000000, // CB_COLOR0_SLICE
+    0x00000000, // CB_COLOR0_VIEW
+    0x00000000, // CB_COLOR0_INFO
+    0x00000000, // CB_COLOR0_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR0_CMASK
+    0x00000000, // CB_COLOR0_CMASK_SLICE
+    0x00000000, // CB_COLOR0_FMASK
+    0x00000000, // CB_COLOR0_FMASK_SLICE
+    0x00000000, // CB_COLOR0_CLEAR_WORD0
+    0x00000000, // CB_COLOR0_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR1_BASE
+    0x00000000, // CB_COLOR1_PITCH
+    0x00000000, // CB_COLOR1_SLICE
+    0x00000000, // CB_COLOR1_VIEW
+    0x00000000, // CB_COLOR1_INFO
+    0x00000000, // CB_COLOR1_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR1_CMASK
+    0x00000000, // CB_COLOR1_CMASK_SLICE
+    0x00000000, // CB_COLOR1_FMASK
+    0x00000000, // CB_COLOR1_FMASK_SLICE
+    0x00000000, // CB_COLOR1_CLEAR_WORD0
+    0x00000000, // CB_COLOR1_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR2_BASE
+    0x00000000, // CB_COLOR2_PITCH
+    0x00000000, // CB_COLOR2_SLICE
+    0x00000000, // CB_COLOR2_VIEW
+    0x00000000, // CB_COLOR2_INFO
+    0x00000000, // CB_COLOR2_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR2_CMASK
+    0x00000000, // CB_COLOR2_CMASK_SLICE
+    0x00000000, // CB_COLOR2_FMASK
+    0x00000000, // CB_COLOR2_FMASK_SLICE
+    0x00000000, // CB_COLOR2_CLEAR_WORD0
+    0x00000000, // CB_COLOR2_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR3_BASE
+    0x00000000, // CB_COLOR3_PITCH
+    0x00000000, // CB_COLOR3_SLICE
+    0x00000000, // CB_COLOR3_VIEW
+    0x00000000, // CB_COLOR3_INFO
+    0x00000000, // CB_COLOR3_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR3_CMASK
+    0x00000000, // CB_COLOR3_CMASK_SLICE
+    0x00000000, // CB_COLOR3_FMASK
+    0x00000000, // CB_COLOR3_FMASK_SLICE
+    0x00000000, // CB_COLOR3_CLEAR_WORD0
+    0x00000000, // CB_COLOR3_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR4_BASE
+    0x00000000, // CB_COLOR4_PITCH
+    0x00000000, // CB_COLOR4_SLICE
+    0x00000000, // CB_COLOR4_VIEW
+    0x00000000, // CB_COLOR4_INFO
+    0x00000000, // CB_COLOR4_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR4_CMASK
+    0x00000000, // CB_COLOR4_CMASK_SLICE
+    0x00000000, // CB_COLOR4_FMASK
+    0x00000000, // CB_COLOR4_FMASK_SLICE
+    0x00000000, // CB_COLOR4_CLEAR_WORD0
+    0x00000000, // CB_COLOR4_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR5_BASE
+    0x00000000, // CB_COLOR5_PITCH
+    0x00000000, // CB_COLOR5_SLICE
+    0x00000000, // CB_COLOR5_VIEW
+    0x00000000, // CB_COLOR5_INFO
+    0x00000000, // CB_COLOR5_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR5_CMASK
+    0x00000000, // CB_COLOR5_CMASK_SLICE
+    0x00000000, // CB_COLOR5_FMASK
+    0x00000000, // CB_COLOR5_FMASK_SLICE
+    0x00000000, // CB_COLOR5_CLEAR_WORD0
+    0x00000000, // CB_COLOR5_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR6_BASE
+    0x00000000, // CB_COLOR6_PITCH
+    0x00000000, // CB_COLOR6_SLICE
+    0x00000000, // CB_COLOR6_VIEW
+    0x00000000, // CB_COLOR6_INFO
+    0x00000000, // CB_COLOR6_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR6_CMASK
+    0x00000000, // CB_COLOR6_CMASK_SLICE
+    0x00000000, // CB_COLOR6_FMASK
+    0x00000000, // CB_COLOR6_FMASK_SLICE
+    0x00000000, // CB_COLOR6_CLEAR_WORD0
+    0x00000000, // CB_COLOR6_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR7_BASE
+    0x00000000, // CB_COLOR7_PITCH
+    0x00000000, // CB_COLOR7_SLICE
+    0x00000000, // CB_COLOR7_VIEW
+    0x00000000, // CB_COLOR7_INFO
+    0x00000000, // CB_COLOR7_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR7_CMASK
+    0x00000000, // CB_COLOR7_CMASK_SLICE
+    0x00000000, // CB_COLOR7_FMASK
+    0x00000000, // CB_COLOR7_FMASK_SLICE
+    0x00000000, // CB_COLOR7_CLEAR_WORD0
+    0x00000000, // CB_COLOR7_CLEAR_WORD1
+};
+static const struct cs_extent_def si_SECT_CONTEXT_defs[] =
+{
+    {si_SECT_CONTEXT_def_1, 0x0000a000, 212 },
+    {si_SECT_CONTEXT_def_2, 0x0000a0d8, 272 },
+    {si_SECT_CONTEXT_def_3, 0x0000a1f5, 6 },
+    {si_SECT_CONTEXT_def_4, 0x0000a200, 157 },
+    {si_SECT_CONTEXT_def_5, 0x0000a2a1, 1 },
+    {si_SECT_CONTEXT_def_6, 0x0000a2a3, 1 },
+    {si_SECT_CONTEXT_def_7, 0x0000a2a5, 233 },
+    { NULL, 0, 0 }
+};
+static const struct cs_section_def si_cs_data[] = {
+    { si_SECT_CONTEXT_defs, SECT_CONTEXT },
+    { NULL, SECT_NONE }
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/clearstate_vi.h b/sys/dev/pci/drm/amd/amdgpu/clearstate_vi.h
new file mode 100644
index 00000000000..1aab9bef934
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/clearstate_vi.h
@@ -0,0 +1,944 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+static const unsigned int vi_SECT_CONTEXT_def_1[] =
+{
+    0x00000000, // DB_RENDER_CONTROL
+    0x00000000, // DB_COUNT_CONTROL
+    0x00000000, // DB_DEPTH_VIEW
+    0x00000000, // DB_RENDER_OVERRIDE
+    0x00000000, // DB_RENDER_OVERRIDE2
+    0x00000000, // DB_HTILE_DATA_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // DB_DEPTH_BOUNDS_MIN
+    0x00000000, // DB_DEPTH_BOUNDS_MAX
+    0x00000000, // DB_STENCIL_CLEAR
+    0x00000000, // DB_DEPTH_CLEAR
+    0x00000000, // PA_SC_SCREEN_SCISSOR_TL
+    0x40004000, // PA_SC_SCREEN_SCISSOR_BR
+    0, // HOLE
+    0x00000000, // DB_DEPTH_INFO
+    0x00000000, // DB_Z_INFO
+    0x00000000, // DB_STENCIL_INFO
+    0x00000000, // DB_Z_READ_BASE
+    0x00000000, // DB_STENCIL_READ_BASE
+    0x00000000, // DB_Z_WRITE_BASE
+    0x00000000, // DB_STENCIL_WRITE_BASE
+    0x00000000, // DB_DEPTH_SIZE
+    0x00000000, // DB_DEPTH_SLICE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // TA_BC_BASE_ADDR
+    0x00000000, // TA_BC_BASE_ADDR_HI
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // COHER_DEST_BASE_HI_0
+    0x00000000, // COHER_DEST_BASE_HI_1
+    0x00000000, // COHER_DEST_BASE_HI_2
+    0x00000000, // COHER_DEST_BASE_HI_3
+    0x00000000, // COHER_DEST_BASE_2
+    0x00000000, // COHER_DEST_BASE_3
+    0x00000000, // PA_SC_WINDOW_OFFSET
+    0x80000000, // PA_SC_WINDOW_SCISSOR_TL
+    0x40004000, // PA_SC_WINDOW_SCISSOR_BR
+    0x0000ffff, // PA_SC_CLIPRECT_RULE
+    0x00000000, // PA_SC_CLIPRECT_0_TL
+    0x40004000, // PA_SC_CLIPRECT_0_BR
+    0x00000000, // PA_SC_CLIPRECT_1_TL
+    0x40004000, // PA_SC_CLIPRECT_1_BR
+    0x00000000, // PA_SC_CLIPRECT_2_TL
+    0x40004000, // PA_SC_CLIPRECT_2_BR
+    0x00000000, // PA_SC_CLIPRECT_3_TL
+    0x40004000, // PA_SC_CLIPRECT_3_BR
+    0xaa99aaaa, // PA_SC_EDGERULE
+    0x00000000, // PA_SU_HARDWARE_SCREEN_OFFSET
+    0xffffffff, // CB_TARGET_MASK
+    0xffffffff, // CB_SHADER_MASK
+    0x80000000, // PA_SC_GENERIC_SCISSOR_TL
+    0x40004000, // PA_SC_GENERIC_SCISSOR_BR
+    0x00000000, // COHER_DEST_BASE_0
+    0x00000000, // COHER_DEST_BASE_1
+    0x80000000, // PA_SC_VPORT_SCISSOR_0_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_0_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_1_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_1_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_2_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_2_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_3_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_3_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_4_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_4_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_5_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_5_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_6_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_6_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_7_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_7_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_8_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_8_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_9_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_9_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_10_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_10_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_11_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_11_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_12_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_12_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_13_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_13_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_14_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_14_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_15_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_15_BR
+    0x00000000, // PA_SC_VPORT_ZMIN_0
+    0x3f800000, // PA_SC_VPORT_ZMAX_0
+    0x00000000, // PA_SC_VPORT_ZMIN_1
+    0x3f800000, // PA_SC_VPORT_ZMAX_1
+    0x00000000, // PA_SC_VPORT_ZMIN_2
+    0x3f800000, // PA_SC_VPORT_ZMAX_2
+    0x00000000, // PA_SC_VPORT_ZMIN_3
+    0x3f800000, // PA_SC_VPORT_ZMAX_3
+    0x00000000, // PA_SC_VPORT_ZMIN_4
+    0x3f800000, // PA_SC_VPORT_ZMAX_4
+    0x00000000, // PA_SC_VPORT_ZMIN_5
+    0x3f800000, // PA_SC_VPORT_ZMAX_5
+    0x00000000, // PA_SC_VPORT_ZMIN_6
+    0x3f800000, // PA_SC_VPORT_ZMAX_6
+    0x00000000, // PA_SC_VPORT_ZMIN_7
+    0x3f800000, // PA_SC_VPORT_ZMAX_7
+    0x00000000, // PA_SC_VPORT_ZMIN_8
+    0x3f800000, // PA_SC_VPORT_ZMAX_8
+    0x00000000, // PA_SC_VPORT_ZMIN_9
+    0x3f800000, // PA_SC_VPORT_ZMAX_9
+    0x00000000, // PA_SC_VPORT_ZMIN_10
+    0x3f800000, // PA_SC_VPORT_ZMAX_10
+    0x00000000, // PA_SC_VPORT_ZMIN_11
+    0x3f800000, // PA_SC_VPORT_ZMAX_11
+    0x00000000, // PA_SC_VPORT_ZMIN_12
+    0x3f800000, // PA_SC_VPORT_ZMAX_12
+    0x00000000, // PA_SC_VPORT_ZMIN_13
+    0x3f800000, // PA_SC_VPORT_ZMAX_13
+    0x00000000, // PA_SC_VPORT_ZMIN_14
+    0x3f800000, // PA_SC_VPORT_ZMAX_14
+    0x00000000, // PA_SC_VPORT_ZMIN_15
+    0x3f800000, // PA_SC_VPORT_ZMAX_15
+};
+static const unsigned int vi_SECT_CONTEXT_def_2[] =
+{
+    0x00000000, // PA_SC_SCREEN_EXTENT_CONTROL
+    0, // HOLE
+    0x00000000, // CP_PERFMON_CNTX_CNTL
+    0x00000000, // CP_RINGID
+    0x00000000, // CP_VMID
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0xffffffff, // VGT_MAX_VTX_INDX
+    0x00000000, // VGT_MIN_VTX_INDX
+    0x00000000, // VGT_INDX_OFFSET
+    0x00000000, // VGT_MULTI_PRIM_IB_RESET_INDX
+    0, // HOLE
+    0x00000000, // CB_BLEND_RED
+    0x00000000, // CB_BLEND_GREEN
+    0x00000000, // CB_BLEND_BLUE
+    0x00000000, // CB_BLEND_ALPHA
+    0x00000000, // CB_DCC_CONTROL
+    0, // HOLE
+    0x00000000, // DB_STENCIL_CONTROL
+    0x00000000, // DB_STENCILREFMASK
+    0x00000000, // DB_STENCILREFMASK_BF
+    0, // HOLE
+    0x00000000, // PA_CL_VPORT_XSCALE
+    0x00000000, // PA_CL_VPORT_XOFFSET
+    0x00000000, // PA_CL_VPORT_YSCALE
+    0x00000000, // PA_CL_VPORT_YOFFSET
+    0x00000000, // PA_CL_VPORT_ZSCALE
+    0x00000000, // PA_CL_VPORT_ZOFFSET
+    0x00000000, // PA_CL_VPORT_XSCALE_1
+    0x00000000, // PA_CL_VPORT_XOFFSET_1
+    0x00000000, // PA_CL_VPORT_YSCALE_1
+    0x00000000, // PA_CL_VPORT_YOFFSET_1
+    0x00000000, // PA_CL_VPORT_ZSCALE_1
+    0x00000000, // PA_CL_VPORT_ZOFFSET_1
+    0x00000000, // PA_CL_VPORT_XSCALE_2
+    0x00000000, // PA_CL_VPORT_XOFFSET_2
+    0x00000000, // PA_CL_VPORT_YSCALE_2
+    0x00000000, // PA_CL_VPORT_YOFFSET_2
+    0x00000000, // PA_CL_VPORT_ZSCALE_2
+    0x00000000, // PA_CL_VPORT_ZOFFSET_2
+    0x00000000, // PA_CL_VPORT_XSCALE_3
+    0x00000000, // PA_CL_VPORT_XOFFSET_3
+    0x00000000, // PA_CL_VPORT_YSCALE_3
+    0x00000000, // PA_CL_VPORT_YOFFSET_3
+    0x00000000, // PA_CL_VPORT_ZSCALE_3
+    0x00000000, // PA_CL_VPORT_ZOFFSET_3
+    0x00000000, // PA_CL_VPORT_XSCALE_4
+    0x00000000, // PA_CL_VPORT_XOFFSET_4
+    0x00000000, // PA_CL_VPORT_YSCALE_4
+    0x00000000, // PA_CL_VPORT_YOFFSET_4
+    0x00000000, // PA_CL_VPORT_ZSCALE_4
+    0x00000000, // PA_CL_VPORT_ZOFFSET_4
+    0x00000000, // PA_CL_VPORT_XSCALE_5
+    0x00000000, // PA_CL_VPORT_XOFFSET_5
+    0x00000000, // PA_CL_VPORT_YSCALE_5
+    0x00000000, // PA_CL_VPORT_YOFFSET_5
+    0x00000000, // PA_CL_VPORT_ZSCALE_5
+    0x00000000, // PA_CL_VPORT_ZOFFSET_5
+    0x00000000, // PA_CL_VPORT_XSCALE_6
+    0x00000000, // PA_CL_VPORT_XOFFSET_6
+    0x00000000, // PA_CL_VPORT_YSCALE_6
+    0x00000000, // PA_CL_VPORT_YOFFSET_6
+    0x00000000, // PA_CL_VPORT_ZSCALE_6
+    0x00000000, // PA_CL_VPORT_ZOFFSET_6
+    0x00000000, // PA_CL_VPORT_XSCALE_7
+    0x00000000, // PA_CL_VPORT_XOFFSET_7
+    0x00000000, // PA_CL_VPORT_YSCALE_7
+    0x00000000, // PA_CL_VPORT_YOFFSET_7
+    0x00000000, // PA_CL_VPORT_ZSCALE_7
+    0x00000000, // PA_CL_VPORT_ZOFFSET_7
+    0x00000000, // PA_CL_VPORT_XSCALE_8
+    0x00000000, // PA_CL_VPORT_XOFFSET_8
+    0x00000000, // PA_CL_VPORT_YSCALE_8
+    0x00000000, // PA_CL_VPORT_YOFFSET_8
+    0x00000000, // PA_CL_VPORT_ZSCALE_8
+    0x00000000, // PA_CL_VPORT_ZOFFSET_8
+    0x00000000, // PA_CL_VPORT_XSCALE_9
+    0x00000000, // PA_CL_VPORT_XOFFSET_9
+    0x00000000, // PA_CL_VPORT_YSCALE_9
+    0x00000000, // PA_CL_VPORT_YOFFSET_9
+    0x00000000, // PA_CL_VPORT_ZSCALE_9
+    0x00000000, // PA_CL_VPORT_ZOFFSET_9
+    0x00000000, // PA_CL_VPORT_XSCALE_10
+    0x00000000, // PA_CL_VPORT_XOFFSET_10
+    0x00000000, // PA_CL_VPORT_YSCALE_10
+    0x00000000, // PA_CL_VPORT_YOFFSET_10
+    0x00000000, // PA_CL_VPORT_ZSCALE_10
+    0x00000000, // PA_CL_VPORT_ZOFFSET_10
+    0x00000000, // PA_CL_VPORT_XSCALE_11
+    0x00000000, // PA_CL_VPORT_XOFFSET_11
+    0x00000000, // PA_CL_VPORT_YSCALE_11
+    0x00000000, // PA_CL_VPORT_YOFFSET_11
+    0x00000000, // PA_CL_VPORT_ZSCALE_11
+    0x00000000, // PA_CL_VPORT_ZOFFSET_11
+    0x00000000, // PA_CL_VPORT_XSCALE_12
+    0x00000000, // PA_CL_VPORT_XOFFSET_12
+    0x00000000, // PA_CL_VPORT_YSCALE_12
+    0x00000000, // PA_CL_VPORT_YOFFSET_12
+    0x00000000, // PA_CL_VPORT_ZSCALE_12
+    0x00000000, // PA_CL_VPORT_ZOFFSET_12
+    0x00000000, // PA_CL_VPORT_XSCALE_13
+    0x00000000, // PA_CL_VPORT_XOFFSET_13
+    0x00000000, // PA_CL_VPORT_YSCALE_13
+    0x00000000, // PA_CL_VPORT_YOFFSET_13
+    0x00000000, // PA_CL_VPORT_ZSCALE_13
+    0x00000000, // PA_CL_VPORT_ZOFFSET_13
+    0x00000000, // PA_CL_VPORT_XSCALE_14
+    0x00000000, // PA_CL_VPORT_XOFFSET_14
+    0x00000000, // PA_CL_VPORT_YSCALE_14
+    0x00000000, // PA_CL_VPORT_YOFFSET_14
+    0x00000000, // PA_CL_VPORT_ZSCALE_14
+    0x00000000, // PA_CL_VPORT_ZOFFSET_14
+    0x00000000, // PA_CL_VPORT_XSCALE_15
+    0x00000000, // PA_CL_VPORT_XOFFSET_15
+    0x00000000, // PA_CL_VPORT_YSCALE_15
+    0x00000000, // PA_CL_VPORT_YOFFSET_15
+    0x00000000, // PA_CL_VPORT_ZSCALE_15
+    0x00000000, // PA_CL_VPORT_ZOFFSET_15
+    0x00000000, // PA_CL_UCP_0_X
+    0x00000000, // PA_CL_UCP_0_Y
+    0x00000000, // PA_CL_UCP_0_Z
+    0x00000000, // PA_CL_UCP_0_W
+    0x00000000, // PA_CL_UCP_1_X
+    0x00000000, // PA_CL_UCP_1_Y
+    0x00000000, // PA_CL_UCP_1_Z
+    0x00000000, // PA_CL_UCP_1_W
+    0x00000000, // PA_CL_UCP_2_X
+    0x00000000, // PA_CL_UCP_2_Y
+    0x00000000, // PA_CL_UCP_2_Z
+    0x00000000, // PA_CL_UCP_2_W
+    0x00000000, // PA_CL_UCP_3_X
+    0x00000000, // PA_CL_UCP_3_Y
+    0x00000000, // PA_CL_UCP_3_Z
+    0x00000000, // PA_CL_UCP_3_W
+    0x00000000, // PA_CL_UCP_4_X
+    0x00000000, // PA_CL_UCP_4_Y
+    0x00000000, // PA_CL_UCP_4_Z
+    0x00000000, // PA_CL_UCP_4_W
+    0x00000000, // PA_CL_UCP_5_X
+    0x00000000, // PA_CL_UCP_5_Y
+    0x00000000, // PA_CL_UCP_5_Z
+    0x00000000, // PA_CL_UCP_5_W
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_CNTL_0
+    0x00000000, // SPI_PS_INPUT_CNTL_1
+    0x00000000, // SPI_PS_INPUT_CNTL_2
+    0x00000000, // SPI_PS_INPUT_CNTL_3
+    0x00000000, // SPI_PS_INPUT_CNTL_4
+    0x00000000, // SPI_PS_INPUT_CNTL_5
+    0x00000000, // SPI_PS_INPUT_CNTL_6
+    0x00000000, // SPI_PS_INPUT_CNTL_7
+    0x00000000, // SPI_PS_INPUT_CNTL_8
+    0x00000000, // SPI_PS_INPUT_CNTL_9
+    0x00000000, // SPI_PS_INPUT_CNTL_10
+    0x00000000, // SPI_PS_INPUT_CNTL_11
+    0x00000000, // SPI_PS_INPUT_CNTL_12
+    0x00000000, // SPI_PS_INPUT_CNTL_13
+    0x00000000, // SPI_PS_INPUT_CNTL_14
+    0x00000000, // SPI_PS_INPUT_CNTL_15
+    0x00000000, // SPI_PS_INPUT_CNTL_16
+    0x00000000, // SPI_PS_INPUT_CNTL_17
+    0x00000000, // SPI_PS_INPUT_CNTL_18
+    0x00000000, // SPI_PS_INPUT_CNTL_19
+    0x00000000, // SPI_PS_INPUT_CNTL_20
+    0x00000000, // SPI_PS_INPUT_CNTL_21
+    0x00000000, // SPI_PS_INPUT_CNTL_22
+    0x00000000, // SPI_PS_INPUT_CNTL_23
+    0x00000000, // SPI_PS_INPUT_CNTL_24
+    0x00000000, // SPI_PS_INPUT_CNTL_25
+    0x00000000, // SPI_PS_INPUT_CNTL_26
+    0x00000000, // SPI_PS_INPUT_CNTL_27
+    0x00000000, // SPI_PS_INPUT_CNTL_28
+    0x00000000, // SPI_PS_INPUT_CNTL_29
+    0x00000000, // SPI_PS_INPUT_CNTL_30
+    0x00000000, // SPI_PS_INPUT_CNTL_31
+    0x00000000, // SPI_VS_OUT_CONFIG
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_ENA
+    0x00000000, // SPI_PS_INPUT_ADDR
+    0x00000000, // SPI_INTERP_CONTROL_0
+    0x00000002, // SPI_PS_IN_CONTROL
+    0, // HOLE
+    0x00000000, // SPI_BARYC_CNTL
+    0, // HOLE
+    0x00000000, // SPI_TMPRING_SIZE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_SHADER_POS_FORMAT
+    0x00000000, // SPI_SHADER_Z_FORMAT
+    0x00000000, // SPI_SHADER_COL_FORMAT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_BLEND0_CONTROL
+    0x00000000, // CB_BLEND1_CONTROL
+    0x00000000, // CB_BLEND2_CONTROL
+    0x00000000, // CB_BLEND3_CONTROL
+    0x00000000, // CB_BLEND4_CONTROL
+    0x00000000, // CB_BLEND5_CONTROL
+    0x00000000, // CB_BLEND6_CONTROL
+    0x00000000, // CB_BLEND7_CONTROL
+};
+static const unsigned int vi_SECT_CONTEXT_def_3[] =
+{
+    0x00000000, // PA_CL_POINT_X_RAD
+    0x00000000, // PA_CL_POINT_Y_RAD
+    0x00000000, // PA_CL_POINT_SIZE
+    0x00000000, // PA_CL_POINT_CULL_RAD
+    0x00000000, // VGT_DMA_BASE_HI
+    0x00000000, // VGT_DMA_BASE
+};
+static const unsigned int vi_SECT_CONTEXT_def_4[] =
+{
+    0x00000000, // DB_DEPTH_CONTROL
+    0x00000000, // DB_EQAA
+    0x00000000, // CB_COLOR_CONTROL
+    0x00000000, // DB_SHADER_CONTROL
+    0x00090000, // PA_CL_CLIP_CNTL
+    0x00000004, // PA_SU_SC_MODE_CNTL
+    0x00000000, // PA_CL_VTE_CNTL
+    0x00000000, // PA_CL_VS_OUT_CNTL
+    0x00000000, // PA_CL_NANINF_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_SCALE
+    0x00000000, // PA_SU_PRIM_FILTER_CNTL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SU_POINT_SIZE
+    0x00000000, // PA_SU_POINT_MINMAX
+    0x00000000, // PA_SU_LINE_CNTL
+    0x00000000, // PA_SC_LINE_STIPPLE
+    0x00000000, // VGT_OUTPUT_PATH_CNTL
+    0x00000000, // VGT_HOS_CNTL
+    0x00000000, // VGT_HOS_MAX_TESS_LEVEL
+    0x00000000, // VGT_HOS_MIN_TESS_LEVEL
+    0x00000000, // VGT_HOS_REUSE_DEPTH
+    0x00000000, // VGT_GROUP_PRIM_TYPE
+    0x00000000, // VGT_GROUP_FIRST_DECR
+    0x00000000, // VGT_GROUP_DECR
+    0x00000000, // VGT_GROUP_VECT_0_CNTL
+    0x00000000, // VGT_GROUP_VECT_1_CNTL
+    0x00000000, // VGT_GROUP_VECT_0_FMT_CNTL
+    0x00000000, // VGT_GROUP_VECT_1_FMT_CNTL
+    0x00000000, // VGT_GS_MODE
+    0x00000000, // VGT_GS_ONCHIP_CNTL
+    0x00000000, // PA_SC_MODE_CNTL_0
+    0x00000000, // PA_SC_MODE_CNTL_1
+    0x00000000, // VGT_ENHANCE
+    0x00000100, // VGT_GS_PER_ES
+    0x00000080, // VGT_ES_PER_GS
+    0x00000002, // VGT_GS_PER_VS
+    0x00000000, // VGT_GSVS_RING_OFFSET_1
+    0x00000000, // VGT_GSVS_RING_OFFSET_2
+    0x00000000, // VGT_GSVS_RING_OFFSET_3
+    0x00000000, // VGT_GS_OUT_PRIM_TYPE
+    0x00000000, // IA_ENHANCE
+};
+static const unsigned int vi_SECT_CONTEXT_def_5[] =
+{
+    0x00000000, // WD_ENHANCE
+    0x00000000, // VGT_PRIMITIVEID_EN
+};
+static const unsigned int vi_SECT_CONTEXT_def_6[] =
+{
+    0x00000000, // VGT_PRIMITIVEID_RESET
+};
+static const unsigned int vi_SECT_CONTEXT_def_7[] =
+{
+    0x00000000, // VGT_MULTI_PRIM_IB_RESET_EN
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_INSTANCE_STEP_RATE_0
+    0x00000000, // VGT_INSTANCE_STEP_RATE_1
+    0x000000ff, // IA_MULTI_VGT_PARAM
+    0x00000000, // VGT_ESGS_RING_ITEMSIZE
+    0x00000000, // VGT_GSVS_RING_ITEMSIZE
+    0x00000000, // VGT_REUSE_OFF
+    0x00000000, // VGT_VTX_CNT_EN
+    0x00000000, // DB_HTILE_SURFACE
+    0x00000000, // DB_SRESULTS_COMPARE_STATE0
+    0x00000000, // DB_SRESULTS_COMPARE_STATE1
+    0x00000000, // DB_PRELOAD_CONTROL
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_0
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_0
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_0
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_1
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_1
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_1
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_2
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_2
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_2
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_3
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_3
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_3
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_OFFSET
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_BUFFER_FILLED_SIZE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_VERTEX_STRIDE
+    0, // HOLE
+    0x00000000, // VGT_GS_MAX_VERT_OUT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_TESS_DISTRIBUTION
+    0x00000000, // VGT_SHADER_STAGES_EN
+    0x00000000, // VGT_LS_HS_CONFIG
+    0x00000000, // VGT_GS_VERT_ITEMSIZE
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_1
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_2
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_3
+    0x00000000, // VGT_TF_PARAM
+    0x00000000, // DB_ALPHA_TO_MASK
+    0x00000000, // VGT_DISPATCH_DRAW_INDEX
+    0x00000000, // PA_SU_POLY_OFFSET_DB_FMT_CNTL
+    0x00000000, // PA_SU_POLY_OFFSET_CLAMP
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_OFFSET
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_OFFSET
+    0x00000000, // VGT_GS_INSTANCE_CNT
+    0x00000000, // VGT_STRMOUT_CONFIG
+    0x00000000, // VGT_STRMOUT_BUFFER_CONFIG
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SC_CENTROID_PRIORITY_0
+    0x00000000, // PA_SC_CENTROID_PRIORITY_1
+    0x00001000, // PA_SC_LINE_CNTL
+    0x00000000, // PA_SC_AA_CONFIG
+    0x00000005, // PA_SU_VTX_CNTL
+    0x3f800000, // PA_CL_GB_VERT_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_VERT_DISC_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_DISC_ADJ
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_3
+    0xffffffff, // PA_SC_AA_MASK_X0Y0_X1Y0
+    0xffffffff, // PA_SC_AA_MASK_X0Y1_X1Y1
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x0000001e, // VGT_VERTEX_REUSE_BLOCK_CNTL
+    0x00000020, // VGT_OUT_DEALLOC_CNTL
+    0x00000000, // CB_COLOR0_BASE
+    0x00000000, // CB_COLOR0_PITCH
+    0x00000000, // CB_COLOR0_SLICE
+    0x00000000, // CB_COLOR0_VIEW
+    0x00000000, // CB_COLOR0_INFO
+    0x00000000, // CB_COLOR0_ATTRIB
+    0x00000000, // CB_COLOR0_DCC_CONTROL
+    0x00000000, // CB_COLOR0_CMASK
+    0x00000000, // CB_COLOR0_CMASK_SLICE
+    0x00000000, // CB_COLOR0_FMASK
+    0x00000000, // CB_COLOR0_FMASK_SLICE
+    0x00000000, // CB_COLOR0_CLEAR_WORD0
+    0x00000000, // CB_COLOR0_CLEAR_WORD1
+    0x00000000, // CB_COLOR0_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR1_BASE
+    0x00000000, // CB_COLOR1_PITCH
+    0x00000000, // CB_COLOR1_SLICE
+    0x00000000, // CB_COLOR1_VIEW
+    0x00000000, // CB_COLOR1_INFO
+    0x00000000, // CB_COLOR1_ATTRIB
+    0x00000000, // CB_COLOR1_DCC_CONTROL
+    0x00000000, // CB_COLOR1_CMASK
+    0x00000000, // CB_COLOR1_CMASK_SLICE
+    0x00000000, // CB_COLOR1_FMASK
+    0x00000000, // CB_COLOR1_FMASK_SLICE
+    0x00000000, // CB_COLOR1_CLEAR_WORD0
+    0x00000000, // CB_COLOR1_CLEAR_WORD1
+    0x00000000, // CB_COLOR1_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR2_BASE
+    0x00000000, // CB_COLOR2_PITCH
+    0x00000000, // CB_COLOR2_SLICE
+    0x00000000, // CB_COLOR2_VIEW
+    0x00000000, // CB_COLOR2_INFO
+    0x00000000, // CB_COLOR2_ATTRIB
+    0x00000000, // CB_COLOR2_DCC_CONTROL
+    0x00000000, // CB_COLOR2_CMASK
+    0x00000000, // CB_COLOR2_CMASK_SLICE
+    0x00000000, // CB_COLOR2_FMASK
+    0x00000000, // CB_COLOR2_FMASK_SLICE
+    0x00000000, // CB_COLOR2_CLEAR_WORD0
+    0x00000000, // CB_COLOR2_CLEAR_WORD1
+    0x00000000, // CB_COLOR2_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR3_BASE
+    0x00000000, // CB_COLOR3_PITCH
+    0x00000000, // CB_COLOR3_SLICE
+    0x00000000, // CB_COLOR3_VIEW
+    0x00000000, // CB_COLOR3_INFO
+    0x00000000, // CB_COLOR3_ATTRIB
+    0x00000000, // CB_COLOR3_DCC_CONTROL
+    0x00000000, // CB_COLOR3_CMASK
+    0x00000000, // CB_COLOR3_CMASK_SLICE
+    0x00000000, // CB_COLOR3_FMASK
+    0x00000000, // CB_COLOR3_FMASK_SLICE
+    0x00000000, // CB_COLOR3_CLEAR_WORD0
+    0x00000000, // CB_COLOR3_CLEAR_WORD1
+    0x00000000, // CB_COLOR3_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR4_BASE
+    0x00000000, // CB_COLOR4_PITCH
+    0x00000000, // CB_COLOR4_SLICE
+    0x00000000, // CB_COLOR4_VIEW
+    0x00000000, // CB_COLOR4_INFO
+    0x00000000, // CB_COLOR4_ATTRIB
+    0x00000000, // CB_COLOR4_DCC_CONTROL
+    0x00000000, // CB_COLOR4_CMASK
+    0x00000000, // CB_COLOR4_CMASK_SLICE
+    0x00000000, // CB_COLOR4_FMASK
+    0x00000000, // CB_COLOR4_FMASK_SLICE
+    0x00000000, // CB_COLOR4_CLEAR_WORD0
+    0x00000000, // CB_COLOR4_CLEAR_WORD1
+    0x00000000, // CB_COLOR4_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR5_BASE
+    0x00000000, // CB_COLOR5_PITCH
+    0x00000000, // CB_COLOR5_SLICE
+    0x00000000, // CB_COLOR5_VIEW
+    0x00000000, // CB_COLOR5_INFO
+    0x00000000, // CB_COLOR5_ATTRIB
+    0x00000000, // CB_COLOR5_DCC_CONTROL
+    0x00000000, // CB_COLOR5_CMASK
+    0x00000000, // CB_COLOR5_CMASK_SLICE
+    0x00000000, // CB_COLOR5_FMASK
+    0x00000000, // CB_COLOR5_FMASK_SLICE
+    0x00000000, // CB_COLOR5_CLEAR_WORD0
+    0x00000000, // CB_COLOR5_CLEAR_WORD1
+    0x00000000, // CB_COLOR5_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR6_BASE
+    0x00000000, // CB_COLOR6_PITCH
+    0x00000000, // CB_COLOR6_SLICE
+    0x00000000, // CB_COLOR6_VIEW
+    0x00000000, // CB_COLOR6_INFO
+    0x00000000, // CB_COLOR6_ATTRIB
+    0x00000000, // CB_COLOR6_DCC_CONTROL
+    0x00000000, // CB_COLOR6_CMASK
+    0x00000000, // CB_COLOR6_CMASK_SLICE
+    0x00000000, // CB_COLOR6_FMASK
+    0x00000000, // CB_COLOR6_FMASK_SLICE
+    0x00000000, // CB_COLOR6_CLEAR_WORD0
+    0x00000000, // CB_COLOR6_CLEAR_WORD1
+    0x00000000, // CB_COLOR6_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR7_BASE
+    0x00000000, // CB_COLOR7_PITCH
+    0x00000000, // CB_COLOR7_SLICE
+    0x00000000, // CB_COLOR7_VIEW
+    0x00000000, // CB_COLOR7_INFO
+    0x00000000, // CB_COLOR7_ATTRIB
+    0x00000000, // CB_COLOR7_DCC_CONTROL
+    0x00000000, // CB_COLOR7_CMASK
+    0x00000000, // CB_COLOR7_CMASK_SLICE
+    0x00000000, // CB_COLOR7_FMASK
+    0x00000000, // CB_COLOR7_FMASK_SLICE
+    0x00000000, // CB_COLOR7_CLEAR_WORD0
+    0x00000000, // CB_COLOR7_CLEAR_WORD1
+};
+static const struct cs_extent_def vi_SECT_CONTEXT_defs[] =
+{
+    {vi_SECT_CONTEXT_def_1, 0x0000a000, 212 },
+    {vi_SECT_CONTEXT_def_2, 0x0000a0d6, 274 },
+    {vi_SECT_CONTEXT_def_3, 0x0000a1f5, 6 },
+    {vi_SECT_CONTEXT_def_4, 0x0000a200, 157 },
+    {vi_SECT_CONTEXT_def_5, 0x0000a2a0, 2 },
+    {vi_SECT_CONTEXT_def_6, 0x0000a2a3, 1 },
+    {vi_SECT_CONTEXT_def_7, 0x0000a2a5, 233 },
+    { 0, 0, 0 }
+};
+static const struct cs_section_def vi_cs_data[] = {
+    { vi_SECT_CONTEXT_defs, SECT_CONTEXT },
+    { 0, SECT_NONE }
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/cz_ih.c b/sys/dev/pci/drm/amd/amdgpu/cz_ih.c
new file mode 100644
index 00000000000..960c29e17da
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/cz_ih.c
@@ -0,0 +1,463 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "vid.h"
+
+#include "oss/oss_3_0_1_d.h"
+#include "oss/oss_3_0_1_sh_mask.h"
+
+#include "bif/bif_5_1_d.h"
+#include "bif/bif_5_1_sh_mask.h"
+
+/*
+ * Interrupts
+ * Starting with r6xx, interrupts are handled via a ring buffer.
+ * Ring buffers are areas of GPU accessible memory that the GPU
+ * writes interrupt vectors into and the host reads vectors out of.
+ * There is a rptr (read pointer) that determines where the
+ * host is currently reading, and a wptr (write pointer)
+ * which determines where the GPU has written.  When the
+ * pointers are equal, the ring is idle.  When the GPU
+ * writes vectors to the ring buffer, it increments the
+ * wptr.  When there is an interrupt, the host then starts
+ * fetching commands and processing them until the pointers are
+ * equal again at which point it updates the rptr.
+ */
+
+static void cz_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * cz_ih_enable_interrupts - Enable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enable the interrupt ring buffer (VI).
+ */
+static void cz_ih_enable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_cntl = RREG32(mmIH_CNTL);
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, ENABLE_INTR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 1);
+	WREG32(mmIH_CNTL, ih_cntl);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	adev->irq.ih.enabled = true;
+}
+
+/**
+ * cz_ih_disable_interrupts - Disable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable the interrupt ring buffer (VI).
+ */
+static void cz_ih_disable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+	u32 ih_cntl = RREG32(mmIH_CNTL);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 0);
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, ENABLE_INTR, 0);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	WREG32(mmIH_CNTL, ih_cntl);
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+	adev->irq.ih.enabled = false;
+	adev->irq.ih.rptr = 0;
+}
+
+/**
+ * cz_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (VI).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int cz_ih_irq_init(struct amdgpu_device *adev)
+{
+	int rb_bufsz;
+	u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
+	u64 wptr_off;
+
+	/* disable irqs */
+	cz_ih_disable_interrupts(adev);
+
+	/* setup interrupt control */
+	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
+	interrupt_cntl = RREG32(mmINTERRUPT_CNTL);
+	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_DUMMY_RD_OVERRIDE, 0);
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_REQ_NONSNOOP_EN, 0);
+	WREG32(mmINTERRUPT_CNTL, interrupt_cntl);
+
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	WREG32(mmIH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
+
+	rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
+	ih_rb_cntl = REG_SET_FIELD(0, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register value is written to memory */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
+
+	/* set the writeback address whether it's enabled or not */
+	wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
+	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
+	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+
+	/* Default settings for IH_CNTL (disabled at first) */
+	ih_cntl = RREG32(mmIH_CNTL);
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, MC_VMID, 0);
+
+	if (adev->irq.msi_enabled)
+		ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, RPTR_REARM, 1);
+	WREG32(mmIH_CNTL, ih_cntl);
+
+	pci_set_master(adev->pdev);
+
+	/* enable interrupts */
+	cz_ih_enable_interrupts(adev);
+
+	return 0;
+}
+
+/**
+ * cz_ih_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw (VI).
+ */
+static void cz_ih_irq_disable(struct amdgpu_device *adev)
+{
+	cz_ih_disable_interrupts(adev);
+
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * cz_ih_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer (VI).  Also check for
+ * ring buffer overflow and deal with it.
+ * Used by cz_irq_process(VI).
+ * Returns the value of the wptr.
+ */
+static u32 cz_ih_get_wptr(struct amdgpu_device *adev)
+{
+	u32 wptr, tmp;
+
+	wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
+
+	if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
+		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+		/* When a ring buffer overflow happen start parsing interrupt
+		 * from the last not overwritten vector (wptr + 16). Hopefully
+		 * this should allow us to catchup.
+		 */
+		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
+		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
+		tmp = RREG32(mmIH_RB_CNTL);
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+		WREG32(mmIH_RB_CNTL, tmp);
+	}
+	return (wptr & adev->irq.ih.ptr_mask);
+}
+
+/**
+ * cz_ih_prescreen_iv - prescreen an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns true if the interrupt vector should be further processed.
+ */
+static bool cz_ih_prescreen_iv(struct amdgpu_device *adev)
+{
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	u16 pasid;
+
+	switch (le32_to_cpu(adev->irq.ih.ring[ring_index]) & 0xff) {
+	case 146:
+	case 147:
+		pasid = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]) >> 16;
+		if (!pasid || amdgpu_vm_pasid_fault_credit(adev, pasid))
+			return true;
+		break;
+	default:
+		/* Not a VM fault */
+		return true;
+	}
+
+	adev->irq.ih.rptr += 16;
+	return false;
+}
+
+/**
+ * cz_ih_decode_iv - decode an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Decodes the interrupt vector at the current rptr
+ * position and also advance the position.
+ */
+static void cz_ih_decode_iv(struct amdgpu_device *adev,
+				 struct amdgpu_iv_entry *entry)
+{
+	/* wptr/rptr are in bytes! */
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	uint32_t dw[4];
+
+	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+
+	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->src_id = dw[0] & 0xff;
+	entry->src_data[0] = dw[1] & 0xfffffff;
+	entry->ring_id = dw[2] & 0xff;
+	entry->vmid = (dw[2] >> 8) & 0xff;
+	entry->pasid = (dw[2] >> 16) & 0xffff;
+
+	/* wptr/rptr are in bytes! */
+	adev->irq.ih.rptr += 16;
+}
+
+/**
+ * cz_ih_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void cz_ih_set_rptr(struct amdgpu_device *adev)
+{
+	WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
+}
+
+static int cz_ih_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret;
+
+	ret = amdgpu_irq_add_domain(adev);
+	if (ret)
+		return ret;
+
+	cz_ih_set_interrupt_funcs(adev);
+
+	return 0;
+}
+
+static int cz_ih_sw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int cz_ih_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_fini(adev);
+	amdgpu_ih_ring_fini(adev);
+	amdgpu_irq_remove_domain(adev);
+
+	return 0;
+}
+
+static int cz_ih_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = cz_ih_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int cz_ih_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	cz_ih_irq_disable(adev);
+
+	return 0;
+}
+
+static int cz_ih_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cz_ih_hw_fini(adev);
+}
+
+static int cz_ih_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cz_ih_hw_init(adev);
+}
+
+static bool cz_ih_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
+		return false;
+
+	return true;
+}
+
+static int cz_ih_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS);
+		if (!REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int cz_ih_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET,
+						SOFT_RESET_IH, 1);
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int cz_ih_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	// TODO
+	return 0;
+}
+
+static int cz_ih_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	// TODO
+	return 0;
+}
+
+static const struct amd_ip_funcs cz_ih_ip_funcs = {
+	.name = "cz_ih",
+	.early_init = cz_ih_early_init,
+	.late_init = NULL,
+	.sw_init = cz_ih_sw_init,
+	.sw_fini = cz_ih_sw_fini,
+	.hw_init = cz_ih_hw_init,
+	.hw_fini = cz_ih_hw_fini,
+	.suspend = cz_ih_suspend,
+	.resume = cz_ih_resume,
+	.is_idle = cz_ih_is_idle,
+	.wait_for_idle = cz_ih_wait_for_idle,
+	.soft_reset = cz_ih_soft_reset,
+	.set_clockgating_state = cz_ih_set_clockgating_state,
+	.set_powergating_state = cz_ih_set_powergating_state,
+};
+
+static const struct amdgpu_ih_funcs cz_ih_funcs = {
+	.get_wptr = cz_ih_get_wptr,
+	.prescreen_iv = cz_ih_prescreen_iv,
+	.decode_iv = cz_ih_decode_iv,
+	.set_rptr = cz_ih_set_rptr
+};
+
+static void cz_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih_funcs == NULL)
+		adev->irq.ih_funcs = &cz_ih_funcs;
+}
+
+const struct amdgpu_ip_block_version cz_ih_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_IH,
+	.major = 3,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &cz_ih_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/cz_ih.h b/sys/dev/pci/drm/amd/amdgpu/cz_ih.h
new file mode 100644
index 00000000000..14be7753221
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/cz_ih.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __CZ_IH_H__
+#define __CZ_IH_H__
+
+extern const struct amdgpu_ip_block_version cz_ih_ip_block;
+
+#endif /* __CZ_IH_H__ */
diff --git a/sys/dev/pci/drm/amd/amdgpu/dce_v10_0.c b/sys/dev/pci/drm/amd/amdgpu/dce_v10_0.c
new file mode 100644
index 00000000000..308f9f238bc
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/dce_v10_0.c
@@ -0,0 +1,3611 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_i2c.h"
+#include "vid.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+#include "atombios_crtc.h"
+#include "atombios_encoders.h"
+#include "amdgpu_pll.h"
+#include "amdgpu_connectors.h"
+#include "dce_v10_0.h"
+
+#include "dce/dce_10_0_d.h"
+#include "dce/dce_10_0_sh_mask.h"
+#include "dce/dce_10_0_enum.h"
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+#include "gmc/gmc_8_1_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+static void dce_v10_0_set_display_funcs(struct amdgpu_device *adev);
+static void dce_v10_0_set_irq_funcs(struct amdgpu_device *adev);
+
+static const u32 crtc_offsets[] =
+{
+	CRTC0_REGISTER_OFFSET,
+	CRTC1_REGISTER_OFFSET,
+	CRTC2_REGISTER_OFFSET,
+	CRTC3_REGISTER_OFFSET,
+	CRTC4_REGISTER_OFFSET,
+	CRTC5_REGISTER_OFFSET,
+	CRTC6_REGISTER_OFFSET
+};
+
+static const u32 hpd_offsets[] =
+{
+	HPD0_REGISTER_OFFSET,
+	HPD1_REGISTER_OFFSET,
+	HPD2_REGISTER_OFFSET,
+	HPD3_REGISTER_OFFSET,
+	HPD4_REGISTER_OFFSET,
+	HPD5_REGISTER_OFFSET
+};
+
+static const uint32_t dig_offsets[] = {
+	DIG0_REGISTER_OFFSET,
+	DIG1_REGISTER_OFFSET,
+	DIG2_REGISTER_OFFSET,
+	DIG3_REGISTER_OFFSET,
+	DIG4_REGISTER_OFFSET,
+	DIG5_REGISTER_OFFSET,
+	DIG6_REGISTER_OFFSET
+};
+
+static const struct {
+	uint32_t        reg;
+	uint32_t        vblank;
+	uint32_t        vline;
+	uint32_t        hpd;
+
+} interrupt_status_offsets[] = { {
+	.reg = mmDISP_INTERRUPT_STATUS,
+	.vblank = DISP_INTERRUPT_STATUS__LB_D1_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS__LB_D1_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS__DC_HPD1_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE__DC_HPD2_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE2,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE2__DC_HPD3_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE3,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE3__DC_HPD4_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE4,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE4__DC_HPD5_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE5,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE5__DC_HPD6_INTERRUPT_MASK
+} };
+
+static const u32 golden_settings_tonga_a11[] =
+{
+	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
+	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
+	mmFBC_MISC, 0x1f311fff, 0x12300000,
+	mmHDMI_CONTROL, 0x31000111, 0x00000011,
+};
+
+static const u32 tonga_mgcg_cgcg_init[] =
+{
+	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
+	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
+};
+
+static const u32 golden_settings_fiji_a10[] =
+{
+	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
+	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
+	mmFBC_MISC, 0x1f311fff, 0x12300000,
+	mmHDMI_CONTROL, 0x31000111, 0x00000011,
+};
+
+static const u32 fiji_mgcg_cgcg_init[] =
+{
+	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
+	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
+};
+
+static void dce_v10_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+		amdgpu_device_program_register_sequence(adev,
+							fiji_mgcg_cgcg_init,
+							ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_fiji_a10,
+							ARRAY_SIZE(golden_settings_fiji_a10));
+		break;
+	case CHIP_TONGA:
+		amdgpu_device_program_register_sequence(adev,
+							tonga_mgcg_cgcg_init,
+							ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_tonga_a11,
+							ARRAY_SIZE(golden_settings_tonga_a11));
+		break;
+	default:
+		break;
+	}
+}
+
+static u32 dce_v10_0_audio_endpt_rreg(struct amdgpu_device *adev,
+				     u32 block_offset, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	r = RREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+
+	return r;
+}
+
+static void dce_v10_0_audio_endpt_wreg(struct amdgpu_device *adev,
+				      u32 block_offset, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset, v);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+}
+
+static u32 dce_v10_0_vblank_get_counter(struct amdgpu_device *adev, int crtc)
+{
+	if (crtc >= adev->mode_info.num_crtc)
+		return 0;
+	else
+		return RREG32(mmCRTC_STATUS_FRAME_COUNT + crtc_offsets[crtc]);
+}
+
+static void dce_v10_0_pageflip_interrupt_init(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Enable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_get(adev, &adev->pageflip_irq, i);
+}
+
+static void dce_v10_0_pageflip_interrupt_fini(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Disable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_put(adev, &adev->pageflip_irq, i);
+}
+
+/**
+ * dce_v10_0_page_flip - pageflip callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @crtc_id: crtc to cleanup pageflip on
+ * @crtc_base: new address of the crtc (GPU MC address)
+ *
+ * Triggers the actual pageflip by updating the primary
+ * surface base address.
+ */
+static void dce_v10_0_page_flip(struct amdgpu_device *adev,
+				int crtc_id, u64 crtc_base, bool async)
+{
+	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+	u32 tmp;
+
+	/* flip at hsync for async, default is vsync */
+	tmp = RREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, GRPH_FLIP_CONTROL,
+			    GRPH_SURFACE_UPDATE_H_RETRACE_EN, async ? 1 : 0);
+	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	/* update the primary scanout address */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(crtc_base));
+	/* writing to the low address triggers the update */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(crtc_base));
+	/* post the write */
+	RREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset);
+}
+
+static int dce_v10_0_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
+					u32 *vbl, u32 *position)
+{
+	if ((crtc < 0) || (crtc >= adev->mode_info.num_crtc))
+		return -EINVAL;
+
+	*vbl = RREG32(mmCRTC_V_BLANK_START_END + crtc_offsets[crtc]);
+	*position = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+
+	return 0;
+}
+
+/**
+ * dce_v10_0_hpd_sense - hpd sense callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Checks if a digital monitor is connected (evergreen+).
+ * Returns true if connected, false if not connected.
+ */
+static bool dce_v10_0_hpd_sense(struct amdgpu_device *adev,
+			       enum amdgpu_hpd_id hpd)
+{
+	bool connected = false;
+
+	if (hpd >= adev->mode_info.num_hpd)
+		return connected;
+
+	if (RREG32(mmDC_HPD_INT_STATUS + hpd_offsets[hpd]) &
+	    DC_HPD_INT_STATUS__DC_HPD_SENSE_MASK)
+		connected = true;
+
+	return connected;
+}
+
+/**
+ * dce_v10_0_hpd_set_polarity - hpd set polarity callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Set the polarity of the hpd pin (evergreen+).
+ */
+static void dce_v10_0_hpd_set_polarity(struct amdgpu_device *adev,
+				      enum amdgpu_hpd_id hpd)
+{
+	u32 tmp;
+	bool connected = dce_v10_0_hpd_sense(adev, hpd);
+
+	if (hpd >= adev->mode_info.num_hpd)
+		return;
+
+	tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+	if (connected)
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_POLARITY, 0);
+	else
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_POLARITY, 1);
+	WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+}
+
+/**
+ * dce_v10_0_hpd_init - hpd setup callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup the hpd pins used by the card (evergreen+).
+ * Enable the pin, set the polarity, and enable the hpd interrupts.
+ */
+static void dce_v10_0_hpd_init(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
+			continue;
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP ||
+		    connector->connector_type == DRM_MODE_CONNECTOR_LVDS) {
+			/* don't try to enable hpd on eDP or LVDS avoid breaking the
+			 * aux dp channel on imac and help (but not completely fix)
+			 * https://bugzilla.redhat.com/show_bug.cgi?id=726143
+			 * also avoid interrupt storms during dpms.
+			 */
+			tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+			tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 0);
+			WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+			continue;
+		}
+
+		tmp = RREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_CONTROL, DC_HPD_EN, 1);
+		WREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+
+		tmp = RREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_TOGGLE_FILT_CNTL,
+				    DC_HPD_CONNECT_INT_DELAY,
+				    AMDGPU_HPD_CONNECT_INT_DELAY_IN_MS);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_TOGGLE_FILT_CNTL,
+				    DC_HPD_DISCONNECT_INT_DELAY,
+				    AMDGPU_HPD_DISCONNECT_INT_DELAY_IN_MS);
+		WREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+
+		dce_v10_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
+		amdgpu_irq_get(adev, &adev->hpd_irq,
+			       amdgpu_connector->hpd.hpd);
+	}
+}
+
+/**
+ * dce_v10_0_hpd_fini - hpd tear down callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the hpd pins used by the card (evergreen+).
+ * Disable the hpd interrupts.
+ */
+static void dce_v10_0_hpd_fini(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
+			continue;
+
+		tmp = RREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_CONTROL, DC_HPD_EN, 0);
+		WREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+
+		amdgpu_irq_put(adev, &adev->hpd_irq,
+			       amdgpu_connector->hpd.hpd);
+	}
+}
+
+static u32 dce_v10_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
+{
+	return mmDC_GPIO_HPD_A;
+}
+
+static bool dce_v10_0_is_display_hung(struct amdgpu_device *adev)
+{
+	u32 crtc_hung = 0;
+	u32 crtc_status[6];
+	u32 i, j, tmp;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
+		if (REG_GET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN)) {
+			crtc_status[i] = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+			crtc_hung |= (1 << i);
+		}
+	}
+
+	for (j = 0; j < 10; j++) {
+		for (i = 0; i < adev->mode_info.num_crtc; i++) {
+			if (crtc_hung & (1 << i)) {
+				tmp = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+				if (tmp != crtc_status[i])
+					crtc_hung &= ~(1 << i);
+			}
+		}
+		if (crtc_hung == 0)
+			return false;
+		udelay(100);
+	}
+
+	return true;
+}
+
+static void dce_v10_0_set_vga_render_state(struct amdgpu_device *adev,
+					   bool render)
+{
+	u32 tmp;
+
+	/* Lockout access through VGA aperture*/
+	tmp = RREG32(mmVGA_HDP_CONTROL);
+	if (render)
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 0);
+	else
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
+	WREG32(mmVGA_HDP_CONTROL, tmp);
+
+	/* disable VGA render */
+	tmp = RREG32(mmVGA_RENDER_CONTROL);
+	if (render)
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
+	WREG32(mmVGA_RENDER_CONTROL, tmp);
+}
+
+static int dce_v10_0_get_num_crtc(struct amdgpu_device *adev)
+{
+	int num_crtc = 0;
+
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+	case CHIP_TONGA:
+		num_crtc = 6;
+		break;
+	default:
+		num_crtc = 0;
+	}
+	return num_crtc;
+}
+
+void dce_v10_0_disable_dce(struct amdgpu_device *adev)
+{
+	/*Disable VGA render and enabled crtc, if has DCE engine*/
+	if (amdgpu_atombios_has_dce_engine_info(adev)) {
+		u32 tmp;
+		int crtc_enabled, i;
+
+		dce_v10_0_set_vga_render_state(adev, false);
+
+		/*Disable crtc*/
+		for (i = 0; i < dce_v10_0_get_num_crtc(adev); i++) {
+			crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]),
+									 CRTC_CONTROL, CRTC_MASTER_EN);
+			if (crtc_enabled) {
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+				tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
+				tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0);
+				WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			}
+		}
+	}
+}
+
+static void dce_v10_0_program_fmt(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	int bpc = 0;
+	u32 tmp = 0;
+	enum amdgpu_connector_dither dither = AMDGPU_FMT_DITHER_DISABLE;
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		bpc = amdgpu_connector_get_monitor_bpc(connector);
+		dither = amdgpu_connector->dither;
+	}
+
+	/* LVDS/eDP FMT is set up by atom */
+	if (amdgpu_encoder->devices & ATOM_DEVICE_LCD_SUPPORT)
+		return;
+
+	/* not needed for analog */
+	if ((amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1) ||
+	    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2))
+		return;
+
+	if (bpc == 0)
+		return;
+
+	switch (bpc) {
+	case 6:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
+			/* XXX sort out optimal dither settings */
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 0);
+		} else {
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 0);
+		}
+		break;
+	case 8:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
+			/* XXX sort out optimal dither settings */
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 1);
+		} else {
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 1);
+		}
+		break;
+	case 10:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
+			/* XXX sort out optimal dither settings */
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 2);
+		} else {
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 2);
+		}
+		break;
+	default:
+		/* not needed */
+		break;
+	}
+
+	WREG32(mmFMT_BIT_DEPTH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+
+/* display watermark setup */
+/**
+ * dce_v10_0_line_buffer_adjust - Set up the line buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @mode: the current display mode on the selected display
+ * controller
+ *
+ * Setup up the line buffer allocation for
+ * the selected display controller (CIK).
+ * Returns the line buffer size in pixels.
+ */
+static u32 dce_v10_0_line_buffer_adjust(struct amdgpu_device *adev,
+				       struct amdgpu_crtc *amdgpu_crtc,
+				       struct drm_display_mode *mode)
+{
+	u32 tmp, buffer_alloc, i, mem_cfg;
+	u32 pipe_offset = amdgpu_crtc->crtc_id;
+	/*
+	 * Line Buffer Setup
+	 * There are 6 line buffers, one for each display controllers.
+	 * There are 3 partitions per LB. Select the number of partitions
+	 * to enable based on the display width.  For display widths larger
+	 * than 4096, you need use to use 2 display controllers and combine
+	 * them using the stereo blender.
+	 */
+	if (amdgpu_crtc->base.enabled && mode) {
+		if (mode->crtc_hdisplay < 1920) {
+			mem_cfg = 1;
+			buffer_alloc = 2;
+		} else if (mode->crtc_hdisplay < 2560) {
+			mem_cfg = 2;
+			buffer_alloc = 2;
+		} else if (mode->crtc_hdisplay < 4096) {
+			mem_cfg = 0;
+			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
+		} else {
+			DRM_DEBUG_KMS("Mode too big for LB!\n");
+			mem_cfg = 0;
+			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
+		}
+	} else {
+		mem_cfg = 1;
+		buffer_alloc = 0;
+	}
+
+	tmp = RREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, LB_MEMORY_CTRL, LB_MEMORY_CONFIG, mem_cfg);
+	WREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset);
+	tmp = REG_SET_FIELD(tmp, PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATED, buffer_alloc);
+	WREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset, tmp);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset);
+		if (REG_GET_FIELD(tmp, PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATION_COMPLETED))
+			break;
+		udelay(1);
+	}
+
+	if (amdgpu_crtc->base.enabled && mode) {
+		switch (mem_cfg) {
+		case 0:
+		default:
+			return 4096 * 2;
+		case 1:
+			return 1920 * 2;
+		case 2:
+			return 2560 * 2;
+		}
+	}
+
+	/* controller not enabled, so no lb used */
+	return 0;
+}
+
+/**
+ * cik_get_number_of_dram_channels - get the number of dram channels
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the number of video ram channels (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the number of dram channels
+ */
+static u32 cik_get_number_of_dram_channels(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(mmMC_SHARED_CHMAP);
+
+	switch (REG_GET_FIELD(tmp, MC_SHARED_CHMAP, NOOFCHAN)) {
+	case 0:
+	default:
+		return 1;
+	case 1:
+		return 2;
+	case 2:
+		return 4;
+	case 3:
+		return 8;
+	case 4:
+		return 3;
+	case 5:
+		return 6;
+	case 6:
+		return 10;
+	case 7:
+		return 12;
+	case 8:
+		return 16;
+	}
+}
+
+struct dce10_wm_params {
+	u32 dram_channels; /* number of dram channels */
+	u32 yclk;          /* bandwidth per dram data pin in kHz */
+	u32 sclk;          /* engine clock in kHz */
+	u32 disp_clk;      /* display clock in kHz */
+	u32 src_width;     /* viewport width */
+	u32 active_time;   /* active display time in ns */
+	u32 blank_time;    /* blank time in ns */
+	bool interlaced;    /* mode is interlaced */
+	fixed20_12 vsc;    /* vertical scale ratio */
+	u32 num_heads;     /* number of active crtcs */
+	u32 bytes_per_pixel; /* bytes per pixel display + overlay */
+	u32 lb_size;       /* line buffer allocated to pipe */
+	u32 vtaps;         /* vertical scaler taps */
+};
+
+/**
+ * dce_v10_0_dram_bandwidth - get the dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the raw dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth in MBytes/s
+ */
+static u32 dce_v10_0_dram_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate raw DRAM Bandwidth */
+	fixed20_12 dram_efficiency; /* 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	dram_efficiency.full = dfixed_const(7);
+	dram_efficiency.full = dfixed_div(dram_efficiency, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, dram_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v10_0_dram_bandwidth_for_display - get the dram bandwidth for display
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dram bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth for display in MBytes/s
+ */
+static u32 dce_v10_0_dram_bandwidth_for_display(struct dce10_wm_params *wm)
+{
+	/* Calculate DRAM Bandwidth and the part allocated to display. */
+	fixed20_12 disp_dram_allocation; /* 0.3 to 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	disp_dram_allocation.full = dfixed_const(3); /* XXX worse case value 0.3 */
+	disp_dram_allocation.full = dfixed_div(disp_dram_allocation, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, disp_dram_allocation);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v10_0_data_return_bandwidth - get the data return bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the data return bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the data return bandwidth in MBytes/s
+ */
+static u32 dce_v10_0_data_return_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the display Data return Bandwidth */
+	fixed20_12 return_efficiency; /* 0.8 */
+	fixed20_12 sclk, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	sclk.full = dfixed_const(wm->sclk);
+	sclk.full = dfixed_div(sclk, a);
+	a.full = dfixed_const(10);
+	return_efficiency.full = dfixed_const(8);
+	return_efficiency.full = dfixed_div(return_efficiency, a);
+	a.full = dfixed_const(32);
+	bandwidth.full = dfixed_mul(a, sclk);
+	bandwidth.full = dfixed_mul(bandwidth, return_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v10_0_dmif_request_bandwidth - get the dmif bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dmif bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dmif bandwidth in MBytes/s
+ */
+static u32 dce_v10_0_dmif_request_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the DMIF Request Bandwidth */
+	fixed20_12 disp_clk_request_efficiency; /* 0.8 */
+	fixed20_12 disp_clk, bandwidth;
+	fixed20_12 a, b;
+
+	a.full = dfixed_const(1000);
+	disp_clk.full = dfixed_const(wm->disp_clk);
+	disp_clk.full = dfixed_div(disp_clk, a);
+	a.full = dfixed_const(32);
+	b.full = dfixed_mul(a, disp_clk);
+
+	a.full = dfixed_const(10);
+	disp_clk_request_efficiency.full = dfixed_const(8);
+	disp_clk_request_efficiency.full = dfixed_div(disp_clk_request_efficiency, a);
+
+	bandwidth.full = dfixed_mul(b, disp_clk_request_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v10_0_available_bandwidth - get the min available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the min available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the min available bandwidth in MBytes/s
+ */
+static u32 dce_v10_0_available_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the Available bandwidth. Display can use this temporarily but not in average. */
+	u32 dram_bandwidth = dce_v10_0_dram_bandwidth(wm);
+	u32 data_return_bandwidth = dce_v10_0_data_return_bandwidth(wm);
+	u32 dmif_req_bandwidth = dce_v10_0_dmif_request_bandwidth(wm);
+
+	return min(dram_bandwidth, min(data_return_bandwidth, dmif_req_bandwidth));
+}
+
+/**
+ * dce_v10_0_average_bandwidth - get the average available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the average available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the average available bandwidth in MBytes/s
+ */
+static u32 dce_v10_0_average_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the display mode Average Bandwidth
+	 * DisplayMode should contain the source and destination dimensions,
+	 * timing, etc.
+	 */
+	fixed20_12 bpp;
+	fixed20_12 line_time;
+	fixed20_12 src_width;
+	fixed20_12 bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	line_time.full = dfixed_const(wm->active_time + wm->blank_time);
+	line_time.full = dfixed_div(line_time, a);
+	bpp.full = dfixed_const(wm->bytes_per_pixel);
+	src_width.full = dfixed_const(wm->src_width);
+	bandwidth.full = dfixed_mul(src_width, bpp);
+	bandwidth.full = dfixed_mul(bandwidth, wm->vsc);
+	bandwidth.full = dfixed_div(bandwidth, line_time);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v10_0_latency_watermark - get the latency watermark
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the latency watermark (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the latency watermark in ns
+ */
+static u32 dce_v10_0_latency_watermark(struct dce10_wm_params *wm)
+{
+	/* First calculate the latency in ns */
+	u32 mc_latency = 2000; /* 2000 ns. */
+	u32 available_bandwidth = dce_v10_0_available_bandwidth(wm);
+	u32 worst_chunk_return_time = (512 * 8 * 1000) / available_bandwidth;
+	u32 cursor_line_pair_return_time = (128 * 4 * 1000) / available_bandwidth;
+	u32 dc_latency = 40000000 / wm->disp_clk; /* dc pipe latency */
+	u32 other_heads_data_return_time = ((wm->num_heads + 1) * worst_chunk_return_time) +
+		(wm->num_heads * cursor_line_pair_return_time);
+	u32 latency = mc_latency + other_heads_data_return_time + dc_latency;
+	u32 max_src_lines_per_dst_line, lb_fill_bw, line_fill_time;
+	u32 tmp, dmif_size = 12288;
+	fixed20_12 a, b, c;
+
+	if (wm->num_heads == 0)
+		return 0;
+
+	a.full = dfixed_const(2);
+	b.full = dfixed_const(1);
+	if ((wm->vsc.full > a.full) ||
+	    ((wm->vsc.full > b.full) && (wm->vtaps >= 3)) ||
+	    (wm->vtaps >= 5) ||
+	    ((wm->vsc.full >= a.full) && wm->interlaced))
+		max_src_lines_per_dst_line = 4;
+	else
+		max_src_lines_per_dst_line = 2;
+
+	a.full = dfixed_const(available_bandwidth);
+	b.full = dfixed_const(wm->num_heads);
+	a.full = dfixed_div(a, b);
+	tmp = div_u64((u64) dmif_size * (u64) wm->disp_clk, mc_latency + 512);
+	tmp = min(dfixed_trunc(a), tmp);
+
+	lb_fill_bw = min(tmp, wm->disp_clk * wm->bytes_per_pixel / 1000);
+
+	a.full = dfixed_const(max_src_lines_per_dst_line * wm->src_width * wm->bytes_per_pixel);
+	b.full = dfixed_const(1000);
+	c.full = dfixed_const(lb_fill_bw);
+	b.full = dfixed_div(c, b);
+	a.full = dfixed_div(a, b);
+	line_fill_time = dfixed_trunc(a);
+
+	if (line_fill_time < wm->active_time)
+		return latency;
+	else
+		return latency + (line_fill_time - wm->active_time);
+
+}
+
+/**
+ * dce_v10_0_average_bandwidth_vs_dram_bandwidth_for_display - check
+ * average and available dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v10_0_average_bandwidth_vs_dram_bandwidth_for_display(struct dce10_wm_params *wm)
+{
+	if (dce_v10_0_average_bandwidth(wm) <=
+	    (dce_v10_0_dram_bandwidth_for_display(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v10_0_average_bandwidth_vs_available_bandwidth - check
+ * average and available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * available bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v10_0_average_bandwidth_vs_available_bandwidth(struct dce10_wm_params *wm)
+{
+	if (dce_v10_0_average_bandwidth(wm) <=
+	    (dce_v10_0_available_bandwidth(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v10_0_check_latency_hiding - check latency hiding
+ *
+ * @wm: watermark calculation data
+ *
+ * Check latency hiding (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v10_0_check_latency_hiding(struct dce10_wm_params *wm)
+{
+	u32 lb_partitions = wm->lb_size / wm->src_width;
+	u32 line_time = wm->active_time + wm->blank_time;
+	u32 latency_tolerant_lines;
+	u32 latency_hiding;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1);
+	if (wm->vsc.full > a.full)
+		latency_tolerant_lines = 1;
+	else {
+		if (lb_partitions <= (wm->vtaps + 1))
+			latency_tolerant_lines = 1;
+		else
+			latency_tolerant_lines = 2;
+	}
+
+	latency_hiding = (latency_tolerant_lines * line_time + wm->blank_time);
+
+	if (dce_v10_0_latency_watermark(wm) <= latency_hiding)
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v10_0_program_watermarks - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @lb_size: line buffer size
+ * @num_heads: number of display controllers in use
+ *
+ * Calculate and program the display watermarks for the
+ * selected display controller (CIK).
+ */
+static void dce_v10_0_program_watermarks(struct amdgpu_device *adev,
+					struct amdgpu_crtc *amdgpu_crtc,
+					u32 lb_size, u32 num_heads)
+{
+	struct drm_display_mode *mode = &amdgpu_crtc->base.mode;
+	struct dce10_wm_params wm_low, wm_high;
+	u32 active_time;
+	u32 line_time = 0;
+	u32 latency_watermark_a = 0, latency_watermark_b = 0;
+	u32 tmp, wm_mask, lb_vblank_lead_lines = 0;
+
+	if (amdgpu_crtc->base.enabled && num_heads && mode) {
+		active_time = (u32) div_u64((u64)mode->crtc_hdisplay * 1000000,
+					    (u32)mode->clock);
+		line_time = (u32) div_u64((u64)mode->crtc_htotal * 1000000,
+					  (u32)mode->clock);
+		line_time = min(line_time, (u32)65535);
+
+		/* watermark for high clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_high.yclk =
+				amdgpu_dpm_get_mclk(adev, false) * 10;
+			wm_high.sclk =
+				amdgpu_dpm_get_sclk(adev, false) * 10;
+		} else {
+			wm_high.yclk = adev->pm.current_mclk * 10;
+			wm_high.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_high.disp_clk = mode->clock;
+		wm_high.src_width = mode->crtc_hdisplay;
+		wm_high.active_time = active_time;
+		wm_high.blank_time = line_time - wm_high.active_time;
+		wm_high.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_high.interlaced = true;
+		wm_high.vsc = amdgpu_crtc->vsc;
+		wm_high.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_high.vtaps = 2;
+		wm_high.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_high.lb_size = lb_size;
+		wm_high.dram_channels = cik_get_number_of_dram_channels(adev);
+		wm_high.num_heads = num_heads;
+
+		/* set for high clocks */
+		latency_watermark_a = min(dce_v10_0_latency_watermark(&wm_high), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v10_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_high) ||
+		    !dce_v10_0_average_bandwidth_vs_available_bandwidth(&wm_high) ||
+		    !dce_v10_0_check_latency_hiding(&wm_high) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+		}
+
+		/* watermark for low clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_low.yclk =
+				amdgpu_dpm_get_mclk(adev, true) * 10;
+			wm_low.sclk =
+				amdgpu_dpm_get_sclk(adev, true) * 10;
+		} else {
+			wm_low.yclk = adev->pm.current_mclk * 10;
+			wm_low.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_low.disp_clk = mode->clock;
+		wm_low.src_width = mode->crtc_hdisplay;
+		wm_low.active_time = active_time;
+		wm_low.blank_time = line_time - wm_low.active_time;
+		wm_low.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_low.interlaced = true;
+		wm_low.vsc = amdgpu_crtc->vsc;
+		wm_low.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_low.vtaps = 2;
+		wm_low.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_low.lb_size = lb_size;
+		wm_low.dram_channels = cik_get_number_of_dram_channels(adev);
+		wm_low.num_heads = num_heads;
+
+		/* set for low clocks */
+		latency_watermark_b = min(dce_v10_0_latency_watermark(&wm_low), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v10_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_low) ||
+		    !dce_v10_0_average_bandwidth_vs_available_bandwidth(&wm_low) ||
+		    !dce_v10_0_check_latency_hiding(&wm_low) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+		}
+		lb_vblank_lead_lines = DIV_ROUND_UP(lb_size, mode->crtc_hdisplay);
+	}
+
+	/* select wm A */
+	wm_mask = RREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(wm_mask, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, 1);
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	tmp = RREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, latency_watermark_a);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, line_time);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	/* select wm B */
+	tmp = REG_SET_FIELD(wm_mask, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, 2);
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	tmp = RREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, latency_watermark_b);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, line_time);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	/* restore original selection */
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, wm_mask);
+
+	/* save values for DPM */
+	amdgpu_crtc->line_time = line_time;
+	amdgpu_crtc->wm_high = latency_watermark_a;
+	amdgpu_crtc->wm_low = latency_watermark_b;
+	/* Save number of lines the linebuffer leads before the scanout */
+	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
+}
+
+/**
+ * dce_v10_0_bandwidth_update - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Calculate and program the display watermarks and line
+ * buffer allocation (CIK).
+ */
+static void dce_v10_0_bandwidth_update(struct amdgpu_device *adev)
+{
+	struct drm_display_mode *mode = NULL;
+	u32 num_heads = 0, lb_size;
+	int i;
+
+	amdgpu_display_update_priority(adev);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i]->base.enabled)
+			num_heads++;
+	}
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		mode = &adev->mode_info.crtcs[i]->base.mode;
+		lb_size = dce_v10_0_line_buffer_adjust(adev, adev->mode_info.crtcs[i], mode);
+		dce_v10_0_program_watermarks(adev, adev->mode_info.crtcs[i],
+					    lb_size, num_heads);
+	}
+}
+
+static void dce_v10_0_audio_get_connected_pins(struct amdgpu_device *adev)
+{
+	int i;
+	u32 offset, tmp;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		offset = adev->mode_info.audio.pin[i].offset;
+		tmp = RREG32_AUDIO_ENDPT(offset,
+					 ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT);
+		if (((tmp &
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY_MASK) >>
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY__SHIFT) == 1)
+			adev->mode_info.audio.pin[i].connected = false;
+		else
+			adev->mode_info.audio.pin[i].connected = true;
+	}
+}
+
+static struct amdgpu_audio_pin *dce_v10_0_audio_get_pin(struct amdgpu_device *adev)
+{
+	int i;
+
+	dce_v10_0_audio_get_connected_pins(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		if (adev->mode_info.audio.pin[i].connected)
+			return &adev->mode_info.audio.pin[i];
+	}
+	DRM_ERROR("No connected audio pins found!\n");
+	return NULL;
+}
+
+static void dce_v10_0_afmt_audio_select_pin(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	tmp = RREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_SRC_CONTROL, AFMT_AUDIO_SRC_SELECT, dig->afmt->pin->id);
+	WREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset, tmp);
+}
+
+static void dce_v10_0_audio_write_latency_fields(struct drm_encoder *encoder,
+						struct drm_display_mode *mode)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u32 tmp;
+	int interlace = 0;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		interlace = 1;
+	if (connector->latency_present[interlace]) {
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    VIDEO_LIPSYNC, connector->video_latency[interlace]);
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    AUDIO_LIPSYNC, connector->audio_latency[interlace]);
+	} else {
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    VIDEO_LIPSYNC, 0);
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    AUDIO_LIPSYNC, 0);
+	}
+	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+			   ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC, tmp);
+}
+
+static void dce_v10_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u32 tmp;
+	u8 *sadb = NULL;
+	int sad_count;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector_edid(connector), &sadb);
+	if (sad_count < 0) {
+		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sad_count);
+		sad_count = 0;
+	}
+
+	/* program the speaker allocation */
+	tmp = RREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+				 ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER);
+	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			    DP_CONNECTION, 0);
+	/* set HDMI mode */
+	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			    HDMI_CONNECTION, 1);
+	if (sad_count)
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+				    SPEAKER_ALLOCATION, sadb[0]);
+	else
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+				    SPEAKER_ALLOCATION, 5); /* stereo */
+	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+			   ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER, tmp);
+
+	kfree(sadb);
+}
+
+static void dce_v10_0_audio_write_sad_regs(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	struct cea_sad *sads;
+	int i, sad_count;
+
+	static const u16 eld_reg_to_type[][2] = {
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0, HDMI_AUDIO_CODING_TYPE_PCM },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR1, HDMI_AUDIO_CODING_TYPE_AC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR2, HDMI_AUDIO_CODING_TYPE_MPEG1 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR3, HDMI_AUDIO_CODING_TYPE_MP3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR4, HDMI_AUDIO_CODING_TYPE_MPEG2 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR5, HDMI_AUDIO_CODING_TYPE_AAC_LC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR6, HDMI_AUDIO_CODING_TYPE_DTS },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR7, HDMI_AUDIO_CODING_TYPE_ATRAC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR9, HDMI_AUDIO_CODING_TYPE_EAC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR10, HDMI_AUDIO_CODING_TYPE_DTS_HD },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR11, HDMI_AUDIO_CODING_TYPE_MLP },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO },
+	};
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_sad(amdgpu_connector_edid(connector), &sads);
+	if (sad_count <= 0) {
+		DRM_ERROR("Couldn't read SADs: %d\n", sad_count);
+		return;
+	}
+	BUG_ON(!sads);
+
+	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
+		u32 tmp = 0;
+		u8 stereo_freqs = 0;
+		int max_channels = -1;
+		int j;
+
+		for (j = 0; j < sad_count; j++) {
+			struct cea_sad *sad = &sads[j];
+
+			if (sad->format == eld_reg_to_type[i][1]) {
+				if (sad->channels > max_channels) {
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							    MAX_CHANNELS, sad->channels);
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							    DESCRIPTOR_BYTE_2, sad->byte2);
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							    SUPPORTED_FREQUENCIES, sad->freq);
+					max_channels = sad->channels;
+				}
+
+				if (sad->format == HDMI_AUDIO_CODING_TYPE_PCM)
+					stereo_freqs |= sad->freq;
+				else
+					break;
+			}
+		}
+
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+				    SUPPORTED_FREQUENCIES_STEREO, stereo_freqs);
+		WREG32_AUDIO_ENDPT(dig->afmt->pin->offset, eld_reg_to_type[i][0], tmp);
+	}
+
+	kfree(sads);
+}
+
+static void dce_v10_0_audio_enable(struct amdgpu_device *adev,
+				  struct amdgpu_audio_pin *pin,
+				  bool enable)
+{
+	if (!pin)
+		return;
+
+	WREG32_AUDIO_ENDPT(pin->offset, ixAZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			   enable ? AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL__AUDIO_ENABLED_MASK : 0);
+}
+
+static const u32 pin_offsets[] =
+{
+	AUD0_REGISTER_OFFSET,
+	AUD1_REGISTER_OFFSET,
+	AUD2_REGISTER_OFFSET,
+	AUD3_REGISTER_OFFSET,
+	AUD4_REGISTER_OFFSET,
+	AUD5_REGISTER_OFFSET,
+	AUD6_REGISTER_OFFSET,
+};
+
+static int dce_v10_0_audio_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!amdgpu_audio)
+		return 0;
+
+	adev->mode_info.audio.enabled = true;
+
+	adev->mode_info.audio.num_pins = 7;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		adev->mode_info.audio.pin[i].channels = -1;
+		adev->mode_info.audio.pin[i].rate = -1;
+		adev->mode_info.audio.pin[i].bits_per_sample = -1;
+		adev->mode_info.audio.pin[i].status_bits = 0;
+		adev->mode_info.audio.pin[i].category_code = 0;
+		adev->mode_info.audio.pin[i].connected = false;
+		adev->mode_info.audio.pin[i].offset = pin_offsets[i];
+		adev->mode_info.audio.pin[i].id = i;
+		/* disable audio.  it will be set up later */
+		/* XXX remove once we switch to ip funcs */
+		dce_v10_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	return 0;
+}
+
+static void dce_v10_0_audio_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!amdgpu_audio)
+		return;
+
+	if (!adev->mode_info.audio.enabled)
+		return;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
+		dce_v10_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+
+	adev->mode_info.audio.enabled = false;
+}
+
+/*
+ * update the N and CTS parameters for a given pixel clock rate
+ */
+static void dce_v10_0_afmt_update_ACR(struct drm_encoder *encoder, uint32_t clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_afmt_acr acr = amdgpu_afmt_acr(clock);
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	tmp = RREG32(mmHDMI_ACR_32_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_0, HDMI_ACR_CTS_32, acr.cts_32khz);
+	WREG32(mmHDMI_ACR_32_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_32_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_1, HDMI_ACR_N_32, acr.n_32khz);
+	WREG32(mmHDMI_ACR_32_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_44_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_0, HDMI_ACR_CTS_44, acr.cts_44_1khz);
+	WREG32(mmHDMI_ACR_44_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_44_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_1, HDMI_ACR_N_44, acr.n_44_1khz);
+	WREG32(mmHDMI_ACR_44_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_48_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_0, HDMI_ACR_CTS_48, acr.cts_48khz);
+	WREG32(mmHDMI_ACR_48_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_48_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_1, HDMI_ACR_N_48, acr.n_48khz);
+	WREG32(mmHDMI_ACR_48_1 + dig->afmt->offset, tmp);
+
+}
+
+/*
+ * build a HDMI Video Info Frame
+ */
+static void dce_v10_0_afmt_update_avi_infoframe(struct drm_encoder *encoder,
+					       void *buffer, size_t size)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	uint8_t *frame = buffer + 3;
+	uint8_t *header = buffer;
+
+	WREG32(mmAFMT_AVI_INFO0 + dig->afmt->offset,
+		frame[0x0] | (frame[0x1] << 8) | (frame[0x2] << 16) | (frame[0x3] << 24));
+	WREG32(mmAFMT_AVI_INFO1 + dig->afmt->offset,
+		frame[0x4] | (frame[0x5] << 8) | (frame[0x6] << 16) | (frame[0x7] << 24));
+	WREG32(mmAFMT_AVI_INFO2 + dig->afmt->offset,
+		frame[0x8] | (frame[0x9] << 8) | (frame[0xA] << 16) | (frame[0xB] << 24));
+	WREG32(mmAFMT_AVI_INFO3 + dig->afmt->offset,
+		frame[0xC] | (frame[0xD] << 8) | (header[1] << 24));
+}
+
+static void dce_v10_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	u32 dto_phase = 24 * 1000;
+	u32 dto_modulo = clock;
+	u32 tmp;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* XXX two dtos; generally use dto0 for hdmi */
+	/* Express [24MHz / target pixel clock] as an exact rational
+	 * number (coefficient of two integer numbers.  DCCG_AUDIO_DTOx_PHASE
+	 * is the numerator, DCCG_AUDIO_DTOx_MODULE is the denominator
+	 */
+	tmp = RREG32(mmDCCG_AUDIO_DTO_SOURCE);
+	tmp = REG_SET_FIELD(tmp, DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL,
+			    amdgpu_crtc->crtc_id);
+	WREG32(mmDCCG_AUDIO_DTO_SOURCE, tmp);
+	WREG32(mmDCCG_AUDIO_DTO0_PHASE, dto_phase);
+	WREG32(mmDCCG_AUDIO_DTO0_MODULE, dto_modulo);
+}
+
+/*
+ * update the info frames with the data from the current display mode
+ */
+static void dce_v10_0_afmt_setmode(struct drm_encoder *encoder,
+				  struct drm_display_mode *mode)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	u8 buffer[HDMI_INFOFRAME_HEADER_SIZE + HDMI_AVI_INFOFRAME_SIZE];
+	struct hdmi_avi_infoframe frame;
+	ssize_t err;
+	u32 tmp;
+	int bpc = 8;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (!dig->afmt->enabled)
+		return;
+
+	/* hdmi deep color mode general control packets setup, if bpc > 8 */
+	if (encoder->crtc) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+		bpc = amdgpu_crtc->bpc;
+	}
+
+	/* disable audio prior to setting up hw */
+	dig->afmt->pin = dce_v10_0_audio_get_pin(adev);
+	dce_v10_0_audio_enable(adev, dig->afmt->pin, false);
+
+	dce_v10_0_audio_set_dto(encoder, mode->clock);
+
+	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1);
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp); /* send null packets when required */
+
+	WREG32(mmAFMT_AUDIO_CRC_CONTROL + dig->afmt->offset, 0x1000);
+
+	tmp = RREG32(mmHDMI_CONTROL + dig->afmt->offset);
+	switch (bpc) {
+	case 0:
+	case 6:
+	case 8:
+	case 16:
+	default:
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 0);
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
+		DRM_DEBUG("%s: Disabling hdmi deep color for %d bpc.\n",
+			  connector->name, bpc);
+		break;
+	case 10:
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 1);
+		DRM_DEBUG("%s: Enabling hdmi deep color 30 for 10 bpc.\n",
+			  connector->name);
+		break;
+	case 12:
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 2);
+		DRM_DEBUG("%s: Enabling hdmi deep color 36 for 12 bpc.\n",
+			  connector->name);
+		break;
+	}
+	WREG32(mmHDMI_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1); /* send null packets when required */
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, 1); /* send general control packets */
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, 1); /* send general control packets every frame */
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	/* enable audio info frames (frames won't be set until audio is enabled) */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
+	/* required for audio info values to be updated */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_CONT, 1);
+	WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	/* required for audio info values to be updated */
+	tmp = REG_SET_FIELD(tmp, AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
+	WREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
+	/* anything other than 0 */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, 2);
+	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
+
+	WREG32(mmHDMI_GC + dig->afmt->offset, 0); /* unset HDMI_GC_AVMUTE */
+
+	tmp = RREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	/* set the default audio delay */
+	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, 1);
+	/* should be suffient for all audio modes and small enough for all hblanks */
+	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_PACKETS_PER_LINE, 3);
+	WREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	/* allow 60958 channel status fields to be updated */
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset);
+	if (bpc > 8)
+		/* clear SW CTS value */
+		tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, 0);
+	else
+		/* select SW CTS value */
+		tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, 1);
+	/* allow hw to sent ACR packets when required */
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, 1);
+	WREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	dce_v10_0_afmt_update_ACR(encoder, mode->clock);
+
+	tmp = RREG32(mmAFMT_60958_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_0, AFMT_60958_CS_CHANNEL_NUMBER_L, 1);
+	WREG32(mmAFMT_60958_0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_60958_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, 2);
+	WREG32(mmAFMT_60958_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_60958_2 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_2, 3);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_3, 4);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_4, 5);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_5, 6);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_6, 7);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_7, 8);
+	WREG32(mmAFMT_60958_2 + dig->afmt->offset, tmp);
+
+	dce_v10_0_audio_write_speaker_allocation(encoder);
+
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL2 + dig->afmt->offset,
+	       (0xff << AFMT_AUDIO_PACKET_CONTROL2__AFMT_AUDIO_CHANNEL_ENABLE__SHIFT));
+
+	dce_v10_0_afmt_audio_select_pin(encoder);
+	dce_v10_0_audio_write_sad_regs(encoder);
+	dce_v10_0_audio_write_latency_fields(encoder, mode);
+
+	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, mode, false);
+	if (err < 0) {
+		DRM_ERROR("failed to setup AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	err = hdmi_avi_infoframe_pack(&frame, buffer, sizeof(buffer));
+	if (err < 0) {
+		DRM_ERROR("failed to pack AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	dce_v10_0_afmt_update_avi_infoframe(encoder, buffer, sizeof(buffer));
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	/* enable AVI info frames */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_SEND, 1);
+	/* required for audio info values to be updated */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_CONT, 1);
+	WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AVI_INFO_LINE, 2);
+	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	/* send audio packets */
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, 1);
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	WREG32(mmAFMT_RAMP_CONTROL0 + dig->afmt->offset, 0x00FFFFFF);
+	WREG32(mmAFMT_RAMP_CONTROL1 + dig->afmt->offset, 0x007FFFFF);
+	WREG32(mmAFMT_RAMP_CONTROL2 + dig->afmt->offset, 0x00000001);
+	WREG32(mmAFMT_RAMP_CONTROL3 + dig->afmt->offset, 0x00000001);
+
+	/* enable audio after to setting up hw */
+	dce_v10_0_audio_enable(adev, dig->afmt->pin, true);
+}
+
+static void dce_v10_0_afmt_enable(struct drm_encoder *encoder, bool enable)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (enable && dig->afmt->enabled)
+		return;
+	if (!enable && !dig->afmt->enabled)
+		return;
+
+	if (!enable && dig->afmt->pin) {
+		dce_v10_0_audio_enable(adev, dig->afmt->pin, false);
+		dig->afmt->pin = NULL;
+	}
+
+	dig->afmt->enabled = enable;
+
+	DRM_DEBUG("%sabling AFMT interface @ 0x%04X for encoder 0x%x\n",
+		  enable ? "En" : "Dis", dig->afmt->offset, amdgpu_encoder->encoder_id);
+}
+
+static int dce_v10_0_afmt_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++)
+		adev->mode_info.afmt[i] = NULL;
+
+	/* DCE10 has audio blocks tied to DIG encoders */
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		adev->mode_info.afmt[i] = kzalloc(sizeof(struct amdgpu_afmt), GFP_KERNEL);
+		if (adev->mode_info.afmt[i]) {
+			adev->mode_info.afmt[i]->offset = dig_offsets[i];
+			adev->mode_info.afmt[i]->id = i;
+		} else {
+			int j;
+			for (j = 0; j < i; j++) {
+				kfree(adev->mode_info.afmt[j]);
+				adev->mode_info.afmt[j] = NULL;
+			}
+			return -ENOMEM;
+		}
+	}
+	return 0;
+}
+
+static void dce_v10_0_afmt_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		kfree(adev->mode_info.afmt[i]);
+		adev->mode_info.afmt[i] = NULL;
+	}
+}
+
+static const u32 vga_control_regs[6] =
+{
+	mmD1VGA_CONTROL,
+	mmD2VGA_CONTROL,
+	mmD3VGA_CONTROL,
+	mmD4VGA_CONTROL,
+	mmD5VGA_CONTROL,
+	mmD6VGA_CONTROL,
+};
+
+static void dce_v10_0_vga_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 vga_control;
+
+	vga_control = RREG32(vga_control_regs[amdgpu_crtc->crtc_id]) & ~1;
+	if (enable)
+		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control | 1);
+	else
+		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control);
+}
+
+static void dce_v10_0_grph_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (enable)
+		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 1);
+	else
+		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 0);
+}
+
+static int dce_v10_0_crtc_do_set_base(struct drm_crtc *crtc,
+				     struct drm_framebuffer *fb,
+				     int x, int y, int atomic)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_framebuffer *target_fb;
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *abo;
+	uint64_t fb_location, tiling_flags;
+	uint32_t fb_format, fb_pitch_pixels;
+	u32 fb_swap = REG_SET_FIELD(0, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP, ENDIAN_NONE);
+	u32 pipe_config;
+	u32 tmp, viewport_w, viewport_h;
+	int r;
+	bool bypass_lut = false;
+	struct drm_format_name_buf format_name;
+
+	/* no fb bound */
+	if (!atomic && !crtc->primary->fb) {
+		DRM_DEBUG_KMS("No FB bound\n");
+		return 0;
+	}
+
+	if (atomic)
+		target_fb = fb;
+	else
+		target_fb = crtc->primary->fb;
+
+	/* If atomic, assume fb object is pinned & idle & fenced and
+	 * just update base pointers
+	 */
+	obj = target_fb->obj[0];
+	abo = gem_to_amdgpu_bo(obj);
+	r = amdgpu_bo_reserve(abo, false);
+	if (unlikely(r != 0))
+		return r;
+
+	if (!atomic) {
+		r = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM);
+		if (unlikely(r != 0)) {
+			amdgpu_bo_unreserve(abo);
+			return -EINVAL;
+		}
+	}
+	fb_location = amdgpu_bo_gpu_offset(abo);
+
+	amdgpu_bo_get_tiling_flags(abo, &tiling_flags);
+	amdgpu_bo_unreserve(abo);
+
+	pipe_config = AMDGPU_TILING_GET(tiling_flags, PIPE_CONFIG);
+
+	switch (target_fb->format->format) {
+	case DRM_FORMAT_C8:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 0);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+		break;
+	case DRM_FORMAT_XRGB4444:
+	case DRM_FORMAT_ARGB4444:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 2);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_XRGB1555:
+	case DRM_FORMAT_ARGB1555:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_BGRX5551:
+	case DRM_FORMAT_BGRA5551:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 5);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_RGB565:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 1);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_XRGB8888:
+	case DRM_FORMAT_ARGB8888:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		break;
+	case DRM_FORMAT_XRGB2101010:
+	case DRM_FORMAT_ARGB2101010:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 1);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	case DRM_FORMAT_BGRX1010102:
+	case DRM_FORMAT_BGRA1010102:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 4);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	default:
+		DRM_ERROR("Unsupported screen format %s\n",
+		          drm_get_format_name(target_fb->format->format, &format_name));
+		return -EINVAL;
+	}
+
+	if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_2D_TILED_THIN1) {
+		unsigned bankw, bankh, mtaspect, tile_split, num_banks;
+
+		bankw = AMDGPU_TILING_GET(tiling_flags, BANK_WIDTH);
+		bankh = AMDGPU_TILING_GET(tiling_flags, BANK_HEIGHT);
+		mtaspect = AMDGPU_TILING_GET(tiling_flags, MACRO_TILE_ASPECT);
+		tile_split = AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT);
+		num_banks = AMDGPU_TILING_GET(tiling_flags, NUM_BANKS);
+
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_NUM_BANKS, num_banks);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_ARRAY_MODE,
+					  ARRAY_2D_TILED_THIN1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_TILE_SPLIT,
+					  tile_split);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_BANK_WIDTH, bankw);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_BANK_HEIGHT, bankh);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_MACRO_TILE_ASPECT,
+					  mtaspect);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_MICRO_TILE_MODE,
+					  ADDR_SURF_MICRO_TILING_DISPLAY);
+	} else if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_1D_TILED_THIN1) {
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_ARRAY_MODE,
+					  ARRAY_1D_TILED_THIN1);
+	}
+
+	fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_PIPE_CONFIG,
+				  pipe_config);
+
+	dce_v10_0_vga_enable(crtc, false);
+
+	/* Make sure surface address is updated at vertical blank rather than
+	 * horizontal blank
+	 */
+	tmp = RREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, GRPH_FLIP_CONTROL,
+			    GRPH_SURFACE_UPDATE_H_RETRACE_EN, 0);
+	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32)fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32) fb_location & GRPH_SECONDARY_SURFACE_ADDRESS__GRPH_SECONDARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
+	WREG32(mmGRPH_SWAP_CNTL + amdgpu_crtc->crtc_offset, fb_swap);
+
+	/*
+	 * The LUT only has 256 slots for indexing by a 8 bpc fb. Bypass the LUT
+	 * for > 8 bpc scanout to avoid truncation of fb indices to 8 msb's, to
+	 * retain the full precision throughout the pipeline.
+	 */
+	tmp = RREG32(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset);
+	if (bypass_lut)
+		tmp = REG_SET_FIELD(tmp, GRPH_LUT_10BIT_BYPASS, GRPH_LUT_10BIT_BYPASS_EN, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, GRPH_LUT_10BIT_BYPASS, GRPH_LUT_10BIT_BYPASS_EN, 0);
+	WREG32(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset, tmp);
+
+	if (bypass_lut)
+		DRM_DEBUG_KMS("Bypassing hardware LUT due to 10 bit fb scanout.\n");
+
+	WREG32(mmGRPH_SURFACE_OFFSET_X + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_SURFACE_OFFSET_Y + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_Y_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_END + amdgpu_crtc->crtc_offset, target_fb->width);
+	WREG32(mmGRPH_Y_END + amdgpu_crtc->crtc_offset, target_fb->height);
+
+	fb_pitch_pixels = target_fb->pitches[0] / target_fb->format->cpp[0];
+	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset, fb_pitch_pixels);
+
+	dce_v10_0_grph_enable(crtc, true);
+
+	WREG32(mmLB_DESKTOP_HEIGHT + amdgpu_crtc->crtc_offset,
+	       target_fb->height);
+
+	x &= ~3;
+	y &= ~1;
+	WREG32(mmVIEWPORT_START + amdgpu_crtc->crtc_offset,
+	       (x << 16) | y);
+	viewport_w = crtc->mode.hdisplay;
+	viewport_h = (crtc->mode.vdisplay + 1) & ~1;
+	WREG32(mmVIEWPORT_SIZE + amdgpu_crtc->crtc_offset,
+	       (viewport_w << 16) | viewport_h);
+
+	/* set pageflip to happen anywhere in vblank interval */
+	WREG32(mmMASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 0);
+
+	if (!atomic && fb && fb != crtc->primary->fb) {
+		abo = gem_to_amdgpu_bo(fb->obj[0]);
+		r = amdgpu_bo_reserve(abo, true);
+		if (unlikely(r != 0))
+			return r;
+		amdgpu_bo_unpin(abo);
+		amdgpu_bo_unreserve(abo);
+	}
+
+	/* Bytes per pixel may have changed */
+	dce_v10_0_bandwidth_update(adev);
+
+	return 0;
+}
+
+static void dce_v10_0_set_interleave(struct drm_crtc *crtc,
+				     struct drm_display_mode *mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	u32 tmp;
+
+	tmp = RREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset);
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		tmp = REG_SET_FIELD(tmp, LB_DATA_FORMAT, INTERLEAVE_EN, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, LB_DATA_FORMAT, INTERLEAVE_EN, 0);
+	WREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static void dce_v10_0_crtc_load_lut(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u16 *r, *g, *b;
+	int i;
+	u32 tmp;
+
+	DRM_DEBUG_KMS("%d\n", amdgpu_crtc->crtc_id);
+
+	tmp = RREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, INPUT_CSC_CONTROL, INPUT_CSC_GRPH_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, INPUT_CSC_CONTROL, INPUT_CSC_OVL_MODE, 0);
+	WREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_BYPASS, 1);
+	WREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmPRESCALE_OVL_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, PRESCALE_OVL_CONTROL, OVL_PRESCALE_BYPASS, 1);
+	WREG32(mmPRESCALE_OVL_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, INPUT_GAMMA_CONTROL, GRPH_INPUT_GAMMA_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, INPUT_GAMMA_CONTROL, OVL_INPUT_GAMMA_MODE, 0);
+	WREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	WREG32(mmDC_LUT_CONTROL + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_BLACK_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_RED + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_WHITE_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_RED + amdgpu_crtc->crtc_offset, 0xffff);
+
+	WREG32(mmDC_LUT_RW_MODE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_WRITE_EN_MASK + amdgpu_crtc->crtc_offset, 0x00000007);
+
+	WREG32(mmDC_LUT_RW_INDEX + amdgpu_crtc->crtc_offset, 0);
+	r = crtc->gamma_store;
+	g = r + crtc->gamma_size;
+	b = g + crtc->gamma_size;
+	for (i = 0; i < 256; i++) {
+		WREG32(mmDC_LUT_30_COLOR + amdgpu_crtc->crtc_offset,
+		       ((*r++ & 0xffc0) << 14) |
+		       ((*g++ & 0xffc0) << 4) |
+		       (*b++ >> 6));
+	}
+
+	tmp = RREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, GRPH_DEGAMMA_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, OVL_DEGAMMA_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, CURSOR_DEGAMMA_MODE, 0);
+	WREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, GAMUT_REMAP_CONTROL, GRPH_GAMUT_REMAP_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, GAMUT_REMAP_CONTROL, OVL_GAMUT_REMAP_MODE, 0);
+	WREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, REGAMMA_CONTROL, GRPH_REGAMMA_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, REGAMMA_CONTROL, OVL_REGAMMA_MODE, 0);
+	WREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, OUTPUT_CSC_CONTROL, OUTPUT_CSC_GRPH_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, OUTPUT_CSC_CONTROL, OUTPUT_CSC_OVL_MODE, 0);
+	WREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	/* XXX match this to the depth of the crtc fmt block, move to modeset? */
+	WREG32(mmDENORM_CONTROL + amdgpu_crtc->crtc_offset, 0);
+	/* XXX this only needs to be programmed once per crtc at startup,
+	 * not sure where the best place for it is
+	 */
+	tmp = RREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, ALPHA_CONTROL, CURSOR_ALPHA_BLND_ENA, 1);
+	WREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static int dce_v10_0_pick_dig_encoder(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		if (dig->linkb)
+			return 1;
+		else
+			return 0;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		if (dig->linkb)
+			return 3;
+		else
+			return 2;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+		if (dig->linkb)
+			return 5;
+		else
+			return 4;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		return 6;
+		break;
+	default:
+		DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
+		return 0;
+	}
+}
+
+/**
+ * dce_v10_0_pick_pll - Allocate a PPLL for use by the crtc.
+ *
+ * @crtc: drm crtc
+ *
+ * Returns the PPLL (Pixel PLL) to be used by the crtc.  For DP monitors
+ * a single PPLL can be used for all DP crtcs/encoders.  For non-DP
+ * monitors a dedicated PPLL must be used.  If a particular board has
+ * an external DP PLL, return ATOM_PPLL_INVALID to skip PLL programming
+ * as there is no need to program the PLL itself.  If we are not able to
+ * allocate a PLL, return ATOM_PPLL_INVALID to skip PLL programming to
+ * avoid messing up an existing monitor.
+ *
+ * Asic specific PLL information
+ *
+ * DCE 10.x
+ * Tonga
+ * - PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP)
+ * CI
+ * - PPLL0, PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP) and DAC
+ *
+ */
+static u32 dce_v10_0_pick_pll(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 pll_in_use;
+	int pll;
+
+	if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder))) {
+		if (adev->clock.dp_extclk)
+			/* skip PPLL programming if using ext clock */
+			return ATOM_PPLL_INVALID;
+		else {
+			/* use the same PPLL for all DP monitors */
+			pll = amdgpu_pll_get_shared_dp_ppll(crtc);
+			if (pll != ATOM_PPLL_INVALID)
+				return pll;
+		}
+	} else {
+		/* use the same PPLL for all monitors with the same clock */
+		pll = amdgpu_pll_get_shared_nondp_ppll(crtc);
+		if (pll != ATOM_PPLL_INVALID)
+			return pll;
+	}
+
+	/* DCE10 has PPLL0, PPLL1, and PPLL2 */
+	pll_in_use = amdgpu_pll_get_use_mask(crtc);
+	if (!(pll_in_use & (1 << ATOM_PPLL2)))
+		return ATOM_PPLL2;
+	if (!(pll_in_use & (1 << ATOM_PPLL1)))
+		return ATOM_PPLL1;
+	if (!(pll_in_use & (1 << ATOM_PPLL0)))
+		return ATOM_PPLL0;
+	DRM_ERROR("unable to allocate a PPLL\n");
+	return ATOM_PPLL_INVALID;
+}
+
+static void dce_v10_0_lock_cursor(struct drm_crtc *crtc, bool lock)
+{
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	uint32_t cur_lock;
+
+	cur_lock = RREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset);
+	if (lock)
+		cur_lock = REG_SET_FIELD(cur_lock, CUR_UPDATE, CURSOR_UPDATE_LOCK, 1);
+	else
+		cur_lock = REG_SET_FIELD(cur_lock, CUR_UPDATE, CURSOR_UPDATE_LOCK, 0);
+	WREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset, cur_lock);
+}
+
+static void dce_v10_0_hide_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	u32 tmp;
+
+	tmp = RREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 0);
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static void dce_v10_0_show_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	u32 tmp;
+
+	WREG32(mmCUR_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(amdgpu_crtc->cursor_addr));
+	WREG32(mmCUR_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(amdgpu_crtc->cursor_addr));
+
+	tmp = RREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 1);
+	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_MODE, 2);
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static int dce_v10_0_cursor_move_locked(struct drm_crtc *crtc,
+					int x, int y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	int xorigin = 0, yorigin = 0;
+
+	amdgpu_crtc->cursor_x = x;
+	amdgpu_crtc->cursor_y = y;
+
+	/* avivo cursor are offset into the total surface */
+	x += crtc->x;
+	y += crtc->y;
+	DRM_DEBUG("x %d y %d c->x %d c->y %d\n", x, y, crtc->x, crtc->y);
+
+	if (x < 0) {
+		xorigin = min(-x, amdgpu_crtc->max_cursor_width - 1);
+		x = 0;
+	}
+	if (y < 0) {
+		yorigin = min(-y, amdgpu_crtc->max_cursor_height - 1);
+		y = 0;
+	}
+
+	WREG32(mmCUR_POSITION + amdgpu_crtc->crtc_offset, (x << 16) | y);
+	WREG32(mmCUR_HOT_SPOT + amdgpu_crtc->crtc_offset, (xorigin << 16) | yorigin);
+	WREG32(mmCUR_SIZE + amdgpu_crtc->crtc_offset,
+	       ((amdgpu_crtc->cursor_width - 1) << 16) | (amdgpu_crtc->cursor_height - 1));
+
+	return 0;
+}
+
+static int dce_v10_0_crtc_cursor_move(struct drm_crtc *crtc,
+				      int x, int y)
+{
+	int ret;
+
+	dce_v10_0_lock_cursor(crtc, true);
+	ret = dce_v10_0_cursor_move_locked(crtc, x, y);
+	dce_v10_0_lock_cursor(crtc, false);
+
+	return ret;
+}
+
+static int dce_v10_0_crtc_cursor_set2(struct drm_crtc *crtc,
+				      struct drm_file *file_priv,
+				      uint32_t handle,
+				      uint32_t width,
+				      uint32_t height,
+				      int32_t hot_x,
+				      int32_t hot_y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *aobj;
+	int ret;
+
+	if (!handle) {
+		/* turn off cursor */
+		dce_v10_0_hide_cursor(crtc);
+		obj = NULL;
+		goto unpin;
+	}
+
+	if ((width > amdgpu_crtc->max_cursor_width) ||
+	    (height > amdgpu_crtc->max_cursor_height)) {
+		DRM_ERROR("bad cursor width or height %d x %d\n", width, height);
+		return -EINVAL;
+	}
+
+	obj = drm_gem_object_lookup(file_priv, handle);
+	if (!obj) {
+		DRM_ERROR("Cannot find cursor object %x for crtc %d\n", handle, amdgpu_crtc->crtc_id);
+		return -ENOENT;
+	}
+
+	aobj = gem_to_amdgpu_bo(obj);
+	ret = amdgpu_bo_reserve(aobj, false);
+	if (ret != 0) {
+		drm_gem_object_put_unlocked(obj);
+		return ret;
+	}
+
+	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM);
+	amdgpu_bo_unreserve(aobj);
+	if (ret) {
+		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
+		drm_gem_object_put_unlocked(obj);
+		return ret;
+	}
+	amdgpu_crtc->cursor_addr = amdgpu_bo_gpu_offset(aobj);
+
+	dce_v10_0_lock_cursor(crtc, true);
+
+	if (width != amdgpu_crtc->cursor_width ||
+	    height != amdgpu_crtc->cursor_height ||
+	    hot_x != amdgpu_crtc->cursor_hot_x ||
+	    hot_y != amdgpu_crtc->cursor_hot_y) {
+		int x, y;
+
+		x = amdgpu_crtc->cursor_x + amdgpu_crtc->cursor_hot_x - hot_x;
+		y = amdgpu_crtc->cursor_y + amdgpu_crtc->cursor_hot_y - hot_y;
+
+		dce_v10_0_cursor_move_locked(crtc, x, y);
+
+		amdgpu_crtc->cursor_width = width;
+		amdgpu_crtc->cursor_height = height;
+		amdgpu_crtc->cursor_hot_x = hot_x;
+		amdgpu_crtc->cursor_hot_y = hot_y;
+	}
+
+	dce_v10_0_show_cursor(crtc);
+	dce_v10_0_lock_cursor(crtc, false);
+
+unpin:
+	if (amdgpu_crtc->cursor_bo) {
+		struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+		ret = amdgpu_bo_reserve(aobj, true);
+		if (likely(ret == 0)) {
+			amdgpu_bo_unpin(aobj);
+			amdgpu_bo_unreserve(aobj);
+		}
+		drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
+	}
+
+	amdgpu_crtc->cursor_bo = obj;
+	return 0;
+}
+
+static void dce_v10_0_cursor_reset(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (amdgpu_crtc->cursor_bo) {
+		dce_v10_0_lock_cursor(crtc, true);
+
+		dce_v10_0_cursor_move_locked(crtc, amdgpu_crtc->cursor_x,
+					     amdgpu_crtc->cursor_y);
+
+		dce_v10_0_show_cursor(crtc);
+
+		dce_v10_0_lock_cursor(crtc, false);
+	}
+}
+
+static int dce_v10_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
+				    u16 *blue, uint32_t size,
+				    struct drm_modeset_acquire_ctx *ctx)
+{
+	dce_v10_0_crtc_load_lut(crtc);
+
+	return 0;
+}
+
+static void dce_v10_0_crtc_destroy(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	drm_crtc_cleanup(crtc);
+	kfree(amdgpu_crtc);
+}
+
+static const struct drm_crtc_funcs dce_v10_0_crtc_funcs = {
+	.cursor_set2 = dce_v10_0_crtc_cursor_set2,
+	.cursor_move = dce_v10_0_crtc_cursor_move,
+	.gamma_set = dce_v10_0_crtc_gamma_set,
+	.set_config = amdgpu_display_crtc_set_config,
+	.destroy = dce_v10_0_crtc_destroy,
+	.page_flip_target = amdgpu_display_crtc_page_flip_target,
+};
+
+static void dce_v10_0_crtc_dpms(struct drm_crtc *crtc, int mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	unsigned type;
+
+	switch (mode) {
+	case DRM_MODE_DPMS_ON:
+		amdgpu_crtc->enabled = true;
+		amdgpu_atombios_crtc_enable(crtc, ATOM_ENABLE);
+		dce_v10_0_vga_enable(crtc, true);
+		amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
+		dce_v10_0_vga_enable(crtc, false);
+		/* Make sure VBLANK and PFLIP interrupts are still enabled */
+		type = amdgpu_display_crtc_idx_to_irq_type(adev,
+						amdgpu_crtc->crtc_id);
+		amdgpu_irq_update(adev, &adev->crtc_irq, type);
+		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
+		drm_crtc_vblank_on(crtc);
+		dce_v10_0_crtc_load_lut(crtc);
+		break;
+	case DRM_MODE_DPMS_STANDBY:
+	case DRM_MODE_DPMS_SUSPEND:
+	case DRM_MODE_DPMS_OFF:
+		drm_crtc_vblank_off(crtc);
+		if (amdgpu_crtc->enabled) {
+			dce_v10_0_vga_enable(crtc, true);
+			amdgpu_atombios_crtc_blank(crtc, ATOM_ENABLE);
+			dce_v10_0_vga_enable(crtc, false);
+		}
+		amdgpu_atombios_crtc_enable(crtc, ATOM_DISABLE);
+		amdgpu_crtc->enabled = false;
+		break;
+	}
+	/* adjust pm to dpms */
+	amdgpu_pm_compute_clocks(adev);
+}
+
+static void dce_v10_0_crtc_prepare(struct drm_crtc *crtc)
+{
+	/* disable crtc pair power gating before programming */
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_DISABLE);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_ENABLE);
+	dce_v10_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+}
+
+static void dce_v10_0_crtc_commit(struct drm_crtc *crtc)
+{
+	dce_v10_0_crtc_dpms(crtc, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_DISABLE);
+}
+
+static void dce_v10_0_crtc_disable(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_atom_ss ss;
+	int i;
+
+	dce_v10_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+	if (crtc->primary->fb) {
+		int r;
+		struct amdgpu_bo *abo;
+
+		abo = gem_to_amdgpu_bo(crtc->primary->fb->obj[0]);
+		r = amdgpu_bo_reserve(abo, true);
+		if (unlikely(r))
+			DRM_ERROR("failed to reserve abo before unpin\n");
+		else {
+			amdgpu_bo_unpin(abo);
+			amdgpu_bo_unreserve(abo);
+		}
+	}
+	/* disable the GRPH */
+	dce_v10_0_grph_enable(crtc, false);
+
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_ENABLE);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i] &&
+		    adev->mode_info.crtcs[i]->enabled &&
+		    i != amdgpu_crtc->crtc_id &&
+		    amdgpu_crtc->pll_id == adev->mode_info.crtcs[i]->pll_id) {
+			/* one other crtc is using this pll don't turn
+			 * off the pll
+			 */
+			goto done;
+		}
+	}
+
+	switch (amdgpu_crtc->pll_id) {
+	case ATOM_PPLL0:
+	case ATOM_PPLL1:
+	case ATOM_PPLL2:
+		/* disable the ppll */
+		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
+					  0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
+		break;
+	default:
+		break;
+	}
+done:
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+}
+
+static int dce_v10_0_crtc_mode_set(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode,
+				  int x, int y, struct drm_framebuffer *old_fb)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (!amdgpu_crtc->adjusted_clock)
+		return -EINVAL;
+
+	amdgpu_atombios_crtc_set_pll(crtc, adjusted_mode);
+	amdgpu_atombios_crtc_set_dtd_timing(crtc, adjusted_mode);
+	dce_v10_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+	amdgpu_atombios_crtc_overscan_setup(crtc, mode, adjusted_mode);
+	amdgpu_atombios_crtc_scaler_setup(crtc);
+	dce_v10_0_cursor_reset(crtc);
+	/* update the hw version fpr dpm */
+	amdgpu_crtc->hw_mode = *adjusted_mode;
+
+	return 0;
+}
+
+static bool dce_v10_0_crtc_mode_fixup(struct drm_crtc *crtc,
+				     const struct drm_display_mode *mode,
+				     struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct drm_encoder *encoder;
+
+	/* assign the encoder to the amdgpu crtc to avoid repeated lookups later */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		if (encoder->crtc == crtc) {
+			amdgpu_crtc->encoder = encoder;
+			amdgpu_crtc->connector = amdgpu_get_connector_for_encoder(encoder);
+			break;
+		}
+	}
+	if ((amdgpu_crtc->encoder == NULL) || (amdgpu_crtc->connector == NULL)) {
+		amdgpu_crtc->encoder = NULL;
+		amdgpu_crtc->connector = NULL;
+		return false;
+	}
+	if (!amdgpu_display_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
+		return false;
+	if (amdgpu_atombios_crtc_prepare_pll(crtc, adjusted_mode))
+		return false;
+	/* pick pll */
+	amdgpu_crtc->pll_id = dce_v10_0_pick_pll(crtc);
+	/* if we can't get a PPLL for a non-DP encoder, fail */
+	if ((amdgpu_crtc->pll_id == ATOM_PPLL_INVALID) &&
+	    !ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder)))
+		return false;
+
+	return true;
+}
+
+static int dce_v10_0_crtc_set_base(struct drm_crtc *crtc, int x, int y,
+				  struct drm_framebuffer *old_fb)
+{
+	return dce_v10_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+}
+
+static int dce_v10_0_crtc_set_base_atomic(struct drm_crtc *crtc,
+					 struct drm_framebuffer *fb,
+					 int x, int y, enum mode_set_atomic state)
+{
+       return dce_v10_0_crtc_do_set_base(crtc, fb, x, y, 1);
+}
+
+static const struct drm_crtc_helper_funcs dce_v10_0_crtc_helper_funcs = {
+	.dpms = dce_v10_0_crtc_dpms,
+	.mode_fixup = dce_v10_0_crtc_mode_fixup,
+	.mode_set = dce_v10_0_crtc_mode_set,
+	.mode_set_base = dce_v10_0_crtc_set_base,
+	.mode_set_base_atomic = dce_v10_0_crtc_set_base_atomic,
+	.prepare = dce_v10_0_crtc_prepare,
+	.commit = dce_v10_0_crtc_commit,
+	.disable = dce_v10_0_crtc_disable,
+};
+
+static int dce_v10_0_crtc_init(struct amdgpu_device *adev, int index)
+{
+	struct amdgpu_crtc *amdgpu_crtc;
+
+	amdgpu_crtc = kzalloc(sizeof(struct amdgpu_crtc) +
+			      (AMDGPUFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
+	if (amdgpu_crtc == NULL)
+		return -ENOMEM;
+
+	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_v10_0_crtc_funcs);
+
+	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
+	amdgpu_crtc->crtc_id = index;
+	adev->mode_info.crtcs[index] = amdgpu_crtc;
+
+	amdgpu_crtc->max_cursor_width = 128;
+	amdgpu_crtc->max_cursor_height = 128;
+	adev->ddev->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
+	adev->ddev->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
+
+	switch (amdgpu_crtc->crtc_id) {
+	case 0:
+	default:
+		amdgpu_crtc->crtc_offset = CRTC0_REGISTER_OFFSET;
+		break;
+	case 1:
+		amdgpu_crtc->crtc_offset = CRTC1_REGISTER_OFFSET;
+		break;
+	case 2:
+		amdgpu_crtc->crtc_offset = CRTC2_REGISTER_OFFSET;
+		break;
+	case 3:
+		amdgpu_crtc->crtc_offset = CRTC3_REGISTER_OFFSET;
+		break;
+	case 4:
+		amdgpu_crtc->crtc_offset = CRTC4_REGISTER_OFFSET;
+		break;
+	case 5:
+		amdgpu_crtc->crtc_offset = CRTC5_REGISTER_OFFSET;
+		break;
+	}
+
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_v10_0_crtc_helper_funcs);
+
+	return 0;
+}
+
+static int dce_v10_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->audio_endpt_rreg = &dce_v10_0_audio_endpt_rreg;
+	adev->audio_endpt_wreg = &dce_v10_0_audio_endpt_wreg;
+
+	dce_v10_0_set_display_funcs(adev);
+
+	adev->mode_info.num_crtc = dce_v10_0_get_num_crtc(adev);
+
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+	case CHIP_TONGA:
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 7;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	dce_v10_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int dce_v10_0_sw_init(void *handle)
+{
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
+		if (r)
+			return r;
+	}
+
+	for (i = VISLANDS30_IV_SRCID_D1_GRPH_PFLIP; i < 20; i += 2) {
+		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i, &adev->pageflip_irq);
+		if (r)
+			return r;
+	}
+
+	/* HPD hotplug */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_HOTPLUG_DETECT_A, &adev->hpd_irq);
+	if (r)
+		return r;
+
+	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
+
+	adev->ddev->mode_config.async_page_flip = true;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	adev->ddev->mode_config.preferred_depth = 24;
+	adev->ddev->mode_config.prefer_shadow = 1;
+
+	adev->ddev->mode_config.fb_base = adev->gmc.aper_base;
+
+	r = amdgpu_display_modeset_create_props(adev);
+	if (r)
+		return r;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	/* allocate crtcs */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = dce_v10_0_crtc_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	if (amdgpu_atombios_get_connector_info_from_object_table(adev))
+		amdgpu_display_print_display_setup(adev->ddev);
+	else
+		return -EINVAL;
+
+	/* setup afmt */
+	r = dce_v10_0_afmt_init(adev);
+	if (r)
+		return r;
+
+	r = dce_v10_0_audio_init(adev);
+	if (r)
+		return r;
+
+	drm_kms_helper_poll_init(adev->ddev);
+
+	adev->mode_info.mode_config_initialized = true;
+	return 0;
+}
+
+static int dce_v10_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	kfree(adev->mode_info.bios_hardcoded_edid);
+
+	drm_kms_helper_poll_fini(adev->ddev);
+
+	dce_v10_0_audio_fini(adev);
+
+	dce_v10_0_afmt_fini(adev);
+
+	drm_mode_config_cleanup(adev->ddev);
+	adev->mode_info.mode_config_initialized = false;
+
+	return 0;
+}
+
+static int dce_v10_0_hw_init(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dce_v10_0_init_golden_registers(adev);
+
+	/* disable vga render */
+	dce_v10_0_set_vga_render_state(adev, false);
+	/* init dig PHYs, disp eng pll */
+	amdgpu_atombios_encoder_init_dig(adev);
+	amdgpu_atombios_crtc_set_disp_eng_pll(adev, adev->clock.default_dispclk);
+
+	/* initialize hpd */
+	dce_v10_0_hpd_init(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v10_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v10_0_pageflip_interrupt_init(adev);
+
+	return 0;
+}
+
+static int dce_v10_0_hw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dce_v10_0_hpd_fini(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v10_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v10_0_pageflip_interrupt_fini(adev);
+
+	return 0;
+}
+
+static int dce_v10_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->mode_info.bl_level =
+		amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+
+	return dce_v10_0_hw_fini(handle);
+}
+
+static int dce_v10_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret;
+
+	amdgpu_atombios_encoder_set_backlight_level_to_reg(adev,
+							   adev->mode_info.bl_level);
+
+	ret = dce_v10_0_hw_init(handle);
+
+	/* turn on the BL */
+	if (adev->mode_info.bl_encoder) {
+		u8 bl_level = amdgpu_display_backlight_get_level(adev,
+								  adev->mode_info.bl_encoder);
+		amdgpu_display_backlight_set_level(adev, adev->mode_info.bl_encoder,
+						    bl_level);
+	}
+
+	return ret;
+}
+
+static bool dce_v10_0_is_idle(void *handle)
+{
+	return true;
+}
+
+static int dce_v10_0_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static bool dce_v10_0_check_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return dce_v10_0_is_display_hung(adev);
+}
+
+static int dce_v10_0_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0, tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (dce_v10_0_is_display_hung(adev))
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+	return 0;
+}
+
+static void dce_v10_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
+						     int crtc,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 lb_interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VBLANK_INTERRUPT_MASK, 0);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VBLANK_INTERRUPT_MASK, 1);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static void dce_v10_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
+						    int crtc,
+						    enum amdgpu_interrupt_state state)
+{
+	u32 lb_interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VLINE_INTERRUPT_MASK, 0);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VLINE_INTERRUPT_MASK, 1);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static int dce_v10_0_set_hpd_irq_state(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *source,
+				       unsigned hpd,
+				       enum amdgpu_interrupt_state state)
+{
+	u32 tmp;
+
+	if (hpd >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hdp %d\n", hpd);
+		return 0;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 0);
+		WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 1);
+		WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v10_0_set_crtc_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CRTC_IRQ_VBLANK1:
+		dce_v10_0_set_crtc_vblank_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK2:
+		dce_v10_0_set_crtc_vblank_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK3:
+		dce_v10_0_set_crtc_vblank_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK4:
+		dce_v10_0_set_crtc_vblank_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK5:
+		dce_v10_0_set_crtc_vblank_interrupt_state(adev, 4, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK6:
+		dce_v10_0_set_crtc_vblank_interrupt_state(adev, 5, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE1:
+		dce_v10_0_set_crtc_vline_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE2:
+		dce_v10_0_set_crtc_vline_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE3:
+		dce_v10_0_set_crtc_vline_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE4:
+		dce_v10_0_set_crtc_vline_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE5:
+		dce_v10_0_set_crtc_vline_interrupt_state(adev, 4, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE6:
+		dce_v10_0_set_crtc_vline_interrupt_state(adev, 5, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int dce_v10_0_set_pageflip_irq_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	u32 reg;
+
+	if (type >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", type);
+		return -EINVAL;
+	}
+
+	reg = RREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type]);
+	if (state == AMDGPU_IRQ_STATE_DISABLE)
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg & ~GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+	else
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg | GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+
+	return 0;
+}
+
+static int dce_v10_0_pageflip_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	unsigned long flags;
+	unsigned crtc_id;
+	struct amdgpu_crtc *amdgpu_crtc;
+	struct amdgpu_flip_work *works;
+
+	crtc_id = (entry->src_id - 8) >> 1;
+	amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	if (crtc_id >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", crtc_id);
+		return -EINVAL;
+	}
+
+	if (RREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id]) &
+	    GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_OCCURRED_MASK)
+		WREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id],
+		       GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_CLEAR_MASK);
+
+	/* IRQ could occur when in initial stage */
+	if (amdgpu_crtc == NULL)
+		return 0;
+
+	spin_lock_irqsave(&adev->ddev->event_lock, flags);
+	works = amdgpu_crtc->pflip_works;
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED) {
+		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
+						 "AMDGPU_FLIP_SUBMITTED(%d)\n",
+						 amdgpu_crtc->pflip_status,
+						 AMDGPU_FLIP_SUBMITTED);
+		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+		return 0;
+	}
+
+	/* page flip completed. clean up */
+	amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE;
+	amdgpu_crtc->pflip_works = NULL;
+
+	/* wakeup usersapce */
+	if (works->event)
+		drm_crtc_send_vblank_event(&amdgpu_crtc->base, works->event);
+
+	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+
+	drm_crtc_vblank_put(&amdgpu_crtc->base);
+	schedule_work(&works->unpin_work);
+
+	return 0;
+}
+
+static void dce_v10_0_hpd_int_ack(struct amdgpu_device *adev,
+				  int hpd)
+{
+	u32 tmp;
+
+	if (hpd >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hdp %d\n", hpd);
+		return;
+	}
+
+	tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+	tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_ACK, 1);
+	WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+}
+
+static void dce_v10_0_crtc_vblank_int_ack(struct amdgpu_device *adev,
+					  int crtc)
+{
+	u32 tmp;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	tmp = RREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc]);
+	tmp = REG_SET_FIELD(tmp, LB_VBLANK_STATUS, VBLANK_ACK, 1);
+	WREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc], tmp);
+}
+
+static void dce_v10_0_crtc_vline_int_ack(struct amdgpu_device *adev,
+					 int crtc)
+{
+	u32 tmp;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	tmp = RREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc]);
+	tmp = REG_SET_FIELD(tmp, LB_VLINE_STATUS, VLINE_ACK, 1);
+	WREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc], tmp);
+}
+
+static int dce_v10_0_crtc_irq(struct amdgpu_device *adev,
+			      struct amdgpu_irq_src *source,
+			      struct amdgpu_iv_entry *entry)
+{
+	unsigned crtc = entry->src_id - 1;
+	uint32_t disp_int = RREG32(interrupt_status_offsets[crtc].reg);
+	unsigned int irq_type = amdgpu_display_crtc_idx_to_irq_type(adev, crtc);
+
+	switch (entry->src_data[0]) {
+	case 0: /* vblank */
+		if (disp_int & interrupt_status_offsets[crtc].vblank)
+			dce_v10_0_crtc_vblank_int_ack(adev, crtc);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		if (amdgpu_irq_enabled(adev, source, irq_type)) {
+			drm_handle_vblank(adev->ddev, crtc);
+		}
+		DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+
+		break;
+	case 1: /* vline */
+		if (disp_int & interrupt_status_offsets[crtc].vline)
+			dce_v10_0_crtc_vline_int_ack(adev, crtc);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		DRM_DEBUG("IH: D%d vline\n", crtc + 1);
+
+		break;
+	default:
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v10_0_hpd_irq(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *source,
+			     struct amdgpu_iv_entry *entry)
+{
+	uint32_t disp_int, mask;
+	unsigned hpd;
+
+	if (entry->src_data[0] >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]);
+		return 0;
+	}
+
+	hpd = entry->src_data[0];
+	disp_int = RREG32(interrupt_status_offsets[hpd].reg);
+	mask = interrupt_status_offsets[hpd].hpd;
+
+	if (disp_int & mask) {
+		dce_v10_0_hpd_int_ack(adev, hpd);
+		schedule_work(&adev->hotplug_work);
+		DRM_DEBUG("IH: HPD%d\n", hpd + 1);
+	}
+
+	return 0;
+}
+
+static int dce_v10_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int dce_v10_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs dce_v10_0_ip_funcs = {
+	.name = "dce_v10_0",
+	.early_init = dce_v10_0_early_init,
+	.late_init = NULL,
+	.sw_init = dce_v10_0_sw_init,
+	.sw_fini = dce_v10_0_sw_fini,
+	.hw_init = dce_v10_0_hw_init,
+	.hw_fini = dce_v10_0_hw_fini,
+	.suspend = dce_v10_0_suspend,
+	.resume = dce_v10_0_resume,
+	.is_idle = dce_v10_0_is_idle,
+	.wait_for_idle = dce_v10_0_wait_for_idle,
+	.check_soft_reset = dce_v10_0_check_soft_reset,
+	.soft_reset = dce_v10_0_soft_reset,
+	.set_clockgating_state = dce_v10_0_set_clockgating_state,
+	.set_powergating_state = dce_v10_0_set_powergating_state,
+};
+
+static void
+dce_v10_0_encoder_mode_set(struct drm_encoder *encoder,
+			  struct drm_display_mode *mode,
+			  struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	amdgpu_encoder->pixel_clock = adjusted_mode->clock;
+
+	/* need to call this here rather than in prepare() since we need some crtc info */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	/* set scaler clears this on some chips */
+	dce_v10_0_set_interleave(encoder->crtc, mode);
+
+	if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) {
+		dce_v10_0_afmt_enable(encoder, true);
+		dce_v10_0_afmt_setmode(encoder, adjusted_mode);
+	}
+}
+
+static void dce_v10_0_encoder_prepare(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+
+	if ((amdgpu_encoder->active_device &
+	     (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) !=
+	     ENCODER_OBJECT_ID_NONE)) {
+		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+		if (dig) {
+			dig->dig_encoder = dce_v10_0_pick_dig_encoder(encoder);
+			if (amdgpu_encoder->active_device & ATOM_DEVICE_DFP_SUPPORT)
+				dig->afmt = adev->mode_info.afmt[dig->dig_encoder];
+		}
+	}
+
+	amdgpu_atombios_scratch_regs_lock(adev, true);
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		/* select the clock/data port if it uses a router */
+		if (amdgpu_connector->router.cd_valid)
+			amdgpu_i2c_router_select_cd_port(amdgpu_connector);
+
+		/* turn eDP panel on for mode set */
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP)
+			amdgpu_atombios_encoder_set_edp_panel_power(connector,
+							     ATOM_TRANSMITTER_ACTION_POWER_ON);
+	}
+
+	/* this is needed for the pll/ss setup to work correctly in some cases */
+	amdgpu_atombios_encoder_set_crtc_source(encoder);
+	/* set up the FMT blocks */
+	dce_v10_0_program_fmt(encoder);
+}
+
+static void dce_v10_0_encoder_commit(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* need to call this here as we need the crtc set up */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_scratch_regs_lock(adev, false);
+}
+
+static void dce_v10_0_encoder_disable(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig;
+
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	if (amdgpu_atombios_encoder_is_digital(encoder)) {
+		if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI)
+			dce_v10_0_afmt_enable(encoder, false);
+		dig = amdgpu_encoder->enc_priv;
+		dig->dig_encoder = -1;
+	}
+	amdgpu_encoder->active_device = 0;
+}
+
+/* these are handled by the primary encoders */
+static void dce_v10_0_ext_prepare(struct drm_encoder *encoder)
+{
+
+}
+
+static void dce_v10_0_ext_commit(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v10_0_ext_mode_set(struct drm_encoder *encoder,
+		      struct drm_display_mode *mode,
+		      struct drm_display_mode *adjusted_mode)
+{
+
+}
+
+static void dce_v10_0_ext_disable(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v10_0_ext_dpms(struct drm_encoder *encoder, int mode)
+{
+
+}
+
+static const struct drm_encoder_helper_funcs dce_v10_0_ext_helper_funcs = {
+	.dpms = dce_v10_0_ext_dpms,
+	.prepare = dce_v10_0_ext_prepare,
+	.mode_set = dce_v10_0_ext_mode_set,
+	.commit = dce_v10_0_ext_commit,
+	.disable = dce_v10_0_ext_disable,
+	/* no detect for TMDS/LVDS yet */
+};
+
+static const struct drm_encoder_helper_funcs dce_v10_0_dig_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v10_0_encoder_prepare,
+	.mode_set = dce_v10_0_encoder_mode_set,
+	.commit = dce_v10_0_encoder_commit,
+	.disable = dce_v10_0_encoder_disable,
+	.detect = amdgpu_atombios_encoder_dig_detect,
+};
+
+static const struct drm_encoder_helper_funcs dce_v10_0_dac_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v10_0_encoder_prepare,
+	.mode_set = dce_v10_0_encoder_mode_set,
+	.commit = dce_v10_0_encoder_commit,
+	.detect = amdgpu_atombios_encoder_dac_detect,
+};
+
+static void dce_v10_0_encoder_destroy(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+		amdgpu_atombios_encoder_fini_backlight(amdgpu_encoder);
+	kfree(amdgpu_encoder->enc_priv);
+	drm_encoder_cleanup(encoder);
+	kfree(amdgpu_encoder);
+}
+
+static const struct drm_encoder_funcs dce_v10_0_encoder_funcs = {
+	.destroy = dce_v10_0_encoder_destroy,
+};
+
+static void dce_v10_0_encoder_add(struct amdgpu_device *adev,
+				 uint32_t encoder_enum,
+				 uint32_t supported_device,
+				 u16 caps)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	/* see if we already added it */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		if (amdgpu_encoder->encoder_enum == encoder_enum) {
+			amdgpu_encoder->devices |= supported_device;
+			return;
+		}
+
+	}
+
+	/* add a new one */
+	amdgpu_encoder = kzalloc(sizeof(struct amdgpu_encoder), GFP_KERNEL);
+	if (!amdgpu_encoder)
+		return;
+
+	encoder = &amdgpu_encoder->base;
+	switch (adev->mode_info.num_crtc) {
+	case 1:
+		encoder->possible_crtcs = 0x1;
+		break;
+	case 2:
+	default:
+		encoder->possible_crtcs = 0x3;
+		break;
+	case 4:
+		encoder->possible_crtcs = 0xf;
+		break;
+	case 6:
+		encoder->possible_crtcs = 0x3f;
+		break;
+	}
+
+	amdgpu_encoder->enc_priv = NULL;
+
+	amdgpu_encoder->encoder_enum = encoder_enum;
+	amdgpu_encoder->encoder_id = (encoder_enum & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+	amdgpu_encoder->devices = supported_device;
+	amdgpu_encoder->rmx_type = RMX_OFF;
+	amdgpu_encoder->underscan_type = UNDERSCAN_OFF;
+	amdgpu_encoder->is_ext_encoder = false;
+	amdgpu_encoder->caps = caps;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+		drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+				 DRM_MODE_ENCODER_DAC, NULL);
+		drm_encoder_helper_add(encoder, &dce_v10_0_dac_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+			amdgpu_encoder->rmx_type = RMX_FULL;
+			drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS, NULL);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_lcd_info(amdgpu_encoder);
+		} else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) {
+			drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC, NULL);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		} else {
+			drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS, NULL);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		}
+		drm_encoder_helper_add(encoder, &dce_v10_0_dig_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_SI170B:
+	case ENCODER_OBJECT_ID_CH7303:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOA:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOB:
+	case ENCODER_OBJECT_ID_TITFP513:
+	case ENCODER_OBJECT_ID_VT1623:
+	case ENCODER_OBJECT_ID_HDMI_SI1930:
+	case ENCODER_OBJECT_ID_TRAVIS:
+	case ENCODER_OBJECT_ID_NUTMEG:
+		/* these are handled by the primary encoders */
+		amdgpu_encoder->is_ext_encoder = true;
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS, NULL);
+		else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC, NULL);
+		else
+			drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS, NULL);
+		drm_encoder_helper_add(encoder, &dce_v10_0_ext_helper_funcs);
+		break;
+	}
+}
+
+static const struct amdgpu_display_funcs dce_v10_0_display_funcs = {
+	.bandwidth_update = &dce_v10_0_bandwidth_update,
+	.vblank_get_counter = &dce_v10_0_vblank_get_counter,
+	.backlight_set_level = &amdgpu_atombios_encoder_set_backlight_level,
+	.backlight_get_level = &amdgpu_atombios_encoder_get_backlight_level,
+	.hpd_sense = &dce_v10_0_hpd_sense,
+	.hpd_set_polarity = &dce_v10_0_hpd_set_polarity,
+	.hpd_get_gpio_reg = &dce_v10_0_hpd_get_gpio_reg,
+	.page_flip = &dce_v10_0_page_flip,
+	.page_flip_get_scanoutpos = &dce_v10_0_crtc_get_scanoutpos,
+	.add_encoder = &dce_v10_0_encoder_add,
+	.add_connector = &amdgpu_connector_add,
+};
+
+static void dce_v10_0_set_display_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.funcs == NULL)
+		adev->mode_info.funcs = &dce_v10_0_display_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs dce_v10_0_crtc_irq_funcs = {
+	.set = dce_v10_0_set_crtc_irq_state,
+	.process = dce_v10_0_crtc_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v10_0_pageflip_irq_funcs = {
+	.set = dce_v10_0_set_pageflip_irq_state,
+	.process = dce_v10_0_pageflip_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v10_0_hpd_irq_funcs = {
+	.set = dce_v10_0_set_hpd_irq_state,
+	.process = dce_v10_0_hpd_irq,
+};
+
+static void dce_v10_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.num_crtc > 0)
+		adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_VLINE1 + adev->mode_info.num_crtc;
+	else
+		adev->crtc_irq.num_types = 0;
+	adev->crtc_irq.funcs = &dce_v10_0_crtc_irq_funcs;
+
+	adev->pageflip_irq.num_types = adev->mode_info.num_crtc;
+	adev->pageflip_irq.funcs = &dce_v10_0_pageflip_irq_funcs;
+
+	adev->hpd_irq.num_types = adev->mode_info.num_hpd;
+	adev->hpd_irq.funcs = &dce_v10_0_hpd_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version dce_v10_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 10,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &dce_v10_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version dce_v10_1_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 10,
+	.minor = 1,
+	.rev = 0,
+	.funcs = &dce_v10_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/dce_v10_0.h b/sys/dev/pci/drm/amd/amdgpu/dce_v10_0.h
new file mode 100644
index 00000000000..7a0747789f1
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/dce_v10_0.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DCE_V10_0_H__
+#define __DCE_V10_0_H__
+
+
+extern const struct amdgpu_ip_block_version dce_v10_0_ip_block;
+extern const struct amdgpu_ip_block_version dce_v10_1_ip_block;
+
+void dce_v10_0_disable_dce(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/dce_v11_0.c b/sys/dev/pci/drm/amd/amdgpu/dce_v11_0.c
new file mode 100644
index 00000000000..76dfb76f790
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/dce_v11_0.c
@@ -0,0 +1,3743 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_i2c.h"
+#include "vid.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+#include "atombios_crtc.h"
+#include "atombios_encoders.h"
+#include "amdgpu_pll.h"
+#include "amdgpu_connectors.h"
+#include "dce_v11_0.h"
+
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+#include "dce/dce_11_0_enum.h"
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+#include "gmc/gmc_8_1_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+static void dce_v11_0_set_display_funcs(struct amdgpu_device *adev);
+static void dce_v11_0_set_irq_funcs(struct amdgpu_device *adev);
+
+static const u32 crtc_offsets[] =
+{
+	CRTC0_REGISTER_OFFSET,
+	CRTC1_REGISTER_OFFSET,
+	CRTC2_REGISTER_OFFSET,
+	CRTC3_REGISTER_OFFSET,
+	CRTC4_REGISTER_OFFSET,
+	CRTC5_REGISTER_OFFSET,
+	CRTC6_REGISTER_OFFSET
+};
+
+static const u32 hpd_offsets[] =
+{
+	HPD0_REGISTER_OFFSET,
+	HPD1_REGISTER_OFFSET,
+	HPD2_REGISTER_OFFSET,
+	HPD3_REGISTER_OFFSET,
+	HPD4_REGISTER_OFFSET,
+	HPD5_REGISTER_OFFSET
+};
+
+static const uint32_t dig_offsets[] = {
+	DIG0_REGISTER_OFFSET,
+	DIG1_REGISTER_OFFSET,
+	DIG2_REGISTER_OFFSET,
+	DIG3_REGISTER_OFFSET,
+	DIG4_REGISTER_OFFSET,
+	DIG5_REGISTER_OFFSET,
+	DIG6_REGISTER_OFFSET,
+	DIG7_REGISTER_OFFSET,
+	DIG8_REGISTER_OFFSET
+};
+
+static const struct {
+	uint32_t        reg;
+	uint32_t        vblank;
+	uint32_t        vline;
+	uint32_t        hpd;
+
+} interrupt_status_offsets[] = { {
+	.reg = mmDISP_INTERRUPT_STATUS,
+	.vblank = DISP_INTERRUPT_STATUS__LB_D1_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS__LB_D1_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS__DC_HPD1_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE__DC_HPD2_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE2,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE2__DC_HPD3_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE3,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE3__DC_HPD4_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE4,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE4__DC_HPD5_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE5,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE5__DC_HPD6_INTERRUPT_MASK
+} };
+
+static const u32 cz_golden_settings_a11[] =
+{
+	mmCRTC_DOUBLE_BUFFER_CONTROL, 0x00010101, 0x00010000,
+	mmFBC_MISC, 0x1f311fff, 0x14300000,
+};
+
+static const u32 cz_mgcg_cgcg_init[] =
+{
+	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
+	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
+};
+
+static const u32 stoney_golden_settings_a11[] =
+{
+	mmCRTC_DOUBLE_BUFFER_CONTROL, 0x00010101, 0x00010000,
+	mmFBC_MISC, 0x1f311fff, 0x14302000,
+};
+
+static const u32 polaris11_golden_settings_a11[] =
+{
+	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
+	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
+	mmFBC_DEBUG1, 0xffffffff, 0x00000008,
+	mmFBC_MISC, 0x9f313fff, 0x14302008,
+	mmHDMI_CONTROL, 0x313f031f, 0x00000011,
+};
+
+static const u32 polaris10_golden_settings_a11[] =
+{
+	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
+	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
+	mmFBC_MISC, 0x9f313fff, 0x14302008,
+	mmHDMI_CONTROL, 0x313f031f, 0x00000011,
+};
+
+static void dce_v11_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_CARRIZO:
+		amdgpu_device_program_register_sequence(adev,
+							cz_mgcg_cgcg_init,
+							ARRAY_SIZE(cz_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							cz_golden_settings_a11,
+							ARRAY_SIZE(cz_golden_settings_a11));
+		break;
+	case CHIP_STONEY:
+		amdgpu_device_program_register_sequence(adev,
+							stoney_golden_settings_a11,
+							ARRAY_SIZE(stoney_golden_settings_a11));
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+		amdgpu_device_program_register_sequence(adev,
+							polaris11_golden_settings_a11,
+							ARRAY_SIZE(polaris11_golden_settings_a11));
+		break;
+	case CHIP_POLARIS10:
+	case CHIP_VEGAM:
+		amdgpu_device_program_register_sequence(adev,
+							polaris10_golden_settings_a11,
+							ARRAY_SIZE(polaris10_golden_settings_a11));
+		break;
+	default:
+		break;
+	}
+}
+
+static u32 dce_v11_0_audio_endpt_rreg(struct amdgpu_device *adev,
+				     u32 block_offset, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	r = RREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+
+	return r;
+}
+
+static void dce_v11_0_audio_endpt_wreg(struct amdgpu_device *adev,
+				      u32 block_offset, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset, v);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+}
+
+static u32 dce_v11_0_vblank_get_counter(struct amdgpu_device *adev, int crtc)
+{
+	if (crtc < 0 || crtc >= adev->mode_info.num_crtc)
+		return 0;
+	else
+		return RREG32(mmCRTC_STATUS_FRAME_COUNT + crtc_offsets[crtc]);
+}
+
+static void dce_v11_0_pageflip_interrupt_init(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Enable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_get(adev, &adev->pageflip_irq, i);
+}
+
+static void dce_v11_0_pageflip_interrupt_fini(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Disable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_put(adev, &adev->pageflip_irq, i);
+}
+
+/**
+ * dce_v11_0_page_flip - pageflip callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @crtc_id: crtc to cleanup pageflip on
+ * @crtc_base: new address of the crtc (GPU MC address)
+ *
+ * Triggers the actual pageflip by updating the primary
+ * surface base address.
+ */
+static void dce_v11_0_page_flip(struct amdgpu_device *adev,
+				int crtc_id, u64 crtc_base, bool async)
+{
+	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+	u32 tmp;
+
+	/* flip immediate for async, default is vsync */
+	tmp = RREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, GRPH_FLIP_CONTROL,
+			    GRPH_SURFACE_UPDATE_IMMEDIATE_EN, async ? 1 : 0);
+	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	/* update the scanout addresses */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(crtc_base));
+	/* writing to the low address triggers the update */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(crtc_base));
+	/* post the write */
+	RREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset);
+}
+
+static int dce_v11_0_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
+					u32 *vbl, u32 *position)
+{
+	if ((crtc < 0) || (crtc >= adev->mode_info.num_crtc))
+		return -EINVAL;
+
+	*vbl = RREG32(mmCRTC_V_BLANK_START_END + crtc_offsets[crtc]);
+	*position = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+
+	return 0;
+}
+
+/**
+ * dce_v11_0_hpd_sense - hpd sense callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Checks if a digital monitor is connected (evergreen+).
+ * Returns true if connected, false if not connected.
+ */
+static bool dce_v11_0_hpd_sense(struct amdgpu_device *adev,
+			       enum amdgpu_hpd_id hpd)
+{
+	bool connected = false;
+
+	if (hpd >= adev->mode_info.num_hpd)
+		return connected;
+
+	if (RREG32(mmDC_HPD_INT_STATUS + hpd_offsets[hpd]) &
+	    DC_HPD_INT_STATUS__DC_HPD_SENSE_MASK)
+		connected = true;
+
+	return connected;
+}
+
+/**
+ * dce_v11_0_hpd_set_polarity - hpd set polarity callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Set the polarity of the hpd pin (evergreen+).
+ */
+static void dce_v11_0_hpd_set_polarity(struct amdgpu_device *adev,
+				      enum amdgpu_hpd_id hpd)
+{
+	u32 tmp;
+	bool connected = dce_v11_0_hpd_sense(adev, hpd);
+
+	if (hpd >= adev->mode_info.num_hpd)
+		return;
+
+	tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+	if (connected)
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_POLARITY, 0);
+	else
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_POLARITY, 1);
+	WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+}
+
+/**
+ * dce_v11_0_hpd_init - hpd setup callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup the hpd pins used by the card (evergreen+).
+ * Enable the pin, set the polarity, and enable the hpd interrupts.
+ */
+static void dce_v11_0_hpd_init(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
+			continue;
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP ||
+		    connector->connector_type == DRM_MODE_CONNECTOR_LVDS) {
+			/* don't try to enable hpd on eDP or LVDS avoid breaking the
+			 * aux dp channel on imac and help (but not completely fix)
+			 * https://bugzilla.redhat.com/show_bug.cgi?id=726143
+			 * also avoid interrupt storms during dpms.
+			 */
+			tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+			tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 0);
+			WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+			continue;
+		}
+
+		tmp = RREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_CONTROL, DC_HPD_EN, 1);
+		WREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+
+		tmp = RREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_TOGGLE_FILT_CNTL,
+				    DC_HPD_CONNECT_INT_DELAY,
+				    AMDGPU_HPD_CONNECT_INT_DELAY_IN_MS);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_TOGGLE_FILT_CNTL,
+				    DC_HPD_DISCONNECT_INT_DELAY,
+				    AMDGPU_HPD_DISCONNECT_INT_DELAY_IN_MS);
+		WREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+
+		dce_v11_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
+		amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
+	}
+}
+
+/**
+ * dce_v11_0_hpd_fini - hpd tear down callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the hpd pins used by the card (evergreen+).
+ * Disable the hpd interrupts.
+ */
+static void dce_v11_0_hpd_fini(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
+			continue;
+
+		tmp = RREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_CONTROL, DC_HPD_EN, 0);
+		WREG32(mmDC_HPD_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+
+		amdgpu_irq_put(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
+	}
+}
+
+static u32 dce_v11_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
+{
+	return mmDC_GPIO_HPD_A;
+}
+
+static bool dce_v11_0_is_display_hung(struct amdgpu_device *adev)
+{
+	u32 crtc_hung = 0;
+	u32 crtc_status[6];
+	u32 i, j, tmp;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
+		if (REG_GET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN)) {
+			crtc_status[i] = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+			crtc_hung |= (1 << i);
+		}
+	}
+
+	for (j = 0; j < 10; j++) {
+		for (i = 0; i < adev->mode_info.num_crtc; i++) {
+			if (crtc_hung & (1 << i)) {
+				tmp = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+				if (tmp != crtc_status[i])
+					crtc_hung &= ~(1 << i);
+			}
+		}
+		if (crtc_hung == 0)
+			return false;
+		udelay(100);
+	}
+
+	return true;
+}
+
+static void dce_v11_0_set_vga_render_state(struct amdgpu_device *adev,
+					   bool render)
+{
+	u32 tmp;
+
+	/* Lockout access through VGA aperture*/
+	tmp = RREG32(mmVGA_HDP_CONTROL);
+	if (render)
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 0);
+	else
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
+	WREG32(mmVGA_HDP_CONTROL, tmp);
+
+	/* disable VGA render */
+	tmp = RREG32(mmVGA_RENDER_CONTROL);
+	if (render)
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
+	WREG32(mmVGA_RENDER_CONTROL, tmp);
+}
+
+static int dce_v11_0_get_num_crtc (struct amdgpu_device *adev)
+{
+	int num_crtc = 0;
+
+	switch (adev->asic_type) {
+	case CHIP_CARRIZO:
+		num_crtc = 3;
+		break;
+	case CHIP_STONEY:
+		num_crtc = 2;
+		break;
+	case CHIP_POLARIS10:
+	case CHIP_VEGAM:
+		num_crtc = 6;
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+		num_crtc = 5;
+		break;
+	default:
+		num_crtc = 0;
+	}
+	return num_crtc;
+}
+
+void dce_v11_0_disable_dce(struct amdgpu_device *adev)
+{
+	/*Disable VGA render and enabled crtc, if has DCE engine*/
+	if (amdgpu_atombios_has_dce_engine_info(adev)) {
+		u32 tmp;
+		int crtc_enabled, i;
+
+		dce_v11_0_set_vga_render_state(adev, false);
+
+		/*Disable crtc*/
+		for (i = 0; i < dce_v11_0_get_num_crtc(adev); i++) {
+			crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]),
+									 CRTC_CONTROL, CRTC_MASTER_EN);
+			if (crtc_enabled) {
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+				tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
+				tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0);
+				WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			}
+		}
+	}
+}
+
+static void dce_v11_0_program_fmt(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	int bpc = 0;
+	u32 tmp = 0;
+	enum amdgpu_connector_dither dither = AMDGPU_FMT_DITHER_DISABLE;
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		bpc = amdgpu_connector_get_monitor_bpc(connector);
+		dither = amdgpu_connector->dither;
+	}
+
+	/* LVDS/eDP FMT is set up by atom */
+	if (amdgpu_encoder->devices & ATOM_DEVICE_LCD_SUPPORT)
+		return;
+
+	/* not needed for analog */
+	if ((amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1) ||
+	    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2))
+		return;
+
+	if (bpc == 0)
+		return;
+
+	switch (bpc) {
+	case 6:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
+			/* XXX sort out optimal dither settings */
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 0);
+		} else {
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 0);
+		}
+		break;
+	case 8:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
+			/* XXX sort out optimal dither settings */
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 1);
+		} else {
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 1);
+		}
+		break;
+	case 10:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
+			/* XXX sort out optimal dither settings */
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 2);
+		} else {
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 2);
+		}
+		break;
+	default:
+		/* not needed */
+		break;
+	}
+
+	WREG32(mmFMT_BIT_DEPTH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+
+/* display watermark setup */
+/**
+ * dce_v11_0_line_buffer_adjust - Set up the line buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @mode: the current display mode on the selected display
+ * controller
+ *
+ * Setup up the line buffer allocation for
+ * the selected display controller (CIK).
+ * Returns the line buffer size in pixels.
+ */
+static u32 dce_v11_0_line_buffer_adjust(struct amdgpu_device *adev,
+				       struct amdgpu_crtc *amdgpu_crtc,
+				       struct drm_display_mode *mode)
+{
+	u32 tmp, buffer_alloc, i, mem_cfg;
+	u32 pipe_offset = amdgpu_crtc->crtc_id;
+	/*
+	 * Line Buffer Setup
+	 * There are 6 line buffers, one for each display controllers.
+	 * There are 3 partitions per LB. Select the number of partitions
+	 * to enable based on the display width.  For display widths larger
+	 * than 4096, you need use to use 2 display controllers and combine
+	 * them using the stereo blender.
+	 */
+	if (amdgpu_crtc->base.enabled && mode) {
+		if (mode->crtc_hdisplay < 1920) {
+			mem_cfg = 1;
+			buffer_alloc = 2;
+		} else if (mode->crtc_hdisplay < 2560) {
+			mem_cfg = 2;
+			buffer_alloc = 2;
+		} else if (mode->crtc_hdisplay < 4096) {
+			mem_cfg = 0;
+			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
+		} else {
+			DRM_DEBUG_KMS("Mode too big for LB!\n");
+			mem_cfg = 0;
+			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
+		}
+	} else {
+		mem_cfg = 1;
+		buffer_alloc = 0;
+	}
+
+	tmp = RREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, LB_MEMORY_CTRL, LB_MEMORY_CONFIG, mem_cfg);
+	WREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset);
+	tmp = REG_SET_FIELD(tmp, PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATED, buffer_alloc);
+	WREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset, tmp);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset);
+		if (REG_GET_FIELD(tmp, PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATION_COMPLETED))
+			break;
+		udelay(1);
+	}
+
+	if (amdgpu_crtc->base.enabled && mode) {
+		switch (mem_cfg) {
+		case 0:
+		default:
+			return 4096 * 2;
+		case 1:
+			return 1920 * 2;
+		case 2:
+			return 2560 * 2;
+		}
+	}
+
+	/* controller not enabled, so no lb used */
+	return 0;
+}
+
+/**
+ * cik_get_number_of_dram_channels - get the number of dram channels
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the number of video ram channels (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the number of dram channels
+ */
+static u32 cik_get_number_of_dram_channels(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(mmMC_SHARED_CHMAP);
+
+	switch (REG_GET_FIELD(tmp, MC_SHARED_CHMAP, NOOFCHAN)) {
+	case 0:
+	default:
+		return 1;
+	case 1:
+		return 2;
+	case 2:
+		return 4;
+	case 3:
+		return 8;
+	case 4:
+		return 3;
+	case 5:
+		return 6;
+	case 6:
+		return 10;
+	case 7:
+		return 12;
+	case 8:
+		return 16;
+	}
+}
+
+struct dce10_wm_params {
+	u32 dram_channels; /* number of dram channels */
+	u32 yclk;          /* bandwidth per dram data pin in kHz */
+	u32 sclk;          /* engine clock in kHz */
+	u32 disp_clk;      /* display clock in kHz */
+	u32 src_width;     /* viewport width */
+	u32 active_time;   /* active display time in ns */
+	u32 blank_time;    /* blank time in ns */
+	bool interlaced;    /* mode is interlaced */
+	fixed20_12 vsc;    /* vertical scale ratio */
+	u32 num_heads;     /* number of active crtcs */
+	u32 bytes_per_pixel; /* bytes per pixel display + overlay */
+	u32 lb_size;       /* line buffer allocated to pipe */
+	u32 vtaps;         /* vertical scaler taps */
+};
+
+/**
+ * dce_v11_0_dram_bandwidth - get the dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the raw dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth in MBytes/s
+ */
+static u32 dce_v11_0_dram_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate raw DRAM Bandwidth */
+	fixed20_12 dram_efficiency; /* 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	dram_efficiency.full = dfixed_const(7);
+	dram_efficiency.full = dfixed_div(dram_efficiency, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, dram_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v11_0_dram_bandwidth_for_display - get the dram bandwidth for display
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dram bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth for display in MBytes/s
+ */
+static u32 dce_v11_0_dram_bandwidth_for_display(struct dce10_wm_params *wm)
+{
+	/* Calculate DRAM Bandwidth and the part allocated to display. */
+	fixed20_12 disp_dram_allocation; /* 0.3 to 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	disp_dram_allocation.full = dfixed_const(3); /* XXX worse case value 0.3 */
+	disp_dram_allocation.full = dfixed_div(disp_dram_allocation, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, disp_dram_allocation);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v11_0_data_return_bandwidth - get the data return bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the data return bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the data return bandwidth in MBytes/s
+ */
+static u32 dce_v11_0_data_return_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the display Data return Bandwidth */
+	fixed20_12 return_efficiency; /* 0.8 */
+	fixed20_12 sclk, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	sclk.full = dfixed_const(wm->sclk);
+	sclk.full = dfixed_div(sclk, a);
+	a.full = dfixed_const(10);
+	return_efficiency.full = dfixed_const(8);
+	return_efficiency.full = dfixed_div(return_efficiency, a);
+	a.full = dfixed_const(32);
+	bandwidth.full = dfixed_mul(a, sclk);
+	bandwidth.full = dfixed_mul(bandwidth, return_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v11_0_dmif_request_bandwidth - get the dmif bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dmif bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dmif bandwidth in MBytes/s
+ */
+static u32 dce_v11_0_dmif_request_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the DMIF Request Bandwidth */
+	fixed20_12 disp_clk_request_efficiency; /* 0.8 */
+	fixed20_12 disp_clk, bandwidth;
+	fixed20_12 a, b;
+
+	a.full = dfixed_const(1000);
+	disp_clk.full = dfixed_const(wm->disp_clk);
+	disp_clk.full = dfixed_div(disp_clk, a);
+	a.full = dfixed_const(32);
+	b.full = dfixed_mul(a, disp_clk);
+
+	a.full = dfixed_const(10);
+	disp_clk_request_efficiency.full = dfixed_const(8);
+	disp_clk_request_efficiency.full = dfixed_div(disp_clk_request_efficiency, a);
+
+	bandwidth.full = dfixed_mul(b, disp_clk_request_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v11_0_available_bandwidth - get the min available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the min available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the min available bandwidth in MBytes/s
+ */
+static u32 dce_v11_0_available_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the Available bandwidth. Display can use this temporarily but not in average. */
+	u32 dram_bandwidth = dce_v11_0_dram_bandwidth(wm);
+	u32 data_return_bandwidth = dce_v11_0_data_return_bandwidth(wm);
+	u32 dmif_req_bandwidth = dce_v11_0_dmif_request_bandwidth(wm);
+
+	return min(dram_bandwidth, min(data_return_bandwidth, dmif_req_bandwidth));
+}
+
+/**
+ * dce_v11_0_average_bandwidth - get the average available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the average available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the average available bandwidth in MBytes/s
+ */
+static u32 dce_v11_0_average_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the display mode Average Bandwidth
+	 * DisplayMode should contain the source and destination dimensions,
+	 * timing, etc.
+	 */
+	fixed20_12 bpp;
+	fixed20_12 line_time;
+	fixed20_12 src_width;
+	fixed20_12 bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	line_time.full = dfixed_const(wm->active_time + wm->blank_time);
+	line_time.full = dfixed_div(line_time, a);
+	bpp.full = dfixed_const(wm->bytes_per_pixel);
+	src_width.full = dfixed_const(wm->src_width);
+	bandwidth.full = dfixed_mul(src_width, bpp);
+	bandwidth.full = dfixed_mul(bandwidth, wm->vsc);
+	bandwidth.full = dfixed_div(bandwidth, line_time);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v11_0_latency_watermark - get the latency watermark
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the latency watermark (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the latency watermark in ns
+ */
+static u32 dce_v11_0_latency_watermark(struct dce10_wm_params *wm)
+{
+	/* First calculate the latency in ns */
+	u32 mc_latency = 2000; /* 2000 ns. */
+	u32 available_bandwidth = dce_v11_0_available_bandwidth(wm);
+	u32 worst_chunk_return_time = (512 * 8 * 1000) / available_bandwidth;
+	u32 cursor_line_pair_return_time = (128 * 4 * 1000) / available_bandwidth;
+	u32 dc_latency = 40000000 / wm->disp_clk; /* dc pipe latency */
+	u32 other_heads_data_return_time = ((wm->num_heads + 1) * worst_chunk_return_time) +
+		(wm->num_heads * cursor_line_pair_return_time);
+	u32 latency = mc_latency + other_heads_data_return_time + dc_latency;
+	u32 max_src_lines_per_dst_line, lb_fill_bw, line_fill_time;
+	u32 tmp, dmif_size = 12288;
+	fixed20_12 a, b, c;
+
+	if (wm->num_heads == 0)
+		return 0;
+
+	a.full = dfixed_const(2);
+	b.full = dfixed_const(1);
+	if ((wm->vsc.full > a.full) ||
+	    ((wm->vsc.full > b.full) && (wm->vtaps >= 3)) ||
+	    (wm->vtaps >= 5) ||
+	    ((wm->vsc.full >= a.full) && wm->interlaced))
+		max_src_lines_per_dst_line = 4;
+	else
+		max_src_lines_per_dst_line = 2;
+
+	a.full = dfixed_const(available_bandwidth);
+	b.full = dfixed_const(wm->num_heads);
+	a.full = dfixed_div(a, b);
+	tmp = div_u64((u64) dmif_size * (u64) wm->disp_clk, mc_latency + 512);
+	tmp = min(dfixed_trunc(a), tmp);
+
+	lb_fill_bw = min(tmp, wm->disp_clk * wm->bytes_per_pixel / 1000);
+
+	a.full = dfixed_const(max_src_lines_per_dst_line * wm->src_width * wm->bytes_per_pixel);
+	b.full = dfixed_const(1000);
+	c.full = dfixed_const(lb_fill_bw);
+	b.full = dfixed_div(c, b);
+	a.full = dfixed_div(a, b);
+	line_fill_time = dfixed_trunc(a);
+
+	if (line_fill_time < wm->active_time)
+		return latency;
+	else
+		return latency + (line_fill_time - wm->active_time);
+
+}
+
+/**
+ * dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display - check
+ * average and available dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display(struct dce10_wm_params *wm)
+{
+	if (dce_v11_0_average_bandwidth(wm) <=
+	    (dce_v11_0_dram_bandwidth_for_display(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v11_0_average_bandwidth_vs_available_bandwidth - check
+ * average and available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * available bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v11_0_average_bandwidth_vs_available_bandwidth(struct dce10_wm_params *wm)
+{
+	if (dce_v11_0_average_bandwidth(wm) <=
+	    (dce_v11_0_available_bandwidth(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v11_0_check_latency_hiding - check latency hiding
+ *
+ * @wm: watermark calculation data
+ *
+ * Check latency hiding (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v11_0_check_latency_hiding(struct dce10_wm_params *wm)
+{
+	u32 lb_partitions = wm->lb_size / wm->src_width;
+	u32 line_time = wm->active_time + wm->blank_time;
+	u32 latency_tolerant_lines;
+	u32 latency_hiding;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1);
+	if (wm->vsc.full > a.full)
+		latency_tolerant_lines = 1;
+	else {
+		if (lb_partitions <= (wm->vtaps + 1))
+			latency_tolerant_lines = 1;
+		else
+			latency_tolerant_lines = 2;
+	}
+
+	latency_hiding = (latency_tolerant_lines * line_time + wm->blank_time);
+
+	if (dce_v11_0_latency_watermark(wm) <= latency_hiding)
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v11_0_program_watermarks - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @lb_size: line buffer size
+ * @num_heads: number of display controllers in use
+ *
+ * Calculate and program the display watermarks for the
+ * selected display controller (CIK).
+ */
+static void dce_v11_0_program_watermarks(struct amdgpu_device *adev,
+					struct amdgpu_crtc *amdgpu_crtc,
+					u32 lb_size, u32 num_heads)
+{
+	struct drm_display_mode *mode = &amdgpu_crtc->base.mode;
+	struct dce10_wm_params wm_low, wm_high;
+	u32 active_time;
+	u32 line_time = 0;
+	u32 latency_watermark_a = 0, latency_watermark_b = 0;
+	u32 tmp, wm_mask, lb_vblank_lead_lines = 0;
+
+	if (amdgpu_crtc->base.enabled && num_heads && mode) {
+		active_time = (u32) div_u64((u64)mode->crtc_hdisplay * 1000000,
+					    (u32)mode->clock);
+		line_time = (u32) div_u64((u64)mode->crtc_htotal * 1000000,
+					  (u32)mode->clock);
+		line_time = min(line_time, (u32)65535);
+
+		/* watermark for high clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_high.yclk =
+				amdgpu_dpm_get_mclk(adev, false) * 10;
+			wm_high.sclk =
+				amdgpu_dpm_get_sclk(adev, false) * 10;
+		} else {
+			wm_high.yclk = adev->pm.current_mclk * 10;
+			wm_high.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_high.disp_clk = mode->clock;
+		wm_high.src_width = mode->crtc_hdisplay;
+		wm_high.active_time = active_time;
+		wm_high.blank_time = line_time - wm_high.active_time;
+		wm_high.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_high.interlaced = true;
+		wm_high.vsc = amdgpu_crtc->vsc;
+		wm_high.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_high.vtaps = 2;
+		wm_high.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_high.lb_size = lb_size;
+		wm_high.dram_channels = cik_get_number_of_dram_channels(adev);
+		wm_high.num_heads = num_heads;
+
+		/* set for high clocks */
+		latency_watermark_a = min(dce_v11_0_latency_watermark(&wm_high), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_high) ||
+		    !dce_v11_0_average_bandwidth_vs_available_bandwidth(&wm_high) ||
+		    !dce_v11_0_check_latency_hiding(&wm_high) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+		}
+
+		/* watermark for low clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_low.yclk =
+				amdgpu_dpm_get_mclk(adev, true) * 10;
+			wm_low.sclk =
+				amdgpu_dpm_get_sclk(adev, true) * 10;
+		} else {
+			wm_low.yclk = adev->pm.current_mclk * 10;
+			wm_low.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_low.disp_clk = mode->clock;
+		wm_low.src_width = mode->crtc_hdisplay;
+		wm_low.active_time = active_time;
+		wm_low.blank_time = line_time - wm_low.active_time;
+		wm_low.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_low.interlaced = true;
+		wm_low.vsc = amdgpu_crtc->vsc;
+		wm_low.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_low.vtaps = 2;
+		wm_low.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_low.lb_size = lb_size;
+		wm_low.dram_channels = cik_get_number_of_dram_channels(adev);
+		wm_low.num_heads = num_heads;
+
+		/* set for low clocks */
+		latency_watermark_b = min(dce_v11_0_latency_watermark(&wm_low), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_low) ||
+		    !dce_v11_0_average_bandwidth_vs_available_bandwidth(&wm_low) ||
+		    !dce_v11_0_check_latency_hiding(&wm_low) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+		}
+		lb_vblank_lead_lines = DIV_ROUND_UP(lb_size, mode->crtc_hdisplay);
+	}
+
+	/* select wm A */
+	wm_mask = RREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(wm_mask, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, 1);
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	tmp = RREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, latency_watermark_a);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, line_time);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	/* select wm B */
+	tmp = REG_SET_FIELD(wm_mask, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, 2);
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	tmp = RREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, latency_watermark_b);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, line_time);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	/* restore original selection */
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, wm_mask);
+
+	/* save values for DPM */
+	amdgpu_crtc->line_time = line_time;
+	amdgpu_crtc->wm_high = latency_watermark_a;
+	amdgpu_crtc->wm_low = latency_watermark_b;
+	/* Save number of lines the linebuffer leads before the scanout */
+	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
+}
+
+/**
+ * dce_v11_0_bandwidth_update - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Calculate and program the display watermarks and line
+ * buffer allocation (CIK).
+ */
+static void dce_v11_0_bandwidth_update(struct amdgpu_device *adev)
+{
+	struct drm_display_mode *mode = NULL;
+	u32 num_heads = 0, lb_size;
+	int i;
+
+	amdgpu_display_update_priority(adev);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i]->base.enabled)
+			num_heads++;
+	}
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		mode = &adev->mode_info.crtcs[i]->base.mode;
+		lb_size = dce_v11_0_line_buffer_adjust(adev, adev->mode_info.crtcs[i], mode);
+		dce_v11_0_program_watermarks(adev, adev->mode_info.crtcs[i],
+					    lb_size, num_heads);
+	}
+}
+
+static void dce_v11_0_audio_get_connected_pins(struct amdgpu_device *adev)
+{
+	int i;
+	u32 offset, tmp;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		offset = adev->mode_info.audio.pin[i].offset;
+		tmp = RREG32_AUDIO_ENDPT(offset,
+					 ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT);
+		if (((tmp &
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY_MASK) >>
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY__SHIFT) == 1)
+			adev->mode_info.audio.pin[i].connected = false;
+		else
+			adev->mode_info.audio.pin[i].connected = true;
+	}
+}
+
+static struct amdgpu_audio_pin *dce_v11_0_audio_get_pin(struct amdgpu_device *adev)
+{
+	int i;
+
+	dce_v11_0_audio_get_connected_pins(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		if (adev->mode_info.audio.pin[i].connected)
+			return &adev->mode_info.audio.pin[i];
+	}
+	DRM_ERROR("No connected audio pins found!\n");
+	return NULL;
+}
+
+static void dce_v11_0_afmt_audio_select_pin(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	tmp = RREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_SRC_CONTROL, AFMT_AUDIO_SRC_SELECT, dig->afmt->pin->id);
+	WREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset, tmp);
+}
+
+static void dce_v11_0_audio_write_latency_fields(struct drm_encoder *encoder,
+						struct drm_display_mode *mode)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u32 tmp;
+	int interlace = 0;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		interlace = 1;
+	if (connector->latency_present[interlace]) {
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    VIDEO_LIPSYNC, connector->video_latency[interlace]);
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    AUDIO_LIPSYNC, connector->audio_latency[interlace]);
+	} else {
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    VIDEO_LIPSYNC, 0);
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    AUDIO_LIPSYNC, 0);
+	}
+	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+			   ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC, tmp);
+}
+
+static void dce_v11_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u32 tmp;
+	u8 *sadb = NULL;
+	int sad_count;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector_edid(connector), &sadb);
+	if (sad_count < 0) {
+		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sad_count);
+		sad_count = 0;
+	}
+
+	/* program the speaker allocation */
+	tmp = RREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+				 ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER);
+	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			    DP_CONNECTION, 0);
+	/* set HDMI mode */
+	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			    HDMI_CONNECTION, 1);
+	if (sad_count)
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+				    SPEAKER_ALLOCATION, sadb[0]);
+	else
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+				    SPEAKER_ALLOCATION, 5); /* stereo */
+	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+			   ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER, tmp);
+
+	kfree(sadb);
+}
+
+static void dce_v11_0_audio_write_sad_regs(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	struct cea_sad *sads;
+	int i, sad_count;
+
+	static const u16 eld_reg_to_type[][2] = {
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0, HDMI_AUDIO_CODING_TYPE_PCM },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR1, HDMI_AUDIO_CODING_TYPE_AC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR2, HDMI_AUDIO_CODING_TYPE_MPEG1 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR3, HDMI_AUDIO_CODING_TYPE_MP3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR4, HDMI_AUDIO_CODING_TYPE_MPEG2 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR5, HDMI_AUDIO_CODING_TYPE_AAC_LC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR6, HDMI_AUDIO_CODING_TYPE_DTS },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR7, HDMI_AUDIO_CODING_TYPE_ATRAC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR9, HDMI_AUDIO_CODING_TYPE_EAC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR10, HDMI_AUDIO_CODING_TYPE_DTS_HD },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR11, HDMI_AUDIO_CODING_TYPE_MLP },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO },
+	};
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_sad(amdgpu_connector_edid(connector), &sads);
+	if (sad_count <= 0) {
+		DRM_ERROR("Couldn't read SADs: %d\n", sad_count);
+		return;
+	}
+	BUG_ON(!sads);
+
+	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
+		u32 tmp = 0;
+		u8 stereo_freqs = 0;
+		int max_channels = -1;
+		int j;
+
+		for (j = 0; j < sad_count; j++) {
+			struct cea_sad *sad = &sads[j];
+
+			if (sad->format == eld_reg_to_type[i][1]) {
+				if (sad->channels > max_channels) {
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							    MAX_CHANNELS, sad->channels);
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							    DESCRIPTOR_BYTE_2, sad->byte2);
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							    SUPPORTED_FREQUENCIES, sad->freq);
+					max_channels = sad->channels;
+				}
+
+				if (sad->format == HDMI_AUDIO_CODING_TYPE_PCM)
+					stereo_freqs |= sad->freq;
+				else
+					break;
+			}
+		}
+
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+				    SUPPORTED_FREQUENCIES_STEREO, stereo_freqs);
+		WREG32_AUDIO_ENDPT(dig->afmt->pin->offset, eld_reg_to_type[i][0], tmp);
+	}
+
+	kfree(sads);
+}
+
+static void dce_v11_0_audio_enable(struct amdgpu_device *adev,
+				  struct amdgpu_audio_pin *pin,
+				  bool enable)
+{
+	if (!pin)
+		return;
+
+	WREG32_AUDIO_ENDPT(pin->offset, ixAZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			   enable ? AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL__AUDIO_ENABLED_MASK : 0);
+}
+
+static const u32 pin_offsets[] =
+{
+	AUD0_REGISTER_OFFSET,
+	AUD1_REGISTER_OFFSET,
+	AUD2_REGISTER_OFFSET,
+	AUD3_REGISTER_OFFSET,
+	AUD4_REGISTER_OFFSET,
+	AUD5_REGISTER_OFFSET,
+	AUD6_REGISTER_OFFSET,
+	AUD7_REGISTER_OFFSET,
+};
+
+static int dce_v11_0_audio_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!amdgpu_audio)
+		return 0;
+
+	adev->mode_info.audio.enabled = true;
+
+	switch (adev->asic_type) {
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		adev->mode_info.audio.num_pins = 7;
+		break;
+	case CHIP_POLARIS10:
+	case CHIP_VEGAM:
+		adev->mode_info.audio.num_pins = 8;
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+		adev->mode_info.audio.num_pins = 6;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		adev->mode_info.audio.pin[i].channels = -1;
+		adev->mode_info.audio.pin[i].rate = -1;
+		adev->mode_info.audio.pin[i].bits_per_sample = -1;
+		adev->mode_info.audio.pin[i].status_bits = 0;
+		adev->mode_info.audio.pin[i].category_code = 0;
+		adev->mode_info.audio.pin[i].connected = false;
+		adev->mode_info.audio.pin[i].offset = pin_offsets[i];
+		adev->mode_info.audio.pin[i].id = i;
+		/* disable audio.  it will be set up later */
+		/* XXX remove once we switch to ip funcs */
+		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	return 0;
+}
+
+static void dce_v11_0_audio_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!amdgpu_audio)
+		return;
+
+	if (!adev->mode_info.audio.enabled)
+		return;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
+		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+
+	adev->mode_info.audio.enabled = false;
+}
+
+/*
+ * update the N and CTS parameters for a given pixel clock rate
+ */
+static void dce_v11_0_afmt_update_ACR(struct drm_encoder *encoder, uint32_t clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_afmt_acr acr = amdgpu_afmt_acr(clock);
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	tmp = RREG32(mmHDMI_ACR_32_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_0, HDMI_ACR_CTS_32, acr.cts_32khz);
+	WREG32(mmHDMI_ACR_32_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_32_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_1, HDMI_ACR_N_32, acr.n_32khz);
+	WREG32(mmHDMI_ACR_32_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_44_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_0, HDMI_ACR_CTS_44, acr.cts_44_1khz);
+	WREG32(mmHDMI_ACR_44_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_44_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_1, HDMI_ACR_N_44, acr.n_44_1khz);
+	WREG32(mmHDMI_ACR_44_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_48_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_0, HDMI_ACR_CTS_48, acr.cts_48khz);
+	WREG32(mmHDMI_ACR_48_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_48_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_1, HDMI_ACR_N_48, acr.n_48khz);
+	WREG32(mmHDMI_ACR_48_1 + dig->afmt->offset, tmp);
+
+}
+
+/*
+ * build a HDMI Video Info Frame
+ */
+static void dce_v11_0_afmt_update_avi_infoframe(struct drm_encoder *encoder,
+					       void *buffer, size_t size)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	uint8_t *frame = buffer + 3;
+	uint8_t *header = buffer;
+
+	WREG32(mmAFMT_AVI_INFO0 + dig->afmt->offset,
+		frame[0x0] | (frame[0x1] << 8) | (frame[0x2] << 16) | (frame[0x3] << 24));
+	WREG32(mmAFMT_AVI_INFO1 + dig->afmt->offset,
+		frame[0x4] | (frame[0x5] << 8) | (frame[0x6] << 16) | (frame[0x7] << 24));
+	WREG32(mmAFMT_AVI_INFO2 + dig->afmt->offset,
+		frame[0x8] | (frame[0x9] << 8) | (frame[0xA] << 16) | (frame[0xB] << 24));
+	WREG32(mmAFMT_AVI_INFO3 + dig->afmt->offset,
+		frame[0xC] | (frame[0xD] << 8) | (header[1] << 24));
+}
+
+static void dce_v11_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	u32 dto_phase = 24 * 1000;
+	u32 dto_modulo = clock;
+	u32 tmp;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* XXX two dtos; generally use dto0 for hdmi */
+	/* Express [24MHz / target pixel clock] as an exact rational
+	 * number (coefficient of two integer numbers.  DCCG_AUDIO_DTOx_PHASE
+	 * is the numerator, DCCG_AUDIO_DTOx_MODULE is the denominator
+	 */
+	tmp = RREG32(mmDCCG_AUDIO_DTO_SOURCE);
+	tmp = REG_SET_FIELD(tmp, DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL,
+			    amdgpu_crtc->crtc_id);
+	WREG32(mmDCCG_AUDIO_DTO_SOURCE, tmp);
+	WREG32(mmDCCG_AUDIO_DTO0_PHASE, dto_phase);
+	WREG32(mmDCCG_AUDIO_DTO0_MODULE, dto_modulo);
+}
+
+/*
+ * update the info frames with the data from the current display mode
+ */
+static void dce_v11_0_afmt_setmode(struct drm_encoder *encoder,
+				  struct drm_display_mode *mode)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	u8 buffer[HDMI_INFOFRAME_HEADER_SIZE + HDMI_AVI_INFOFRAME_SIZE];
+	struct hdmi_avi_infoframe frame;
+	ssize_t err;
+	u32 tmp;
+	int bpc = 8;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (!dig->afmt->enabled)
+		return;
+
+	/* hdmi deep color mode general control packets setup, if bpc > 8 */
+	if (encoder->crtc) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+		bpc = amdgpu_crtc->bpc;
+	}
+
+	/* disable audio prior to setting up hw */
+	dig->afmt->pin = dce_v11_0_audio_get_pin(adev);
+	dce_v11_0_audio_enable(adev, dig->afmt->pin, false);
+
+	dce_v11_0_audio_set_dto(encoder, mode->clock);
+
+	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1);
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp); /* send null packets when required */
+
+	WREG32(mmAFMT_AUDIO_CRC_CONTROL + dig->afmt->offset, 0x1000);
+
+	tmp = RREG32(mmHDMI_CONTROL + dig->afmt->offset);
+	switch (bpc) {
+	case 0:
+	case 6:
+	case 8:
+	case 16:
+	default:
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 0);
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
+		DRM_DEBUG("%s: Disabling hdmi deep color for %d bpc.\n",
+			  connector->name, bpc);
+		break;
+	case 10:
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 1);
+		DRM_DEBUG("%s: Enabling hdmi deep color 30 for 10 bpc.\n",
+			  connector->name);
+		break;
+	case 12:
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 2);
+		DRM_DEBUG("%s: Enabling hdmi deep color 36 for 12 bpc.\n",
+			  connector->name);
+		break;
+	}
+	WREG32(mmHDMI_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1); /* send null packets when required */
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, 1); /* send general control packets */
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, 1); /* send general control packets every frame */
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	/* enable audio info frames (frames won't be set until audio is enabled) */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
+	/* required for audio info values to be updated */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_CONT, 1);
+	WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	/* required for audio info values to be updated */
+	tmp = REG_SET_FIELD(tmp, AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
+	WREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
+	/* anything other than 0 */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, 2);
+	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
+
+	WREG32(mmHDMI_GC + dig->afmt->offset, 0); /* unset HDMI_GC_AVMUTE */
+
+	tmp = RREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	/* set the default audio delay */
+	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, 1);
+	/* should be suffient for all audio modes and small enough for all hblanks */
+	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_PACKETS_PER_LINE, 3);
+	WREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	/* allow 60958 channel status fields to be updated */
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset);
+	if (bpc > 8)
+		/* clear SW CTS value */
+		tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, 0);
+	else
+		/* select SW CTS value */
+		tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, 1);
+	/* allow hw to sent ACR packets when required */
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, 1);
+	WREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	dce_v11_0_afmt_update_ACR(encoder, mode->clock);
+
+	tmp = RREG32(mmAFMT_60958_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_0, AFMT_60958_CS_CHANNEL_NUMBER_L, 1);
+	WREG32(mmAFMT_60958_0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_60958_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, 2);
+	WREG32(mmAFMT_60958_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_60958_2 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_2, 3);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_3, 4);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_4, 5);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_5, 6);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_6, 7);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_7, 8);
+	WREG32(mmAFMT_60958_2 + dig->afmt->offset, tmp);
+
+	dce_v11_0_audio_write_speaker_allocation(encoder);
+
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL2 + dig->afmt->offset,
+	       (0xff << AFMT_AUDIO_PACKET_CONTROL2__AFMT_AUDIO_CHANNEL_ENABLE__SHIFT));
+
+	dce_v11_0_afmt_audio_select_pin(encoder);
+	dce_v11_0_audio_write_sad_regs(encoder);
+	dce_v11_0_audio_write_latency_fields(encoder, mode);
+
+	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, mode, false);
+	if (err < 0) {
+		DRM_ERROR("failed to setup AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	err = hdmi_avi_infoframe_pack(&frame, buffer, sizeof(buffer));
+	if (err < 0) {
+		DRM_ERROR("failed to pack AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	dce_v11_0_afmt_update_avi_infoframe(encoder, buffer, sizeof(buffer));
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	/* enable AVI info frames */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_SEND, 1);
+	/* required for audio info values to be updated */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_CONT, 1);
+	WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AVI_INFO_LINE, 2);
+	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	/* send audio packets */
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, 1);
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	WREG32(mmAFMT_RAMP_CONTROL0 + dig->afmt->offset, 0x00FFFFFF);
+	WREG32(mmAFMT_RAMP_CONTROL1 + dig->afmt->offset, 0x007FFFFF);
+	WREG32(mmAFMT_RAMP_CONTROL2 + dig->afmt->offset, 0x00000001);
+	WREG32(mmAFMT_RAMP_CONTROL3 + dig->afmt->offset, 0x00000001);
+
+	/* enable audio after to setting up hw */
+	dce_v11_0_audio_enable(adev, dig->afmt->pin, true);
+}
+
+static void dce_v11_0_afmt_enable(struct drm_encoder *encoder, bool enable)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (enable && dig->afmt->enabled)
+		return;
+	if (!enable && !dig->afmt->enabled)
+		return;
+
+	if (!enable && dig->afmt->pin) {
+		dce_v11_0_audio_enable(adev, dig->afmt->pin, false);
+		dig->afmt->pin = NULL;
+	}
+
+	dig->afmt->enabled = enable;
+
+	DRM_DEBUG("%sabling AFMT interface @ 0x%04X for encoder 0x%x\n",
+		  enable ? "En" : "Dis", dig->afmt->offset, amdgpu_encoder->encoder_id);
+}
+
+static int dce_v11_0_afmt_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++)
+		adev->mode_info.afmt[i] = NULL;
+
+	/* DCE11 has audio blocks tied to DIG encoders */
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		adev->mode_info.afmt[i] = kzalloc(sizeof(struct amdgpu_afmt), GFP_KERNEL);
+		if (adev->mode_info.afmt[i]) {
+			adev->mode_info.afmt[i]->offset = dig_offsets[i];
+			adev->mode_info.afmt[i]->id = i;
+		} else {
+			int j;
+			for (j = 0; j < i; j++) {
+				kfree(adev->mode_info.afmt[j]);
+				adev->mode_info.afmt[j] = NULL;
+			}
+			return -ENOMEM;
+		}
+	}
+	return 0;
+}
+
+static void dce_v11_0_afmt_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		kfree(adev->mode_info.afmt[i]);
+		adev->mode_info.afmt[i] = NULL;
+	}
+}
+
+static const u32 vga_control_regs[6] =
+{
+	mmD1VGA_CONTROL,
+	mmD2VGA_CONTROL,
+	mmD3VGA_CONTROL,
+	mmD4VGA_CONTROL,
+	mmD5VGA_CONTROL,
+	mmD6VGA_CONTROL,
+};
+
+static void dce_v11_0_vga_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 vga_control;
+
+	vga_control = RREG32(vga_control_regs[amdgpu_crtc->crtc_id]) & ~1;
+	if (enable)
+		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control | 1);
+	else
+		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control);
+}
+
+static void dce_v11_0_grph_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (enable)
+		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 1);
+	else
+		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 0);
+}
+
+static int dce_v11_0_crtc_do_set_base(struct drm_crtc *crtc,
+				     struct drm_framebuffer *fb,
+				     int x, int y, int atomic)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_framebuffer *target_fb;
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *abo;
+	uint64_t fb_location, tiling_flags;
+	uint32_t fb_format, fb_pitch_pixels;
+	u32 fb_swap = REG_SET_FIELD(0, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP, ENDIAN_NONE);
+	u32 pipe_config;
+	u32 tmp, viewport_w, viewport_h;
+	int r;
+	bool bypass_lut = false;
+	struct drm_format_name_buf format_name;
+
+	/* no fb bound */
+	if (!atomic && !crtc->primary->fb) {
+		DRM_DEBUG_KMS("No FB bound\n");
+		return 0;
+	}
+
+	if (atomic)
+		target_fb = fb;
+	else
+		target_fb = crtc->primary->fb;
+
+	/* If atomic, assume fb object is pinned & idle & fenced and
+	 * just update base pointers
+	 */
+	obj = target_fb->obj[0];
+	abo = gem_to_amdgpu_bo(obj);
+	r = amdgpu_bo_reserve(abo, false);
+	if (unlikely(r != 0))
+		return r;
+
+	if (!atomic) {
+		r = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM);
+		if (unlikely(r != 0)) {
+			amdgpu_bo_unreserve(abo);
+			return -EINVAL;
+		}
+	}
+	fb_location = amdgpu_bo_gpu_offset(abo);
+
+	amdgpu_bo_get_tiling_flags(abo, &tiling_flags);
+	amdgpu_bo_unreserve(abo);
+
+	pipe_config = AMDGPU_TILING_GET(tiling_flags, PIPE_CONFIG);
+
+	switch (target_fb->format->format) {
+	case DRM_FORMAT_C8:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 0);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+		break;
+	case DRM_FORMAT_XRGB4444:
+	case DRM_FORMAT_ARGB4444:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 2);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_XRGB1555:
+	case DRM_FORMAT_ARGB1555:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_BGRX5551:
+	case DRM_FORMAT_BGRA5551:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 5);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_RGB565:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 1);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_XRGB8888:
+	case DRM_FORMAT_ARGB8888:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		break;
+	case DRM_FORMAT_XRGB2101010:
+	case DRM_FORMAT_ARGB2101010:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 1);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	case DRM_FORMAT_BGRX1010102:
+	case DRM_FORMAT_BGRA1010102:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 4);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	default:
+		DRM_ERROR("Unsupported screen format %s\n",
+		          drm_get_format_name(target_fb->format->format, &format_name));
+		return -EINVAL;
+	}
+
+	if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_2D_TILED_THIN1) {
+		unsigned bankw, bankh, mtaspect, tile_split, num_banks;
+
+		bankw = AMDGPU_TILING_GET(tiling_flags, BANK_WIDTH);
+		bankh = AMDGPU_TILING_GET(tiling_flags, BANK_HEIGHT);
+		mtaspect = AMDGPU_TILING_GET(tiling_flags, MACRO_TILE_ASPECT);
+		tile_split = AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT);
+		num_banks = AMDGPU_TILING_GET(tiling_flags, NUM_BANKS);
+
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_NUM_BANKS, num_banks);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_ARRAY_MODE,
+					  ARRAY_2D_TILED_THIN1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_TILE_SPLIT,
+					  tile_split);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_BANK_WIDTH, bankw);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_BANK_HEIGHT, bankh);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_MACRO_TILE_ASPECT,
+					  mtaspect);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_MICRO_TILE_MODE,
+					  ADDR_SURF_MICRO_TILING_DISPLAY);
+	} else if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_1D_TILED_THIN1) {
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_ARRAY_MODE,
+					  ARRAY_1D_TILED_THIN1);
+	}
+
+	fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_PIPE_CONFIG,
+				  pipe_config);
+
+	dce_v11_0_vga_enable(crtc, false);
+
+	/* Make sure surface address is updated at vertical blank rather than
+	 * horizontal blank
+	 */
+	tmp = RREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, GRPH_FLIP_CONTROL,
+			    GRPH_SURFACE_UPDATE_H_RETRACE_EN, 0);
+	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32)fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32) fb_location & GRPH_SECONDARY_SURFACE_ADDRESS__GRPH_SECONDARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
+	WREG32(mmGRPH_SWAP_CNTL + amdgpu_crtc->crtc_offset, fb_swap);
+
+	/*
+	 * The LUT only has 256 slots for indexing by a 8 bpc fb. Bypass the LUT
+	 * for > 8 bpc scanout to avoid truncation of fb indices to 8 msb's, to
+	 * retain the full precision throughout the pipeline.
+	 */
+	tmp = RREG32(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset);
+	if (bypass_lut)
+		tmp = REG_SET_FIELD(tmp, GRPH_LUT_10BIT_BYPASS, GRPH_LUT_10BIT_BYPASS_EN, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, GRPH_LUT_10BIT_BYPASS, GRPH_LUT_10BIT_BYPASS_EN, 0);
+	WREG32(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset, tmp);
+
+	if (bypass_lut)
+		DRM_DEBUG_KMS("Bypassing hardware LUT due to 10 bit fb scanout.\n");
+
+	WREG32(mmGRPH_SURFACE_OFFSET_X + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_SURFACE_OFFSET_Y + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_Y_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_END + amdgpu_crtc->crtc_offset, target_fb->width);
+	WREG32(mmGRPH_Y_END + amdgpu_crtc->crtc_offset, target_fb->height);
+
+	fb_pitch_pixels = target_fb->pitches[0] / target_fb->format->cpp[0];
+	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset, fb_pitch_pixels);
+
+	dce_v11_0_grph_enable(crtc, true);
+
+	WREG32(mmLB_DESKTOP_HEIGHT + amdgpu_crtc->crtc_offset,
+	       target_fb->height);
+
+	x &= ~3;
+	y &= ~1;
+	WREG32(mmVIEWPORT_START + amdgpu_crtc->crtc_offset,
+	       (x << 16) | y);
+	viewport_w = crtc->mode.hdisplay;
+	viewport_h = (crtc->mode.vdisplay + 1) & ~1;
+	WREG32(mmVIEWPORT_SIZE + amdgpu_crtc->crtc_offset,
+	       (viewport_w << 16) | viewport_h);
+
+	/* set pageflip to happen anywhere in vblank interval */
+	WREG32(mmCRTC_MASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 0);
+
+	if (!atomic && fb && fb != crtc->primary->fb) {
+		abo = gem_to_amdgpu_bo(fb->obj[0]);
+		r = amdgpu_bo_reserve(abo, true);
+		if (unlikely(r != 0))
+			return r;
+		amdgpu_bo_unpin(abo);
+		amdgpu_bo_unreserve(abo);
+	}
+
+	/* Bytes per pixel may have changed */
+	dce_v11_0_bandwidth_update(adev);
+
+	return 0;
+}
+
+static void dce_v11_0_set_interleave(struct drm_crtc *crtc,
+				     struct drm_display_mode *mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	u32 tmp;
+
+	tmp = RREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset);
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		tmp = REG_SET_FIELD(tmp, LB_DATA_FORMAT, INTERLEAVE_EN, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, LB_DATA_FORMAT, INTERLEAVE_EN, 0);
+	WREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static void dce_v11_0_crtc_load_lut(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u16 *r, *g, *b;
+	int i;
+	u32 tmp;
+
+	DRM_DEBUG_KMS("%d\n", amdgpu_crtc->crtc_id);
+
+	tmp = RREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, INPUT_CSC_CONTROL, INPUT_CSC_GRPH_MODE, 0);
+	WREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_BYPASS, 1);
+	WREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, INPUT_GAMMA_CONTROL, GRPH_INPUT_GAMMA_MODE, 0);
+	WREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	WREG32(mmDC_LUT_CONTROL + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_BLACK_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_RED + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_WHITE_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_RED + amdgpu_crtc->crtc_offset, 0xffff);
+
+	WREG32(mmDC_LUT_RW_MODE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_WRITE_EN_MASK + amdgpu_crtc->crtc_offset, 0x00000007);
+
+	WREG32(mmDC_LUT_RW_INDEX + amdgpu_crtc->crtc_offset, 0);
+	r = crtc->gamma_store;
+	g = r + crtc->gamma_size;
+	b = g + crtc->gamma_size;
+	for (i = 0; i < 256; i++) {
+		WREG32(mmDC_LUT_30_COLOR + amdgpu_crtc->crtc_offset,
+		       ((*r++ & 0xffc0) << 14) |
+		       ((*g++ & 0xffc0) << 4) |
+		       (*b++ >> 6));
+	}
+
+	tmp = RREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, GRPH_DEGAMMA_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, CURSOR_DEGAMMA_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, CURSOR2_DEGAMMA_MODE, 0);
+	WREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, GAMUT_REMAP_CONTROL, GRPH_GAMUT_REMAP_MODE, 0);
+	WREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, REGAMMA_CONTROL, GRPH_REGAMMA_MODE, 0);
+	WREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, OUTPUT_CSC_CONTROL, OUTPUT_CSC_GRPH_MODE, 0);
+	WREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	/* XXX match this to the depth of the crtc fmt block, move to modeset? */
+	WREG32(mmDENORM_CONTROL + amdgpu_crtc->crtc_offset, 0);
+	/* XXX this only needs to be programmed once per crtc at startup,
+	 * not sure where the best place for it is
+	 */
+	tmp = RREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, ALPHA_CONTROL, CURSOR_ALPHA_BLND_ENA, 1);
+	WREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static int dce_v11_0_pick_dig_encoder(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		if (dig->linkb)
+			return 1;
+		else
+			return 0;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		if (dig->linkb)
+			return 3;
+		else
+			return 2;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+		if (dig->linkb)
+			return 5;
+		else
+			return 4;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		return 6;
+		break;
+	default:
+		DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
+		return 0;
+	}
+}
+
+/**
+ * dce_v11_0_pick_pll - Allocate a PPLL for use by the crtc.
+ *
+ * @crtc: drm crtc
+ *
+ * Returns the PPLL (Pixel PLL) to be used by the crtc.  For DP monitors
+ * a single PPLL can be used for all DP crtcs/encoders.  For non-DP
+ * monitors a dedicated PPLL must be used.  If a particular board has
+ * an external DP PLL, return ATOM_PPLL_INVALID to skip PLL programming
+ * as there is no need to program the PLL itself.  If we are not able to
+ * allocate a PLL, return ATOM_PPLL_INVALID to skip PLL programming to
+ * avoid messing up an existing monitor.
+ *
+ * Asic specific PLL information
+ *
+ * DCE 10.x
+ * Tonga
+ * - PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP)
+ * CI
+ * - PPLL0, PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP) and DAC
+ *
+ */
+static u32 dce_v11_0_pick_pll(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 pll_in_use;
+	int pll;
+
+	if ((adev->asic_type == CHIP_POLARIS10) ||
+	    (adev->asic_type == CHIP_POLARIS11) ||
+	    (adev->asic_type == CHIP_POLARIS12) ||
+	    (adev->asic_type == CHIP_VEGAM)) {
+		struct amdgpu_encoder *amdgpu_encoder =
+			to_amdgpu_encoder(amdgpu_crtc->encoder);
+		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder)))
+			return ATOM_DP_DTO;
+
+		switch (amdgpu_encoder->encoder_id) {
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+			if (dig->linkb)
+				return ATOM_COMBOPHY_PLL1;
+			else
+				return ATOM_COMBOPHY_PLL0;
+			break;
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+			if (dig->linkb)
+				return ATOM_COMBOPHY_PLL3;
+			else
+				return ATOM_COMBOPHY_PLL2;
+			break;
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+			if (dig->linkb)
+				return ATOM_COMBOPHY_PLL5;
+			else
+				return ATOM_COMBOPHY_PLL4;
+			break;
+		default:
+			DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
+			return ATOM_PPLL_INVALID;
+		}
+	}
+
+	if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder))) {
+		if (adev->clock.dp_extclk)
+			/* skip PPLL programming if using ext clock */
+			return ATOM_PPLL_INVALID;
+		else {
+			/* use the same PPLL for all DP monitors */
+			pll = amdgpu_pll_get_shared_dp_ppll(crtc);
+			if (pll != ATOM_PPLL_INVALID)
+				return pll;
+		}
+	} else {
+		/* use the same PPLL for all monitors with the same clock */
+		pll = amdgpu_pll_get_shared_nondp_ppll(crtc);
+		if (pll != ATOM_PPLL_INVALID)
+			return pll;
+	}
+
+	/* XXX need to determine what plls are available on each DCE11 part */
+	pll_in_use = amdgpu_pll_get_use_mask(crtc);
+	if (adev->asic_type == CHIP_CARRIZO || adev->asic_type == CHIP_STONEY) {
+		if (!(pll_in_use & (1 << ATOM_PPLL1)))
+			return ATOM_PPLL1;
+		if (!(pll_in_use & (1 << ATOM_PPLL0)))
+			return ATOM_PPLL0;
+		DRM_ERROR("unable to allocate a PPLL\n");
+		return ATOM_PPLL_INVALID;
+	} else {
+		if (!(pll_in_use & (1 << ATOM_PPLL2)))
+			return ATOM_PPLL2;
+		if (!(pll_in_use & (1 << ATOM_PPLL1)))
+			return ATOM_PPLL1;
+		if (!(pll_in_use & (1 << ATOM_PPLL0)))
+			return ATOM_PPLL0;
+		DRM_ERROR("unable to allocate a PPLL\n");
+		return ATOM_PPLL_INVALID;
+	}
+	return ATOM_PPLL_INVALID;
+}
+
+static void dce_v11_0_lock_cursor(struct drm_crtc *crtc, bool lock)
+{
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	uint32_t cur_lock;
+
+	cur_lock = RREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset);
+	if (lock)
+		cur_lock = REG_SET_FIELD(cur_lock, CUR_UPDATE, CURSOR_UPDATE_LOCK, 1);
+	else
+		cur_lock = REG_SET_FIELD(cur_lock, CUR_UPDATE, CURSOR_UPDATE_LOCK, 0);
+	WREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset, cur_lock);
+}
+
+static void dce_v11_0_hide_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	u32 tmp;
+
+	tmp = RREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 0);
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static void dce_v11_0_show_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	u32 tmp;
+
+	WREG32(mmCUR_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(amdgpu_crtc->cursor_addr));
+	WREG32(mmCUR_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(amdgpu_crtc->cursor_addr));
+
+	tmp = RREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 1);
+	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_MODE, 2);
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static int dce_v11_0_cursor_move_locked(struct drm_crtc *crtc,
+					int x, int y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	int xorigin = 0, yorigin = 0;
+
+	amdgpu_crtc->cursor_x = x;
+	amdgpu_crtc->cursor_y = y;
+
+	/* avivo cursor are offset into the total surface */
+	x += crtc->x;
+	y += crtc->y;
+	DRM_DEBUG("x %d y %d c->x %d c->y %d\n", x, y, crtc->x, crtc->y);
+
+	if (x < 0) {
+		xorigin = min(-x, amdgpu_crtc->max_cursor_width - 1);
+		x = 0;
+	}
+	if (y < 0) {
+		yorigin = min(-y, amdgpu_crtc->max_cursor_height - 1);
+		y = 0;
+	}
+
+	WREG32(mmCUR_POSITION + amdgpu_crtc->crtc_offset, (x << 16) | y);
+	WREG32(mmCUR_HOT_SPOT + amdgpu_crtc->crtc_offset, (xorigin << 16) | yorigin);
+	WREG32(mmCUR_SIZE + amdgpu_crtc->crtc_offset,
+	       ((amdgpu_crtc->cursor_width - 1) << 16) | (amdgpu_crtc->cursor_height - 1));
+
+	return 0;
+}
+
+static int dce_v11_0_crtc_cursor_move(struct drm_crtc *crtc,
+				      int x, int y)
+{
+	int ret;
+
+	dce_v11_0_lock_cursor(crtc, true);
+	ret = dce_v11_0_cursor_move_locked(crtc, x, y);
+	dce_v11_0_lock_cursor(crtc, false);
+
+	return ret;
+}
+
+static int dce_v11_0_crtc_cursor_set2(struct drm_crtc *crtc,
+				      struct drm_file *file_priv,
+				      uint32_t handle,
+				      uint32_t width,
+				      uint32_t height,
+				      int32_t hot_x,
+				      int32_t hot_y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *aobj;
+	int ret;
+
+	if (!handle) {
+		/* turn off cursor */
+		dce_v11_0_hide_cursor(crtc);
+		obj = NULL;
+		goto unpin;
+	}
+
+	if ((width > amdgpu_crtc->max_cursor_width) ||
+	    (height > amdgpu_crtc->max_cursor_height)) {
+		DRM_ERROR("bad cursor width or height %d x %d\n", width, height);
+		return -EINVAL;
+	}
+
+	obj = drm_gem_object_lookup(file_priv, handle);
+	if (!obj) {
+		DRM_ERROR("Cannot find cursor object %x for crtc %d\n", handle, amdgpu_crtc->crtc_id);
+		return -ENOENT;
+	}
+
+	aobj = gem_to_amdgpu_bo(obj);
+	ret = amdgpu_bo_reserve(aobj, false);
+	if (ret != 0) {
+		drm_gem_object_put_unlocked(obj);
+		return ret;
+	}
+
+	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM);
+	amdgpu_bo_unreserve(aobj);
+	if (ret) {
+		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
+		drm_gem_object_put_unlocked(obj);
+		return ret;
+	}
+	amdgpu_crtc->cursor_addr = amdgpu_bo_gpu_offset(aobj);
+
+	dce_v11_0_lock_cursor(crtc, true);
+
+	if (width != amdgpu_crtc->cursor_width ||
+	    height != amdgpu_crtc->cursor_height ||
+	    hot_x != amdgpu_crtc->cursor_hot_x ||
+	    hot_y != amdgpu_crtc->cursor_hot_y) {
+		int x, y;
+
+		x = amdgpu_crtc->cursor_x + amdgpu_crtc->cursor_hot_x - hot_x;
+		y = amdgpu_crtc->cursor_y + amdgpu_crtc->cursor_hot_y - hot_y;
+
+		dce_v11_0_cursor_move_locked(crtc, x, y);
+
+		amdgpu_crtc->cursor_width = width;
+		amdgpu_crtc->cursor_height = height;
+		amdgpu_crtc->cursor_hot_x = hot_x;
+		amdgpu_crtc->cursor_hot_y = hot_y;
+	}
+
+	dce_v11_0_show_cursor(crtc);
+	dce_v11_0_lock_cursor(crtc, false);
+
+unpin:
+	if (amdgpu_crtc->cursor_bo) {
+		struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+		ret = amdgpu_bo_reserve(aobj, true);
+		if (likely(ret == 0)) {
+			amdgpu_bo_unpin(aobj);
+			amdgpu_bo_unreserve(aobj);
+		}
+		drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
+	}
+
+	amdgpu_crtc->cursor_bo = obj;
+	return 0;
+}
+
+static void dce_v11_0_cursor_reset(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (amdgpu_crtc->cursor_bo) {
+		dce_v11_0_lock_cursor(crtc, true);
+
+		dce_v11_0_cursor_move_locked(crtc, amdgpu_crtc->cursor_x,
+					     amdgpu_crtc->cursor_y);
+
+		dce_v11_0_show_cursor(crtc);
+
+		dce_v11_0_lock_cursor(crtc, false);
+	}
+}
+
+static int dce_v11_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
+				    u16 *blue, uint32_t size,
+				    struct drm_modeset_acquire_ctx *ctx)
+{
+	dce_v11_0_crtc_load_lut(crtc);
+
+	return 0;
+}
+
+static void dce_v11_0_crtc_destroy(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	drm_crtc_cleanup(crtc);
+	kfree(amdgpu_crtc);
+}
+
+static const struct drm_crtc_funcs dce_v11_0_crtc_funcs = {
+	.cursor_set2 = dce_v11_0_crtc_cursor_set2,
+	.cursor_move = dce_v11_0_crtc_cursor_move,
+	.gamma_set = dce_v11_0_crtc_gamma_set,
+	.set_config = amdgpu_display_crtc_set_config,
+	.destroy = dce_v11_0_crtc_destroy,
+	.page_flip_target = amdgpu_display_crtc_page_flip_target,
+};
+
+static void dce_v11_0_crtc_dpms(struct drm_crtc *crtc, int mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	unsigned type;
+
+	switch (mode) {
+	case DRM_MODE_DPMS_ON:
+		amdgpu_crtc->enabled = true;
+		amdgpu_atombios_crtc_enable(crtc, ATOM_ENABLE);
+		dce_v11_0_vga_enable(crtc, true);
+		amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
+		dce_v11_0_vga_enable(crtc, false);
+		/* Make sure VBLANK and PFLIP interrupts are still enabled */
+		type = amdgpu_display_crtc_idx_to_irq_type(adev,
+						amdgpu_crtc->crtc_id);
+		amdgpu_irq_update(adev, &adev->crtc_irq, type);
+		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
+		drm_crtc_vblank_on(crtc);
+		dce_v11_0_crtc_load_lut(crtc);
+		break;
+	case DRM_MODE_DPMS_STANDBY:
+	case DRM_MODE_DPMS_SUSPEND:
+	case DRM_MODE_DPMS_OFF:
+		drm_crtc_vblank_off(crtc);
+		if (amdgpu_crtc->enabled) {
+			dce_v11_0_vga_enable(crtc, true);
+			amdgpu_atombios_crtc_blank(crtc, ATOM_ENABLE);
+			dce_v11_0_vga_enable(crtc, false);
+		}
+		amdgpu_atombios_crtc_enable(crtc, ATOM_DISABLE);
+		amdgpu_crtc->enabled = false;
+		break;
+	}
+	/* adjust pm to dpms */
+	amdgpu_pm_compute_clocks(adev);
+}
+
+static void dce_v11_0_crtc_prepare(struct drm_crtc *crtc)
+{
+	/* disable crtc pair power gating before programming */
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_DISABLE);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_ENABLE);
+	dce_v11_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+}
+
+static void dce_v11_0_crtc_commit(struct drm_crtc *crtc)
+{
+	dce_v11_0_crtc_dpms(crtc, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_DISABLE);
+}
+
+static void dce_v11_0_crtc_disable(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_atom_ss ss;
+	int i;
+
+	dce_v11_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+	if (crtc->primary->fb) {
+		int r;
+		struct amdgpu_bo *abo;
+
+		abo = gem_to_amdgpu_bo(crtc->primary->fb->obj[0]);
+		r = amdgpu_bo_reserve(abo, true);
+		if (unlikely(r))
+			DRM_ERROR("failed to reserve abo before unpin\n");
+		else {
+			amdgpu_bo_unpin(abo);
+			amdgpu_bo_unreserve(abo);
+		}
+	}
+	/* disable the GRPH */
+	dce_v11_0_grph_enable(crtc, false);
+
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_ENABLE);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i] &&
+		    adev->mode_info.crtcs[i]->enabled &&
+		    i != amdgpu_crtc->crtc_id &&
+		    amdgpu_crtc->pll_id == adev->mode_info.crtcs[i]->pll_id) {
+			/* one other crtc is using this pll don't turn
+			 * off the pll
+			 */
+			goto done;
+		}
+	}
+
+	switch (amdgpu_crtc->pll_id) {
+	case ATOM_PPLL0:
+	case ATOM_PPLL1:
+	case ATOM_PPLL2:
+		/* disable the ppll */
+		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
+						 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
+		break;
+	case ATOM_COMBOPHY_PLL0:
+	case ATOM_COMBOPHY_PLL1:
+	case ATOM_COMBOPHY_PLL2:
+	case ATOM_COMBOPHY_PLL3:
+	case ATOM_COMBOPHY_PLL4:
+	case ATOM_COMBOPHY_PLL5:
+		/* disable the ppll */
+		amdgpu_atombios_crtc_program_pll(crtc, ATOM_CRTC_INVALID, amdgpu_crtc->pll_id,
+						 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
+		break;
+	default:
+		break;
+	}
+done:
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+}
+
+static int dce_v11_0_crtc_mode_set(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode,
+				  int x, int y, struct drm_framebuffer *old_fb)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (!amdgpu_crtc->adjusted_clock)
+		return -EINVAL;
+
+	if ((adev->asic_type == CHIP_POLARIS10) ||
+	    (adev->asic_type == CHIP_POLARIS11) ||
+	    (adev->asic_type == CHIP_POLARIS12) ||
+	    (adev->asic_type == CHIP_VEGAM)) {
+		struct amdgpu_encoder *amdgpu_encoder =
+			to_amdgpu_encoder(amdgpu_crtc->encoder);
+		int encoder_mode =
+			amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder);
+
+		/* SetPixelClock calculates the plls and ss values now */
+		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id,
+						 amdgpu_crtc->pll_id,
+						 encoder_mode, amdgpu_encoder->encoder_id,
+						 adjusted_mode->clock, 0, 0, 0, 0,
+						 amdgpu_crtc->bpc, amdgpu_crtc->ss_enabled, &amdgpu_crtc->ss);
+	} else {
+		amdgpu_atombios_crtc_set_pll(crtc, adjusted_mode);
+	}
+	amdgpu_atombios_crtc_set_dtd_timing(crtc, adjusted_mode);
+	dce_v11_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+	amdgpu_atombios_crtc_overscan_setup(crtc, mode, adjusted_mode);
+	amdgpu_atombios_crtc_scaler_setup(crtc);
+	dce_v11_0_cursor_reset(crtc);
+	/* update the hw version fpr dpm */
+	amdgpu_crtc->hw_mode = *adjusted_mode;
+
+	return 0;
+}
+
+static bool dce_v11_0_crtc_mode_fixup(struct drm_crtc *crtc,
+				     const struct drm_display_mode *mode,
+				     struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct drm_encoder *encoder;
+
+	/* assign the encoder to the amdgpu crtc to avoid repeated lookups later */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		if (encoder->crtc == crtc) {
+			amdgpu_crtc->encoder = encoder;
+			amdgpu_crtc->connector = amdgpu_get_connector_for_encoder(encoder);
+			break;
+		}
+	}
+	if ((amdgpu_crtc->encoder == NULL) || (amdgpu_crtc->connector == NULL)) {
+		amdgpu_crtc->encoder = NULL;
+		amdgpu_crtc->connector = NULL;
+		return false;
+	}
+	if (!amdgpu_display_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
+		return false;
+	if (amdgpu_atombios_crtc_prepare_pll(crtc, adjusted_mode))
+		return false;
+	/* pick pll */
+	amdgpu_crtc->pll_id = dce_v11_0_pick_pll(crtc);
+	/* if we can't get a PPLL for a non-DP encoder, fail */
+	if ((amdgpu_crtc->pll_id == ATOM_PPLL_INVALID) &&
+	    !ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder)))
+		return false;
+
+	return true;
+}
+
+static int dce_v11_0_crtc_set_base(struct drm_crtc *crtc, int x, int y,
+				  struct drm_framebuffer *old_fb)
+{
+	return dce_v11_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+}
+
+static int dce_v11_0_crtc_set_base_atomic(struct drm_crtc *crtc,
+					 struct drm_framebuffer *fb,
+					 int x, int y, enum mode_set_atomic state)
+{
+       return dce_v11_0_crtc_do_set_base(crtc, fb, x, y, 1);
+}
+
+static const struct drm_crtc_helper_funcs dce_v11_0_crtc_helper_funcs = {
+	.dpms = dce_v11_0_crtc_dpms,
+	.mode_fixup = dce_v11_0_crtc_mode_fixup,
+	.mode_set = dce_v11_0_crtc_mode_set,
+	.mode_set_base = dce_v11_0_crtc_set_base,
+	.mode_set_base_atomic = dce_v11_0_crtc_set_base_atomic,
+	.prepare = dce_v11_0_crtc_prepare,
+	.commit = dce_v11_0_crtc_commit,
+	.disable = dce_v11_0_crtc_disable,
+};
+
+static int dce_v11_0_crtc_init(struct amdgpu_device *adev, int index)
+{
+	struct amdgpu_crtc *amdgpu_crtc;
+
+	amdgpu_crtc = kzalloc(sizeof(struct amdgpu_crtc) +
+			      (AMDGPUFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
+	if (amdgpu_crtc == NULL)
+		return -ENOMEM;
+
+	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_v11_0_crtc_funcs);
+
+	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
+	amdgpu_crtc->crtc_id = index;
+	adev->mode_info.crtcs[index] = amdgpu_crtc;
+
+	amdgpu_crtc->max_cursor_width = 128;
+	amdgpu_crtc->max_cursor_height = 128;
+	adev->ddev->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
+	adev->ddev->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
+
+	switch (amdgpu_crtc->crtc_id) {
+	case 0:
+	default:
+		amdgpu_crtc->crtc_offset = CRTC0_REGISTER_OFFSET;
+		break;
+	case 1:
+		amdgpu_crtc->crtc_offset = CRTC1_REGISTER_OFFSET;
+		break;
+	case 2:
+		amdgpu_crtc->crtc_offset = CRTC2_REGISTER_OFFSET;
+		break;
+	case 3:
+		amdgpu_crtc->crtc_offset = CRTC3_REGISTER_OFFSET;
+		break;
+	case 4:
+		amdgpu_crtc->crtc_offset = CRTC4_REGISTER_OFFSET;
+		break;
+	case 5:
+		amdgpu_crtc->crtc_offset = CRTC5_REGISTER_OFFSET;
+		break;
+	}
+
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_v11_0_crtc_helper_funcs);
+
+	return 0;
+}
+
+static int dce_v11_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->audio_endpt_rreg = &dce_v11_0_audio_endpt_rreg;
+	adev->audio_endpt_wreg = &dce_v11_0_audio_endpt_wreg;
+
+	dce_v11_0_set_display_funcs(adev);
+
+	adev->mode_info.num_crtc = dce_v11_0_get_num_crtc(adev);
+
+	switch (adev->asic_type) {
+	case CHIP_CARRIZO:
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 9;
+		break;
+	case CHIP_STONEY:
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 9;
+		break;
+	case CHIP_POLARIS10:
+	case CHIP_VEGAM:
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 6;
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+		adev->mode_info.num_hpd = 5;
+		adev->mode_info.num_dig = 5;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	dce_v11_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int dce_v11_0_sw_init(void *handle)
+{
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
+		if (r)
+			return r;
+	}
+
+	for (i = VISLANDS30_IV_SRCID_D1_GRPH_PFLIP; i < 20; i += 2) {
+		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i, &adev->pageflip_irq);
+		if (r)
+			return r;
+	}
+
+	/* HPD hotplug */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_HOTPLUG_DETECT_A, &adev->hpd_irq);
+	if (r)
+		return r;
+
+	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
+
+	adev->ddev->mode_config.async_page_flip = true;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	adev->ddev->mode_config.preferred_depth = 24;
+	adev->ddev->mode_config.prefer_shadow = 1;
+
+	adev->ddev->mode_config.fb_base = adev->gmc.aper_base;
+
+	r = amdgpu_display_modeset_create_props(adev);
+	if (r)
+		return r;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+
+	/* allocate crtcs */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = dce_v11_0_crtc_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	if (amdgpu_atombios_get_connector_info_from_object_table(adev))
+		amdgpu_display_print_display_setup(adev->ddev);
+	else
+		return -EINVAL;
+
+	/* setup afmt */
+	r = dce_v11_0_afmt_init(adev);
+	if (r)
+		return r;
+
+	r = dce_v11_0_audio_init(adev);
+	if (r)
+		return r;
+
+	drm_kms_helper_poll_init(adev->ddev);
+
+	adev->mode_info.mode_config_initialized = true;
+	return 0;
+}
+
+static int dce_v11_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	kfree(adev->mode_info.bios_hardcoded_edid);
+
+	drm_kms_helper_poll_fini(adev->ddev);
+
+	dce_v11_0_audio_fini(adev);
+
+	dce_v11_0_afmt_fini(adev);
+
+	drm_mode_config_cleanup(adev->ddev);
+	adev->mode_info.mode_config_initialized = false;
+
+	return 0;
+}
+
+static int dce_v11_0_hw_init(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dce_v11_0_init_golden_registers(adev);
+
+	/* disable vga render */
+	dce_v11_0_set_vga_render_state(adev, false);
+	/* init dig PHYs, disp eng pll */
+	amdgpu_atombios_crtc_powergate_init(adev);
+	amdgpu_atombios_encoder_init_dig(adev);
+	if ((adev->asic_type == CHIP_POLARIS10) ||
+	    (adev->asic_type == CHIP_POLARIS11) ||
+	    (adev->asic_type == CHIP_POLARIS12) ||
+	    (adev->asic_type == CHIP_VEGAM)) {
+		amdgpu_atombios_crtc_set_dce_clock(adev, adev->clock.default_dispclk,
+						   DCE_CLOCK_TYPE_DISPCLK, ATOM_GCK_DFS);
+		amdgpu_atombios_crtc_set_dce_clock(adev, 0,
+						   DCE_CLOCK_TYPE_DPREFCLK, ATOM_GCK_DFS);
+	} else {
+		amdgpu_atombios_crtc_set_disp_eng_pll(adev, adev->clock.default_dispclk);
+	}
+
+	/* initialize hpd */
+	dce_v11_0_hpd_init(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v11_0_pageflip_interrupt_init(adev);
+
+	return 0;
+}
+
+static int dce_v11_0_hw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dce_v11_0_hpd_fini(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v11_0_pageflip_interrupt_fini(adev);
+
+	return 0;
+}
+
+static int dce_v11_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->mode_info.bl_level =
+		amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+
+	return dce_v11_0_hw_fini(handle);
+}
+
+static int dce_v11_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret;
+
+	amdgpu_atombios_encoder_set_backlight_level_to_reg(adev,
+							   adev->mode_info.bl_level);
+
+	ret = dce_v11_0_hw_init(handle);
+
+	/* turn on the BL */
+	if (adev->mode_info.bl_encoder) {
+		u8 bl_level = amdgpu_display_backlight_get_level(adev,
+								  adev->mode_info.bl_encoder);
+		amdgpu_display_backlight_set_level(adev, adev->mode_info.bl_encoder,
+						    bl_level);
+	}
+
+	return ret;
+}
+
+static bool dce_v11_0_is_idle(void *handle)
+{
+	return true;
+}
+
+static int dce_v11_0_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static int dce_v11_0_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0, tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (dce_v11_0_is_display_hung(adev))
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+	return 0;
+}
+
+static void dce_v11_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
+						     int crtc,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 lb_interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VBLANK_INTERRUPT_MASK, 0);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VBLANK_INTERRUPT_MASK, 1);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static void dce_v11_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
+						    int crtc,
+						    enum amdgpu_interrupt_state state)
+{
+	u32 lb_interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VLINE_INTERRUPT_MASK, 0);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VLINE_INTERRUPT_MASK, 1);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static int dce_v11_0_set_hpd_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned hpd,
+					enum amdgpu_interrupt_state state)
+{
+	u32 tmp;
+
+	if (hpd >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hdp %d\n", hpd);
+		return 0;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 0);
+		WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 1);
+		WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v11_0_set_crtc_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CRTC_IRQ_VBLANK1:
+		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK2:
+		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK3:
+		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK4:
+		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK5:
+		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 4, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK6:
+		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 5, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE1:
+		dce_v11_0_set_crtc_vline_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE2:
+		dce_v11_0_set_crtc_vline_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE3:
+		dce_v11_0_set_crtc_vline_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE4:
+		dce_v11_0_set_crtc_vline_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE5:
+		dce_v11_0_set_crtc_vline_interrupt_state(adev, 4, state);
+		break;
+	 case AMDGPU_CRTC_IRQ_VLINE6:
+		dce_v11_0_set_crtc_vline_interrupt_state(adev, 5, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int dce_v11_0_set_pageflip_irq_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	u32 reg;
+
+	if (type >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", type);
+		return -EINVAL;
+	}
+
+	reg = RREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type]);
+	if (state == AMDGPU_IRQ_STATE_DISABLE)
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg & ~GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+	else
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg | GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+
+	return 0;
+}
+
+static int dce_v11_0_pageflip_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	unsigned long flags;
+	unsigned crtc_id;
+	struct amdgpu_crtc *amdgpu_crtc;
+	struct amdgpu_flip_work *works;
+
+	crtc_id = (entry->src_id - 8) >> 1;
+	amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	if (crtc_id >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", crtc_id);
+		return -EINVAL;
+	}
+
+	if (RREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id]) &
+	    GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_OCCURRED_MASK)
+		WREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id],
+		       GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_CLEAR_MASK);
+
+	/* IRQ could occur when in initial stage */
+	if(amdgpu_crtc == NULL)
+		return 0;
+
+	spin_lock_irqsave(&adev->ddev->event_lock, flags);
+	works = amdgpu_crtc->pflip_works;
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED){
+		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
+						 "AMDGPU_FLIP_SUBMITTED(%d)\n",
+						 amdgpu_crtc->pflip_status,
+						 AMDGPU_FLIP_SUBMITTED);
+		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+		return 0;
+	}
+
+	/* page flip completed. clean up */
+	amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE;
+	amdgpu_crtc->pflip_works = NULL;
+
+	/* wakeup usersapce */
+	if(works->event)
+		drm_crtc_send_vblank_event(&amdgpu_crtc->base, works->event);
+
+	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+
+	drm_crtc_vblank_put(&amdgpu_crtc->base);
+	schedule_work(&works->unpin_work);
+
+	return 0;
+}
+
+static void dce_v11_0_hpd_int_ack(struct amdgpu_device *adev,
+				  int hpd)
+{
+	u32 tmp;
+
+	if (hpd >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hdp %d\n", hpd);
+		return;
+	}
+
+	tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+	tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_ACK, 1);
+	WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+}
+
+static void dce_v11_0_crtc_vblank_int_ack(struct amdgpu_device *adev,
+					  int crtc)
+{
+	u32 tmp;
+
+	if (crtc < 0 || crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	tmp = RREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc]);
+	tmp = REG_SET_FIELD(tmp, LB_VBLANK_STATUS, VBLANK_ACK, 1);
+	WREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc], tmp);
+}
+
+static void dce_v11_0_crtc_vline_int_ack(struct amdgpu_device *adev,
+					 int crtc)
+{
+	u32 tmp;
+
+	if (crtc < 0 || crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	tmp = RREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc]);
+	tmp = REG_SET_FIELD(tmp, LB_VLINE_STATUS, VLINE_ACK, 1);
+	WREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc], tmp);
+}
+
+static int dce_v11_0_crtc_irq(struct amdgpu_device *adev,
+				struct amdgpu_irq_src *source,
+				struct amdgpu_iv_entry *entry)
+{
+	unsigned crtc = entry->src_id - 1;
+	uint32_t disp_int = RREG32(interrupt_status_offsets[crtc].reg);
+	unsigned int irq_type = amdgpu_display_crtc_idx_to_irq_type(adev,
+								    crtc);
+
+	switch (entry->src_data[0]) {
+	case 0: /* vblank */
+		if (disp_int & interrupt_status_offsets[crtc].vblank)
+			dce_v11_0_crtc_vblank_int_ack(adev, crtc);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		if (amdgpu_irq_enabled(adev, source, irq_type)) {
+			drm_handle_vblank(adev->ddev, crtc);
+		}
+		DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+
+		break;
+	case 1: /* vline */
+		if (disp_int & interrupt_status_offsets[crtc].vline)
+			dce_v11_0_crtc_vline_int_ack(adev, crtc);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		DRM_DEBUG("IH: D%d vline\n", crtc + 1);
+
+		break;
+	default:
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v11_0_hpd_irq(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *source,
+			     struct amdgpu_iv_entry *entry)
+{
+	uint32_t disp_int, mask;
+	unsigned hpd;
+
+	if (entry->src_data[0] >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]);
+		return 0;
+	}
+
+	hpd = entry->src_data[0];
+	disp_int = RREG32(interrupt_status_offsets[hpd].reg);
+	mask = interrupt_status_offsets[hpd].hpd;
+
+	if (disp_int & mask) {
+		dce_v11_0_hpd_int_ack(adev, hpd);
+		schedule_work(&adev->hotplug_work);
+		DRM_DEBUG("IH: HPD%d\n", hpd + 1);
+	}
+
+	return 0;
+}
+
+static int dce_v11_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int dce_v11_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs dce_v11_0_ip_funcs = {
+	.name = "dce_v11_0",
+	.early_init = dce_v11_0_early_init,
+	.late_init = NULL,
+	.sw_init = dce_v11_0_sw_init,
+	.sw_fini = dce_v11_0_sw_fini,
+	.hw_init = dce_v11_0_hw_init,
+	.hw_fini = dce_v11_0_hw_fini,
+	.suspend = dce_v11_0_suspend,
+	.resume = dce_v11_0_resume,
+	.is_idle = dce_v11_0_is_idle,
+	.wait_for_idle = dce_v11_0_wait_for_idle,
+	.soft_reset = dce_v11_0_soft_reset,
+	.set_clockgating_state = dce_v11_0_set_clockgating_state,
+	.set_powergating_state = dce_v11_0_set_powergating_state,
+};
+
+static void
+dce_v11_0_encoder_mode_set(struct drm_encoder *encoder,
+			  struct drm_display_mode *mode,
+			  struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	amdgpu_encoder->pixel_clock = adjusted_mode->clock;
+
+	/* need to call this here rather than in prepare() since we need some crtc info */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	/* set scaler clears this on some chips */
+	dce_v11_0_set_interleave(encoder->crtc, mode);
+
+	if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) {
+		dce_v11_0_afmt_enable(encoder, true);
+		dce_v11_0_afmt_setmode(encoder, adjusted_mode);
+	}
+}
+
+static void dce_v11_0_encoder_prepare(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+
+	if ((amdgpu_encoder->active_device &
+	     (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) !=
+	     ENCODER_OBJECT_ID_NONE)) {
+		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+		if (dig) {
+			dig->dig_encoder = dce_v11_0_pick_dig_encoder(encoder);
+			if (amdgpu_encoder->active_device & ATOM_DEVICE_DFP_SUPPORT)
+				dig->afmt = adev->mode_info.afmt[dig->dig_encoder];
+		}
+	}
+
+	amdgpu_atombios_scratch_regs_lock(adev, true);
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		/* select the clock/data port if it uses a router */
+		if (amdgpu_connector->router.cd_valid)
+			amdgpu_i2c_router_select_cd_port(amdgpu_connector);
+
+		/* turn eDP panel on for mode set */
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP)
+			amdgpu_atombios_encoder_set_edp_panel_power(connector,
+							     ATOM_TRANSMITTER_ACTION_POWER_ON);
+	}
+
+	/* this is needed for the pll/ss setup to work correctly in some cases */
+	amdgpu_atombios_encoder_set_crtc_source(encoder);
+	/* set up the FMT blocks */
+	dce_v11_0_program_fmt(encoder);
+}
+
+static void dce_v11_0_encoder_commit(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* need to call this here as we need the crtc set up */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_scratch_regs_lock(adev, false);
+}
+
+static void dce_v11_0_encoder_disable(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig;
+
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	if (amdgpu_atombios_encoder_is_digital(encoder)) {
+		if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI)
+			dce_v11_0_afmt_enable(encoder, false);
+		dig = amdgpu_encoder->enc_priv;
+		dig->dig_encoder = -1;
+	}
+	amdgpu_encoder->active_device = 0;
+}
+
+/* these are handled by the primary encoders */
+static void dce_v11_0_ext_prepare(struct drm_encoder *encoder)
+{
+
+}
+
+static void dce_v11_0_ext_commit(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v11_0_ext_mode_set(struct drm_encoder *encoder,
+		      struct drm_display_mode *mode,
+		      struct drm_display_mode *adjusted_mode)
+{
+
+}
+
+static void dce_v11_0_ext_disable(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v11_0_ext_dpms(struct drm_encoder *encoder, int mode)
+{
+
+}
+
+static const struct drm_encoder_helper_funcs dce_v11_0_ext_helper_funcs = {
+	.dpms = dce_v11_0_ext_dpms,
+	.prepare = dce_v11_0_ext_prepare,
+	.mode_set = dce_v11_0_ext_mode_set,
+	.commit = dce_v11_0_ext_commit,
+	.disable = dce_v11_0_ext_disable,
+	/* no detect for TMDS/LVDS yet */
+};
+
+static const struct drm_encoder_helper_funcs dce_v11_0_dig_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v11_0_encoder_prepare,
+	.mode_set = dce_v11_0_encoder_mode_set,
+	.commit = dce_v11_0_encoder_commit,
+	.disable = dce_v11_0_encoder_disable,
+	.detect = amdgpu_atombios_encoder_dig_detect,
+};
+
+static const struct drm_encoder_helper_funcs dce_v11_0_dac_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v11_0_encoder_prepare,
+	.mode_set = dce_v11_0_encoder_mode_set,
+	.commit = dce_v11_0_encoder_commit,
+	.detect = amdgpu_atombios_encoder_dac_detect,
+};
+
+static void dce_v11_0_encoder_destroy(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+		amdgpu_atombios_encoder_fini_backlight(amdgpu_encoder);
+	kfree(amdgpu_encoder->enc_priv);
+	drm_encoder_cleanup(encoder);
+	kfree(amdgpu_encoder);
+}
+
+static const struct drm_encoder_funcs dce_v11_0_encoder_funcs = {
+	.destroy = dce_v11_0_encoder_destroy,
+};
+
+static void dce_v11_0_encoder_add(struct amdgpu_device *adev,
+				 uint32_t encoder_enum,
+				 uint32_t supported_device,
+				 u16 caps)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	/* see if we already added it */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		if (amdgpu_encoder->encoder_enum == encoder_enum) {
+			amdgpu_encoder->devices |= supported_device;
+			return;
+		}
+
+	}
+
+	/* add a new one */
+	amdgpu_encoder = kzalloc(sizeof(struct amdgpu_encoder), GFP_KERNEL);
+	if (!amdgpu_encoder)
+		return;
+
+	encoder = &amdgpu_encoder->base;
+	switch (adev->mode_info.num_crtc) {
+	case 1:
+		encoder->possible_crtcs = 0x1;
+		break;
+	case 2:
+	default:
+		encoder->possible_crtcs = 0x3;
+		break;
+	case 3:
+		encoder->possible_crtcs = 0x7;
+		break;
+	case 4:
+		encoder->possible_crtcs = 0xf;
+		break;
+	case 5:
+		encoder->possible_crtcs = 0x1f;
+		break;
+	case 6:
+		encoder->possible_crtcs = 0x3f;
+		break;
+	}
+
+	amdgpu_encoder->enc_priv = NULL;
+
+	amdgpu_encoder->encoder_enum = encoder_enum;
+	amdgpu_encoder->encoder_id = (encoder_enum & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+	amdgpu_encoder->devices = supported_device;
+	amdgpu_encoder->rmx_type = RMX_OFF;
+	amdgpu_encoder->underscan_type = UNDERSCAN_OFF;
+	amdgpu_encoder->is_ext_encoder = false;
+	amdgpu_encoder->caps = caps;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+		drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+				 DRM_MODE_ENCODER_DAC, NULL);
+		drm_encoder_helper_add(encoder, &dce_v11_0_dac_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+			amdgpu_encoder->rmx_type = RMX_FULL;
+			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS, NULL);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_lcd_info(amdgpu_encoder);
+		} else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) {
+			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC, NULL);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		} else {
+			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS, NULL);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		}
+		drm_encoder_helper_add(encoder, &dce_v11_0_dig_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_SI170B:
+	case ENCODER_OBJECT_ID_CH7303:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOA:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOB:
+	case ENCODER_OBJECT_ID_TITFP513:
+	case ENCODER_OBJECT_ID_VT1623:
+	case ENCODER_OBJECT_ID_HDMI_SI1930:
+	case ENCODER_OBJECT_ID_TRAVIS:
+	case ENCODER_OBJECT_ID_NUTMEG:
+		/* these are handled by the primary encoders */
+		amdgpu_encoder->is_ext_encoder = true;
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS, NULL);
+		else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC, NULL);
+		else
+			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS, NULL);
+		drm_encoder_helper_add(encoder, &dce_v11_0_ext_helper_funcs);
+		break;
+	}
+}
+
+static const struct amdgpu_display_funcs dce_v11_0_display_funcs = {
+	.bandwidth_update = &dce_v11_0_bandwidth_update,
+	.vblank_get_counter = &dce_v11_0_vblank_get_counter,
+	.backlight_set_level = &amdgpu_atombios_encoder_set_backlight_level,
+	.backlight_get_level = &amdgpu_atombios_encoder_get_backlight_level,
+	.hpd_sense = &dce_v11_0_hpd_sense,
+	.hpd_set_polarity = &dce_v11_0_hpd_set_polarity,
+	.hpd_get_gpio_reg = &dce_v11_0_hpd_get_gpio_reg,
+	.page_flip = &dce_v11_0_page_flip,
+	.page_flip_get_scanoutpos = &dce_v11_0_crtc_get_scanoutpos,
+	.add_encoder = &dce_v11_0_encoder_add,
+	.add_connector = &amdgpu_connector_add,
+};
+
+static void dce_v11_0_set_display_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.funcs == NULL)
+		adev->mode_info.funcs = &dce_v11_0_display_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs dce_v11_0_crtc_irq_funcs = {
+	.set = dce_v11_0_set_crtc_irq_state,
+	.process = dce_v11_0_crtc_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v11_0_pageflip_irq_funcs = {
+	.set = dce_v11_0_set_pageflip_irq_state,
+	.process = dce_v11_0_pageflip_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v11_0_hpd_irq_funcs = {
+	.set = dce_v11_0_set_hpd_irq_state,
+	.process = dce_v11_0_hpd_irq,
+};
+
+static void dce_v11_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.num_crtc > 0)
+		adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_VLINE1 + adev->mode_info.num_crtc;
+	else
+		adev->crtc_irq.num_types = 0;
+	adev->crtc_irq.funcs = &dce_v11_0_crtc_irq_funcs;
+
+	adev->pageflip_irq.num_types = adev->mode_info.num_crtc;
+	adev->pageflip_irq.funcs = &dce_v11_0_pageflip_irq_funcs;
+
+	adev->hpd_irq.num_types = adev->mode_info.num_hpd;
+	adev->hpd_irq.funcs = &dce_v11_0_hpd_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version dce_v11_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 11,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &dce_v11_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version dce_v11_2_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 11,
+	.minor = 2,
+	.rev = 0,
+	.funcs = &dce_v11_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/dce_v11_0.h b/sys/dev/pci/drm/amd/amdgpu/dce_v11_0.h
new file mode 100644
index 00000000000..0d878ca3acb
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/dce_v11_0.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DCE_V11_0_H__
+#define __DCE_V11_0_H__
+
+extern const struct amdgpu_ip_block_version dce_v11_0_ip_block;
+extern const struct amdgpu_ip_block_version dce_v11_2_ip_block;
+
+void dce_v11_0_disable_dce(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/dce_v6_0.c b/sys/dev/pci/drm/amd/amdgpu/dce_v6_0.c
new file mode 100644
index 00000000000..c9adc627305
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/dce_v6_0.c
@@ -0,0 +1,3418 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_i2c.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+#include "atombios_crtc.h"
+#include "atombios_encoders.h"
+#include "amdgpu_pll.h"
+#include "amdgpu_connectors.h"
+
+#include "bif/bif_3_0_d.h"
+#include "bif/bif_3_0_sh_mask.h"
+#include "oss/oss_1_0_d.h"
+#include "oss/oss_1_0_sh_mask.h"
+#include "gca/gfx_6_0_d.h"
+#include "gca/gfx_6_0_sh_mask.h"
+#include "gmc/gmc_6_0_d.h"
+#include "gmc/gmc_6_0_sh_mask.h"
+#include "dce/dce_6_0_d.h"
+#include "dce/dce_6_0_sh_mask.h"
+#include "gca/gfx_7_2_enum.h"
+#include "dce_v6_0.h"
+#include "si_enums.h"
+
+static void dce_v6_0_set_display_funcs(struct amdgpu_device *adev);
+static void dce_v6_0_set_irq_funcs(struct amdgpu_device *adev);
+
+static const u32 crtc_offsets[6] =
+{
+	SI_CRTC0_REGISTER_OFFSET,
+	SI_CRTC1_REGISTER_OFFSET,
+	SI_CRTC2_REGISTER_OFFSET,
+	SI_CRTC3_REGISTER_OFFSET,
+	SI_CRTC4_REGISTER_OFFSET,
+	SI_CRTC5_REGISTER_OFFSET
+};
+
+static const u32 hpd_offsets[] =
+{
+	mmDC_HPD1_INT_STATUS - mmDC_HPD1_INT_STATUS,
+	mmDC_HPD2_INT_STATUS - mmDC_HPD1_INT_STATUS,
+	mmDC_HPD3_INT_STATUS - mmDC_HPD1_INT_STATUS,
+	mmDC_HPD4_INT_STATUS - mmDC_HPD1_INT_STATUS,
+	mmDC_HPD5_INT_STATUS - mmDC_HPD1_INT_STATUS,
+	mmDC_HPD6_INT_STATUS - mmDC_HPD1_INT_STATUS,
+};
+
+static const uint32_t dig_offsets[] = {
+	SI_CRTC0_REGISTER_OFFSET,
+	SI_CRTC1_REGISTER_OFFSET,
+	SI_CRTC2_REGISTER_OFFSET,
+	SI_CRTC3_REGISTER_OFFSET,
+	SI_CRTC4_REGISTER_OFFSET,
+	SI_CRTC5_REGISTER_OFFSET,
+	(0x13830 - 0x7030) >> 2,
+};
+
+static const struct {
+	uint32_t	reg;
+	uint32_t	vblank;
+	uint32_t	vline;
+	uint32_t	hpd;
+
+} interrupt_status_offsets[6] = { {
+	.reg = mmDISP_INTERRUPT_STATUS,
+	.vblank = DISP_INTERRUPT_STATUS__LB_D1_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS__LB_D1_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS__DC_HPD1_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE__DC_HPD2_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE2,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE2__DC_HPD3_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE3,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE3__DC_HPD4_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE4,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE4__DC_HPD5_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE5,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE5__DC_HPD6_INTERRUPT_MASK
+} };
+
+static u32 dce_v6_0_audio_endpt_rreg(struct amdgpu_device *adev,
+				     u32 block_offset, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	r = RREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+
+	return r;
+}
+
+static void dce_v6_0_audio_endpt_wreg(struct amdgpu_device *adev,
+				      u32 block_offset, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset,
+		reg | AZALIA_F0_CODEC_ENDPOINT_INDEX__AZALIA_ENDPOINT_REG_WRITE_EN_MASK);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset, v);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+}
+
+static u32 dce_v6_0_vblank_get_counter(struct amdgpu_device *adev, int crtc)
+{
+	if (crtc >= adev->mode_info.num_crtc)
+		return 0;
+	else
+		return RREG32(mmCRTC_STATUS_FRAME_COUNT + crtc_offsets[crtc]);
+}
+
+static void dce_v6_0_pageflip_interrupt_init(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Enable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_get(adev, &adev->pageflip_irq, i);
+}
+
+static void dce_v6_0_pageflip_interrupt_fini(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Disable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_put(adev, &adev->pageflip_irq, i);
+}
+
+/**
+ * dce_v6_0_page_flip - pageflip callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @crtc_id: crtc to cleanup pageflip on
+ * @crtc_base: new address of the crtc (GPU MC address)
+ *
+ * Does the actual pageflip (evergreen+).
+ * During vblank we take the crtc lock and wait for the update_pending
+ * bit to go high, when it does, we release the lock, and allow the
+ * double buffered update to take place.
+ * Returns the current update pending status.
+ */
+static void dce_v6_0_page_flip(struct amdgpu_device *adev,
+			       int crtc_id, u64 crtc_base, bool async)
+{
+	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	/* flip at hsync for async, default is vsync */
+	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, async ?
+	       GRPH_FLIP_CONTROL__GRPH_SURFACE_UPDATE_H_RETRACE_EN_MASK : 0);
+	/* update the scanout addresses */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(crtc_base));
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32)crtc_base);
+
+	/* post the write */
+	RREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset);
+}
+
+static int dce_v6_0_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
+					u32 *vbl, u32 *position)
+{
+	if ((crtc < 0) || (crtc >= adev->mode_info.num_crtc))
+		return -EINVAL;
+	*vbl = RREG32(mmCRTC_V_BLANK_START_END + crtc_offsets[crtc]);
+	*position = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+
+	return 0;
+
+}
+
+/**
+ * dce_v6_0_hpd_sense - hpd sense callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Checks if a digital monitor is connected (evergreen+).
+ * Returns true if connected, false if not connected.
+ */
+static bool dce_v6_0_hpd_sense(struct amdgpu_device *adev,
+			       enum amdgpu_hpd_id hpd)
+{
+	bool connected = false;
+
+	if (hpd >= adev->mode_info.num_hpd)
+		return connected;
+
+	if (RREG32(mmDC_HPD1_INT_STATUS + hpd_offsets[hpd]) & DC_HPD1_INT_STATUS__DC_HPD1_SENSE_MASK)
+		connected = true;
+
+	return connected;
+}
+
+/**
+ * dce_v6_0_hpd_set_polarity - hpd set polarity callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Set the polarity of the hpd pin (evergreen+).
+ */
+static void dce_v6_0_hpd_set_polarity(struct amdgpu_device *adev,
+				      enum amdgpu_hpd_id hpd)
+{
+	u32 tmp;
+	bool connected = dce_v6_0_hpd_sense(adev, hpd);
+
+	if (hpd >= adev->mode_info.num_hpd)
+		return;
+
+	tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
+	if (connected)
+		tmp &= ~DC_HPD1_INT_CONTROL__DC_HPD1_INT_POLARITY_MASK;
+	else
+		tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_POLARITY_MASK;
+	WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
+}
+
+/**
+ * dce_v6_0_hpd_init - hpd setup callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup the hpd pins used by the card (evergreen+).
+ * Enable the pin, set the polarity, and enable the hpd interrupts.
+ */
+static void dce_v6_0_hpd_init(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
+			continue;
+
+		tmp = RREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+		tmp |= DC_HPD1_CONTROL__DC_HPD1_EN_MASK;
+		WREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP ||
+		    connector->connector_type == DRM_MODE_CONNECTOR_LVDS) {
+			/* don't try to enable hpd on eDP or LVDS avoid breaking the
+			 * aux dp channel on imac and help (but not completely fix)
+			 * https://bugzilla.redhat.com/show_bug.cgi?id=726143
+			 * also avoid interrupt storms during dpms.
+			 */
+			tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+			tmp &= ~DC_HPD1_INT_CONTROL__DC_HPD1_INT_EN_MASK;
+			WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+			continue;
+		}
+
+		dce_v6_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
+		amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
+	}
+
+}
+
+/**
+ * dce_v6_0_hpd_fini - hpd tear down callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the hpd pins used by the card (evergreen+).
+ * Disable the hpd interrupts.
+ */
+static void dce_v6_0_hpd_fini(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
+			continue;
+
+		tmp = RREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+		tmp &= ~DC_HPD1_CONTROL__DC_HPD1_EN_MASK;
+		WREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], 0);
+
+		amdgpu_irq_put(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
+	}
+}
+
+static u32 dce_v6_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
+{
+	return mmDC_GPIO_HPD_A;
+}
+
+static void dce_v6_0_set_vga_render_state(struct amdgpu_device *adev,
+					  bool render)
+{
+	if (!render)
+		WREG32(mmVGA_RENDER_CONTROL,
+			RREG32(mmVGA_RENDER_CONTROL) & VGA_VSTATUS_CNTL);
+
+}
+
+static int dce_v6_0_get_num_crtc(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+		return 6;
+	case CHIP_OLAND:
+		return 2;
+	default:
+		return 0;
+	}
+}
+
+void dce_v6_0_disable_dce(struct amdgpu_device *adev)
+{
+	/*Disable VGA render and enabled crtc, if has DCE engine*/
+	if (amdgpu_atombios_has_dce_engine_info(adev)) {
+		u32 tmp;
+		int crtc_enabled, i;
+
+		dce_v6_0_set_vga_render_state(adev, false);
+
+		/*Disable crtc*/
+		for (i = 0; i < dce_v6_0_get_num_crtc(adev); i++) {
+			crtc_enabled = RREG32(mmCRTC_CONTROL + crtc_offsets[i]) &
+				CRTC_CONTROL__CRTC_MASTER_EN_MASK;
+			if (crtc_enabled) {
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+				tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
+				tmp &= ~CRTC_CONTROL__CRTC_MASTER_EN_MASK;
+				WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			}
+		}
+	}
+}
+
+static void dce_v6_0_program_fmt(struct drm_encoder *encoder)
+{
+
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	int bpc = 0;
+	u32 tmp = 0;
+	enum amdgpu_connector_dither dither = AMDGPU_FMT_DITHER_DISABLE;
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		bpc = amdgpu_connector_get_monitor_bpc(connector);
+		dither = amdgpu_connector->dither;
+	}
+
+	/* LVDS FMT is set up by atom */
+	if (amdgpu_encoder->devices & ATOM_DEVICE_LCD_SUPPORT)
+		return;
+
+	if (bpc == 0)
+		return;
+
+
+	switch (bpc) {
+	case 6:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE)
+			/* XXX sort out optimal dither settings */
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_FRAME_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_HIGHPASS_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_EN_MASK);
+		else
+			tmp |= FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_EN_MASK;
+		break;
+	case 8:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE)
+			/* XXX sort out optimal dither settings */
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_FRAME_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_HIGHPASS_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_RGB_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_EN_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_DEPTH_MASK);
+		else
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_EN_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_DEPTH_MASK);
+		break;
+	case 10:
+	default:
+		/* not needed */
+		break;
+	}
+
+	WREG32(mmFMT_BIT_DEPTH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+/**
+ * cik_get_number_of_dram_channels - get the number of dram channels
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the number of video ram channels (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the number of dram channels
+ */
+static u32 si_get_number_of_dram_channels(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(mmMC_SHARED_CHMAP);
+
+	switch ((tmp & MC_SHARED_CHMAP__NOOFCHAN_MASK) >> MC_SHARED_CHMAP__NOOFCHAN__SHIFT) {
+	case 0:
+	default:
+		return 1;
+	case 1:
+		return 2;
+	case 2:
+		return 4;
+	case 3:
+		return 8;
+	case 4:
+		return 3;
+	case 5:
+		return 6;
+	case 6:
+		return 10;
+	case 7:
+		return 12;
+	case 8:
+		return 16;
+	}
+}
+
+struct dce6_wm_params {
+	u32 dram_channels; /* number of dram channels */
+	u32 yclk;          /* bandwidth per dram data pin in kHz */
+	u32 sclk;          /* engine clock in kHz */
+	u32 disp_clk;      /* display clock in kHz */
+	u32 src_width;     /* viewport width */
+	u32 active_time;   /* active display time in ns */
+	u32 blank_time;    /* blank time in ns */
+	bool interlaced;    /* mode is interlaced */
+	fixed20_12 vsc;    /* vertical scale ratio */
+	u32 num_heads;     /* number of active crtcs */
+	u32 bytes_per_pixel; /* bytes per pixel display + overlay */
+	u32 lb_size;       /* line buffer allocated to pipe */
+	u32 vtaps;         /* vertical scaler taps */
+};
+
+/**
+ * dce_v6_0_dram_bandwidth - get the dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the raw dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth in MBytes/s
+ */
+static u32 dce_v6_0_dram_bandwidth(struct dce6_wm_params *wm)
+{
+	/* Calculate raw DRAM Bandwidth */
+	fixed20_12 dram_efficiency; /* 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	dram_efficiency.full = dfixed_const(7);
+	dram_efficiency.full = dfixed_div(dram_efficiency, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, dram_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v6_0_dram_bandwidth_for_display - get the dram bandwidth for display
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dram bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth for display in MBytes/s
+ */
+static u32 dce_v6_0_dram_bandwidth_for_display(struct dce6_wm_params *wm)
+{
+	/* Calculate DRAM Bandwidth and the part allocated to display. */
+	fixed20_12 disp_dram_allocation; /* 0.3 to 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	disp_dram_allocation.full = dfixed_const(3); /* XXX worse case value 0.3 */
+	disp_dram_allocation.full = dfixed_div(disp_dram_allocation, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, disp_dram_allocation);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v6_0_data_return_bandwidth - get the data return bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the data return bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the data return bandwidth in MBytes/s
+ */
+static u32 dce_v6_0_data_return_bandwidth(struct dce6_wm_params *wm)
+{
+	/* Calculate the display Data return Bandwidth */
+	fixed20_12 return_efficiency; /* 0.8 */
+	fixed20_12 sclk, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	sclk.full = dfixed_const(wm->sclk);
+	sclk.full = dfixed_div(sclk, a);
+	a.full = dfixed_const(10);
+	return_efficiency.full = dfixed_const(8);
+	return_efficiency.full = dfixed_div(return_efficiency, a);
+	a.full = dfixed_const(32);
+	bandwidth.full = dfixed_mul(a, sclk);
+	bandwidth.full = dfixed_mul(bandwidth, return_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v6_0_dmif_request_bandwidth - get the dmif bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dmif bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dmif bandwidth in MBytes/s
+ */
+static u32 dce_v6_0_dmif_request_bandwidth(struct dce6_wm_params *wm)
+{
+	/* Calculate the DMIF Request Bandwidth */
+	fixed20_12 disp_clk_request_efficiency; /* 0.8 */
+	fixed20_12 disp_clk, bandwidth;
+	fixed20_12 a, b;
+
+	a.full = dfixed_const(1000);
+	disp_clk.full = dfixed_const(wm->disp_clk);
+	disp_clk.full = dfixed_div(disp_clk, a);
+	a.full = dfixed_const(32);
+	b.full = dfixed_mul(a, disp_clk);
+
+	a.full = dfixed_const(10);
+	disp_clk_request_efficiency.full = dfixed_const(8);
+	disp_clk_request_efficiency.full = dfixed_div(disp_clk_request_efficiency, a);
+
+	bandwidth.full = dfixed_mul(b, disp_clk_request_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v6_0_available_bandwidth - get the min available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the min available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the min available bandwidth in MBytes/s
+ */
+static u32 dce_v6_0_available_bandwidth(struct dce6_wm_params *wm)
+{
+	/* Calculate the Available bandwidth. Display can use this temporarily but not in average. */
+	u32 dram_bandwidth = dce_v6_0_dram_bandwidth(wm);
+	u32 data_return_bandwidth = dce_v6_0_data_return_bandwidth(wm);
+	u32 dmif_req_bandwidth = dce_v6_0_dmif_request_bandwidth(wm);
+
+	return min(dram_bandwidth, min(data_return_bandwidth, dmif_req_bandwidth));
+}
+
+/**
+ * dce_v6_0_average_bandwidth - get the average available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the average available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the average available bandwidth in MBytes/s
+ */
+static u32 dce_v6_0_average_bandwidth(struct dce6_wm_params *wm)
+{
+	/* Calculate the display mode Average Bandwidth
+	 * DisplayMode should contain the source and destination dimensions,
+	 * timing, etc.
+	 */
+	fixed20_12 bpp;
+	fixed20_12 line_time;
+	fixed20_12 src_width;
+	fixed20_12 bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	line_time.full = dfixed_const(wm->active_time + wm->blank_time);
+	line_time.full = dfixed_div(line_time, a);
+	bpp.full = dfixed_const(wm->bytes_per_pixel);
+	src_width.full = dfixed_const(wm->src_width);
+	bandwidth.full = dfixed_mul(src_width, bpp);
+	bandwidth.full = dfixed_mul(bandwidth, wm->vsc);
+	bandwidth.full = dfixed_div(bandwidth, line_time);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v6_0_latency_watermark - get the latency watermark
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the latency watermark (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the latency watermark in ns
+ */
+static u32 dce_v6_0_latency_watermark(struct dce6_wm_params *wm)
+{
+	/* First calculate the latency in ns */
+	u32 mc_latency = 2000; /* 2000 ns. */
+	u32 available_bandwidth = dce_v6_0_available_bandwidth(wm);
+	u32 worst_chunk_return_time = (512 * 8 * 1000) / available_bandwidth;
+	u32 cursor_line_pair_return_time = (128 * 4 * 1000) / available_bandwidth;
+	u32 dc_latency = 40000000 / wm->disp_clk; /* dc pipe latency */
+	u32 other_heads_data_return_time = ((wm->num_heads + 1) * worst_chunk_return_time) +
+		(wm->num_heads * cursor_line_pair_return_time);
+	u32 latency = mc_latency + other_heads_data_return_time + dc_latency;
+	u32 max_src_lines_per_dst_line, lb_fill_bw, line_fill_time;
+	u32 tmp, dmif_size = 12288;
+	fixed20_12 a, b, c;
+
+	if (wm->num_heads == 0)
+		return 0;
+
+	a.full = dfixed_const(2);
+	b.full = dfixed_const(1);
+	if ((wm->vsc.full > a.full) ||
+	    ((wm->vsc.full > b.full) && (wm->vtaps >= 3)) ||
+	    (wm->vtaps >= 5) ||
+	    ((wm->vsc.full >= a.full) && wm->interlaced))
+		max_src_lines_per_dst_line = 4;
+	else
+		max_src_lines_per_dst_line = 2;
+
+	a.full = dfixed_const(available_bandwidth);
+	b.full = dfixed_const(wm->num_heads);
+	a.full = dfixed_div(a, b);
+	tmp = div_u64((u64) dmif_size * (u64) wm->disp_clk, mc_latency + 512);
+	tmp = min(dfixed_trunc(a), tmp);
+
+	lb_fill_bw = min(tmp, wm->disp_clk * wm->bytes_per_pixel / 1000);
+
+	a.full = dfixed_const(max_src_lines_per_dst_line * wm->src_width * wm->bytes_per_pixel);
+	b.full = dfixed_const(1000);
+	c.full = dfixed_const(lb_fill_bw);
+	b.full = dfixed_div(c, b);
+	a.full = dfixed_div(a, b);
+	line_fill_time = dfixed_trunc(a);
+
+	if (line_fill_time < wm->active_time)
+		return latency;
+	else
+		return latency + (line_fill_time - wm->active_time);
+
+}
+
+/**
+ * dce_v6_0_average_bandwidth_vs_dram_bandwidth_for_display - check
+ * average and available dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v6_0_average_bandwidth_vs_dram_bandwidth_for_display(struct dce6_wm_params *wm)
+{
+	if (dce_v6_0_average_bandwidth(wm) <=
+	    (dce_v6_0_dram_bandwidth_for_display(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v6_0_average_bandwidth_vs_available_bandwidth - check
+ * average and available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * available bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v6_0_average_bandwidth_vs_available_bandwidth(struct dce6_wm_params *wm)
+{
+	if (dce_v6_0_average_bandwidth(wm) <=
+	    (dce_v6_0_available_bandwidth(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v6_0_check_latency_hiding - check latency hiding
+ *
+ * @wm: watermark calculation data
+ *
+ * Check latency hiding (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v6_0_check_latency_hiding(struct dce6_wm_params *wm)
+{
+	u32 lb_partitions = wm->lb_size / wm->src_width;
+	u32 line_time = wm->active_time + wm->blank_time;
+	u32 latency_tolerant_lines;
+	u32 latency_hiding;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1);
+	if (wm->vsc.full > a.full)
+		latency_tolerant_lines = 1;
+	else {
+		if (lb_partitions <= (wm->vtaps + 1))
+			latency_tolerant_lines = 1;
+		else
+			latency_tolerant_lines = 2;
+	}
+
+	latency_hiding = (latency_tolerant_lines * line_time + wm->blank_time);
+
+	if (dce_v6_0_latency_watermark(wm) <= latency_hiding)
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v6_0_program_watermarks - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @lb_size: line buffer size
+ * @num_heads: number of display controllers in use
+ *
+ * Calculate and program the display watermarks for the
+ * selected display controller (CIK).
+ */
+static void dce_v6_0_program_watermarks(struct amdgpu_device *adev,
+					struct amdgpu_crtc *amdgpu_crtc,
+					u32 lb_size, u32 num_heads)
+{
+	struct drm_display_mode *mode = &amdgpu_crtc->base.mode;
+	struct dce6_wm_params wm_low, wm_high;
+	u32 dram_channels;
+	u32 active_time;
+	u32 line_time = 0;
+	u32 latency_watermark_a = 0, latency_watermark_b = 0;
+	u32 priority_a_mark = 0, priority_b_mark = 0;
+	u32 priority_a_cnt = PRIORITY_OFF;
+	u32 priority_b_cnt = PRIORITY_OFF;
+	u32 tmp, arb_control3, lb_vblank_lead_lines = 0;
+	fixed20_12 a, b, c;
+
+	if (amdgpu_crtc->base.enabled && num_heads && mode) {
+		active_time = (u32) div_u64((u64)mode->crtc_hdisplay * 1000000,
+					    (u32)mode->clock);
+		line_time = (u32) div_u64((u64)mode->crtc_htotal * 1000000,
+					  (u32)mode->clock);
+		line_time = min(line_time, (u32)65535);
+		priority_a_cnt = 0;
+		priority_b_cnt = 0;
+
+		dram_channels = si_get_number_of_dram_channels(adev);
+
+		/* watermark for high clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_high.yclk =
+				amdgpu_dpm_get_mclk(adev, false) * 10;
+			wm_high.sclk =
+				amdgpu_dpm_get_sclk(adev, false) * 10;
+		} else {
+			wm_high.yclk = adev->pm.current_mclk * 10;
+			wm_high.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_high.disp_clk = mode->clock;
+		wm_high.src_width = mode->crtc_hdisplay;
+		wm_high.active_time = active_time;
+		wm_high.blank_time = line_time - wm_high.active_time;
+		wm_high.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_high.interlaced = true;
+		wm_high.vsc = amdgpu_crtc->vsc;
+		wm_high.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_high.vtaps = 2;
+		wm_high.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_high.lb_size = lb_size;
+		wm_high.dram_channels = dram_channels;
+		wm_high.num_heads = num_heads;
+
+		if (adev->pm.dpm_enabled) {
+		/* watermark for low clocks */
+			wm_low.yclk =
+				amdgpu_dpm_get_mclk(adev, true) * 10;
+			wm_low.sclk =
+				amdgpu_dpm_get_sclk(adev, true) * 10;
+		} else {
+			wm_low.yclk = adev->pm.current_mclk * 10;
+			wm_low.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_low.disp_clk = mode->clock;
+		wm_low.src_width = mode->crtc_hdisplay;
+		wm_low.active_time = active_time;
+		wm_low.blank_time = line_time - wm_low.active_time;
+		wm_low.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_low.interlaced = true;
+		wm_low.vsc = amdgpu_crtc->vsc;
+		wm_low.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_low.vtaps = 2;
+		wm_low.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_low.lb_size = lb_size;
+		wm_low.dram_channels = dram_channels;
+		wm_low.num_heads = num_heads;
+
+		/* set for high clocks */
+		latency_watermark_a = min(dce_v6_0_latency_watermark(&wm_high), (u32)65535);
+		/* set for low clocks */
+		latency_watermark_b = min(dce_v6_0_latency_watermark(&wm_low), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v6_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_high) ||
+		    !dce_v6_0_average_bandwidth_vs_available_bandwidth(&wm_high) ||
+		    !dce_v6_0_check_latency_hiding(&wm_high) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+			priority_a_cnt |= PRIORITY_ALWAYS_ON;
+			priority_b_cnt |= PRIORITY_ALWAYS_ON;
+		}
+		if (!dce_v6_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_low) ||
+		    !dce_v6_0_average_bandwidth_vs_available_bandwidth(&wm_low) ||
+		    !dce_v6_0_check_latency_hiding(&wm_low) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+			priority_a_cnt |= PRIORITY_ALWAYS_ON;
+			priority_b_cnt |= PRIORITY_ALWAYS_ON;
+		}
+
+		a.full = dfixed_const(1000);
+		b.full = dfixed_const(mode->clock);
+		b.full = dfixed_div(b, a);
+		c.full = dfixed_const(latency_watermark_a);
+		c.full = dfixed_mul(c, b);
+		c.full = dfixed_mul(c, amdgpu_crtc->hsc);
+		c.full = dfixed_div(c, a);
+		a.full = dfixed_const(16);
+		c.full = dfixed_div(c, a);
+		priority_a_mark = dfixed_trunc(c);
+		priority_a_cnt |= priority_a_mark & PRIORITY_MARK_MASK;
+
+		a.full = dfixed_const(1000);
+		b.full = dfixed_const(mode->clock);
+		b.full = dfixed_div(b, a);
+		c.full = dfixed_const(latency_watermark_b);
+		c.full = dfixed_mul(c, b);
+		c.full = dfixed_mul(c, amdgpu_crtc->hsc);
+		c.full = dfixed_div(c, a);
+		a.full = dfixed_const(16);
+		c.full = dfixed_div(c, a);
+		priority_b_mark = dfixed_trunc(c);
+		priority_b_cnt |= priority_b_mark & PRIORITY_MARK_MASK;
+
+		lb_vblank_lead_lines = DIV_ROUND_UP(lb_size, mode->crtc_hdisplay);
+	}
+
+	/* select wm A */
+	arb_control3 = RREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset);
+	tmp = arb_control3;
+	tmp &= ~LATENCY_WATERMARK_MASK(3);
+	tmp |= LATENCY_WATERMARK_MASK(1);
+	WREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset, tmp);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((latency_watermark_a << DPG_PIPE_URGENCY_CONTROL__URGENCY_LOW_WATERMARK__SHIFT)  |
+		(line_time << DPG_PIPE_URGENCY_CONTROL__URGENCY_HIGH_WATERMARK__SHIFT)));
+	/* select wm B */
+	tmp = RREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset);
+	tmp &= ~LATENCY_WATERMARK_MASK(3);
+	tmp |= LATENCY_WATERMARK_MASK(2);
+	WREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset, tmp);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((latency_watermark_b << DPG_PIPE_URGENCY_CONTROL__URGENCY_LOW_WATERMARK__SHIFT) |
+		(line_time << DPG_PIPE_URGENCY_CONTROL__URGENCY_HIGH_WATERMARK__SHIFT)));
+	/* restore original selection */
+	WREG32(mmDPG_PIPE_ARBITRATION_CONTROL3 + amdgpu_crtc->crtc_offset, arb_control3);
+
+	/* write the priority marks */
+	WREG32(mmPRIORITY_A_CNT + amdgpu_crtc->crtc_offset, priority_a_cnt);
+	WREG32(mmPRIORITY_B_CNT + amdgpu_crtc->crtc_offset, priority_b_cnt);
+
+	/* save values for DPM */
+	amdgpu_crtc->line_time = line_time;
+	amdgpu_crtc->wm_high = latency_watermark_a;
+
+	/* Save number of lines the linebuffer leads before the scanout */
+	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
+}
+
+/* watermark setup */
+static u32 dce_v6_0_line_buffer_adjust(struct amdgpu_device *adev,
+				   struct amdgpu_crtc *amdgpu_crtc,
+				   struct drm_display_mode *mode,
+				   struct drm_display_mode *other_mode)
+{
+	u32 tmp, buffer_alloc, i;
+	u32 pipe_offset = amdgpu_crtc->crtc_id * 0x8;
+	/*
+	 * Line Buffer Setup
+	 * There are 3 line buffers, each one shared by 2 display controllers.
+	 * mmDC_LB_MEMORY_SPLIT controls how that line buffer is shared between
+	 * the display controllers.  The paritioning is done via one of four
+	 * preset allocations specified in bits 21:20:
+	 *  0 - half lb
+	 *  2 - whole lb, other crtc must be disabled
+	 */
+	/* this can get tricky if we have two large displays on a paired group
+	 * of crtcs.  Ideally for multiple large displays we'd assign them to
+	 * non-linked crtcs for maximum line buffer allocation.
+	 */
+	if (amdgpu_crtc->base.enabled && mode) {
+		if (other_mode) {
+			tmp = 0; /* 1/2 */
+			buffer_alloc = 1;
+		} else {
+			tmp = 2; /* whole */
+			buffer_alloc = 2;
+		}
+	} else {
+		tmp = 0;
+		buffer_alloc = 0;
+	}
+
+	WREG32(mmDC_LB_MEMORY_SPLIT + amdgpu_crtc->crtc_offset,
+	       DC_LB_MEMORY_CONFIG(tmp));
+
+	WREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset,
+	       (buffer_alloc << PIPE0_DMIF_BUFFER_CONTROL__DMIF_BUFFERS_ALLOCATED__SHIFT));
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset) &
+		    PIPE0_DMIF_BUFFER_CONTROL__DMIF_BUFFERS_ALLOCATION_COMPLETED_MASK)
+			break;
+		udelay(1);
+	}
+
+	if (amdgpu_crtc->base.enabled && mode) {
+		switch (tmp) {
+		case 0:
+		default:
+			return 4096 * 2;
+		case 2:
+			return 8192 * 2;
+		}
+	}
+
+	/* controller not enabled, so no lb used */
+	return 0;
+}
+
+
+/**
+ *
+ * dce_v6_0_bandwidth_update - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Calculate and program the display watermarks and line
+ * buffer allocation (CIK).
+ */
+static void dce_v6_0_bandwidth_update(struct amdgpu_device *adev)
+{
+	struct drm_display_mode *mode0 = NULL;
+	struct drm_display_mode *mode1 = NULL;
+	u32 num_heads = 0, lb_size;
+	int i;
+
+	if (!adev->mode_info.mode_config_initialized)
+		return;
+
+	amdgpu_display_update_priority(adev);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i]->base.enabled)
+			num_heads++;
+	}
+	for (i = 0; i < adev->mode_info.num_crtc; i += 2) {
+		mode0 = &adev->mode_info.crtcs[i]->base.mode;
+		mode1 = &adev->mode_info.crtcs[i+1]->base.mode;
+		lb_size = dce_v6_0_line_buffer_adjust(adev, adev->mode_info.crtcs[i], mode0, mode1);
+		dce_v6_0_program_watermarks(adev, adev->mode_info.crtcs[i], lb_size, num_heads);
+		lb_size = dce_v6_0_line_buffer_adjust(adev, adev->mode_info.crtcs[i+1], mode1, mode0);
+		dce_v6_0_program_watermarks(adev, adev->mode_info.crtcs[i+1], lb_size, num_heads);
+	}
+}
+
+static void dce_v6_0_audio_get_connected_pins(struct amdgpu_device *adev)
+{
+	int i;
+	u32 tmp;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		tmp = RREG32_AUDIO_ENDPT(adev->mode_info.audio.pin[i].offset,
+				ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT);
+		if (REG_GET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT,
+					PORT_CONNECTIVITY))
+			adev->mode_info.audio.pin[i].connected = false;
+		else
+			adev->mode_info.audio.pin[i].connected = true;
+	}
+
+}
+
+static struct amdgpu_audio_pin *dce_v6_0_audio_get_pin(struct amdgpu_device *adev)
+{
+	int i;
+
+	dce_v6_0_audio_get_connected_pins(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		if (adev->mode_info.audio.pin[i].connected)
+			return &adev->mode_info.audio.pin[i];
+	}
+	DRM_ERROR("No connected audio pins found!\n");
+	return NULL;
+}
+
+static void dce_v6_0_audio_select_pin(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	WREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset,
+	       REG_SET_FIELD(0, AFMT_AUDIO_SRC_CONTROL, AFMT_AUDIO_SRC_SELECT,
+		             dig->afmt->pin->id));
+}
+
+static void dce_v6_0_audio_write_latency_fields(struct drm_encoder *encoder,
+						struct drm_display_mode *mode)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	int interlace = 0;
+	u32 tmp;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		interlace = 1;
+
+	if (connector->latency_present[interlace]) {
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				VIDEO_LIPSYNC, connector->video_latency[interlace]);
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				AUDIO_LIPSYNC, connector->audio_latency[interlace]);
+	} else {
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				VIDEO_LIPSYNC, 0);
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				AUDIO_LIPSYNC, 0);
+	}
+	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+			   ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC, tmp);
+}
+
+static void dce_v6_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u8 *sadb = NULL;
+	int sad_count;
+	u32 tmp;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector_edid(connector), &sadb);
+	if (sad_count < 0) {
+		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sad_count);
+		sad_count = 0;
+	}
+
+	/* program the speaker allocation */
+	tmp = RREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+			ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER);
+	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			HDMI_CONNECTION, 0);
+	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			DP_CONNECTION, 0);
+
+	if (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort)
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+				DP_CONNECTION, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+				HDMI_CONNECTION, 1);
+
+	if (sad_count)
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+				SPEAKER_ALLOCATION, sadb[0]);
+	else
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+				SPEAKER_ALLOCATION, 5); /* stereo */
+
+	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+			ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER, tmp);
+
+	kfree(sadb);
+}
+
+static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	struct cea_sad *sads;
+	int i, sad_count;
+
+	static const u16 eld_reg_to_type[][2] = {
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0, HDMI_AUDIO_CODING_TYPE_PCM },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR1, HDMI_AUDIO_CODING_TYPE_AC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR2, HDMI_AUDIO_CODING_TYPE_MPEG1 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR3, HDMI_AUDIO_CODING_TYPE_MP3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR4, HDMI_AUDIO_CODING_TYPE_MPEG2 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR5, HDMI_AUDIO_CODING_TYPE_AAC_LC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR6, HDMI_AUDIO_CODING_TYPE_DTS },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR7, HDMI_AUDIO_CODING_TYPE_ATRAC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR9, HDMI_AUDIO_CODING_TYPE_EAC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR10, HDMI_AUDIO_CODING_TYPE_DTS_HD },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR11, HDMI_AUDIO_CODING_TYPE_MLP },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO },
+	};
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_sad(amdgpu_connector_edid(connector), &sads);
+	if (sad_count <= 0) {
+		DRM_ERROR("Couldn't read SADs: %d\n", sad_count);
+		return;
+	}
+
+	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
+		u32 tmp = 0;
+		u8 stereo_freqs = 0;
+		int max_channels = -1;
+		int j;
+
+		for (j = 0; j < sad_count; j++) {
+			struct cea_sad *sad = &sads[j];
+
+			if (sad->format == eld_reg_to_type[i][1]) {
+				if (sad->channels > max_channels) {
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							MAX_CHANNELS, sad->channels);
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							DESCRIPTOR_BYTE_2, sad->byte2);
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							SUPPORTED_FREQUENCIES, sad->freq);
+					max_channels = sad->channels;
+				}
+
+				if (sad->format == HDMI_AUDIO_CODING_TYPE_PCM)
+					stereo_freqs |= sad->freq;
+				else
+					break;
+			}
+		}
+
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+				SUPPORTED_FREQUENCIES_STEREO, stereo_freqs);
+		WREG32_AUDIO_ENDPT(dig->afmt->pin->offset, eld_reg_to_type[i][0], tmp);
+	}
+
+	kfree(sads);
+
+}
+
+static void dce_v6_0_audio_enable(struct amdgpu_device *adev,
+				  struct amdgpu_audio_pin *pin,
+				  bool enable)
+{
+	if (!pin)
+		return;
+
+	WREG32_AUDIO_ENDPT(pin->offset, ixAZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			enable ? AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL__AUDIO_ENABLED_MASK : 0);
+}
+
+static const u32 pin_offsets[7] =
+{
+	(0x1780 - 0x1780),
+	(0x1786 - 0x1780),
+	(0x178c - 0x1780),
+	(0x1792 - 0x1780),
+	(0x1798 - 0x1780),
+	(0x179d - 0x1780),
+	(0x17a4 - 0x1780),
+};
+
+static int dce_v6_0_audio_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!amdgpu_audio)
+		return 0;
+
+	adev->mode_info.audio.enabled = true;
+
+	switch (adev->asic_type) {
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+	default:
+		adev->mode_info.audio.num_pins = 6;
+		break;
+	case CHIP_OLAND:
+		adev->mode_info.audio.num_pins = 2;
+		break;
+	}
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		adev->mode_info.audio.pin[i].channels = -1;
+		adev->mode_info.audio.pin[i].rate = -1;
+		adev->mode_info.audio.pin[i].bits_per_sample = -1;
+		adev->mode_info.audio.pin[i].status_bits = 0;
+		adev->mode_info.audio.pin[i].category_code = 0;
+		adev->mode_info.audio.pin[i].connected = false;
+		adev->mode_info.audio.pin[i].offset = pin_offsets[i];
+		adev->mode_info.audio.pin[i].id = i;
+		dce_v6_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	return 0;
+}
+
+static void dce_v6_0_audio_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!amdgpu_audio)
+		return;
+
+	if (!adev->mode_info.audio.enabled)
+		return;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
+		dce_v6_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+
+	adev->mode_info.audio.enabled = false;
+}
+
+static void dce_v6_0_audio_set_vbi_packet(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1);
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, 1);
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, 1);
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp);
+}
+
+static void dce_v6_0_audio_set_acr(struct drm_encoder *encoder,
+				   uint32_t clock, int bpc)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_afmt_acr acr = amdgpu_afmt_acr(clock);
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	tmp = RREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, 1);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE,
+			bpc > 8 ? 0 : 1);
+	WREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_32_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_0, HDMI_ACR_CTS_32, acr.cts_32khz);
+	WREG32(mmHDMI_ACR_32_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_32_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_1, HDMI_ACR_N_32, acr.n_32khz);
+	WREG32(mmHDMI_ACR_32_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_44_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_0, HDMI_ACR_CTS_44, acr.cts_44_1khz);
+	WREG32(mmHDMI_ACR_44_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_44_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_1, HDMI_ACR_N_44, acr.n_44_1khz);
+	WREG32(mmHDMI_ACR_44_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_48_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_0, HDMI_ACR_CTS_48, acr.cts_48khz);
+	WREG32(mmHDMI_ACR_48_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_48_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_1, HDMI_ACR_N_48, acr.n_48khz);
+	WREG32(mmHDMI_ACR_48_1 + dig->afmt->offset, tmp);
+}
+
+static void dce_v6_0_audio_set_avi_infoframe(struct drm_encoder *encoder,
+					       struct drm_display_mode *mode)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct hdmi_avi_infoframe frame;
+	u8 buffer[HDMI_INFOFRAME_HEADER_SIZE + HDMI_AVI_INFOFRAME_SIZE];
+	uint8_t *payload = buffer + 3;
+	uint8_t *header = buffer;
+	ssize_t err;
+	u32 tmp;
+
+	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, mode, false);
+	if (err < 0) {
+		DRM_ERROR("failed to setup AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	err = hdmi_avi_infoframe_pack(&frame, buffer, sizeof(buffer));
+	if (err < 0) {
+		DRM_ERROR("failed to pack AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	WREG32(mmAFMT_AVI_INFO0 + dig->afmt->offset,
+	       payload[0x0] | (payload[0x1] << 8) | (payload[0x2] << 16) | (payload[0x3] << 24));
+	WREG32(mmAFMT_AVI_INFO1 + dig->afmt->offset,
+	       payload[0x4] | (payload[0x5] << 8) | (payload[0x6] << 16) | (payload[0x7] << 24));
+	WREG32(mmAFMT_AVI_INFO2 + dig->afmt->offset,
+	       payload[0x8] | (payload[0x9] << 8) | (payload[0xA] << 16) | (payload[0xB] << 24));
+	WREG32(mmAFMT_AVI_INFO3 + dig->afmt->offset,
+	       payload[0xC] | (payload[0xD] << 8) | (header[1] << 24));
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
+	/* anything other than 0 */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1,
+			HDMI_AUDIO_INFO_LINE, 2);
+	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
+}
+
+static void dce_v6_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	int em = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+	u32 tmp;
+
+	/*
+	 * Two dtos: generally use dto0 for hdmi, dto1 for dp.
+	 * Express [24MHz / target pixel clock] as an exact rational
+	 * number (coefficient of two integer numbers.  DCCG_AUDIO_DTOx_PHASE
+	 * is the numerator, DCCG_AUDIO_DTOx_MODULE is the denominator
+	 */
+	tmp = RREG32(mmDCCG_AUDIO_DTO_SOURCE);
+	tmp = REG_SET_FIELD(tmp, DCCG_AUDIO_DTO_SOURCE,
+			DCCG_AUDIO_DTO0_SOURCE_SEL, amdgpu_crtc->crtc_id);
+	if (em == ATOM_ENCODER_MODE_HDMI) {
+		tmp = REG_SET_FIELD(tmp, DCCG_AUDIO_DTO_SOURCE,
+				DCCG_AUDIO_DTO_SEL, 0);
+	} else if (ENCODER_MODE_IS_DP(em)) {
+		tmp = REG_SET_FIELD(tmp, DCCG_AUDIO_DTO_SOURCE,
+				DCCG_AUDIO_DTO_SEL, 1);
+	}
+	WREG32(mmDCCG_AUDIO_DTO_SOURCE, tmp);
+	if (em == ATOM_ENCODER_MODE_HDMI) {
+		WREG32(mmDCCG_AUDIO_DTO0_PHASE, 24000);
+		WREG32(mmDCCG_AUDIO_DTO0_MODULE, clock);
+	} else if (ENCODER_MODE_IS_DP(em)) {
+		WREG32(mmDCCG_AUDIO_DTO1_PHASE, 24000);
+		WREG32(mmDCCG_AUDIO_DTO1_MODULE, clock);
+	}
+}
+
+static void dce_v6_0_audio_set_packet(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	tmp = RREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
+	WREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_60958_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_0, AFMT_60958_CS_CHANNEL_NUMBER_L, 1);
+	WREG32(mmAFMT_60958_0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_60958_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, 2);
+	WREG32(mmAFMT_60958_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_60958_2 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_2, 3);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_3, 4);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_4, 5);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_5, 6);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_6, 7);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_7, 8);
+	WREG32(mmAFMT_60958_2 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL2 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL2, AFMT_AUDIO_CHANNEL_ENABLE, 0xff);
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL2 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, 1);
+	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_PACKETS_PER_LINE, 3);
+	WREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_RESET_FIFO_WHEN_AUDIO_DIS, 1);
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+}
+
+static void dce_v6_0_audio_set_mute(struct drm_encoder *encoder, bool mute)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	tmp = RREG32(mmHDMI_GC + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_GC, HDMI_GC_AVMUTE, mute ? 1 : 0);
+	WREG32(mmHDMI_GC + dig->afmt->offset, tmp);
+}
+
+static void dce_v6_0_audio_hdmi_enable(struct drm_encoder *encoder, bool enable)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	if (enable) {
+		tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
+		tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_SEND, 1);
+		tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_CONT, 1);
+		tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
+		tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_CONT, 1);
+		WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+		tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
+		tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AVI_INFO_LINE, 2);
+		WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
+
+		tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+		tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, 1);
+		WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+	} else {
+		tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
+		tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_SEND, 0);
+		tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_CONT, 0);
+		tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 0);
+		tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_CONT, 0);
+		WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+		tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+		tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, 0);
+		WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+	}
+}
+
+static void dce_v6_0_audio_dp_enable(struct drm_encoder *encoder, bool enable)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	if (enable) {
+		tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+		tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, 1);
+		WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+		tmp = RREG32(mmDP_SEC_TIMESTAMP + dig->afmt->offset);
+		tmp = REG_SET_FIELD(tmp, DP_SEC_TIMESTAMP, DP_SEC_TIMESTAMP_MODE, 1);
+		WREG32(mmDP_SEC_TIMESTAMP + dig->afmt->offset, tmp);
+
+		tmp = RREG32(mmDP_SEC_CNTL + dig->afmt->offset);
+		tmp = REG_SET_FIELD(tmp, DP_SEC_CNTL, DP_SEC_ASP_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, DP_SEC_CNTL, DP_SEC_ATP_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, DP_SEC_CNTL, DP_SEC_AIP_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, DP_SEC_CNTL, DP_SEC_STREAM_ENABLE, 1);
+		WREG32(mmDP_SEC_CNTL + dig->afmt->offset, tmp);
+	} else {
+		WREG32(mmDP_SEC_CNTL + dig->afmt->offset, 0);
+	}
+}
+
+static void dce_v6_0_afmt_setmode(struct drm_encoder *encoder,
+				  struct drm_display_mode *mode)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	int em = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+	int bpc = 8;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	if (!dig->afmt->enabled)
+		return;
+
+	dig->afmt->pin = dce_v6_0_audio_get_pin(adev);
+	if (!dig->afmt->pin)
+		return;
+
+	if (encoder->crtc) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+		bpc = amdgpu_crtc->bpc;
+	}
+
+	/* disable audio before setting up hw */
+	dce_v6_0_audio_enable(adev, dig->afmt->pin, false);
+
+	dce_v6_0_audio_set_mute(encoder, true);
+	dce_v6_0_audio_write_speaker_allocation(encoder);
+	dce_v6_0_audio_write_sad_regs(encoder);
+	dce_v6_0_audio_write_latency_fields(encoder, mode);
+	if (em == ATOM_ENCODER_MODE_HDMI) {
+		dce_v6_0_audio_set_dto(encoder, mode->clock);
+		dce_v6_0_audio_set_vbi_packet(encoder);
+		dce_v6_0_audio_set_acr(encoder, mode->clock, bpc);
+	} else if (ENCODER_MODE_IS_DP(em)) {
+		dce_v6_0_audio_set_dto(encoder, adev->clock.default_dispclk * 10);
+	}
+	dce_v6_0_audio_set_packet(encoder);
+	dce_v6_0_audio_select_pin(encoder);
+	dce_v6_0_audio_set_avi_infoframe(encoder, mode);
+	dce_v6_0_audio_set_mute(encoder, false);
+	if (em == ATOM_ENCODER_MODE_HDMI) {
+		dce_v6_0_audio_hdmi_enable(encoder, 1);
+	} else if (ENCODER_MODE_IS_DP(em)) {
+		dce_v6_0_audio_dp_enable(encoder, 1);
+	}
+
+	/* enable audio after setting up hw */
+	dce_v6_0_audio_enable(adev, dig->afmt->pin, true);
+}
+
+static void dce_v6_0_afmt_enable(struct drm_encoder *encoder, bool enable)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (enable && dig->afmt->enabled)
+		return;
+
+	if (!enable && !dig->afmt->enabled)
+		return;
+
+	if (!enable && dig->afmt->pin) {
+		dce_v6_0_audio_enable(adev, dig->afmt->pin, false);
+		dig->afmt->pin = NULL;
+	}
+
+	dig->afmt->enabled = enable;
+
+	DRM_DEBUG("%sabling AFMT interface @ 0x%04X for encoder 0x%x\n",
+		  enable ? "En" : "Dis", dig->afmt->offset, amdgpu_encoder->encoder_id);
+}
+
+static int dce_v6_0_afmt_init(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++)
+		adev->mode_info.afmt[i] = NULL;
+
+	/* DCE6 has audio blocks tied to DIG encoders */
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		adev->mode_info.afmt[i] = kzalloc(sizeof(struct amdgpu_afmt), GFP_KERNEL);
+		if (adev->mode_info.afmt[i]) {
+			adev->mode_info.afmt[i]->offset = dig_offsets[i];
+			adev->mode_info.afmt[i]->id = i;
+		} else {
+			for (j = 0; j < i; j++) {
+				kfree(adev->mode_info.afmt[j]);
+				adev->mode_info.afmt[j] = NULL;
+			}
+			DRM_ERROR("Out of memory allocating afmt table\n");
+			return -ENOMEM;
+		}
+	}
+	return 0;
+}
+
+static void dce_v6_0_afmt_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		kfree(adev->mode_info.afmt[i]);
+		adev->mode_info.afmt[i] = NULL;
+	}
+}
+
+static const u32 vga_control_regs[6] =
+{
+	mmD1VGA_CONTROL,
+	mmD2VGA_CONTROL,
+	mmD3VGA_CONTROL,
+	mmD4VGA_CONTROL,
+	mmD5VGA_CONTROL,
+	mmD6VGA_CONTROL,
+};
+
+static void dce_v6_0_vga_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 vga_control;
+
+	vga_control = RREG32(vga_control_regs[amdgpu_crtc->crtc_id]) & ~1;
+	WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control | (enable ? 1 : 0));
+}
+
+static void dce_v6_0_grph_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, enable ? 1 : 0);
+}
+
+static int dce_v6_0_crtc_do_set_base(struct drm_crtc *crtc,
+				     struct drm_framebuffer *fb,
+				     int x, int y, int atomic)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_framebuffer *target_fb;
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *abo;
+	uint64_t fb_location, tiling_flags;
+	uint32_t fb_format, fb_pitch_pixels, pipe_config;
+	u32 fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_NONE);
+	u32 viewport_w, viewport_h;
+	int r;
+	bool bypass_lut = false;
+	struct drm_format_name_buf format_name;
+
+	/* no fb bound */
+	if (!atomic && !crtc->primary->fb) {
+		DRM_DEBUG_KMS("No FB bound\n");
+		return 0;
+	}
+
+	if (atomic)
+		target_fb = fb;
+	else
+		target_fb = crtc->primary->fb;
+
+	/* If atomic, assume fb object is pinned & idle & fenced and
+	 * just update base pointers
+	 */
+	obj = target_fb->obj[0];
+	abo = gem_to_amdgpu_bo(obj);
+	r = amdgpu_bo_reserve(abo, false);
+	if (unlikely(r != 0))
+		return r;
+
+	if (!atomic) {
+		r = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM);
+		if (unlikely(r != 0)) {
+			amdgpu_bo_unreserve(abo);
+			return -EINVAL;
+		}
+	}
+	fb_location = amdgpu_bo_gpu_offset(abo);
+
+	amdgpu_bo_get_tiling_flags(abo, &tiling_flags);
+	amdgpu_bo_unreserve(abo);
+
+	switch (target_fb->format->format) {
+	case DRM_FORMAT_C8:
+		fb_format = (GRPH_DEPTH(GRPH_DEPTH_8BPP) |
+			     GRPH_FORMAT(GRPH_FORMAT_INDEXED));
+		break;
+	case DRM_FORMAT_XRGB4444:
+	case DRM_FORMAT_ARGB4444:
+		fb_format = (GRPH_DEPTH(GRPH_DEPTH_16BPP) |
+			     GRPH_FORMAT(GRPH_FORMAT_ARGB4444));
+#ifdef __BIG_ENDIAN
+		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_XRGB1555:
+	case DRM_FORMAT_ARGB1555:
+		fb_format = (GRPH_DEPTH(GRPH_DEPTH_16BPP) |
+			     GRPH_FORMAT(GRPH_FORMAT_ARGB1555));
+#ifdef __BIG_ENDIAN
+		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_BGRX5551:
+	case DRM_FORMAT_BGRA5551:
+		fb_format = (GRPH_DEPTH(GRPH_DEPTH_16BPP) |
+			     GRPH_FORMAT(GRPH_FORMAT_BGRA5551));
+#ifdef __BIG_ENDIAN
+		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_RGB565:
+		fb_format = (GRPH_DEPTH(GRPH_DEPTH_16BPP) |
+			     GRPH_FORMAT(GRPH_FORMAT_ARGB565));
+#ifdef __BIG_ENDIAN
+		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_XRGB8888:
+	case DRM_FORMAT_ARGB8888:
+		fb_format = (GRPH_DEPTH(GRPH_DEPTH_32BPP) |
+			     GRPH_FORMAT(GRPH_FORMAT_ARGB8888));
+#ifdef __BIG_ENDIAN
+		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN32);
+#endif
+		break;
+	case DRM_FORMAT_XRGB2101010:
+	case DRM_FORMAT_ARGB2101010:
+		fb_format = (GRPH_DEPTH(GRPH_DEPTH_32BPP) |
+			     GRPH_FORMAT(GRPH_FORMAT_ARGB2101010));
+#ifdef __BIG_ENDIAN
+		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN32);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	case DRM_FORMAT_BGRX1010102:
+	case DRM_FORMAT_BGRA1010102:
+		fb_format = (GRPH_DEPTH(GRPH_DEPTH_32BPP) |
+			     GRPH_FORMAT(GRPH_FORMAT_BGRA1010102));
+#ifdef __BIG_ENDIAN
+		fb_swap = GRPH_ENDIAN_SWAP(GRPH_ENDIAN_8IN32);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	default:
+		DRM_ERROR("Unsupported screen format %s\n",
+		          drm_get_format_name(target_fb->format->format, &format_name));
+		return -EINVAL;
+	}
+
+	if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_2D_TILED_THIN1) {
+		unsigned bankw, bankh, mtaspect, tile_split, num_banks;
+
+		bankw = AMDGPU_TILING_GET(tiling_flags, BANK_WIDTH);
+		bankh = AMDGPU_TILING_GET(tiling_flags, BANK_HEIGHT);
+		mtaspect = AMDGPU_TILING_GET(tiling_flags, MACRO_TILE_ASPECT);
+		tile_split = AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT);
+		num_banks = AMDGPU_TILING_GET(tiling_flags, NUM_BANKS);
+
+		fb_format |= GRPH_NUM_BANKS(num_banks);
+		fb_format |= GRPH_ARRAY_MODE(GRPH_ARRAY_2D_TILED_THIN1);
+		fb_format |= GRPH_TILE_SPLIT(tile_split);
+		fb_format |= GRPH_BANK_WIDTH(bankw);
+		fb_format |= GRPH_BANK_HEIGHT(bankh);
+		fb_format |= GRPH_MACRO_TILE_ASPECT(mtaspect);
+	} else if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_1D_TILED_THIN1) {
+		fb_format |= GRPH_ARRAY_MODE(GRPH_ARRAY_1D_TILED_THIN1);
+	}
+
+	pipe_config = AMDGPU_TILING_GET(tiling_flags, PIPE_CONFIG);
+	fb_format |= GRPH_PIPE_CONFIG(pipe_config);
+
+	dce_v6_0_vga_enable(crtc, false);
+
+	/* Make sure surface address is updated at vertical blank rather than
+	 * horizontal blank
+	 */
+	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32)fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32) fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
+	WREG32(mmGRPH_SWAP_CNTL + amdgpu_crtc->crtc_offset, fb_swap);
+
+	/*
+	 * The LUT only has 256 slots for indexing by a 8 bpc fb. Bypass the LUT
+	 * for > 8 bpc scanout to avoid truncation of fb indices to 8 msb's, to
+	 * retain the full precision throughout the pipeline.
+	 */
+	WREG32_P(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset,
+		 (bypass_lut ? GRPH_LUT_10BIT_BYPASS__GRPH_LUT_10BIT_BYPASS_EN_MASK : 0),
+		 ~GRPH_LUT_10BIT_BYPASS__GRPH_LUT_10BIT_BYPASS_EN_MASK);
+
+	if (bypass_lut)
+		DRM_DEBUG_KMS("Bypassing hardware LUT due to 10 bit fb scanout.\n");
+
+	WREG32(mmGRPH_SURFACE_OFFSET_X + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_SURFACE_OFFSET_Y + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_Y_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_END + amdgpu_crtc->crtc_offset, target_fb->width);
+	WREG32(mmGRPH_Y_END + amdgpu_crtc->crtc_offset, target_fb->height);
+
+	fb_pitch_pixels = target_fb->pitches[0] / target_fb->format->cpp[0];
+	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset, fb_pitch_pixels);
+
+	dce_v6_0_grph_enable(crtc, true);
+
+	WREG32(mmDESKTOP_HEIGHT + amdgpu_crtc->crtc_offset,
+		       target_fb->height);
+	x &= ~3;
+	y &= ~1;
+	WREG32(mmVIEWPORT_START + amdgpu_crtc->crtc_offset,
+	       (x << 16) | y);
+	viewport_w = crtc->mode.hdisplay;
+	viewport_h = (crtc->mode.vdisplay + 1) & ~1;
+
+	WREG32(mmVIEWPORT_SIZE + amdgpu_crtc->crtc_offset,
+	       (viewport_w << 16) | viewport_h);
+
+	/* set pageflip to happen anywhere in vblank interval */
+	WREG32(mmMASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 0);
+
+	if (!atomic && fb && fb != crtc->primary->fb) {
+		abo = gem_to_amdgpu_bo(fb->obj[0]);
+		r = amdgpu_bo_reserve(abo, true);
+		if (unlikely(r != 0))
+			return r;
+		amdgpu_bo_unpin(abo);
+		amdgpu_bo_unreserve(abo);
+	}
+
+	/* Bytes per pixel may have changed */
+	dce_v6_0_bandwidth_update(adev);
+
+	return 0;
+
+}
+
+static void dce_v6_0_set_interleave(struct drm_crtc *crtc,
+				    struct drm_display_mode *mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		WREG32(mmDATA_FORMAT + amdgpu_crtc->crtc_offset,
+		       INTERLEAVE_EN);
+	else
+		WREG32(mmDATA_FORMAT + amdgpu_crtc->crtc_offset, 0);
+}
+
+static void dce_v6_0_crtc_load_lut(struct drm_crtc *crtc)
+{
+
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u16 *r, *g, *b;
+	int i;
+
+	DRM_DEBUG_KMS("%d\n", amdgpu_crtc->crtc_id);
+
+	WREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((0 << INPUT_CSC_CONTROL__INPUT_CSC_GRPH_MODE__SHIFT) |
+		(0 << INPUT_CSC_CONTROL__INPUT_CSC_OVL_MODE__SHIFT)));
+	WREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset,
+	       PRESCALE_GRPH_CONTROL__GRPH_PRESCALE_BYPASS_MASK);
+	WREG32(mmPRESCALE_OVL_CONTROL + amdgpu_crtc->crtc_offset,
+	       PRESCALE_OVL_CONTROL__OVL_PRESCALE_BYPASS_MASK);
+	WREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((0 << INPUT_GAMMA_CONTROL__GRPH_INPUT_GAMMA_MODE__SHIFT) |
+		(0 << INPUT_GAMMA_CONTROL__OVL_INPUT_GAMMA_MODE__SHIFT)));
+
+	WREG32(mmDC_LUT_CONTROL + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_BLACK_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_RED + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_WHITE_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_RED + amdgpu_crtc->crtc_offset, 0xffff);
+
+	WREG32(mmDC_LUT_RW_MODE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_WRITE_EN_MASK + amdgpu_crtc->crtc_offset, 0x00000007);
+
+	WREG32(mmDC_LUT_RW_INDEX + amdgpu_crtc->crtc_offset, 0);
+	r = crtc->gamma_store;
+	g = r + crtc->gamma_size;
+	b = g + crtc->gamma_size;
+	for (i = 0; i < 256; i++) {
+		WREG32(mmDC_LUT_30_COLOR + amdgpu_crtc->crtc_offset,
+		       ((*r++ & 0xffc0) << 14) |
+		       ((*g++ & 0xffc0) << 4) |
+		       (*b++ >> 6));
+	}
+
+	WREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((0 << DEGAMMA_CONTROL__GRPH_DEGAMMA_MODE__SHIFT) |
+		(0 << DEGAMMA_CONTROL__OVL_DEGAMMA_MODE__SHIFT) |
+		ICON_DEGAMMA_MODE(0) |
+		(0 << DEGAMMA_CONTROL__CURSOR_DEGAMMA_MODE__SHIFT)));
+	WREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((0 << GAMUT_REMAP_CONTROL__GRPH_GAMUT_REMAP_MODE__SHIFT) |
+		(0 << GAMUT_REMAP_CONTROL__OVL_GAMUT_REMAP_MODE__SHIFT)));
+	WREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((0 << REGAMMA_CONTROL__GRPH_REGAMMA_MODE__SHIFT) |
+		(0 << REGAMMA_CONTROL__OVL_REGAMMA_MODE__SHIFT)));
+	WREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((0 << OUTPUT_CSC_CONTROL__OUTPUT_CSC_GRPH_MODE__SHIFT) |
+		(0 << OUTPUT_CSC_CONTROL__OUTPUT_CSC_OVL_MODE__SHIFT)));
+	/* XXX match this to the depth of the crtc fmt block, move to modeset? */
+	WREG32(0x1a50 + amdgpu_crtc->crtc_offset, 0);
+
+
+}
+
+static int dce_v6_0_pick_dig_encoder(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		return dig->linkb ? 1 : 0;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		return dig->linkb ? 3 : 2;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+		return dig->linkb ? 5 : 4;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		return 6;
+	default:
+		DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
+		return 0;
+	}
+}
+
+/**
+ * dce_v6_0_pick_pll - Allocate a PPLL for use by the crtc.
+ *
+ * @crtc: drm crtc
+ *
+ * Returns the PPLL (Pixel PLL) to be used by the crtc.  For DP monitors
+ * a single PPLL can be used for all DP crtcs/encoders.  For non-DP
+ * monitors a dedicated PPLL must be used.  If a particular board has
+ * an external DP PLL, return ATOM_PPLL_INVALID to skip PLL programming
+ * as there is no need to program the PLL itself.  If we are not able to
+ * allocate a PLL, return ATOM_PPLL_INVALID to skip PLL programming to
+ * avoid messing up an existing monitor.
+ *
+ *
+ */
+static u32 dce_v6_0_pick_pll(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 pll_in_use;
+	int pll;
+
+	if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder))) {
+		if (adev->clock.dp_extclk)
+			/* skip PPLL programming if using ext clock */
+			return ATOM_PPLL_INVALID;
+		else
+			return ATOM_PPLL0;
+	} else {
+		/* use the same PPLL for all monitors with the same clock */
+		pll = amdgpu_pll_get_shared_nondp_ppll(crtc);
+		if (pll != ATOM_PPLL_INVALID)
+			return pll;
+	}
+
+	/*  PPLL1, and PPLL2 */
+	pll_in_use = amdgpu_pll_get_use_mask(crtc);
+	if (!(pll_in_use & (1 << ATOM_PPLL2)))
+		return ATOM_PPLL2;
+	if (!(pll_in_use & (1 << ATOM_PPLL1)))
+		return ATOM_PPLL1;
+	DRM_ERROR("unable to allocate a PPLL\n");
+	return ATOM_PPLL_INVALID;
+}
+
+static void dce_v6_0_lock_cursor(struct drm_crtc *crtc, bool lock)
+{
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	uint32_t cur_lock;
+
+	cur_lock = RREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset);
+	if (lock)
+		cur_lock |= CUR_UPDATE__CURSOR_UPDATE_LOCK_MASK;
+	else
+		cur_lock &= ~CUR_UPDATE__CURSOR_UPDATE_LOCK_MASK;
+	WREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset, cur_lock);
+}
+
+static void dce_v6_0_hide_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
+		   (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
+		   (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
+
+
+}
+
+static void dce_v6_0_show_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+
+	WREG32(mmCUR_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(amdgpu_crtc->cursor_addr));
+	WREG32(mmCUR_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(amdgpu_crtc->cursor_addr));
+
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
+		   CUR_CONTROL__CURSOR_EN_MASK |
+		   (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
+		   (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
+
+}
+
+static int dce_v6_0_cursor_move_locked(struct drm_crtc *crtc,
+				       int x, int y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	int xorigin = 0, yorigin = 0;
+
+	int w = amdgpu_crtc->cursor_width;
+
+	amdgpu_crtc->cursor_x = x;
+	amdgpu_crtc->cursor_y = y;
+
+	/* avivo cursor are offset into the total surface */
+	x += crtc->x;
+	y += crtc->y;
+	DRM_DEBUG("x %d y %d c->x %d c->y %d\n", x, y, crtc->x, crtc->y);
+
+	if (x < 0) {
+		xorigin = min(-x, amdgpu_crtc->max_cursor_width - 1);
+		x = 0;
+	}
+	if (y < 0) {
+		yorigin = min(-y, amdgpu_crtc->max_cursor_height - 1);
+		y = 0;
+	}
+
+	WREG32(mmCUR_POSITION + amdgpu_crtc->crtc_offset, (x << 16) | y);
+	WREG32(mmCUR_HOT_SPOT + amdgpu_crtc->crtc_offset, (xorigin << 16) | yorigin);
+	WREG32(mmCUR_SIZE + amdgpu_crtc->crtc_offset,
+	       ((w - 1) << 16) | (amdgpu_crtc->cursor_height - 1));
+
+	return 0;
+}
+
+static int dce_v6_0_crtc_cursor_move(struct drm_crtc *crtc,
+				     int x, int y)
+{
+	int ret;
+
+	dce_v6_0_lock_cursor(crtc, true);
+	ret = dce_v6_0_cursor_move_locked(crtc, x, y);
+	dce_v6_0_lock_cursor(crtc, false);
+
+	return ret;
+}
+
+static int dce_v6_0_crtc_cursor_set2(struct drm_crtc *crtc,
+				     struct drm_file *file_priv,
+				     uint32_t handle,
+				     uint32_t width,
+				     uint32_t height,
+				     int32_t hot_x,
+				     int32_t hot_y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *aobj;
+	int ret;
+
+	if (!handle) {
+		/* turn off cursor */
+		dce_v6_0_hide_cursor(crtc);
+		obj = NULL;
+		goto unpin;
+	}
+
+	if ((width > amdgpu_crtc->max_cursor_width) ||
+	    (height > amdgpu_crtc->max_cursor_height)) {
+		DRM_ERROR("bad cursor width or height %d x %d\n", width, height);
+		return -EINVAL;
+	}
+
+	obj = drm_gem_object_lookup(file_priv, handle);
+	if (!obj) {
+		DRM_ERROR("Cannot find cursor object %x for crtc %d\n", handle, amdgpu_crtc->crtc_id);
+		return -ENOENT;
+	}
+
+	aobj = gem_to_amdgpu_bo(obj);
+	ret = amdgpu_bo_reserve(aobj, false);
+	if (ret != 0) {
+		drm_gem_object_put_unlocked(obj);
+		return ret;
+	}
+
+	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM);
+	amdgpu_bo_unreserve(aobj);
+	if (ret) {
+		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
+		drm_gem_object_put_unlocked(obj);
+		return ret;
+	}
+	amdgpu_crtc->cursor_addr = amdgpu_bo_gpu_offset(aobj);
+
+	dce_v6_0_lock_cursor(crtc, true);
+
+	if (width != amdgpu_crtc->cursor_width ||
+	    height != amdgpu_crtc->cursor_height ||
+	    hot_x != amdgpu_crtc->cursor_hot_x ||
+	    hot_y != amdgpu_crtc->cursor_hot_y) {
+		int x, y;
+
+		x = amdgpu_crtc->cursor_x + amdgpu_crtc->cursor_hot_x - hot_x;
+		y = amdgpu_crtc->cursor_y + amdgpu_crtc->cursor_hot_y - hot_y;
+
+		dce_v6_0_cursor_move_locked(crtc, x, y);
+
+		amdgpu_crtc->cursor_width = width;
+		amdgpu_crtc->cursor_height = height;
+		amdgpu_crtc->cursor_hot_x = hot_x;
+		amdgpu_crtc->cursor_hot_y = hot_y;
+	}
+
+	dce_v6_0_show_cursor(crtc);
+	dce_v6_0_lock_cursor(crtc, false);
+
+unpin:
+	if (amdgpu_crtc->cursor_bo) {
+		struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+		ret = amdgpu_bo_reserve(aobj, true);
+		if (likely(ret == 0)) {
+			amdgpu_bo_unpin(aobj);
+			amdgpu_bo_unreserve(aobj);
+		}
+		drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
+	}
+
+	amdgpu_crtc->cursor_bo = obj;
+	return 0;
+}
+
+static void dce_v6_0_cursor_reset(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (amdgpu_crtc->cursor_bo) {
+		dce_v6_0_lock_cursor(crtc, true);
+
+		dce_v6_0_cursor_move_locked(crtc, amdgpu_crtc->cursor_x,
+					    amdgpu_crtc->cursor_y);
+
+		dce_v6_0_show_cursor(crtc);
+		dce_v6_0_lock_cursor(crtc, false);
+	}
+}
+
+static int dce_v6_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
+				   u16 *blue, uint32_t size,
+				   struct drm_modeset_acquire_ctx *ctx)
+{
+	dce_v6_0_crtc_load_lut(crtc);
+
+	return 0;
+}
+
+static void dce_v6_0_crtc_destroy(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	drm_crtc_cleanup(crtc);
+	kfree(amdgpu_crtc);
+}
+
+static const struct drm_crtc_funcs dce_v6_0_crtc_funcs = {
+	.cursor_set2 = dce_v6_0_crtc_cursor_set2,
+	.cursor_move = dce_v6_0_crtc_cursor_move,
+	.gamma_set = dce_v6_0_crtc_gamma_set,
+	.set_config = amdgpu_display_crtc_set_config,
+	.destroy = dce_v6_0_crtc_destroy,
+	.page_flip_target = amdgpu_display_crtc_page_flip_target,
+};
+
+static void dce_v6_0_crtc_dpms(struct drm_crtc *crtc, int mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	unsigned type;
+
+	switch (mode) {
+	case DRM_MODE_DPMS_ON:
+		amdgpu_crtc->enabled = true;
+		amdgpu_atombios_crtc_enable(crtc, ATOM_ENABLE);
+		amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
+		/* Make sure VBLANK and PFLIP interrupts are still enabled */
+		type = amdgpu_display_crtc_idx_to_irq_type(adev,
+						amdgpu_crtc->crtc_id);
+		amdgpu_irq_update(adev, &adev->crtc_irq, type);
+		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
+		drm_crtc_vblank_on(crtc);
+		dce_v6_0_crtc_load_lut(crtc);
+		break;
+	case DRM_MODE_DPMS_STANDBY:
+	case DRM_MODE_DPMS_SUSPEND:
+	case DRM_MODE_DPMS_OFF:
+		drm_crtc_vblank_off(crtc);
+		if (amdgpu_crtc->enabled)
+			amdgpu_atombios_crtc_blank(crtc, ATOM_ENABLE);
+		amdgpu_atombios_crtc_enable(crtc, ATOM_DISABLE);
+		amdgpu_crtc->enabled = false;
+		break;
+	}
+	/* adjust pm to dpms */
+	amdgpu_pm_compute_clocks(adev);
+}
+
+static void dce_v6_0_crtc_prepare(struct drm_crtc *crtc)
+{
+	/* disable crtc pair power gating before programming */
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_DISABLE);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_ENABLE);
+	dce_v6_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+}
+
+static void dce_v6_0_crtc_commit(struct drm_crtc *crtc)
+{
+	dce_v6_0_crtc_dpms(crtc, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_DISABLE);
+}
+
+static void dce_v6_0_crtc_disable(struct drm_crtc *crtc)
+{
+
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_atom_ss ss;
+	int i;
+
+	dce_v6_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+	if (crtc->primary->fb) {
+		int r;
+		struct amdgpu_bo *abo;
+
+		abo = gem_to_amdgpu_bo(crtc->primary->fb->obj[0]);
+		r = amdgpu_bo_reserve(abo, true);
+		if (unlikely(r))
+			DRM_ERROR("failed to reserve abo before unpin\n");
+		else {
+			amdgpu_bo_unpin(abo);
+			amdgpu_bo_unreserve(abo);
+		}
+	}
+	/* disable the GRPH */
+	dce_v6_0_grph_enable(crtc, false);
+
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_ENABLE);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i] &&
+		    adev->mode_info.crtcs[i]->enabled &&
+		    i != amdgpu_crtc->crtc_id &&
+		    amdgpu_crtc->pll_id == adev->mode_info.crtcs[i]->pll_id) {
+			/* one other crtc is using this pll don't turn
+			 * off the pll
+			 */
+			goto done;
+		}
+	}
+
+	switch (amdgpu_crtc->pll_id) {
+	case ATOM_PPLL1:
+	case ATOM_PPLL2:
+		/* disable the ppll */
+		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
+						 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
+		break;
+	default:
+		break;
+	}
+done:
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+}
+
+static int dce_v6_0_crtc_mode_set(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode,
+				  int x, int y, struct drm_framebuffer *old_fb)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (!amdgpu_crtc->adjusted_clock)
+		return -EINVAL;
+
+	amdgpu_atombios_crtc_set_pll(crtc, adjusted_mode);
+	amdgpu_atombios_crtc_set_dtd_timing(crtc, adjusted_mode);
+	dce_v6_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+	amdgpu_atombios_crtc_overscan_setup(crtc, mode, adjusted_mode);
+	amdgpu_atombios_crtc_scaler_setup(crtc);
+	dce_v6_0_cursor_reset(crtc);
+	/* update the hw version fpr dpm */
+	amdgpu_crtc->hw_mode = *adjusted_mode;
+
+	return 0;
+}
+
+static bool dce_v6_0_crtc_mode_fixup(struct drm_crtc *crtc,
+				     const struct drm_display_mode *mode,
+				     struct drm_display_mode *adjusted_mode)
+{
+
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct drm_encoder *encoder;
+
+	/* assign the encoder to the amdgpu crtc to avoid repeated lookups later */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		if (encoder->crtc == crtc) {
+			amdgpu_crtc->encoder = encoder;
+			amdgpu_crtc->connector = amdgpu_get_connector_for_encoder(encoder);
+			break;
+		}
+	}
+	if ((amdgpu_crtc->encoder == NULL) || (amdgpu_crtc->connector == NULL)) {
+		amdgpu_crtc->encoder = NULL;
+		amdgpu_crtc->connector = NULL;
+		return false;
+	}
+	if (!amdgpu_display_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
+		return false;
+	if (amdgpu_atombios_crtc_prepare_pll(crtc, adjusted_mode))
+		return false;
+	/* pick pll */
+	amdgpu_crtc->pll_id = dce_v6_0_pick_pll(crtc);
+	/* if we can't get a PPLL for a non-DP encoder, fail */
+	if ((amdgpu_crtc->pll_id == ATOM_PPLL_INVALID) &&
+	    !ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder)))
+		return false;
+
+	return true;
+}
+
+static int dce_v6_0_crtc_set_base(struct drm_crtc *crtc, int x, int y,
+				  struct drm_framebuffer *old_fb)
+{
+	return dce_v6_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+}
+
+static int dce_v6_0_crtc_set_base_atomic(struct drm_crtc *crtc,
+					 struct drm_framebuffer *fb,
+					 int x, int y, enum mode_set_atomic state)
+{
+       return dce_v6_0_crtc_do_set_base(crtc, fb, x, y, 1);
+}
+
+static const struct drm_crtc_helper_funcs dce_v6_0_crtc_helper_funcs = {
+	.dpms = dce_v6_0_crtc_dpms,
+	.mode_fixup = dce_v6_0_crtc_mode_fixup,
+	.mode_set = dce_v6_0_crtc_mode_set,
+	.mode_set_base = dce_v6_0_crtc_set_base,
+	.mode_set_base_atomic = dce_v6_0_crtc_set_base_atomic,
+	.prepare = dce_v6_0_crtc_prepare,
+	.commit = dce_v6_0_crtc_commit,
+	.disable = dce_v6_0_crtc_disable,
+};
+
+static int dce_v6_0_crtc_init(struct amdgpu_device *adev, int index)
+{
+	struct amdgpu_crtc *amdgpu_crtc;
+
+	amdgpu_crtc = kzalloc(sizeof(struct amdgpu_crtc) +
+			      (AMDGPUFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
+	if (amdgpu_crtc == NULL)
+		return -ENOMEM;
+
+	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_v6_0_crtc_funcs);
+
+	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
+	amdgpu_crtc->crtc_id = index;
+	adev->mode_info.crtcs[index] = amdgpu_crtc;
+
+	amdgpu_crtc->max_cursor_width = CURSOR_WIDTH;
+	amdgpu_crtc->max_cursor_height = CURSOR_HEIGHT;
+	adev->ddev->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
+	adev->ddev->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
+
+	amdgpu_crtc->crtc_offset = crtc_offsets[amdgpu_crtc->crtc_id];
+
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_v6_0_crtc_helper_funcs);
+
+	return 0;
+}
+
+static int dce_v6_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->audio_endpt_rreg = &dce_v6_0_audio_endpt_rreg;
+	adev->audio_endpt_wreg = &dce_v6_0_audio_endpt_wreg;
+
+	dce_v6_0_set_display_funcs(adev);
+
+	adev->mode_info.num_crtc = dce_v6_0_get_num_crtc(adev);
+
+	switch (adev->asic_type) {
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 6;
+		break;
+	case CHIP_OLAND:
+		adev->mode_info.num_hpd = 2;
+		adev->mode_info.num_dig = 2;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	dce_v6_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int dce_v6_0_sw_init(void *handle)
+{
+	int r, i;
+	bool ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
+		if (r)
+			return r;
+	}
+
+	for (i = 8; i < 20; i += 2) {
+		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i, &adev->pageflip_irq);
+		if (r)
+			return r;
+	}
+
+	/* HPD hotplug */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 42, &adev->hpd_irq);
+	if (r)
+		return r;
+
+	adev->mode_info.mode_config_initialized = true;
+
+	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
+	adev->ddev->mode_config.async_page_flip = true;
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+	adev->ddev->mode_config.preferred_depth = 24;
+	adev->ddev->mode_config.prefer_shadow = 1;
+	adev->ddev->mode_config.fb_base = adev->gmc.aper_base;
+
+	r = amdgpu_display_modeset_create_props(adev);
+	if (r)
+		return r;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	/* allocate crtcs */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = dce_v6_0_crtc_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	ret = amdgpu_atombios_get_connector_info_from_object_table(adev);
+	if (ret)
+		amdgpu_display_print_display_setup(adev->ddev);
+	else
+		return -EINVAL;
+
+	/* setup afmt */
+	r = dce_v6_0_afmt_init(adev);
+	if (r)
+		return r;
+
+	r = dce_v6_0_audio_init(adev);
+	if (r)
+		return r;
+
+	drm_kms_helper_poll_init(adev->ddev);
+
+	return r;
+}
+
+static int dce_v6_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	kfree(adev->mode_info.bios_hardcoded_edid);
+
+	drm_kms_helper_poll_fini(adev->ddev);
+
+	dce_v6_0_audio_fini(adev);
+	dce_v6_0_afmt_fini(adev);
+
+	drm_mode_config_cleanup(adev->ddev);
+	adev->mode_info.mode_config_initialized = false;
+
+	return 0;
+}
+
+static int dce_v6_0_hw_init(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* disable vga render */
+	dce_v6_0_set_vga_render_state(adev, false);
+	/* init dig PHYs, disp eng pll */
+	amdgpu_atombios_encoder_init_dig(adev);
+	amdgpu_atombios_crtc_set_disp_eng_pll(adev, adev->clock.default_dispclk);
+
+	/* initialize hpd */
+	dce_v6_0_hpd_init(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v6_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v6_0_pageflip_interrupt_init(adev);
+
+	return 0;
+}
+
+static int dce_v6_0_hw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dce_v6_0_hpd_fini(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v6_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v6_0_pageflip_interrupt_fini(adev);
+
+	return 0;
+}
+
+static int dce_v6_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->mode_info.bl_level =
+		amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+
+	return dce_v6_0_hw_fini(handle);
+}
+
+static int dce_v6_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret;
+
+	amdgpu_atombios_encoder_set_backlight_level_to_reg(adev,
+							   adev->mode_info.bl_level);
+
+	ret = dce_v6_0_hw_init(handle);
+
+	/* turn on the BL */
+	if (adev->mode_info.bl_encoder) {
+		u8 bl_level = amdgpu_display_backlight_get_level(adev,
+								  adev->mode_info.bl_encoder);
+		amdgpu_display_backlight_set_level(adev, adev->mode_info.bl_encoder,
+						    bl_level);
+	}
+
+	return ret;
+}
+
+static bool dce_v6_0_is_idle(void *handle)
+{
+	return true;
+}
+
+static int dce_v6_0_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static int dce_v6_0_soft_reset(void *handle)
+{
+	DRM_INFO("xxxx: dce_v6_0_soft_reset --- no impl!!\n");
+	return 0;
+}
+
+static void dce_v6_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
+						     int crtc,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 reg_block, interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (crtc) {
+	case 0:
+		reg_block = SI_CRTC0_REGISTER_OFFSET;
+		break;
+	case 1:
+		reg_block = SI_CRTC1_REGISTER_OFFSET;
+		break;
+	case 2:
+		reg_block = SI_CRTC2_REGISTER_OFFSET;
+		break;
+	case 3:
+		reg_block = SI_CRTC3_REGISTER_OFFSET;
+		break;
+	case 4:
+		reg_block = SI_CRTC4_REGISTER_OFFSET;
+		break;
+	case 5:
+		reg_block = SI_CRTC5_REGISTER_OFFSET;
+		break;
+	default:
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		interrupt_mask = RREG32(mmINT_MASK + reg_block);
+		interrupt_mask &= ~VBLANK_INT_MASK;
+		WREG32(mmINT_MASK + reg_block, interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		interrupt_mask = RREG32(mmINT_MASK + reg_block);
+		interrupt_mask |= VBLANK_INT_MASK;
+		WREG32(mmINT_MASK + reg_block, interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static void dce_v6_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
+						    int crtc,
+						    enum amdgpu_interrupt_state state)
+{
+
+}
+
+static int dce_v6_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	u32 dc_hpd_int_cntl;
+
+	if (type >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hdp %d\n", type);
+		return 0;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		dc_hpd_int_cntl = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type]);
+		dc_hpd_int_cntl &= ~DC_HPDx_INT_EN;
+		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type], dc_hpd_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		dc_hpd_int_cntl = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type]);
+		dc_hpd_int_cntl |= DC_HPDx_INT_EN;
+		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type], dc_hpd_int_cntl);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v6_0_set_crtc_interrupt_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CRTC_IRQ_VBLANK1:
+		dce_v6_0_set_crtc_vblank_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK2:
+		dce_v6_0_set_crtc_vblank_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK3:
+		dce_v6_0_set_crtc_vblank_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK4:
+		dce_v6_0_set_crtc_vblank_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK5:
+		dce_v6_0_set_crtc_vblank_interrupt_state(adev, 4, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK6:
+		dce_v6_0_set_crtc_vblank_interrupt_state(adev, 5, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE1:
+		dce_v6_0_set_crtc_vline_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE2:
+		dce_v6_0_set_crtc_vline_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE3:
+		dce_v6_0_set_crtc_vline_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE4:
+		dce_v6_0_set_crtc_vline_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE5:
+		dce_v6_0_set_crtc_vline_interrupt_state(adev, 4, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE6:
+		dce_v6_0_set_crtc_vline_interrupt_state(adev, 5, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int dce_v6_0_crtc_irq(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *source,
+			     struct amdgpu_iv_entry *entry)
+{
+	unsigned crtc = entry->src_id - 1;
+	uint32_t disp_int = RREG32(interrupt_status_offsets[crtc].reg);
+	unsigned int irq_type = amdgpu_display_crtc_idx_to_irq_type(adev,
+								    crtc);
+
+	switch (entry->src_data[0]) {
+	case 0: /* vblank */
+		if (disp_int & interrupt_status_offsets[crtc].vblank)
+			WREG32(mmVBLANK_STATUS + crtc_offsets[crtc], VBLANK_ACK);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		if (amdgpu_irq_enabled(adev, source, irq_type)) {
+			drm_handle_vblank(adev->ddev, crtc);
+		}
+		DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+		break;
+	case 1: /* vline */
+		if (disp_int & interrupt_status_offsets[crtc].vline)
+			WREG32(mmVLINE_STATUS + crtc_offsets[crtc], VLINE_ACK);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		DRM_DEBUG("IH: D%d vline\n", crtc + 1);
+		break;
+	default:
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v6_0_set_pageflip_interrupt_state(struct amdgpu_device *adev,
+						 struct amdgpu_irq_src *src,
+						 unsigned type,
+						 enum amdgpu_interrupt_state state)
+{
+	u32 reg;
+
+	if (type >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", type);
+		return -EINVAL;
+	}
+
+	reg = RREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type]);
+	if (state == AMDGPU_IRQ_STATE_DISABLE)
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg & ~GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+	else
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg | GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+
+	return 0;
+}
+
+static int dce_v6_0_pageflip_irq(struct amdgpu_device *adev,
+				 struct amdgpu_irq_src *source,
+				 struct amdgpu_iv_entry *entry)
+{
+		unsigned long flags;
+	unsigned crtc_id;
+	struct amdgpu_crtc *amdgpu_crtc;
+	struct amdgpu_flip_work *works;
+
+	crtc_id = (entry->src_id - 8) >> 1;
+	amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	if (crtc_id >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", crtc_id);
+		return -EINVAL;
+	}
+
+	if (RREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id]) &
+	    GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_OCCURRED_MASK)
+		WREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id],
+		       GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_CLEAR_MASK);
+
+	/* IRQ could occur when in initial stage */
+	if (amdgpu_crtc == NULL)
+		return 0;
+
+	spin_lock_irqsave(&adev->ddev->event_lock, flags);
+	works = amdgpu_crtc->pflip_works;
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED){
+		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
+						"AMDGPU_FLIP_SUBMITTED(%d)\n",
+						amdgpu_crtc->pflip_status,
+						AMDGPU_FLIP_SUBMITTED);
+		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+		return 0;
+	}
+
+	/* page flip completed. clean up */
+	amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE;
+	amdgpu_crtc->pflip_works = NULL;
+
+	/* wakeup usersapce */
+	if (works->event)
+		drm_crtc_send_vblank_event(&amdgpu_crtc->base, works->event);
+
+	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+
+	drm_crtc_vblank_put(&amdgpu_crtc->base);
+	schedule_work(&works->unpin_work);
+
+	return 0;
+}
+
+static int dce_v6_0_hpd_irq(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *source,
+			    struct amdgpu_iv_entry *entry)
+{
+	uint32_t disp_int, mask, tmp;
+	unsigned hpd;
+
+	if (entry->src_data[0] >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]);
+		return 0;
+	}
+
+	hpd = entry->src_data[0];
+	disp_int = RREG32(interrupt_status_offsets[hpd].reg);
+	mask = interrupt_status_offsets[hpd].hpd;
+
+	if (disp_int & mask) {
+		tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
+		tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK;
+		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
+		schedule_work(&adev->hotplug_work);
+		DRM_DEBUG("IH: HPD%d\n", hpd + 1);
+	}
+
+	return 0;
+
+}
+
+static int dce_v6_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int dce_v6_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs dce_v6_0_ip_funcs = {
+	.name = "dce_v6_0",
+	.early_init = dce_v6_0_early_init,
+	.late_init = NULL,
+	.sw_init = dce_v6_0_sw_init,
+	.sw_fini = dce_v6_0_sw_fini,
+	.hw_init = dce_v6_0_hw_init,
+	.hw_fini = dce_v6_0_hw_fini,
+	.suspend = dce_v6_0_suspend,
+	.resume = dce_v6_0_resume,
+	.is_idle = dce_v6_0_is_idle,
+	.wait_for_idle = dce_v6_0_wait_for_idle,
+	.soft_reset = dce_v6_0_soft_reset,
+	.set_clockgating_state = dce_v6_0_set_clockgating_state,
+	.set_powergating_state = dce_v6_0_set_powergating_state,
+};
+
+static void
+dce_v6_0_encoder_mode_set(struct drm_encoder *encoder,
+			  struct drm_display_mode *mode,
+			  struct drm_display_mode *adjusted_mode)
+{
+
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	int em = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+	amdgpu_encoder->pixel_clock = adjusted_mode->clock;
+
+	/* need to call this here rather than in prepare() since we need some crtc info */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	/* set scaler clears this on some chips */
+	dce_v6_0_set_interleave(encoder->crtc, mode);
+
+	if (em == ATOM_ENCODER_MODE_HDMI || ENCODER_MODE_IS_DP(em)) {
+		dce_v6_0_afmt_enable(encoder, true);
+		dce_v6_0_afmt_setmode(encoder, adjusted_mode);
+	}
+}
+
+static void dce_v6_0_encoder_prepare(struct drm_encoder *encoder)
+{
+
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+
+	if ((amdgpu_encoder->active_device &
+	     (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) !=
+	     ENCODER_OBJECT_ID_NONE)) {
+		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+		if (dig) {
+			dig->dig_encoder = dce_v6_0_pick_dig_encoder(encoder);
+			if (amdgpu_encoder->active_device & ATOM_DEVICE_DFP_SUPPORT)
+				dig->afmt = adev->mode_info.afmt[dig->dig_encoder];
+		}
+	}
+
+	amdgpu_atombios_scratch_regs_lock(adev, true);
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		/* select the clock/data port if it uses a router */
+		if (amdgpu_connector->router.cd_valid)
+			amdgpu_i2c_router_select_cd_port(amdgpu_connector);
+
+		/* turn eDP panel on for mode set */
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP)
+			amdgpu_atombios_encoder_set_edp_panel_power(connector,
+							     ATOM_TRANSMITTER_ACTION_POWER_ON);
+	}
+
+	/* this is needed for the pll/ss setup to work correctly in some cases */
+	amdgpu_atombios_encoder_set_crtc_source(encoder);
+	/* set up the FMT blocks */
+	dce_v6_0_program_fmt(encoder);
+}
+
+static void dce_v6_0_encoder_commit(struct drm_encoder *encoder)
+{
+
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* need to call this here as we need the crtc set up */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_scratch_regs_lock(adev, false);
+}
+
+static void dce_v6_0_encoder_disable(struct drm_encoder *encoder)
+{
+
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig;
+	int em = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	if (amdgpu_atombios_encoder_is_digital(encoder)) {
+		if (em == ATOM_ENCODER_MODE_HDMI || ENCODER_MODE_IS_DP(em))
+			dce_v6_0_afmt_enable(encoder, false);
+		dig = amdgpu_encoder->enc_priv;
+		dig->dig_encoder = -1;
+	}
+	amdgpu_encoder->active_device = 0;
+}
+
+/* these are handled by the primary encoders */
+static void dce_v6_0_ext_prepare(struct drm_encoder *encoder)
+{
+
+}
+
+static void dce_v6_0_ext_commit(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v6_0_ext_mode_set(struct drm_encoder *encoder,
+		      struct drm_display_mode *mode,
+		      struct drm_display_mode *adjusted_mode)
+{
+
+}
+
+static void dce_v6_0_ext_disable(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v6_0_ext_dpms(struct drm_encoder *encoder, int mode)
+{
+
+}
+
+static bool dce_v6_0_ext_mode_fixup(struct drm_encoder *encoder,
+				    const struct drm_display_mode *mode,
+				    struct drm_display_mode *adjusted_mode)
+{
+	return true;
+}
+
+static const struct drm_encoder_helper_funcs dce_v6_0_ext_helper_funcs = {
+	.dpms = dce_v6_0_ext_dpms,
+	.mode_fixup = dce_v6_0_ext_mode_fixup,
+	.prepare = dce_v6_0_ext_prepare,
+	.mode_set = dce_v6_0_ext_mode_set,
+	.commit = dce_v6_0_ext_commit,
+	.disable = dce_v6_0_ext_disable,
+	/* no detect for TMDS/LVDS yet */
+};
+
+static const struct drm_encoder_helper_funcs dce_v6_0_dig_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v6_0_encoder_prepare,
+	.mode_set = dce_v6_0_encoder_mode_set,
+	.commit = dce_v6_0_encoder_commit,
+	.disable = dce_v6_0_encoder_disable,
+	.detect = amdgpu_atombios_encoder_dig_detect,
+};
+
+static const struct drm_encoder_helper_funcs dce_v6_0_dac_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v6_0_encoder_prepare,
+	.mode_set = dce_v6_0_encoder_mode_set,
+	.commit = dce_v6_0_encoder_commit,
+	.detect = amdgpu_atombios_encoder_dac_detect,
+};
+
+static void dce_v6_0_encoder_destroy(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+		amdgpu_atombios_encoder_fini_backlight(amdgpu_encoder);
+	kfree(amdgpu_encoder->enc_priv);
+	drm_encoder_cleanup(encoder);
+	kfree(amdgpu_encoder);
+}
+
+static const struct drm_encoder_funcs dce_v6_0_encoder_funcs = {
+	.destroy = dce_v6_0_encoder_destroy,
+};
+
+static void dce_v6_0_encoder_add(struct amdgpu_device *adev,
+				 uint32_t encoder_enum,
+				 uint32_t supported_device,
+				 u16 caps)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	/* see if we already added it */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		if (amdgpu_encoder->encoder_enum == encoder_enum) {
+			amdgpu_encoder->devices |= supported_device;
+			return;
+		}
+
+	}
+
+	/* add a new one */
+	amdgpu_encoder = kzalloc(sizeof(struct amdgpu_encoder), GFP_KERNEL);
+	if (!amdgpu_encoder)
+		return;
+
+	encoder = &amdgpu_encoder->base;
+	switch (adev->mode_info.num_crtc) {
+	case 1:
+		encoder->possible_crtcs = 0x1;
+		break;
+	case 2:
+	default:
+		encoder->possible_crtcs = 0x3;
+		break;
+	case 4:
+		encoder->possible_crtcs = 0xf;
+		break;
+	case 6:
+		encoder->possible_crtcs = 0x3f;
+		break;
+	}
+
+	amdgpu_encoder->enc_priv = NULL;
+	amdgpu_encoder->encoder_enum = encoder_enum;
+	amdgpu_encoder->encoder_id = (encoder_enum & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+	amdgpu_encoder->devices = supported_device;
+	amdgpu_encoder->rmx_type = RMX_OFF;
+	amdgpu_encoder->underscan_type = UNDERSCAN_OFF;
+	amdgpu_encoder->is_ext_encoder = false;
+	amdgpu_encoder->caps = caps;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+		drm_encoder_init(dev, encoder, &dce_v6_0_encoder_funcs,
+				 DRM_MODE_ENCODER_DAC, NULL);
+		drm_encoder_helper_add(encoder, &dce_v6_0_dac_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+			amdgpu_encoder->rmx_type = RMX_FULL;
+			drm_encoder_init(dev, encoder, &dce_v6_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS, NULL);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_lcd_info(amdgpu_encoder);
+		} else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) {
+			drm_encoder_init(dev, encoder, &dce_v6_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC, NULL);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		} else {
+			drm_encoder_init(dev, encoder, &dce_v6_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS, NULL);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		}
+		drm_encoder_helper_add(encoder, &dce_v6_0_dig_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_SI170B:
+	case ENCODER_OBJECT_ID_CH7303:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOA:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOB:
+	case ENCODER_OBJECT_ID_TITFP513:
+	case ENCODER_OBJECT_ID_VT1623:
+	case ENCODER_OBJECT_ID_HDMI_SI1930:
+	case ENCODER_OBJECT_ID_TRAVIS:
+	case ENCODER_OBJECT_ID_NUTMEG:
+		/* these are handled by the primary encoders */
+		amdgpu_encoder->is_ext_encoder = true;
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v6_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS, NULL);
+		else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v6_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC, NULL);
+		else
+			drm_encoder_init(dev, encoder, &dce_v6_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS, NULL);
+		drm_encoder_helper_add(encoder, &dce_v6_0_ext_helper_funcs);
+		break;
+	}
+}
+
+static const struct amdgpu_display_funcs dce_v6_0_display_funcs = {
+	.bandwidth_update = &dce_v6_0_bandwidth_update,
+	.vblank_get_counter = &dce_v6_0_vblank_get_counter,
+	.backlight_set_level = &amdgpu_atombios_encoder_set_backlight_level,
+	.backlight_get_level = &amdgpu_atombios_encoder_get_backlight_level,
+	.hpd_sense = &dce_v6_0_hpd_sense,
+	.hpd_set_polarity = &dce_v6_0_hpd_set_polarity,
+	.hpd_get_gpio_reg = &dce_v6_0_hpd_get_gpio_reg,
+	.page_flip = &dce_v6_0_page_flip,
+	.page_flip_get_scanoutpos = &dce_v6_0_crtc_get_scanoutpos,
+	.add_encoder = &dce_v6_0_encoder_add,
+	.add_connector = &amdgpu_connector_add,
+};
+
+static void dce_v6_0_set_display_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.funcs == NULL)
+		adev->mode_info.funcs = &dce_v6_0_display_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs dce_v6_0_crtc_irq_funcs = {
+	.set = dce_v6_0_set_crtc_interrupt_state,
+	.process = dce_v6_0_crtc_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v6_0_pageflip_irq_funcs = {
+	.set = dce_v6_0_set_pageflip_interrupt_state,
+	.process = dce_v6_0_pageflip_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v6_0_hpd_irq_funcs = {
+	.set = dce_v6_0_set_hpd_interrupt_state,
+	.process = dce_v6_0_hpd_irq,
+};
+
+static void dce_v6_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.num_crtc > 0)
+		adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_VLINE1 + adev->mode_info.num_crtc;
+	else
+		adev->crtc_irq.num_types = 0;
+	adev->crtc_irq.funcs = &dce_v6_0_crtc_irq_funcs;
+
+	adev->pageflip_irq.num_types = adev->mode_info.num_crtc;
+	adev->pageflip_irq.funcs = &dce_v6_0_pageflip_irq_funcs;
+
+	adev->hpd_irq.num_types = adev->mode_info.num_hpd;
+	adev->hpd_irq.funcs = &dce_v6_0_hpd_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version dce_v6_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 6,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &dce_v6_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version dce_v6_4_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 6,
+	.minor = 4,
+	.rev = 0,
+	.funcs = &dce_v6_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/dce_v6_0.h b/sys/dev/pci/drm/amd/amdgpu/dce_v6_0.h
new file mode 100644
index 00000000000..7b546b596de
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/dce_v6_0.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DCE_V6_0_H__
+#define __DCE_V6_0_H__
+
+extern const struct amdgpu_ip_block_version dce_v6_0_ip_block;
+extern const struct amdgpu_ip_block_version dce_v6_4_ip_block;
+
+void dce_v6_0_disable_dce(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/dce_v8_0.c b/sys/dev/pci/drm/amd/amdgpu/dce_v8_0.c
new file mode 100644
index 00000000000..50cd03beac7
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/dce_v8_0.c
@@ -0,0 +1,3527 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_i2c.h"
+#include "cikd.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+#include "atombios_crtc.h"
+#include "atombios_encoders.h"
+#include "amdgpu_pll.h"
+#include "amdgpu_connectors.h"
+#include "dce_v8_0.h"
+
+#include "dce/dce_8_0_d.h"
+#include "dce/dce_8_0_sh_mask.h"
+
+#include "gca/gfx_7_2_enum.h"
+
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+static void dce_v8_0_set_display_funcs(struct amdgpu_device *adev);
+static void dce_v8_0_set_irq_funcs(struct amdgpu_device *adev);
+
+static const u32 crtc_offsets[6] =
+{
+	CRTC0_REGISTER_OFFSET,
+	CRTC1_REGISTER_OFFSET,
+	CRTC2_REGISTER_OFFSET,
+	CRTC3_REGISTER_OFFSET,
+	CRTC4_REGISTER_OFFSET,
+	CRTC5_REGISTER_OFFSET
+};
+
+static const u32 hpd_offsets[] =
+{
+	HPD0_REGISTER_OFFSET,
+	HPD1_REGISTER_OFFSET,
+	HPD2_REGISTER_OFFSET,
+	HPD3_REGISTER_OFFSET,
+	HPD4_REGISTER_OFFSET,
+	HPD5_REGISTER_OFFSET
+};
+
+static const uint32_t dig_offsets[] = {
+	CRTC0_REGISTER_OFFSET,
+	CRTC1_REGISTER_OFFSET,
+	CRTC2_REGISTER_OFFSET,
+	CRTC3_REGISTER_OFFSET,
+	CRTC4_REGISTER_OFFSET,
+	CRTC5_REGISTER_OFFSET,
+	(0x13830 - 0x7030) >> 2,
+};
+
+static const struct {
+	uint32_t	reg;
+	uint32_t	vblank;
+	uint32_t	vline;
+	uint32_t	hpd;
+
+} interrupt_status_offsets[6] = { {
+	.reg = mmDISP_INTERRUPT_STATUS,
+	.vblank = DISP_INTERRUPT_STATUS__LB_D1_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS__LB_D1_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS__DC_HPD1_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE__DC_HPD2_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE2,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE2__DC_HPD3_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE3,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE3__DC_HPD4_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE4,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE4__DC_HPD5_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE5,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE5__DC_HPD6_INTERRUPT_MASK
+} };
+
+static u32 dce_v8_0_audio_endpt_rreg(struct amdgpu_device *adev,
+				     u32 block_offset, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	r = RREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+
+	return r;
+}
+
+static void dce_v8_0_audio_endpt_wreg(struct amdgpu_device *adev,
+				      u32 block_offset, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset, v);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+}
+
+static u32 dce_v8_0_vblank_get_counter(struct amdgpu_device *adev, int crtc)
+{
+	if (crtc >= adev->mode_info.num_crtc)
+		return 0;
+	else
+		return RREG32(mmCRTC_STATUS_FRAME_COUNT + crtc_offsets[crtc]);
+}
+
+static void dce_v8_0_pageflip_interrupt_init(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Enable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_get(adev, &adev->pageflip_irq, i);
+}
+
+static void dce_v8_0_pageflip_interrupt_fini(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Disable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_put(adev, &adev->pageflip_irq, i);
+}
+
+/**
+ * dce_v8_0_page_flip - pageflip callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @crtc_id: crtc to cleanup pageflip on
+ * @crtc_base: new address of the crtc (GPU MC address)
+ *
+ * Triggers the actual pageflip by updating the primary
+ * surface base address.
+ */
+static void dce_v8_0_page_flip(struct amdgpu_device *adev,
+			       int crtc_id, u64 crtc_base, bool async)
+{
+	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	/* flip at hsync for async, default is vsync */
+	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, async ?
+	       GRPH_FLIP_CONTROL__GRPH_SURFACE_UPDATE_H_RETRACE_EN_MASK : 0);
+	/* update the primary scanout addresses */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(crtc_base));
+	/* writing to the low address triggers the update */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(crtc_base));
+	/* post the write */
+	RREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset);
+}
+
+static int dce_v8_0_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
+					u32 *vbl, u32 *position)
+{
+	if ((crtc < 0) || (crtc >= adev->mode_info.num_crtc))
+		return -EINVAL;
+
+	*vbl = RREG32(mmCRTC_V_BLANK_START_END + crtc_offsets[crtc]);
+	*position = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+
+	return 0;
+}
+
+/**
+ * dce_v8_0_hpd_sense - hpd sense callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Checks if a digital monitor is connected (evergreen+).
+ * Returns true if connected, false if not connected.
+ */
+static bool dce_v8_0_hpd_sense(struct amdgpu_device *adev,
+			       enum amdgpu_hpd_id hpd)
+{
+	bool connected = false;
+
+	if (hpd >= adev->mode_info.num_hpd)
+		return connected;
+
+	if (RREG32(mmDC_HPD1_INT_STATUS + hpd_offsets[hpd]) &
+	    DC_HPD1_INT_STATUS__DC_HPD1_SENSE_MASK)
+		connected = true;
+
+	return connected;
+}
+
+/**
+ * dce_v8_0_hpd_set_polarity - hpd set polarity callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Set the polarity of the hpd pin (evergreen+).
+ */
+static void dce_v8_0_hpd_set_polarity(struct amdgpu_device *adev,
+				      enum amdgpu_hpd_id hpd)
+{
+	u32 tmp;
+	bool connected = dce_v8_0_hpd_sense(adev, hpd);
+
+	if (hpd >= adev->mode_info.num_hpd)
+		return;
+
+	tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
+	if (connected)
+		tmp &= ~DC_HPD1_INT_CONTROL__DC_HPD1_INT_POLARITY_MASK;
+	else
+		tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_POLARITY_MASK;
+	WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
+}
+
+/**
+ * dce_v8_0_hpd_init - hpd setup callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup the hpd pins used by the card (evergreen+).
+ * Enable the pin, set the polarity, and enable the hpd interrupts.
+ */
+static void dce_v8_0_hpd_init(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
+			continue;
+
+		tmp = RREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+		tmp |= DC_HPD1_CONTROL__DC_HPD1_EN_MASK;
+		WREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP ||
+		    connector->connector_type == DRM_MODE_CONNECTOR_LVDS) {
+			/* don't try to enable hpd on eDP or LVDS avoid breaking the
+			 * aux dp channel on imac and help (but not completely fix)
+			 * https://bugzilla.redhat.com/show_bug.cgi?id=726143
+			 * also avoid interrupt storms during dpms.
+			 */
+			tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+			tmp &= ~DC_HPD1_INT_CONTROL__DC_HPD1_INT_EN_MASK;
+			WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+			continue;
+		}
+
+		dce_v8_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
+		amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
+	}
+}
+
+/**
+ * dce_v8_0_hpd_fini - hpd tear down callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the hpd pins used by the card (evergreen+).
+ * Disable the hpd interrupts.
+ */
+static void dce_v8_0_hpd_fini(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
+			continue;
+
+		tmp = RREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd]);
+		tmp &= ~DC_HPD1_CONTROL__DC_HPD1_EN_MASK;
+		WREG32(mmDC_HPD1_CONTROL + hpd_offsets[amdgpu_connector->hpd.hpd], 0);
+
+		amdgpu_irq_put(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
+	}
+}
+
+static u32 dce_v8_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
+{
+	return mmDC_GPIO_HPD_A;
+}
+
+static bool dce_v8_0_is_display_hung(struct amdgpu_device *adev)
+{
+	u32 crtc_hung = 0;
+	u32 crtc_status[6];
+	u32 i, j, tmp;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (RREG32(mmCRTC_CONTROL + crtc_offsets[i]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK) {
+			crtc_status[i] = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+			crtc_hung |= (1 << i);
+		}
+	}
+
+	for (j = 0; j < 10; j++) {
+		for (i = 0; i < adev->mode_info.num_crtc; i++) {
+			if (crtc_hung & (1 << i)) {
+				tmp = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+				if (tmp != crtc_status[i])
+					crtc_hung &= ~(1 << i);
+			}
+		}
+		if (crtc_hung == 0)
+			return false;
+		udelay(100);
+	}
+
+	return true;
+}
+
+static void dce_v8_0_set_vga_render_state(struct amdgpu_device *adev,
+					  bool render)
+{
+	u32 tmp;
+
+	/* Lockout access through VGA aperture*/
+	tmp = RREG32(mmVGA_HDP_CONTROL);
+	if (render)
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 0);
+	else
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
+	WREG32(mmVGA_HDP_CONTROL, tmp);
+
+	/* disable VGA render */
+	tmp = RREG32(mmVGA_RENDER_CONTROL);
+	if (render)
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
+	WREG32(mmVGA_RENDER_CONTROL, tmp);
+}
+
+static int dce_v8_0_get_num_crtc(struct amdgpu_device *adev)
+{
+	int num_crtc = 0;
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+		num_crtc = 6;
+		break;
+	case CHIP_KAVERI:
+		num_crtc = 4;
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		num_crtc = 2;
+		break;
+	default:
+		num_crtc = 0;
+	}
+	return num_crtc;
+}
+
+void dce_v8_0_disable_dce(struct amdgpu_device *adev)
+{
+	/*Disable VGA render and enabled crtc, if has DCE engine*/
+	if (amdgpu_atombios_has_dce_engine_info(adev)) {
+		u32 tmp;
+		int crtc_enabled, i;
+
+		dce_v8_0_set_vga_render_state(adev, false);
+
+		/*Disable crtc*/
+		for (i = 0; i < dce_v8_0_get_num_crtc(adev); i++) {
+			crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]),
+									 CRTC_CONTROL, CRTC_MASTER_EN);
+			if (crtc_enabled) {
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+				tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
+				tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0);
+				WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			}
+		}
+	}
+}
+
+static void dce_v8_0_program_fmt(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	int bpc = 0;
+	u32 tmp = 0;
+	enum amdgpu_connector_dither dither = AMDGPU_FMT_DITHER_DISABLE;
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		bpc = amdgpu_connector_get_monitor_bpc(connector);
+		dither = amdgpu_connector->dither;
+	}
+
+	/* LVDS/eDP FMT is set up by atom */
+	if (amdgpu_encoder->devices & ATOM_DEVICE_LCD_SUPPORT)
+		return;
+
+	/* not needed for analog */
+	if ((amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1) ||
+	    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2))
+		return;
+
+	if (bpc == 0)
+		return;
+
+	switch (bpc) {
+	case 6:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE)
+			/* XXX sort out optimal dither settings */
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_FRAME_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_HIGHPASS_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_EN_MASK |
+				(0 << FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_DEPTH__SHIFT));
+		else
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_EN_MASK |
+			(0 << FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_DEPTH__SHIFT));
+		break;
+	case 8:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE)
+			/* XXX sort out optimal dither settings */
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_FRAME_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_HIGHPASS_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_RGB_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_EN_MASK |
+				(1 << FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_DEPTH__SHIFT));
+		else
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_EN_MASK |
+			(1 << FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_DEPTH__SHIFT));
+		break;
+	case 10:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE)
+			/* XXX sort out optimal dither settings */
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_FRAME_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_HIGHPASS_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_RGB_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_EN_MASK |
+				(2 << FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_DEPTH__SHIFT));
+		else
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_EN_MASK |
+			(2 << FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_DEPTH__SHIFT));
+		break;
+	default:
+		/* not needed */
+		break;
+	}
+
+	WREG32(mmFMT_BIT_DEPTH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+
+/* display watermark setup */
+/**
+ * dce_v8_0_line_buffer_adjust - Set up the line buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @mode: the current display mode on the selected display
+ * controller
+ *
+ * Setup up the line buffer allocation for
+ * the selected display controller (CIK).
+ * Returns the line buffer size in pixels.
+ */
+static u32 dce_v8_0_line_buffer_adjust(struct amdgpu_device *adev,
+				       struct amdgpu_crtc *amdgpu_crtc,
+				       struct drm_display_mode *mode)
+{
+	u32 tmp, buffer_alloc, i;
+	u32 pipe_offset = amdgpu_crtc->crtc_id * 0x8;
+	/*
+	 * Line Buffer Setup
+	 * There are 6 line buffers, one for each display controllers.
+	 * There are 3 partitions per LB. Select the number of partitions
+	 * to enable based on the display width.  For display widths larger
+	 * than 4096, you need use to use 2 display controllers and combine
+	 * them using the stereo blender.
+	 */
+	if (amdgpu_crtc->base.enabled && mode) {
+		if (mode->crtc_hdisplay < 1920) {
+			tmp = 1;
+			buffer_alloc = 2;
+		} else if (mode->crtc_hdisplay < 2560) {
+			tmp = 2;
+			buffer_alloc = 2;
+		} else if (mode->crtc_hdisplay < 4096) {
+			tmp = 0;
+			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
+		} else {
+			DRM_DEBUG_KMS("Mode too big for LB!\n");
+			tmp = 0;
+			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
+		}
+	} else {
+		tmp = 1;
+		buffer_alloc = 0;
+	}
+
+	WREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset,
+	      (tmp << LB_MEMORY_CTRL__LB_MEMORY_CONFIG__SHIFT) |
+	      (0x6B0 << LB_MEMORY_CTRL__LB_MEMORY_SIZE__SHIFT));
+
+	WREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset,
+	       (buffer_alloc << PIPE0_DMIF_BUFFER_CONTROL__DMIF_BUFFERS_ALLOCATED__SHIFT));
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset) &
+		    PIPE0_DMIF_BUFFER_CONTROL__DMIF_BUFFERS_ALLOCATION_COMPLETED_MASK)
+			break;
+		udelay(1);
+	}
+
+	if (amdgpu_crtc->base.enabled && mode) {
+		switch (tmp) {
+		case 0:
+		default:
+			return 4096 * 2;
+		case 1:
+			return 1920 * 2;
+		case 2:
+			return 2560 * 2;
+		}
+	}
+
+	/* controller not enabled, so no lb used */
+	return 0;
+}
+
+/**
+ * cik_get_number_of_dram_channels - get the number of dram channels
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the number of video ram channels (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the number of dram channels
+ */
+static u32 cik_get_number_of_dram_channels(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(mmMC_SHARED_CHMAP);
+
+	switch ((tmp & MC_SHARED_CHMAP__NOOFCHAN_MASK) >> MC_SHARED_CHMAP__NOOFCHAN__SHIFT) {
+	case 0:
+	default:
+		return 1;
+	case 1:
+		return 2;
+	case 2:
+		return 4;
+	case 3:
+		return 8;
+	case 4:
+		return 3;
+	case 5:
+		return 6;
+	case 6:
+		return 10;
+	case 7:
+		return 12;
+	case 8:
+		return 16;
+	}
+}
+
+struct dce8_wm_params {
+	u32 dram_channels; /* number of dram channels */
+	u32 yclk;          /* bandwidth per dram data pin in kHz */
+	u32 sclk;          /* engine clock in kHz */
+	u32 disp_clk;      /* display clock in kHz */
+	u32 src_width;     /* viewport width */
+	u32 active_time;   /* active display time in ns */
+	u32 blank_time;    /* blank time in ns */
+	bool interlaced;    /* mode is interlaced */
+	fixed20_12 vsc;    /* vertical scale ratio */
+	u32 num_heads;     /* number of active crtcs */
+	u32 bytes_per_pixel; /* bytes per pixel display + overlay */
+	u32 lb_size;       /* line buffer allocated to pipe */
+	u32 vtaps;         /* vertical scaler taps */
+};
+
+/**
+ * dce_v8_0_dram_bandwidth - get the dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the raw dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth in MBytes/s
+ */
+static u32 dce_v8_0_dram_bandwidth(struct dce8_wm_params *wm)
+{
+	/* Calculate raw DRAM Bandwidth */
+	fixed20_12 dram_efficiency; /* 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	dram_efficiency.full = dfixed_const(7);
+	dram_efficiency.full = dfixed_div(dram_efficiency, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, dram_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v8_0_dram_bandwidth_for_display - get the dram bandwidth for display
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dram bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth for display in MBytes/s
+ */
+static u32 dce_v8_0_dram_bandwidth_for_display(struct dce8_wm_params *wm)
+{
+	/* Calculate DRAM Bandwidth and the part allocated to display. */
+	fixed20_12 disp_dram_allocation; /* 0.3 to 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	disp_dram_allocation.full = dfixed_const(3); /* XXX worse case value 0.3 */
+	disp_dram_allocation.full = dfixed_div(disp_dram_allocation, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, disp_dram_allocation);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v8_0_data_return_bandwidth - get the data return bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the data return bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the data return bandwidth in MBytes/s
+ */
+static u32 dce_v8_0_data_return_bandwidth(struct dce8_wm_params *wm)
+{
+	/* Calculate the display Data return Bandwidth */
+	fixed20_12 return_efficiency; /* 0.8 */
+	fixed20_12 sclk, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	sclk.full = dfixed_const(wm->sclk);
+	sclk.full = dfixed_div(sclk, a);
+	a.full = dfixed_const(10);
+	return_efficiency.full = dfixed_const(8);
+	return_efficiency.full = dfixed_div(return_efficiency, a);
+	a.full = dfixed_const(32);
+	bandwidth.full = dfixed_mul(a, sclk);
+	bandwidth.full = dfixed_mul(bandwidth, return_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v8_0_dmif_request_bandwidth - get the dmif bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dmif bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dmif bandwidth in MBytes/s
+ */
+static u32 dce_v8_0_dmif_request_bandwidth(struct dce8_wm_params *wm)
+{
+	/* Calculate the DMIF Request Bandwidth */
+	fixed20_12 disp_clk_request_efficiency; /* 0.8 */
+	fixed20_12 disp_clk, bandwidth;
+	fixed20_12 a, b;
+
+	a.full = dfixed_const(1000);
+	disp_clk.full = dfixed_const(wm->disp_clk);
+	disp_clk.full = dfixed_div(disp_clk, a);
+	a.full = dfixed_const(32);
+	b.full = dfixed_mul(a, disp_clk);
+
+	a.full = dfixed_const(10);
+	disp_clk_request_efficiency.full = dfixed_const(8);
+	disp_clk_request_efficiency.full = dfixed_div(disp_clk_request_efficiency, a);
+
+	bandwidth.full = dfixed_mul(b, disp_clk_request_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v8_0_available_bandwidth - get the min available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the min available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the min available bandwidth in MBytes/s
+ */
+static u32 dce_v8_0_available_bandwidth(struct dce8_wm_params *wm)
+{
+	/* Calculate the Available bandwidth. Display can use this temporarily but not in average. */
+	u32 dram_bandwidth = dce_v8_0_dram_bandwidth(wm);
+	u32 data_return_bandwidth = dce_v8_0_data_return_bandwidth(wm);
+	u32 dmif_req_bandwidth = dce_v8_0_dmif_request_bandwidth(wm);
+
+	return min(dram_bandwidth, min(data_return_bandwidth, dmif_req_bandwidth));
+}
+
+/**
+ * dce_v8_0_average_bandwidth - get the average available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the average available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the average available bandwidth in MBytes/s
+ */
+static u32 dce_v8_0_average_bandwidth(struct dce8_wm_params *wm)
+{
+	/* Calculate the display mode Average Bandwidth
+	 * DisplayMode should contain the source and destination dimensions,
+	 * timing, etc.
+	 */
+	fixed20_12 bpp;
+	fixed20_12 line_time;
+	fixed20_12 src_width;
+	fixed20_12 bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	line_time.full = dfixed_const(wm->active_time + wm->blank_time);
+	line_time.full = dfixed_div(line_time, a);
+	bpp.full = dfixed_const(wm->bytes_per_pixel);
+	src_width.full = dfixed_const(wm->src_width);
+	bandwidth.full = dfixed_mul(src_width, bpp);
+	bandwidth.full = dfixed_mul(bandwidth, wm->vsc);
+	bandwidth.full = dfixed_div(bandwidth, line_time);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v8_0_latency_watermark - get the latency watermark
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the latency watermark (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the latency watermark in ns
+ */
+static u32 dce_v8_0_latency_watermark(struct dce8_wm_params *wm)
+{
+	/* First calculate the latency in ns */
+	u32 mc_latency = 2000; /* 2000 ns. */
+	u32 available_bandwidth = dce_v8_0_available_bandwidth(wm);
+	u32 worst_chunk_return_time = (512 * 8 * 1000) / available_bandwidth;
+	u32 cursor_line_pair_return_time = (128 * 4 * 1000) / available_bandwidth;
+	u32 dc_latency = 40000000 / wm->disp_clk; /* dc pipe latency */
+	u32 other_heads_data_return_time = ((wm->num_heads + 1) * worst_chunk_return_time) +
+		(wm->num_heads * cursor_line_pair_return_time);
+	u32 latency = mc_latency + other_heads_data_return_time + dc_latency;
+	u32 max_src_lines_per_dst_line, lb_fill_bw, line_fill_time;
+	u32 tmp, dmif_size = 12288;
+	fixed20_12 a, b, c;
+
+	if (wm->num_heads == 0)
+		return 0;
+
+	a.full = dfixed_const(2);
+	b.full = dfixed_const(1);
+	if ((wm->vsc.full > a.full) ||
+	    ((wm->vsc.full > b.full) && (wm->vtaps >= 3)) ||
+	    (wm->vtaps >= 5) ||
+	    ((wm->vsc.full >= a.full) && wm->interlaced))
+		max_src_lines_per_dst_line = 4;
+	else
+		max_src_lines_per_dst_line = 2;
+
+	a.full = dfixed_const(available_bandwidth);
+	b.full = dfixed_const(wm->num_heads);
+	a.full = dfixed_div(a, b);
+	tmp = div_u64((u64) dmif_size * (u64) wm->disp_clk, mc_latency + 512);
+	tmp = min(dfixed_trunc(a), tmp);
+
+	lb_fill_bw = min(tmp, wm->disp_clk * wm->bytes_per_pixel / 1000);
+
+	a.full = dfixed_const(max_src_lines_per_dst_line * wm->src_width * wm->bytes_per_pixel);
+	b.full = dfixed_const(1000);
+	c.full = dfixed_const(lb_fill_bw);
+	b.full = dfixed_div(c, b);
+	a.full = dfixed_div(a, b);
+	line_fill_time = dfixed_trunc(a);
+
+	if (line_fill_time < wm->active_time)
+		return latency;
+	else
+		return latency + (line_fill_time - wm->active_time);
+
+}
+
+/**
+ * dce_v8_0_average_bandwidth_vs_dram_bandwidth_for_display - check
+ * average and available dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v8_0_average_bandwidth_vs_dram_bandwidth_for_display(struct dce8_wm_params *wm)
+{
+	if (dce_v8_0_average_bandwidth(wm) <=
+	    (dce_v8_0_dram_bandwidth_for_display(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v8_0_average_bandwidth_vs_available_bandwidth - check
+ * average and available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * available bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v8_0_average_bandwidth_vs_available_bandwidth(struct dce8_wm_params *wm)
+{
+	if (dce_v8_0_average_bandwidth(wm) <=
+	    (dce_v8_0_available_bandwidth(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v8_0_check_latency_hiding - check latency hiding
+ *
+ * @wm: watermark calculation data
+ *
+ * Check latency hiding (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v8_0_check_latency_hiding(struct dce8_wm_params *wm)
+{
+	u32 lb_partitions = wm->lb_size / wm->src_width;
+	u32 line_time = wm->active_time + wm->blank_time;
+	u32 latency_tolerant_lines;
+	u32 latency_hiding;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1);
+	if (wm->vsc.full > a.full)
+		latency_tolerant_lines = 1;
+	else {
+		if (lb_partitions <= (wm->vtaps + 1))
+			latency_tolerant_lines = 1;
+		else
+			latency_tolerant_lines = 2;
+	}
+
+	latency_hiding = (latency_tolerant_lines * line_time + wm->blank_time);
+
+	if (dce_v8_0_latency_watermark(wm) <= latency_hiding)
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v8_0_program_watermarks - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @lb_size: line buffer size
+ * @num_heads: number of display controllers in use
+ *
+ * Calculate and program the display watermarks for the
+ * selected display controller (CIK).
+ */
+static void dce_v8_0_program_watermarks(struct amdgpu_device *adev,
+					struct amdgpu_crtc *amdgpu_crtc,
+					u32 lb_size, u32 num_heads)
+{
+	struct drm_display_mode *mode = &amdgpu_crtc->base.mode;
+	struct dce8_wm_params wm_low, wm_high;
+	u32 active_time;
+	u32 line_time = 0;
+	u32 latency_watermark_a = 0, latency_watermark_b = 0;
+	u32 tmp, wm_mask, lb_vblank_lead_lines = 0;
+
+	if (amdgpu_crtc->base.enabled && num_heads && mode) {
+		active_time = (u32) div_u64((u64)mode->crtc_hdisplay * 1000000,
+					    (u32)mode->clock);
+		line_time = (u32) div_u64((u64)mode->crtc_htotal * 1000000,
+					  (u32)mode->clock);
+		line_time = min(line_time, (u32)65535);
+
+		/* watermark for high clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_high.yclk =
+				amdgpu_dpm_get_mclk(adev, false) * 10;
+			wm_high.sclk =
+				amdgpu_dpm_get_sclk(adev, false) * 10;
+		} else {
+			wm_high.yclk = adev->pm.current_mclk * 10;
+			wm_high.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_high.disp_clk = mode->clock;
+		wm_high.src_width = mode->crtc_hdisplay;
+		wm_high.active_time = active_time;
+		wm_high.blank_time = line_time - wm_high.active_time;
+		wm_high.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_high.interlaced = true;
+		wm_high.vsc = amdgpu_crtc->vsc;
+		wm_high.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_high.vtaps = 2;
+		wm_high.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_high.lb_size = lb_size;
+		wm_high.dram_channels = cik_get_number_of_dram_channels(adev);
+		wm_high.num_heads = num_heads;
+
+		/* set for high clocks */
+		latency_watermark_a = min(dce_v8_0_latency_watermark(&wm_high), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v8_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_high) ||
+		    !dce_v8_0_average_bandwidth_vs_available_bandwidth(&wm_high) ||
+		    !dce_v8_0_check_latency_hiding(&wm_high) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+		}
+
+		/* watermark for low clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_low.yclk =
+				amdgpu_dpm_get_mclk(adev, true) * 10;
+			wm_low.sclk =
+				amdgpu_dpm_get_sclk(adev, true) * 10;
+		} else {
+			wm_low.yclk = adev->pm.current_mclk * 10;
+			wm_low.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_low.disp_clk = mode->clock;
+		wm_low.src_width = mode->crtc_hdisplay;
+		wm_low.active_time = active_time;
+		wm_low.blank_time = line_time - wm_low.active_time;
+		wm_low.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_low.interlaced = true;
+		wm_low.vsc = amdgpu_crtc->vsc;
+		wm_low.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_low.vtaps = 2;
+		wm_low.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_low.lb_size = lb_size;
+		wm_low.dram_channels = cik_get_number_of_dram_channels(adev);
+		wm_low.num_heads = num_heads;
+
+		/* set for low clocks */
+		latency_watermark_b = min(dce_v8_0_latency_watermark(&wm_low), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v8_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_low) ||
+		    !dce_v8_0_average_bandwidth_vs_available_bandwidth(&wm_low) ||
+		    !dce_v8_0_check_latency_hiding(&wm_low) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+		}
+		lb_vblank_lead_lines = DIV_ROUND_UP(lb_size, mode->crtc_hdisplay);
+	}
+
+	/* select wm A */
+	wm_mask = RREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = wm_mask;
+	tmp &= ~(3 << DPG_WATERMARK_MASK_CONTROL__URGENCY_WATERMARK_MASK__SHIFT);
+	tmp |= (1 << DPG_WATERMARK_MASK_CONTROL__URGENCY_WATERMARK_MASK__SHIFT);
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((latency_watermark_a << DPG_PIPE_URGENCY_CONTROL__URGENCY_LOW_WATERMARK__SHIFT) |
+		(line_time << DPG_PIPE_URGENCY_CONTROL__URGENCY_HIGH_WATERMARK__SHIFT)));
+	/* select wm B */
+	tmp = RREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp &= ~(3 << DPG_WATERMARK_MASK_CONTROL__URGENCY_WATERMARK_MASK__SHIFT);
+	tmp |= (2 << DPG_WATERMARK_MASK_CONTROL__URGENCY_WATERMARK_MASK__SHIFT);
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((latency_watermark_b << DPG_PIPE_URGENCY_CONTROL__URGENCY_LOW_WATERMARK__SHIFT) |
+		(line_time << DPG_PIPE_URGENCY_CONTROL__URGENCY_HIGH_WATERMARK__SHIFT)));
+	/* restore original selection */
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, wm_mask);
+
+	/* save values for DPM */
+	amdgpu_crtc->line_time = line_time;
+	amdgpu_crtc->wm_high = latency_watermark_a;
+	amdgpu_crtc->wm_low = latency_watermark_b;
+	/* Save number of lines the linebuffer leads before the scanout */
+	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
+}
+
+/**
+ * dce_v8_0_bandwidth_update - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Calculate and program the display watermarks and line
+ * buffer allocation (CIK).
+ */
+static void dce_v8_0_bandwidth_update(struct amdgpu_device *adev)
+{
+	struct drm_display_mode *mode = NULL;
+	u32 num_heads = 0, lb_size;
+	int i;
+
+	amdgpu_display_update_priority(adev);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i]->base.enabled)
+			num_heads++;
+	}
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		mode = &adev->mode_info.crtcs[i]->base.mode;
+		lb_size = dce_v8_0_line_buffer_adjust(adev, adev->mode_info.crtcs[i], mode);
+		dce_v8_0_program_watermarks(adev, adev->mode_info.crtcs[i],
+					    lb_size, num_heads);
+	}
+}
+
+static void dce_v8_0_audio_get_connected_pins(struct amdgpu_device *adev)
+{
+	int i;
+	u32 offset, tmp;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		offset = adev->mode_info.audio.pin[i].offset;
+		tmp = RREG32_AUDIO_ENDPT(offset,
+					 ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT);
+		if (((tmp &
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY_MASK) >>
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY__SHIFT) == 1)
+			adev->mode_info.audio.pin[i].connected = false;
+		else
+			adev->mode_info.audio.pin[i].connected = true;
+	}
+}
+
+static struct amdgpu_audio_pin *dce_v8_0_audio_get_pin(struct amdgpu_device *adev)
+{
+	int i;
+
+	dce_v8_0_audio_get_connected_pins(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		if (adev->mode_info.audio.pin[i].connected)
+			return &adev->mode_info.audio.pin[i];
+	}
+	DRM_ERROR("No connected audio pins found!\n");
+	return NULL;
+}
+
+static void dce_v8_0_afmt_audio_select_pin(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 offset;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	offset = dig->afmt->offset;
+
+	WREG32(mmAFMT_AUDIO_SRC_CONTROL + offset,
+	       (dig->afmt->pin->id << AFMT_AUDIO_SRC_CONTROL__AFMT_AUDIO_SRC_SELECT__SHIFT));
+}
+
+static void dce_v8_0_audio_write_latency_fields(struct drm_encoder *encoder,
+						struct drm_display_mode *mode)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u32 tmp = 0, offset;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	offset = dig->afmt->pin->offset;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE) {
+		if (connector->latency_present[1])
+			tmp =
+			(connector->video_latency[1] <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__VIDEO_LIPSYNC__SHIFT) |
+			(connector->audio_latency[1] <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__AUDIO_LIPSYNC__SHIFT);
+		else
+			tmp =
+			(0 <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__VIDEO_LIPSYNC__SHIFT) |
+			(0 <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__AUDIO_LIPSYNC__SHIFT);
+	} else {
+		if (connector->latency_present[0])
+			tmp =
+			(connector->video_latency[0] <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__VIDEO_LIPSYNC__SHIFT) |
+			(connector->audio_latency[0] <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__AUDIO_LIPSYNC__SHIFT);
+		else
+			tmp =
+			(0 <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__VIDEO_LIPSYNC__SHIFT) |
+			(0 <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__AUDIO_LIPSYNC__SHIFT);
+
+	}
+	WREG32_AUDIO_ENDPT(offset, ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC, tmp);
+}
+
+static void dce_v8_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u32 offset, tmp;
+	u8 *sadb = NULL;
+	int sad_count;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	offset = dig->afmt->pin->offset;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector_edid(connector), &sadb);
+	if (sad_count < 0) {
+		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sad_count);
+		sad_count = 0;
+	}
+
+	/* program the speaker allocation */
+	tmp = RREG32_AUDIO_ENDPT(offset, ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER);
+	tmp &= ~(AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER__DP_CONNECTION_MASK |
+		AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER__SPEAKER_ALLOCATION_MASK);
+	/* set HDMI mode */
+	tmp |= AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER__HDMI_CONNECTION_MASK;
+	if (sad_count)
+		tmp |= (sadb[0] << AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER__SPEAKER_ALLOCATION__SHIFT);
+	else
+		tmp |= (5 << AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER__SPEAKER_ALLOCATION__SHIFT); /* stereo */
+	WREG32_AUDIO_ENDPT(offset, ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER, tmp);
+
+	kfree(sadb);
+}
+
+static void dce_v8_0_audio_write_sad_regs(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 offset;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	struct cea_sad *sads;
+	int i, sad_count;
+
+	static const u16 eld_reg_to_type[][2] = {
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0, HDMI_AUDIO_CODING_TYPE_PCM },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR1, HDMI_AUDIO_CODING_TYPE_AC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR2, HDMI_AUDIO_CODING_TYPE_MPEG1 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR3, HDMI_AUDIO_CODING_TYPE_MP3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR4, HDMI_AUDIO_CODING_TYPE_MPEG2 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR5, HDMI_AUDIO_CODING_TYPE_AAC_LC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR6, HDMI_AUDIO_CODING_TYPE_DTS },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR7, HDMI_AUDIO_CODING_TYPE_ATRAC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR9, HDMI_AUDIO_CODING_TYPE_EAC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR10, HDMI_AUDIO_CODING_TYPE_DTS_HD },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR11, HDMI_AUDIO_CODING_TYPE_MLP },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO },
+	};
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	offset = dig->afmt->pin->offset;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_sad(amdgpu_connector_edid(connector), &sads);
+	if (sad_count <= 0) {
+		DRM_ERROR("Couldn't read SADs: %d\n", sad_count);
+		return;
+	}
+	BUG_ON(!sads);
+
+	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
+		u32 value = 0;
+		u8 stereo_freqs = 0;
+		int max_channels = -1;
+		int j;
+
+		for (j = 0; j < sad_count; j++) {
+			struct cea_sad *sad = &sads[j];
+
+			if (sad->format == eld_reg_to_type[i][1]) {
+				if (sad->channels > max_channels) {
+					value = (sad->channels <<
+						 AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__MAX_CHANNELS__SHIFT) |
+					        (sad->byte2 <<
+						 AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__DESCRIPTOR_BYTE_2__SHIFT) |
+					        (sad->freq <<
+						 AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__SUPPORTED_FREQUENCIES__SHIFT);
+					max_channels = sad->channels;
+				}
+
+				if (sad->format == HDMI_AUDIO_CODING_TYPE_PCM)
+					stereo_freqs |= sad->freq;
+				else
+					break;
+			}
+		}
+
+		value |= (stereo_freqs <<
+			AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__SUPPORTED_FREQUENCIES_STEREO__SHIFT);
+
+		WREG32_AUDIO_ENDPT(offset, eld_reg_to_type[i][0], value);
+	}
+
+	kfree(sads);
+}
+
+static void dce_v8_0_audio_enable(struct amdgpu_device *adev,
+				  struct amdgpu_audio_pin *pin,
+				  bool enable)
+{
+	if (!pin)
+		return;
+
+	WREG32_AUDIO_ENDPT(pin->offset, ixAZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+		enable ? AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL__AUDIO_ENABLED_MASK : 0);
+}
+
+static const u32 pin_offsets[7] =
+{
+	(0x1780 - 0x1780),
+	(0x1786 - 0x1780),
+	(0x178c - 0x1780),
+	(0x1792 - 0x1780),
+	(0x1798 - 0x1780),
+	(0x179d - 0x1780),
+	(0x17a4 - 0x1780),
+};
+
+static int dce_v8_0_audio_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!amdgpu_audio)
+		return 0;
+
+	adev->mode_info.audio.enabled = true;
+
+	if (adev->asic_type == CHIP_KAVERI) /* KV: 4 streams, 7 endpoints */
+		adev->mode_info.audio.num_pins = 7;
+	else if ((adev->asic_type == CHIP_KABINI) ||
+		 (adev->asic_type == CHIP_MULLINS)) /* KB/ML: 2 streams, 3 endpoints */
+		adev->mode_info.audio.num_pins = 3;
+	else if ((adev->asic_type == CHIP_BONAIRE) ||
+		 (adev->asic_type == CHIP_HAWAII))/* BN/HW: 6 streams, 7 endpoints */
+		adev->mode_info.audio.num_pins = 7;
+	else
+		adev->mode_info.audio.num_pins = 3;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		adev->mode_info.audio.pin[i].channels = -1;
+		adev->mode_info.audio.pin[i].rate = -1;
+		adev->mode_info.audio.pin[i].bits_per_sample = -1;
+		adev->mode_info.audio.pin[i].status_bits = 0;
+		adev->mode_info.audio.pin[i].category_code = 0;
+		adev->mode_info.audio.pin[i].connected = false;
+		adev->mode_info.audio.pin[i].offset = pin_offsets[i];
+		adev->mode_info.audio.pin[i].id = i;
+		/* disable audio.  it will be set up later */
+		/* XXX remove once we switch to ip funcs */
+		dce_v8_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	return 0;
+}
+
+static void dce_v8_0_audio_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!amdgpu_audio)
+		return;
+
+	if (!adev->mode_info.audio.enabled)
+		return;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
+		dce_v8_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+
+	adev->mode_info.audio.enabled = false;
+}
+
+/*
+ * update the N and CTS parameters for a given pixel clock rate
+ */
+static void dce_v8_0_afmt_update_ACR(struct drm_encoder *encoder, uint32_t clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_afmt_acr acr = amdgpu_afmt_acr(clock);
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	uint32_t offset = dig->afmt->offset;
+
+	WREG32(mmHDMI_ACR_32_0 + offset, (acr.cts_32khz << HDMI_ACR_32_0__HDMI_ACR_CTS_32__SHIFT));
+	WREG32(mmHDMI_ACR_32_1 + offset, acr.n_32khz);
+
+	WREG32(mmHDMI_ACR_44_0 + offset, (acr.cts_44_1khz << HDMI_ACR_44_0__HDMI_ACR_CTS_44__SHIFT));
+	WREG32(mmHDMI_ACR_44_1 + offset, acr.n_44_1khz);
+
+	WREG32(mmHDMI_ACR_48_0 + offset, (acr.cts_48khz << HDMI_ACR_48_0__HDMI_ACR_CTS_48__SHIFT));
+	WREG32(mmHDMI_ACR_48_1 + offset, acr.n_48khz);
+}
+
+/*
+ * build a HDMI Video Info Frame
+ */
+static void dce_v8_0_afmt_update_avi_infoframe(struct drm_encoder *encoder,
+					       void *buffer, size_t size)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	uint32_t offset = dig->afmt->offset;
+	uint8_t *frame = buffer + 3;
+	uint8_t *header = buffer;
+
+	WREG32(mmAFMT_AVI_INFO0 + offset,
+		frame[0x0] | (frame[0x1] << 8) | (frame[0x2] << 16) | (frame[0x3] << 24));
+	WREG32(mmAFMT_AVI_INFO1 + offset,
+		frame[0x4] | (frame[0x5] << 8) | (frame[0x6] << 16) | (frame[0x7] << 24));
+	WREG32(mmAFMT_AVI_INFO2 + offset,
+		frame[0x8] | (frame[0x9] << 8) | (frame[0xA] << 16) | (frame[0xB] << 24));
+	WREG32(mmAFMT_AVI_INFO3 + offset,
+		frame[0xC] | (frame[0xD] << 8) | (header[1] << 24));
+}
+
+static void dce_v8_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	u32 dto_phase = 24 * 1000;
+	u32 dto_modulo = clock;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* XXX two dtos; generally use dto0 for hdmi */
+	/* Express [24MHz / target pixel clock] as an exact rational
+	 * number (coefficient of two integer numbers.  DCCG_AUDIO_DTOx_PHASE
+	 * is the numerator, DCCG_AUDIO_DTOx_MODULE is the denominator
+	 */
+	WREG32(mmDCCG_AUDIO_DTO_SOURCE, (amdgpu_crtc->crtc_id << DCCG_AUDIO_DTO_SOURCE__DCCG_AUDIO_DTO0_SOURCE_SEL__SHIFT));
+	WREG32(mmDCCG_AUDIO_DTO0_PHASE, dto_phase);
+	WREG32(mmDCCG_AUDIO_DTO0_MODULE, dto_modulo);
+}
+
+/*
+ * update the info frames with the data from the current display mode
+ */
+static void dce_v8_0_afmt_setmode(struct drm_encoder *encoder,
+				  struct drm_display_mode *mode)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	u8 buffer[HDMI_INFOFRAME_HEADER_SIZE + HDMI_AVI_INFOFRAME_SIZE];
+	struct hdmi_avi_infoframe frame;
+	uint32_t offset, val;
+	ssize_t err;
+	int bpc = 8;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (!dig->afmt->enabled)
+		return;
+
+	offset = dig->afmt->offset;
+
+	/* hdmi deep color mode general control packets setup, if bpc > 8 */
+	if (encoder->crtc) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+		bpc = amdgpu_crtc->bpc;
+	}
+
+	/* disable audio prior to setting up hw */
+	dig->afmt->pin = dce_v8_0_audio_get_pin(adev);
+	dce_v8_0_audio_enable(adev, dig->afmt->pin, false);
+
+	dce_v8_0_audio_set_dto(encoder, mode->clock);
+
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + offset,
+	       HDMI_VBI_PACKET_CONTROL__HDMI_NULL_SEND_MASK); /* send null packets when required */
+
+	WREG32(mmAFMT_AUDIO_CRC_CONTROL + offset, 0x1000);
+
+	val = RREG32(mmHDMI_CONTROL + offset);
+	val &= ~HDMI_CONTROL__HDMI_DEEP_COLOR_ENABLE_MASK;
+	val &= ~HDMI_CONTROL__HDMI_DEEP_COLOR_DEPTH_MASK;
+
+	switch (bpc) {
+	case 0:
+	case 6:
+	case 8:
+	case 16:
+	default:
+		DRM_DEBUG("%s: Disabling hdmi deep color for %d bpc.\n",
+			  connector->name, bpc);
+		break;
+	case 10:
+		val |= HDMI_CONTROL__HDMI_DEEP_COLOR_ENABLE_MASK;
+		val |= 1 << HDMI_CONTROL__HDMI_DEEP_COLOR_DEPTH__SHIFT;
+		DRM_DEBUG("%s: Enabling hdmi deep color 30 for 10 bpc.\n",
+			  connector->name);
+		break;
+	case 12:
+		val |= HDMI_CONTROL__HDMI_DEEP_COLOR_ENABLE_MASK;
+		val |= 2 << HDMI_CONTROL__HDMI_DEEP_COLOR_DEPTH__SHIFT;
+		DRM_DEBUG("%s: Enabling hdmi deep color 36 for 12 bpc.\n",
+			  connector->name);
+		break;
+	}
+
+	WREG32(mmHDMI_CONTROL + offset, val);
+
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + offset,
+	       HDMI_VBI_PACKET_CONTROL__HDMI_NULL_SEND_MASK | /* send null packets when required */
+	       HDMI_VBI_PACKET_CONTROL__HDMI_GC_SEND_MASK | /* send general control packets */
+	       HDMI_VBI_PACKET_CONTROL__HDMI_GC_CONT_MASK); /* send general control packets every frame */
+
+	WREG32(mmHDMI_INFOFRAME_CONTROL0 + offset,
+	       HDMI_INFOFRAME_CONTROL0__HDMI_AUDIO_INFO_SEND_MASK | /* enable audio info frames (frames won't be set until audio is enabled) */
+	       HDMI_INFOFRAME_CONTROL0__HDMI_AUDIO_INFO_CONT_MASK); /* required for audio info values to be updated */
+
+	WREG32(mmAFMT_INFOFRAME_CONTROL0 + offset,
+	       AFMT_INFOFRAME_CONTROL0__AFMT_AUDIO_INFO_UPDATE_MASK); /* required for audio info values to be updated */
+
+	WREG32(mmHDMI_INFOFRAME_CONTROL1 + offset,
+	       (2 << HDMI_INFOFRAME_CONTROL1__HDMI_AUDIO_INFO_LINE__SHIFT)); /* anything other than 0 */
+
+	WREG32(mmHDMI_GC + offset, 0); /* unset HDMI_GC_AVMUTE */
+
+	WREG32(mmHDMI_AUDIO_PACKET_CONTROL + offset,
+	       (1 << HDMI_AUDIO_PACKET_CONTROL__HDMI_AUDIO_DELAY_EN__SHIFT) | /* set the default audio delay */
+	       (3 << HDMI_AUDIO_PACKET_CONTROL__HDMI_AUDIO_PACKETS_PER_LINE__SHIFT)); /* should be suffient for all audio modes and small enough for all hblanks */
+
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + offset,
+	       AFMT_AUDIO_PACKET_CONTROL__AFMT_60958_CS_UPDATE_MASK); /* allow 60958 channel status fields to be updated */
+
+	/* fglrx clears sth in AFMT_AUDIO_PACKET_CONTROL2 here */
+
+	if (bpc > 8)
+		WREG32(mmHDMI_ACR_PACKET_CONTROL + offset,
+		       HDMI_ACR_PACKET_CONTROL__HDMI_ACR_AUTO_SEND_MASK); /* allow hw to sent ACR packets when required */
+	else
+		WREG32(mmHDMI_ACR_PACKET_CONTROL + offset,
+		       HDMI_ACR_PACKET_CONTROL__HDMI_ACR_SOURCE_MASK | /* select SW CTS value */
+		       HDMI_ACR_PACKET_CONTROL__HDMI_ACR_AUTO_SEND_MASK); /* allow hw to sent ACR packets when required */
+
+	dce_v8_0_afmt_update_ACR(encoder, mode->clock);
+
+	WREG32(mmAFMT_60958_0 + offset,
+	       (1 << AFMT_60958_0__AFMT_60958_CS_CHANNEL_NUMBER_L__SHIFT));
+
+	WREG32(mmAFMT_60958_1 + offset,
+	       (2 << AFMT_60958_1__AFMT_60958_CS_CHANNEL_NUMBER_R__SHIFT));
+
+	WREG32(mmAFMT_60958_2 + offset,
+	       (3 << AFMT_60958_2__AFMT_60958_CS_CHANNEL_NUMBER_2__SHIFT) |
+	       (4 << AFMT_60958_2__AFMT_60958_CS_CHANNEL_NUMBER_3__SHIFT) |
+	       (5 << AFMT_60958_2__AFMT_60958_CS_CHANNEL_NUMBER_4__SHIFT) |
+	       (6 << AFMT_60958_2__AFMT_60958_CS_CHANNEL_NUMBER_5__SHIFT) |
+	       (7 << AFMT_60958_2__AFMT_60958_CS_CHANNEL_NUMBER_6__SHIFT) |
+	       (8 << AFMT_60958_2__AFMT_60958_CS_CHANNEL_NUMBER_7__SHIFT));
+
+	dce_v8_0_audio_write_speaker_allocation(encoder);
+
+
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL2 + offset,
+	       (0xff << AFMT_AUDIO_PACKET_CONTROL2__AFMT_AUDIO_CHANNEL_ENABLE__SHIFT));
+
+	dce_v8_0_afmt_audio_select_pin(encoder);
+	dce_v8_0_audio_write_sad_regs(encoder);
+	dce_v8_0_audio_write_latency_fields(encoder, mode);
+
+	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, mode, false);
+	if (err < 0) {
+		DRM_ERROR("failed to setup AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	err = hdmi_avi_infoframe_pack(&frame, buffer, sizeof(buffer));
+	if (err < 0) {
+		DRM_ERROR("failed to pack AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	dce_v8_0_afmt_update_avi_infoframe(encoder, buffer, sizeof(buffer));
+
+	WREG32_OR(mmHDMI_INFOFRAME_CONTROL0 + offset,
+		  HDMI_INFOFRAME_CONTROL0__HDMI_AVI_INFO_SEND_MASK | /* enable AVI info frames */
+		  HDMI_INFOFRAME_CONTROL0__HDMI_AVI_INFO_CONT_MASK); /* required for audio info values to be updated */
+
+	WREG32_P(mmHDMI_INFOFRAME_CONTROL1 + offset,
+		 (2 << HDMI_INFOFRAME_CONTROL1__HDMI_AVI_INFO_LINE__SHIFT), /* anything other than 0 */
+		 ~HDMI_INFOFRAME_CONTROL1__HDMI_AVI_INFO_LINE_MASK);
+
+	WREG32_OR(mmAFMT_AUDIO_PACKET_CONTROL + offset,
+		  AFMT_AUDIO_PACKET_CONTROL__AFMT_AUDIO_SAMPLE_SEND_MASK); /* send audio packets */
+
+	WREG32(mmAFMT_RAMP_CONTROL0 + offset, 0x00FFFFFF);
+	WREG32(mmAFMT_RAMP_CONTROL1 + offset, 0x007FFFFF);
+	WREG32(mmAFMT_RAMP_CONTROL2 + offset, 0x00000001);
+	WREG32(mmAFMT_RAMP_CONTROL3 + offset, 0x00000001);
+
+	/* enable audio after setting up hw */
+	dce_v8_0_audio_enable(adev, dig->afmt->pin, true);
+}
+
+static void dce_v8_0_afmt_enable(struct drm_encoder *encoder, bool enable)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (enable && dig->afmt->enabled)
+		return;
+	if (!enable && !dig->afmt->enabled)
+		return;
+
+	if (!enable && dig->afmt->pin) {
+		dce_v8_0_audio_enable(adev, dig->afmt->pin, false);
+		dig->afmt->pin = NULL;
+	}
+
+	dig->afmt->enabled = enable;
+
+	DRM_DEBUG("%sabling AFMT interface @ 0x%04X for encoder 0x%x\n",
+		  enable ? "En" : "Dis", dig->afmt->offset, amdgpu_encoder->encoder_id);
+}
+
+static int dce_v8_0_afmt_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++)
+		adev->mode_info.afmt[i] = NULL;
+
+	/* DCE8 has audio blocks tied to DIG encoders */
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		adev->mode_info.afmt[i] = kzalloc(sizeof(struct amdgpu_afmt), GFP_KERNEL);
+		if (adev->mode_info.afmt[i]) {
+			adev->mode_info.afmt[i]->offset = dig_offsets[i];
+			adev->mode_info.afmt[i]->id = i;
+		} else {
+			int j;
+			for (j = 0; j < i; j++) {
+				kfree(adev->mode_info.afmt[j]);
+				adev->mode_info.afmt[j] = NULL;
+			}
+			return -ENOMEM;
+		}
+	}
+	return 0;
+}
+
+static void dce_v8_0_afmt_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		kfree(adev->mode_info.afmt[i]);
+		adev->mode_info.afmt[i] = NULL;
+	}
+}
+
+static const u32 vga_control_regs[6] =
+{
+	mmD1VGA_CONTROL,
+	mmD2VGA_CONTROL,
+	mmD3VGA_CONTROL,
+	mmD4VGA_CONTROL,
+	mmD5VGA_CONTROL,
+	mmD6VGA_CONTROL,
+};
+
+static void dce_v8_0_vga_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 vga_control;
+
+	vga_control = RREG32(vga_control_regs[amdgpu_crtc->crtc_id]) & ~1;
+	if (enable)
+		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control | 1);
+	else
+		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control);
+}
+
+static void dce_v8_0_grph_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (enable)
+		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 1);
+	else
+		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 0);
+}
+
+static int dce_v8_0_crtc_do_set_base(struct drm_crtc *crtc,
+				     struct drm_framebuffer *fb,
+				     int x, int y, int atomic)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_framebuffer *target_fb;
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *abo;
+	uint64_t fb_location, tiling_flags;
+	uint32_t fb_format, fb_pitch_pixels;
+	u32 fb_swap = (GRPH_ENDIAN_NONE << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+	u32 pipe_config;
+	u32 viewport_w, viewport_h;
+	int r;
+	bool bypass_lut = false;
+	struct drm_format_name_buf format_name;
+
+	/* no fb bound */
+	if (!atomic && !crtc->primary->fb) {
+		DRM_DEBUG_KMS("No FB bound\n");
+		return 0;
+	}
+
+	if (atomic)
+		target_fb = fb;
+	else
+		target_fb = crtc->primary->fb;
+
+	/* If atomic, assume fb object is pinned & idle & fenced and
+	 * just update base pointers
+	 */
+	obj = target_fb->obj[0];
+	abo = gem_to_amdgpu_bo(obj);
+	r = amdgpu_bo_reserve(abo, false);
+	if (unlikely(r != 0))
+		return r;
+
+	if (!atomic) {
+		r = amdgpu_bo_pin(abo, AMDGPU_GEM_DOMAIN_VRAM);
+		if (unlikely(r != 0)) {
+			amdgpu_bo_unreserve(abo);
+			return -EINVAL;
+		}
+	}
+	fb_location = amdgpu_bo_gpu_offset(abo);
+
+	amdgpu_bo_get_tiling_flags(abo, &tiling_flags);
+	amdgpu_bo_unreserve(abo);
+
+	pipe_config = AMDGPU_TILING_GET(tiling_flags, PIPE_CONFIG);
+
+	switch (target_fb->format->format) {
+	case DRM_FORMAT_C8:
+		fb_format = ((GRPH_DEPTH_8BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_INDEXED << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+		break;
+	case DRM_FORMAT_XRGB4444:
+	case DRM_FORMAT_ARGB4444:
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB4444 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		break;
+	case DRM_FORMAT_XRGB1555:
+	case DRM_FORMAT_ARGB1555:
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB1555 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		break;
+	case DRM_FORMAT_BGRX5551:
+	case DRM_FORMAT_BGRA5551:
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_BGRA5551 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		break;
+	case DRM_FORMAT_RGB565:
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB565 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		break;
+	case DRM_FORMAT_XRGB8888:
+	case DRM_FORMAT_ARGB8888:
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB8888 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		break;
+	case DRM_FORMAT_XRGB2101010:
+	case DRM_FORMAT_ARGB2101010:
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB2101010 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	case DRM_FORMAT_BGRX1010102:
+	case DRM_FORMAT_BGRA1010102:
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_BGRA1010102 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	default:
+		DRM_ERROR("Unsupported screen format %s\n",
+		          drm_get_format_name(target_fb->format->format, &format_name));
+		return -EINVAL;
+	}
+
+	if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_2D_TILED_THIN1) {
+		unsigned bankw, bankh, mtaspect, tile_split, num_banks;
+
+		bankw = AMDGPU_TILING_GET(tiling_flags, BANK_WIDTH);
+		bankh = AMDGPU_TILING_GET(tiling_flags, BANK_HEIGHT);
+		mtaspect = AMDGPU_TILING_GET(tiling_flags, MACRO_TILE_ASPECT);
+		tile_split = AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT);
+		num_banks = AMDGPU_TILING_GET(tiling_flags, NUM_BANKS);
+
+		fb_format |= (num_banks << GRPH_CONTROL__GRPH_NUM_BANKS__SHIFT);
+		fb_format |= (GRPH_ARRAY_2D_TILED_THIN1 << GRPH_CONTROL__GRPH_ARRAY_MODE__SHIFT);
+		fb_format |= (tile_split << GRPH_CONTROL__GRPH_TILE_SPLIT__SHIFT);
+		fb_format |= (bankw << GRPH_CONTROL__GRPH_BANK_WIDTH__SHIFT);
+		fb_format |= (bankh << GRPH_CONTROL__GRPH_BANK_HEIGHT__SHIFT);
+		fb_format |= (mtaspect << GRPH_CONTROL__GRPH_MACRO_TILE_ASPECT__SHIFT);
+		fb_format |= (DISPLAY_MICRO_TILING << GRPH_CONTROL__GRPH_MICRO_TILE_MODE__SHIFT);
+	} else if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_1D_TILED_THIN1) {
+		fb_format |= (GRPH_ARRAY_1D_TILED_THIN1 << GRPH_CONTROL__GRPH_ARRAY_MODE__SHIFT);
+	}
+
+	fb_format |= (pipe_config << GRPH_CONTROL__GRPH_PIPE_CONFIG__SHIFT);
+
+	dce_v8_0_vga_enable(crtc, false);
+
+	/* Make sure surface address is updated at vertical blank rather than
+	 * horizontal blank
+	 */
+	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32)fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32) fb_location & GRPH_SECONDARY_SURFACE_ADDRESS__GRPH_SECONDARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
+	WREG32(mmGRPH_SWAP_CNTL + amdgpu_crtc->crtc_offset, fb_swap);
+
+	/*
+	 * The LUT only has 256 slots for indexing by a 8 bpc fb. Bypass the LUT
+	 * for > 8 bpc scanout to avoid truncation of fb indices to 8 msb's, to
+	 * retain the full precision throughout the pipeline.
+	 */
+	WREG32_P(mmGRPH_LUT_10BIT_BYPASS_CONTROL + amdgpu_crtc->crtc_offset,
+		 (bypass_lut ? LUT_10BIT_BYPASS_EN : 0),
+		 ~LUT_10BIT_BYPASS_EN);
+
+	if (bypass_lut)
+		DRM_DEBUG_KMS("Bypassing hardware LUT due to 10 bit fb scanout.\n");
+
+	WREG32(mmGRPH_SURFACE_OFFSET_X + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_SURFACE_OFFSET_Y + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_Y_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_END + amdgpu_crtc->crtc_offset, target_fb->width);
+	WREG32(mmGRPH_Y_END + amdgpu_crtc->crtc_offset, target_fb->height);
+
+	fb_pitch_pixels = target_fb->pitches[0] / target_fb->format->cpp[0];
+	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset, fb_pitch_pixels);
+
+	dce_v8_0_grph_enable(crtc, true);
+
+	WREG32(mmLB_DESKTOP_HEIGHT + amdgpu_crtc->crtc_offset,
+	       target_fb->height);
+
+	x &= ~3;
+	y &= ~1;
+	WREG32(mmVIEWPORT_START + amdgpu_crtc->crtc_offset,
+	       (x << 16) | y);
+	viewport_w = crtc->mode.hdisplay;
+	viewport_h = (crtc->mode.vdisplay + 1) & ~1;
+	WREG32(mmVIEWPORT_SIZE + amdgpu_crtc->crtc_offset,
+	       (viewport_w << 16) | viewport_h);
+
+	/* set pageflip to happen anywhere in vblank interval */
+	WREG32(mmMASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 0);
+
+	if (!atomic && fb && fb != crtc->primary->fb) {
+		abo = gem_to_amdgpu_bo(fb->obj[0]);
+		r = amdgpu_bo_reserve(abo, true);
+		if (unlikely(r != 0))
+			return r;
+		amdgpu_bo_unpin(abo);
+		amdgpu_bo_unreserve(abo);
+	}
+
+	/* Bytes per pixel may have changed */
+	dce_v8_0_bandwidth_update(adev);
+
+	return 0;
+}
+
+static void dce_v8_0_set_interleave(struct drm_crtc *crtc,
+				    struct drm_display_mode *mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		WREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset,
+		       LB_DATA_FORMAT__INTERLEAVE_EN__SHIFT);
+	else
+		WREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset, 0);
+}
+
+static void dce_v8_0_crtc_load_lut(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u16 *r, *g, *b;
+	int i;
+
+	DRM_DEBUG_KMS("%d\n", amdgpu_crtc->crtc_id);
+
+	WREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((INPUT_CSC_BYPASS << INPUT_CSC_CONTROL__INPUT_CSC_GRPH_MODE__SHIFT) |
+		(INPUT_CSC_BYPASS << INPUT_CSC_CONTROL__INPUT_CSC_OVL_MODE__SHIFT)));
+	WREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset,
+	       PRESCALE_GRPH_CONTROL__GRPH_PRESCALE_BYPASS_MASK);
+	WREG32(mmPRESCALE_OVL_CONTROL + amdgpu_crtc->crtc_offset,
+	       PRESCALE_OVL_CONTROL__OVL_PRESCALE_BYPASS_MASK);
+	WREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((INPUT_GAMMA_USE_LUT << INPUT_GAMMA_CONTROL__GRPH_INPUT_GAMMA_MODE__SHIFT) |
+		(INPUT_GAMMA_USE_LUT << INPUT_GAMMA_CONTROL__OVL_INPUT_GAMMA_MODE__SHIFT)));
+
+	WREG32(mmDC_LUT_CONTROL + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_BLACK_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_RED + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_WHITE_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_RED + amdgpu_crtc->crtc_offset, 0xffff);
+
+	WREG32(mmDC_LUT_RW_MODE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_WRITE_EN_MASK + amdgpu_crtc->crtc_offset, 0x00000007);
+
+	WREG32(mmDC_LUT_RW_INDEX + amdgpu_crtc->crtc_offset, 0);
+	r = crtc->gamma_store;
+	g = r + crtc->gamma_size;
+	b = g + crtc->gamma_size;
+	for (i = 0; i < 256; i++) {
+		WREG32(mmDC_LUT_30_COLOR + amdgpu_crtc->crtc_offset,
+		       ((*r++ & 0xffc0) << 14) |
+		       ((*g++ & 0xffc0) << 4) |
+		       (*b++ >> 6));
+	}
+
+	WREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((DEGAMMA_BYPASS << DEGAMMA_CONTROL__GRPH_DEGAMMA_MODE__SHIFT) |
+		(DEGAMMA_BYPASS << DEGAMMA_CONTROL__OVL_DEGAMMA_MODE__SHIFT) |
+		(DEGAMMA_BYPASS << DEGAMMA_CONTROL__CURSOR_DEGAMMA_MODE__SHIFT)));
+	WREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((GAMUT_REMAP_BYPASS << GAMUT_REMAP_CONTROL__GRPH_GAMUT_REMAP_MODE__SHIFT) |
+		(GAMUT_REMAP_BYPASS << GAMUT_REMAP_CONTROL__OVL_GAMUT_REMAP_MODE__SHIFT)));
+	WREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((REGAMMA_BYPASS << REGAMMA_CONTROL__GRPH_REGAMMA_MODE__SHIFT) |
+		(REGAMMA_BYPASS << REGAMMA_CONTROL__OVL_REGAMMA_MODE__SHIFT)));
+	WREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((OUTPUT_CSC_BYPASS << OUTPUT_CSC_CONTROL__OUTPUT_CSC_GRPH_MODE__SHIFT) |
+		(OUTPUT_CSC_BYPASS << OUTPUT_CSC_CONTROL__OUTPUT_CSC_OVL_MODE__SHIFT)));
+	/* XXX match this to the depth of the crtc fmt block, move to modeset? */
+	WREG32(0x1a50 + amdgpu_crtc->crtc_offset, 0);
+	/* XXX this only needs to be programmed once per crtc at startup,
+	 * not sure where the best place for it is
+	 */
+	WREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset,
+	       ALPHA_CONTROL__CURSOR_ALPHA_BLND_ENA_MASK);
+}
+
+static int dce_v8_0_pick_dig_encoder(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		if (dig->linkb)
+			return 1;
+		else
+			return 0;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		if (dig->linkb)
+			return 3;
+		else
+			return 2;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+		if (dig->linkb)
+			return 5;
+		else
+			return 4;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		return 6;
+		break;
+	default:
+		DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
+		return 0;
+	}
+}
+
+/**
+ * dce_v8_0_pick_pll - Allocate a PPLL for use by the crtc.
+ *
+ * @crtc: drm crtc
+ *
+ * Returns the PPLL (Pixel PLL) to be used by the crtc.  For DP monitors
+ * a single PPLL can be used for all DP crtcs/encoders.  For non-DP
+ * monitors a dedicated PPLL must be used.  If a particular board has
+ * an external DP PLL, return ATOM_PPLL_INVALID to skip PLL programming
+ * as there is no need to program the PLL itself.  If we are not able to
+ * allocate a PLL, return ATOM_PPLL_INVALID to skip PLL programming to
+ * avoid messing up an existing monitor.
+ *
+ * Asic specific PLL information
+ *
+ * DCE 8.x
+ * KB/KV
+ * - PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP)
+ * CI
+ * - PPLL0, PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP) and DAC
+ *
+ */
+static u32 dce_v8_0_pick_pll(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 pll_in_use;
+	int pll;
+
+	if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder))) {
+		if (adev->clock.dp_extclk)
+			/* skip PPLL programming if using ext clock */
+			return ATOM_PPLL_INVALID;
+		else {
+			/* use the same PPLL for all DP monitors */
+			pll = amdgpu_pll_get_shared_dp_ppll(crtc);
+			if (pll != ATOM_PPLL_INVALID)
+				return pll;
+		}
+	} else {
+		/* use the same PPLL for all monitors with the same clock */
+		pll = amdgpu_pll_get_shared_nondp_ppll(crtc);
+		if (pll != ATOM_PPLL_INVALID)
+			return pll;
+	}
+	/* otherwise, pick one of the plls */
+	if ((adev->asic_type == CHIP_KABINI) ||
+	    (adev->asic_type == CHIP_MULLINS)) {
+		/* KB/ML has PPLL1 and PPLL2 */
+		pll_in_use = amdgpu_pll_get_use_mask(crtc);
+		if (!(pll_in_use & (1 << ATOM_PPLL2)))
+			return ATOM_PPLL2;
+		if (!(pll_in_use & (1 << ATOM_PPLL1)))
+			return ATOM_PPLL1;
+		DRM_ERROR("unable to allocate a PPLL\n");
+		return ATOM_PPLL_INVALID;
+	} else {
+		/* CI/KV has PPLL0, PPLL1, and PPLL2 */
+		pll_in_use = amdgpu_pll_get_use_mask(crtc);
+		if (!(pll_in_use & (1 << ATOM_PPLL2)))
+			return ATOM_PPLL2;
+		if (!(pll_in_use & (1 << ATOM_PPLL1)))
+			return ATOM_PPLL1;
+		if (!(pll_in_use & (1 << ATOM_PPLL0)))
+			return ATOM_PPLL0;
+		DRM_ERROR("unable to allocate a PPLL\n");
+		return ATOM_PPLL_INVALID;
+	}
+	return ATOM_PPLL_INVALID;
+}
+
+static void dce_v8_0_lock_cursor(struct drm_crtc *crtc, bool lock)
+{
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	uint32_t cur_lock;
+
+	cur_lock = RREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset);
+	if (lock)
+		cur_lock |= CUR_UPDATE__CURSOR_UPDATE_LOCK_MASK;
+	else
+		cur_lock &= ~CUR_UPDATE__CURSOR_UPDATE_LOCK_MASK;
+	WREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset, cur_lock);
+}
+
+static void dce_v8_0_hide_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
+		   (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
+		   (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
+}
+
+static void dce_v8_0_show_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+
+	WREG32(mmCUR_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(amdgpu_crtc->cursor_addr));
+	WREG32(mmCUR_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(amdgpu_crtc->cursor_addr));
+
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
+		   CUR_CONTROL__CURSOR_EN_MASK |
+		   (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
+		   (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
+}
+
+static int dce_v8_0_cursor_move_locked(struct drm_crtc *crtc,
+				       int x, int y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	int xorigin = 0, yorigin = 0;
+
+	amdgpu_crtc->cursor_x = x;
+	amdgpu_crtc->cursor_y = y;
+
+	/* avivo cursor are offset into the total surface */
+	x += crtc->x;
+	y += crtc->y;
+	DRM_DEBUG("x %d y %d c->x %d c->y %d\n", x, y, crtc->x, crtc->y);
+
+	if (x < 0) {
+		xorigin = min(-x, amdgpu_crtc->max_cursor_width - 1);
+		x = 0;
+	}
+	if (y < 0) {
+		yorigin = min(-y, amdgpu_crtc->max_cursor_height - 1);
+		y = 0;
+	}
+
+	WREG32(mmCUR_POSITION + amdgpu_crtc->crtc_offset, (x << 16) | y);
+	WREG32(mmCUR_HOT_SPOT + amdgpu_crtc->crtc_offset, (xorigin << 16) | yorigin);
+	WREG32(mmCUR_SIZE + amdgpu_crtc->crtc_offset,
+	       ((amdgpu_crtc->cursor_width - 1) << 16) | (amdgpu_crtc->cursor_height - 1));
+
+	return 0;
+}
+
+static int dce_v8_0_crtc_cursor_move(struct drm_crtc *crtc,
+				     int x, int y)
+{
+	int ret;
+
+	dce_v8_0_lock_cursor(crtc, true);
+	ret = dce_v8_0_cursor_move_locked(crtc, x, y);
+	dce_v8_0_lock_cursor(crtc, false);
+
+	return ret;
+}
+
+static int dce_v8_0_crtc_cursor_set2(struct drm_crtc *crtc,
+				     struct drm_file *file_priv,
+				     uint32_t handle,
+				     uint32_t width,
+				     uint32_t height,
+				     int32_t hot_x,
+				     int32_t hot_y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *aobj;
+	int ret;
+
+	if (!handle) {
+		/* turn off cursor */
+		dce_v8_0_hide_cursor(crtc);
+		obj = NULL;
+		goto unpin;
+	}
+
+	if ((width > amdgpu_crtc->max_cursor_width) ||
+	    (height > amdgpu_crtc->max_cursor_height)) {
+		DRM_ERROR("bad cursor width or height %d x %d\n", width, height);
+		return -EINVAL;
+	}
+
+	obj = drm_gem_object_lookup(file_priv, handle);
+	if (!obj) {
+		DRM_ERROR("Cannot find cursor object %x for crtc %d\n", handle, amdgpu_crtc->crtc_id);
+		return -ENOENT;
+	}
+
+	aobj = gem_to_amdgpu_bo(obj);
+	ret = amdgpu_bo_reserve(aobj, false);
+	if (ret != 0) {
+		drm_gem_object_put_unlocked(obj);
+		return ret;
+	}
+
+	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM);
+	amdgpu_bo_unreserve(aobj);
+	if (ret) {
+		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
+		drm_gem_object_put_unlocked(obj);
+		return ret;
+	}
+	amdgpu_crtc->cursor_addr = amdgpu_bo_gpu_offset(aobj);
+
+	dce_v8_0_lock_cursor(crtc, true);
+
+	if (width != amdgpu_crtc->cursor_width ||
+	    height != amdgpu_crtc->cursor_height ||
+	    hot_x != amdgpu_crtc->cursor_hot_x ||
+	    hot_y != amdgpu_crtc->cursor_hot_y) {
+		int x, y;
+
+		x = amdgpu_crtc->cursor_x + amdgpu_crtc->cursor_hot_x - hot_x;
+		y = amdgpu_crtc->cursor_y + amdgpu_crtc->cursor_hot_y - hot_y;
+
+		dce_v8_0_cursor_move_locked(crtc, x, y);
+
+		amdgpu_crtc->cursor_width = width;
+		amdgpu_crtc->cursor_height = height;
+		amdgpu_crtc->cursor_hot_x = hot_x;
+		amdgpu_crtc->cursor_hot_y = hot_y;
+	}
+
+	dce_v8_0_show_cursor(crtc);
+	dce_v8_0_lock_cursor(crtc, false);
+
+unpin:
+	if (amdgpu_crtc->cursor_bo) {
+		struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+		ret = amdgpu_bo_reserve(aobj, true);
+		if (likely(ret == 0)) {
+			amdgpu_bo_unpin(aobj);
+			amdgpu_bo_unreserve(aobj);
+		}
+		drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
+	}
+
+	amdgpu_crtc->cursor_bo = obj;
+	return 0;
+}
+
+static void dce_v8_0_cursor_reset(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (amdgpu_crtc->cursor_bo) {
+		dce_v8_0_lock_cursor(crtc, true);
+
+		dce_v8_0_cursor_move_locked(crtc, amdgpu_crtc->cursor_x,
+					    amdgpu_crtc->cursor_y);
+
+		dce_v8_0_show_cursor(crtc);
+
+		dce_v8_0_lock_cursor(crtc, false);
+	}
+}
+
+static int dce_v8_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
+				   u16 *blue, uint32_t size,
+				   struct drm_modeset_acquire_ctx *ctx)
+{
+	dce_v8_0_crtc_load_lut(crtc);
+
+	return 0;
+}
+
+static void dce_v8_0_crtc_destroy(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	drm_crtc_cleanup(crtc);
+	kfree(amdgpu_crtc);
+}
+
+static const struct drm_crtc_funcs dce_v8_0_crtc_funcs = {
+	.cursor_set2 = dce_v8_0_crtc_cursor_set2,
+	.cursor_move = dce_v8_0_crtc_cursor_move,
+	.gamma_set = dce_v8_0_crtc_gamma_set,
+	.set_config = amdgpu_display_crtc_set_config,
+	.destroy = dce_v8_0_crtc_destroy,
+	.page_flip_target = amdgpu_display_crtc_page_flip_target,
+};
+
+static void dce_v8_0_crtc_dpms(struct drm_crtc *crtc, int mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	unsigned type;
+
+	switch (mode) {
+	case DRM_MODE_DPMS_ON:
+		amdgpu_crtc->enabled = true;
+		amdgpu_atombios_crtc_enable(crtc, ATOM_ENABLE);
+		dce_v8_0_vga_enable(crtc, true);
+		amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
+		dce_v8_0_vga_enable(crtc, false);
+		/* Make sure VBLANK and PFLIP interrupts are still enabled */
+		type = amdgpu_display_crtc_idx_to_irq_type(adev,
+						amdgpu_crtc->crtc_id);
+		amdgpu_irq_update(adev, &adev->crtc_irq, type);
+		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
+		drm_crtc_vblank_on(crtc);
+		dce_v8_0_crtc_load_lut(crtc);
+		break;
+	case DRM_MODE_DPMS_STANDBY:
+	case DRM_MODE_DPMS_SUSPEND:
+	case DRM_MODE_DPMS_OFF:
+		drm_crtc_vblank_off(crtc);
+		if (amdgpu_crtc->enabled) {
+			dce_v8_0_vga_enable(crtc, true);
+			amdgpu_atombios_crtc_blank(crtc, ATOM_ENABLE);
+			dce_v8_0_vga_enable(crtc, false);
+		}
+		amdgpu_atombios_crtc_enable(crtc, ATOM_DISABLE);
+		amdgpu_crtc->enabled = false;
+		break;
+	}
+	/* adjust pm to dpms */
+	amdgpu_pm_compute_clocks(adev);
+}
+
+static void dce_v8_0_crtc_prepare(struct drm_crtc *crtc)
+{
+	/* disable crtc pair power gating before programming */
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_DISABLE);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_ENABLE);
+	dce_v8_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+}
+
+static void dce_v8_0_crtc_commit(struct drm_crtc *crtc)
+{
+	dce_v8_0_crtc_dpms(crtc, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_DISABLE);
+}
+
+static void dce_v8_0_crtc_disable(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_atom_ss ss;
+	int i;
+
+	dce_v8_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+	if (crtc->primary->fb) {
+		int r;
+		struct amdgpu_bo *abo;
+
+		abo = gem_to_amdgpu_bo(crtc->primary->fb->obj[0]);
+		r = amdgpu_bo_reserve(abo, true);
+		if (unlikely(r))
+			DRM_ERROR("failed to reserve abo before unpin\n");
+		else {
+			amdgpu_bo_unpin(abo);
+			amdgpu_bo_unreserve(abo);
+		}
+	}
+	/* disable the GRPH */
+	dce_v8_0_grph_enable(crtc, false);
+
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_ENABLE);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i] &&
+		    adev->mode_info.crtcs[i]->enabled &&
+		    i != amdgpu_crtc->crtc_id &&
+		    amdgpu_crtc->pll_id == adev->mode_info.crtcs[i]->pll_id) {
+			/* one other crtc is using this pll don't turn
+			 * off the pll
+			 */
+			goto done;
+		}
+	}
+
+	switch (amdgpu_crtc->pll_id) {
+	case ATOM_PPLL1:
+	case ATOM_PPLL2:
+		/* disable the ppll */
+		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
+                                                 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
+		break;
+	case ATOM_PPLL0:
+		/* disable the ppll */
+		if ((adev->asic_type == CHIP_KAVERI) ||
+		    (adev->asic_type == CHIP_BONAIRE) ||
+		    (adev->asic_type == CHIP_HAWAII))
+			amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
+						  0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
+		break;
+	default:
+		break;
+	}
+done:
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+}
+
+static int dce_v8_0_crtc_mode_set(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode,
+				  int x, int y, struct drm_framebuffer *old_fb)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (!amdgpu_crtc->adjusted_clock)
+		return -EINVAL;
+
+	amdgpu_atombios_crtc_set_pll(crtc, adjusted_mode);
+	amdgpu_atombios_crtc_set_dtd_timing(crtc, adjusted_mode);
+	dce_v8_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+	amdgpu_atombios_crtc_overscan_setup(crtc, mode, adjusted_mode);
+	amdgpu_atombios_crtc_scaler_setup(crtc);
+	dce_v8_0_cursor_reset(crtc);
+	/* update the hw version fpr dpm */
+	amdgpu_crtc->hw_mode = *adjusted_mode;
+
+	return 0;
+}
+
+static bool dce_v8_0_crtc_mode_fixup(struct drm_crtc *crtc,
+				     const struct drm_display_mode *mode,
+				     struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct drm_encoder *encoder;
+
+	/* assign the encoder to the amdgpu crtc to avoid repeated lookups later */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		if (encoder->crtc == crtc) {
+			amdgpu_crtc->encoder = encoder;
+			amdgpu_crtc->connector = amdgpu_get_connector_for_encoder(encoder);
+			break;
+		}
+	}
+	if ((amdgpu_crtc->encoder == NULL) || (amdgpu_crtc->connector == NULL)) {
+		amdgpu_crtc->encoder = NULL;
+		amdgpu_crtc->connector = NULL;
+		return false;
+	}
+	if (!amdgpu_display_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
+		return false;
+	if (amdgpu_atombios_crtc_prepare_pll(crtc, adjusted_mode))
+		return false;
+	/* pick pll */
+	amdgpu_crtc->pll_id = dce_v8_0_pick_pll(crtc);
+	/* if we can't get a PPLL for a non-DP encoder, fail */
+	if ((amdgpu_crtc->pll_id == ATOM_PPLL_INVALID) &&
+	    !ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder)))
+		return false;
+
+	return true;
+}
+
+static int dce_v8_0_crtc_set_base(struct drm_crtc *crtc, int x, int y,
+				  struct drm_framebuffer *old_fb)
+{
+	return dce_v8_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+}
+
+static int dce_v8_0_crtc_set_base_atomic(struct drm_crtc *crtc,
+					 struct drm_framebuffer *fb,
+					 int x, int y, enum mode_set_atomic state)
+{
+       return dce_v8_0_crtc_do_set_base(crtc, fb, x, y, 1);
+}
+
+static const struct drm_crtc_helper_funcs dce_v8_0_crtc_helper_funcs = {
+	.dpms = dce_v8_0_crtc_dpms,
+	.mode_fixup = dce_v8_0_crtc_mode_fixup,
+	.mode_set = dce_v8_0_crtc_mode_set,
+	.mode_set_base = dce_v8_0_crtc_set_base,
+	.mode_set_base_atomic = dce_v8_0_crtc_set_base_atomic,
+	.prepare = dce_v8_0_crtc_prepare,
+	.commit = dce_v8_0_crtc_commit,
+	.disable = dce_v8_0_crtc_disable,
+};
+
+static int dce_v8_0_crtc_init(struct amdgpu_device *adev, int index)
+{
+	struct amdgpu_crtc *amdgpu_crtc;
+
+	amdgpu_crtc = kzalloc(sizeof(struct amdgpu_crtc) +
+			      (AMDGPUFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
+	if (amdgpu_crtc == NULL)
+		return -ENOMEM;
+
+	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_v8_0_crtc_funcs);
+
+	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
+	amdgpu_crtc->crtc_id = index;
+	adev->mode_info.crtcs[index] = amdgpu_crtc;
+
+	amdgpu_crtc->max_cursor_width = CIK_CURSOR_WIDTH;
+	amdgpu_crtc->max_cursor_height = CIK_CURSOR_HEIGHT;
+	adev->ddev->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
+	adev->ddev->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
+
+	amdgpu_crtc->crtc_offset = crtc_offsets[amdgpu_crtc->crtc_id];
+
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_v8_0_crtc_helper_funcs);
+
+	return 0;
+}
+
+static int dce_v8_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->audio_endpt_rreg = &dce_v8_0_audio_endpt_rreg;
+	adev->audio_endpt_wreg = &dce_v8_0_audio_endpt_wreg;
+
+	dce_v8_0_set_display_funcs(adev);
+
+	adev->mode_info.num_crtc = dce_v8_0_get_num_crtc(adev);
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 6;
+		break;
+	case CHIP_KAVERI:
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 7;
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 6; /* ? */
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	dce_v8_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int dce_v8_0_sw_init(void *handle)
+{
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + 1, &adev->crtc_irq);
+		if (r)
+			return r;
+	}
+
+	for (i = 8; i < 20; i += 2) {
+		r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i, &adev->pageflip_irq);
+		if (r)
+			return r;
+	}
+
+	/* HPD hotplug */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 42, &adev->hpd_irq);
+	if (r)
+		return r;
+
+	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
+
+	adev->ddev->mode_config.async_page_flip = true;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	adev->ddev->mode_config.preferred_depth = 24;
+	adev->ddev->mode_config.prefer_shadow = 1;
+
+	adev->ddev->mode_config.fb_base = adev->gmc.aper_base;
+
+	r = amdgpu_display_modeset_create_props(adev);
+	if (r)
+		return r;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	/* allocate crtcs */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = dce_v8_0_crtc_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	if (amdgpu_atombios_get_connector_info_from_object_table(adev))
+		amdgpu_display_print_display_setup(adev->ddev);
+	else
+		return -EINVAL;
+
+	/* setup afmt */
+	r = dce_v8_0_afmt_init(adev);
+	if (r)
+		return r;
+
+	r = dce_v8_0_audio_init(adev);
+	if (r)
+		return r;
+
+	drm_kms_helper_poll_init(adev->ddev);
+
+	adev->mode_info.mode_config_initialized = true;
+	return 0;
+}
+
+static int dce_v8_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	kfree(adev->mode_info.bios_hardcoded_edid);
+
+	drm_kms_helper_poll_fini(adev->ddev);
+
+	dce_v8_0_audio_fini(adev);
+
+	dce_v8_0_afmt_fini(adev);
+
+	drm_mode_config_cleanup(adev->ddev);
+	adev->mode_info.mode_config_initialized = false;
+
+	return 0;
+}
+
+static int dce_v8_0_hw_init(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* disable vga render */
+	dce_v8_0_set_vga_render_state(adev, false);
+	/* init dig PHYs, disp eng pll */
+	amdgpu_atombios_encoder_init_dig(adev);
+	amdgpu_atombios_crtc_set_disp_eng_pll(adev, adev->clock.default_dispclk);
+
+	/* initialize hpd */
+	dce_v8_0_hpd_init(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v8_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v8_0_pageflip_interrupt_init(adev);
+
+	return 0;
+}
+
+static int dce_v8_0_hw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dce_v8_0_hpd_fini(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v8_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v8_0_pageflip_interrupt_fini(adev);
+
+	return 0;
+}
+
+static int dce_v8_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->mode_info.bl_level =
+		amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+
+	return dce_v8_0_hw_fini(handle);
+}
+
+static int dce_v8_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret;
+
+	amdgpu_atombios_encoder_set_backlight_level_to_reg(adev,
+							   adev->mode_info.bl_level);
+
+	ret = dce_v8_0_hw_init(handle);
+
+	/* turn on the BL */
+	if (adev->mode_info.bl_encoder) {
+		u8 bl_level = amdgpu_display_backlight_get_level(adev,
+								  adev->mode_info.bl_encoder);
+		amdgpu_display_backlight_set_level(adev, adev->mode_info.bl_encoder,
+						    bl_level);
+	}
+
+	return ret;
+}
+
+static bool dce_v8_0_is_idle(void *handle)
+{
+	return true;
+}
+
+static int dce_v8_0_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static int dce_v8_0_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0, tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (dce_v8_0_is_display_hung(adev))
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+	return 0;
+}
+
+static void dce_v8_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
+						     int crtc,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 reg_block, lb_interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (crtc) {
+	case 0:
+		reg_block = CRTC0_REGISTER_OFFSET;
+		break;
+	case 1:
+		reg_block = CRTC1_REGISTER_OFFSET;
+		break;
+	case 2:
+		reg_block = CRTC2_REGISTER_OFFSET;
+		break;
+	case 3:
+		reg_block = CRTC3_REGISTER_OFFSET;
+		break;
+	case 4:
+		reg_block = CRTC4_REGISTER_OFFSET;
+		break;
+	case 5:
+		reg_block = CRTC5_REGISTER_OFFSET;
+		break;
+	default:
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + reg_block);
+		lb_interrupt_mask &= ~LB_INTERRUPT_MASK__VBLANK_INTERRUPT_MASK_MASK;
+		WREG32(mmLB_INTERRUPT_MASK + reg_block, lb_interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + reg_block);
+		lb_interrupt_mask |= LB_INTERRUPT_MASK__VBLANK_INTERRUPT_MASK_MASK;
+		WREG32(mmLB_INTERRUPT_MASK + reg_block, lb_interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static void dce_v8_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
+						    int crtc,
+						    enum amdgpu_interrupt_state state)
+{
+	u32 reg_block, lb_interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (crtc) {
+	case 0:
+		reg_block = CRTC0_REGISTER_OFFSET;
+		break;
+	case 1:
+		reg_block = CRTC1_REGISTER_OFFSET;
+		break;
+	case 2:
+		reg_block = CRTC2_REGISTER_OFFSET;
+		break;
+	case 3:
+		reg_block = CRTC3_REGISTER_OFFSET;
+		break;
+	case 4:
+		reg_block = CRTC4_REGISTER_OFFSET;
+		break;
+	case 5:
+		reg_block = CRTC5_REGISTER_OFFSET;
+		break;
+	default:
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + reg_block);
+		lb_interrupt_mask &= ~LB_INTERRUPT_MASK__VLINE_INTERRUPT_MASK_MASK;
+		WREG32(mmLB_INTERRUPT_MASK + reg_block, lb_interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + reg_block);
+		lb_interrupt_mask |= LB_INTERRUPT_MASK__VLINE_INTERRUPT_MASK_MASK;
+		WREG32(mmLB_INTERRUPT_MASK + reg_block, lb_interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static int dce_v8_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	u32 dc_hpd_int_cntl;
+
+	if (type >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hdp %d\n", type);
+		return 0;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		dc_hpd_int_cntl = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type]);
+		dc_hpd_int_cntl &= ~DC_HPD1_INT_CONTROL__DC_HPD1_INT_EN_MASK;
+		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type], dc_hpd_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		dc_hpd_int_cntl = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type]);
+		dc_hpd_int_cntl |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_EN_MASK;
+		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[type], dc_hpd_int_cntl);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v8_0_set_crtc_interrupt_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CRTC_IRQ_VBLANK1:
+		dce_v8_0_set_crtc_vblank_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK2:
+		dce_v8_0_set_crtc_vblank_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK3:
+		dce_v8_0_set_crtc_vblank_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK4:
+		dce_v8_0_set_crtc_vblank_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK5:
+		dce_v8_0_set_crtc_vblank_interrupt_state(adev, 4, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK6:
+		dce_v8_0_set_crtc_vblank_interrupt_state(adev, 5, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE1:
+		dce_v8_0_set_crtc_vline_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE2:
+		dce_v8_0_set_crtc_vline_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE3:
+		dce_v8_0_set_crtc_vline_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE4:
+		dce_v8_0_set_crtc_vline_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE5:
+		dce_v8_0_set_crtc_vline_interrupt_state(adev, 4, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE6:
+		dce_v8_0_set_crtc_vline_interrupt_state(adev, 5, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int dce_v8_0_crtc_irq(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *source,
+			     struct amdgpu_iv_entry *entry)
+{
+	unsigned crtc = entry->src_id - 1;
+	uint32_t disp_int = RREG32(interrupt_status_offsets[crtc].reg);
+	unsigned int irq_type = amdgpu_display_crtc_idx_to_irq_type(adev,
+								    crtc);
+
+	switch (entry->src_data[0]) {
+	case 0: /* vblank */
+		if (disp_int & interrupt_status_offsets[crtc].vblank)
+			WREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc], LB_VBLANK_STATUS__VBLANK_ACK_MASK);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		if (amdgpu_irq_enabled(adev, source, irq_type)) {
+			drm_handle_vblank(adev->ddev, crtc);
+		}
+		DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+		break;
+	case 1: /* vline */
+		if (disp_int & interrupt_status_offsets[crtc].vline)
+			WREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc], LB_VLINE_STATUS__VLINE_ACK_MASK);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		DRM_DEBUG("IH: D%d vline\n", crtc + 1);
+		break;
+	default:
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v8_0_set_pageflip_interrupt_state(struct amdgpu_device *adev,
+						 struct amdgpu_irq_src *src,
+						 unsigned type,
+						 enum amdgpu_interrupt_state state)
+{
+	u32 reg;
+
+	if (type >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", type);
+		return -EINVAL;
+	}
+
+	reg = RREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type]);
+	if (state == AMDGPU_IRQ_STATE_DISABLE)
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg & ~GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+	else
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg | GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+
+	return 0;
+}
+
+static int dce_v8_0_pageflip_irq(struct amdgpu_device *adev,
+				struct amdgpu_irq_src *source,
+				struct amdgpu_iv_entry *entry)
+{
+	unsigned long flags;
+	unsigned crtc_id;
+	struct amdgpu_crtc *amdgpu_crtc;
+	struct amdgpu_flip_work *works;
+
+	crtc_id = (entry->src_id - 8) >> 1;
+	amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	if (crtc_id >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", crtc_id);
+		return -EINVAL;
+	}
+
+	if (RREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id]) &
+	    GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_OCCURRED_MASK)
+		WREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id],
+		       GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_CLEAR_MASK);
+
+	/* IRQ could occur when in initial stage */
+	if (amdgpu_crtc == NULL)
+		return 0;
+
+	spin_lock_irqsave(&adev->ddev->event_lock, flags);
+	works = amdgpu_crtc->pflip_works;
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED){
+		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
+						"AMDGPU_FLIP_SUBMITTED(%d)\n",
+						amdgpu_crtc->pflip_status,
+						AMDGPU_FLIP_SUBMITTED);
+		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+		return 0;
+	}
+
+	/* page flip completed. clean up */
+	amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE;
+	amdgpu_crtc->pflip_works = NULL;
+
+	/* wakeup usersapce */
+	if (works->event)
+		drm_crtc_send_vblank_event(&amdgpu_crtc->base, works->event);
+
+	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+
+	drm_crtc_vblank_put(&amdgpu_crtc->base);
+	schedule_work(&works->unpin_work);
+
+	return 0;
+}
+
+static int dce_v8_0_hpd_irq(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *source,
+			    struct amdgpu_iv_entry *entry)
+{
+	uint32_t disp_int, mask, tmp;
+	unsigned hpd;
+
+	if (entry->src_data[0] >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]);
+		return 0;
+	}
+
+	hpd = entry->src_data[0];
+	disp_int = RREG32(interrupt_status_offsets[hpd].reg);
+	mask = interrupt_status_offsets[hpd].hpd;
+
+	if (disp_int & mask) {
+		tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
+		tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK;
+		WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
+		schedule_work(&adev->hotplug_work);
+		DRM_DEBUG("IH: HPD%d\n", hpd + 1);
+	}
+
+	return 0;
+
+}
+
+static int dce_v8_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int dce_v8_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs dce_v8_0_ip_funcs = {
+	.name = "dce_v8_0",
+	.early_init = dce_v8_0_early_init,
+	.late_init = NULL,
+	.sw_init = dce_v8_0_sw_init,
+	.sw_fini = dce_v8_0_sw_fini,
+	.hw_init = dce_v8_0_hw_init,
+	.hw_fini = dce_v8_0_hw_fini,
+	.suspend = dce_v8_0_suspend,
+	.resume = dce_v8_0_resume,
+	.is_idle = dce_v8_0_is_idle,
+	.wait_for_idle = dce_v8_0_wait_for_idle,
+	.soft_reset = dce_v8_0_soft_reset,
+	.set_clockgating_state = dce_v8_0_set_clockgating_state,
+	.set_powergating_state = dce_v8_0_set_powergating_state,
+};
+
+static void
+dce_v8_0_encoder_mode_set(struct drm_encoder *encoder,
+			  struct drm_display_mode *mode,
+			  struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	amdgpu_encoder->pixel_clock = adjusted_mode->clock;
+
+	/* need to call this here rather than in prepare() since we need some crtc info */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	/* set scaler clears this on some chips */
+	dce_v8_0_set_interleave(encoder->crtc, mode);
+
+	if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) {
+		dce_v8_0_afmt_enable(encoder, true);
+		dce_v8_0_afmt_setmode(encoder, adjusted_mode);
+	}
+}
+
+static void dce_v8_0_encoder_prepare(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+
+	if ((amdgpu_encoder->active_device &
+	     (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) !=
+	     ENCODER_OBJECT_ID_NONE)) {
+		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+		if (dig) {
+			dig->dig_encoder = dce_v8_0_pick_dig_encoder(encoder);
+			if (amdgpu_encoder->active_device & ATOM_DEVICE_DFP_SUPPORT)
+				dig->afmt = adev->mode_info.afmt[dig->dig_encoder];
+		}
+	}
+
+	amdgpu_atombios_scratch_regs_lock(adev, true);
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		/* select the clock/data port if it uses a router */
+		if (amdgpu_connector->router.cd_valid)
+			amdgpu_i2c_router_select_cd_port(amdgpu_connector);
+
+		/* turn eDP panel on for mode set */
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP)
+			amdgpu_atombios_encoder_set_edp_panel_power(connector,
+							     ATOM_TRANSMITTER_ACTION_POWER_ON);
+	}
+
+	/* this is needed for the pll/ss setup to work correctly in some cases */
+	amdgpu_atombios_encoder_set_crtc_source(encoder);
+	/* set up the FMT blocks */
+	dce_v8_0_program_fmt(encoder);
+}
+
+static void dce_v8_0_encoder_commit(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* need to call this here as we need the crtc set up */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_scratch_regs_lock(adev, false);
+}
+
+static void dce_v8_0_encoder_disable(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig;
+
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	if (amdgpu_atombios_encoder_is_digital(encoder)) {
+		if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI)
+			dce_v8_0_afmt_enable(encoder, false);
+		dig = amdgpu_encoder->enc_priv;
+		dig->dig_encoder = -1;
+	}
+	amdgpu_encoder->active_device = 0;
+}
+
+/* these are handled by the primary encoders */
+static void dce_v8_0_ext_prepare(struct drm_encoder *encoder)
+{
+
+}
+
+static void dce_v8_0_ext_commit(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v8_0_ext_mode_set(struct drm_encoder *encoder,
+		      struct drm_display_mode *mode,
+		      struct drm_display_mode *adjusted_mode)
+{
+
+}
+
+static void dce_v8_0_ext_disable(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v8_0_ext_dpms(struct drm_encoder *encoder, int mode)
+{
+
+}
+
+static const struct drm_encoder_helper_funcs dce_v8_0_ext_helper_funcs = {
+	.dpms = dce_v8_0_ext_dpms,
+	.prepare = dce_v8_0_ext_prepare,
+	.mode_set = dce_v8_0_ext_mode_set,
+	.commit = dce_v8_0_ext_commit,
+	.disable = dce_v8_0_ext_disable,
+	/* no detect for TMDS/LVDS yet */
+};
+
+static const struct drm_encoder_helper_funcs dce_v8_0_dig_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v8_0_encoder_prepare,
+	.mode_set = dce_v8_0_encoder_mode_set,
+	.commit = dce_v8_0_encoder_commit,
+	.disable = dce_v8_0_encoder_disable,
+	.detect = amdgpu_atombios_encoder_dig_detect,
+};
+
+static const struct drm_encoder_helper_funcs dce_v8_0_dac_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v8_0_encoder_prepare,
+	.mode_set = dce_v8_0_encoder_mode_set,
+	.commit = dce_v8_0_encoder_commit,
+	.detect = amdgpu_atombios_encoder_dac_detect,
+};
+
+static void dce_v8_0_encoder_destroy(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+		amdgpu_atombios_encoder_fini_backlight(amdgpu_encoder);
+	kfree(amdgpu_encoder->enc_priv);
+	drm_encoder_cleanup(encoder);
+	kfree(amdgpu_encoder);
+}
+
+static const struct drm_encoder_funcs dce_v8_0_encoder_funcs = {
+	.destroy = dce_v8_0_encoder_destroy,
+};
+
+static void dce_v8_0_encoder_add(struct amdgpu_device *adev,
+				 uint32_t encoder_enum,
+				 uint32_t supported_device,
+				 u16 caps)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	/* see if we already added it */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		if (amdgpu_encoder->encoder_enum == encoder_enum) {
+			amdgpu_encoder->devices |= supported_device;
+			return;
+		}
+
+	}
+
+	/* add a new one */
+	amdgpu_encoder = kzalloc(sizeof(struct amdgpu_encoder), GFP_KERNEL);
+	if (!amdgpu_encoder)
+		return;
+
+	encoder = &amdgpu_encoder->base;
+	switch (adev->mode_info.num_crtc) {
+	case 1:
+		encoder->possible_crtcs = 0x1;
+		break;
+	case 2:
+	default:
+		encoder->possible_crtcs = 0x3;
+		break;
+	case 4:
+		encoder->possible_crtcs = 0xf;
+		break;
+	case 6:
+		encoder->possible_crtcs = 0x3f;
+		break;
+	}
+
+	amdgpu_encoder->enc_priv = NULL;
+
+	amdgpu_encoder->encoder_enum = encoder_enum;
+	amdgpu_encoder->encoder_id = (encoder_enum & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+	amdgpu_encoder->devices = supported_device;
+	amdgpu_encoder->rmx_type = RMX_OFF;
+	amdgpu_encoder->underscan_type = UNDERSCAN_OFF;
+	amdgpu_encoder->is_ext_encoder = false;
+	amdgpu_encoder->caps = caps;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+		drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+				 DRM_MODE_ENCODER_DAC, NULL);
+		drm_encoder_helper_add(encoder, &dce_v8_0_dac_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+			amdgpu_encoder->rmx_type = RMX_FULL;
+			drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS, NULL);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_lcd_info(amdgpu_encoder);
+		} else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) {
+			drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC, NULL);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		} else {
+			drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS, NULL);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		}
+		drm_encoder_helper_add(encoder, &dce_v8_0_dig_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_SI170B:
+	case ENCODER_OBJECT_ID_CH7303:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOA:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOB:
+	case ENCODER_OBJECT_ID_TITFP513:
+	case ENCODER_OBJECT_ID_VT1623:
+	case ENCODER_OBJECT_ID_HDMI_SI1930:
+	case ENCODER_OBJECT_ID_TRAVIS:
+	case ENCODER_OBJECT_ID_NUTMEG:
+		/* these are handled by the primary encoders */
+		amdgpu_encoder->is_ext_encoder = true;
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS, NULL);
+		else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC, NULL);
+		else
+			drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS, NULL);
+		drm_encoder_helper_add(encoder, &dce_v8_0_ext_helper_funcs);
+		break;
+	}
+}
+
+static const struct amdgpu_display_funcs dce_v8_0_display_funcs = {
+	.bandwidth_update = &dce_v8_0_bandwidth_update,
+	.vblank_get_counter = &dce_v8_0_vblank_get_counter,
+	.backlight_set_level = &amdgpu_atombios_encoder_set_backlight_level,
+	.backlight_get_level = &amdgpu_atombios_encoder_get_backlight_level,
+	.hpd_sense = &dce_v8_0_hpd_sense,
+	.hpd_set_polarity = &dce_v8_0_hpd_set_polarity,
+	.hpd_get_gpio_reg = &dce_v8_0_hpd_get_gpio_reg,
+	.page_flip = &dce_v8_0_page_flip,
+	.page_flip_get_scanoutpos = &dce_v8_0_crtc_get_scanoutpos,
+	.add_encoder = &dce_v8_0_encoder_add,
+	.add_connector = &amdgpu_connector_add,
+};
+
+static void dce_v8_0_set_display_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.funcs == NULL)
+		adev->mode_info.funcs = &dce_v8_0_display_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs dce_v8_0_crtc_irq_funcs = {
+	.set = dce_v8_0_set_crtc_interrupt_state,
+	.process = dce_v8_0_crtc_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v8_0_pageflip_irq_funcs = {
+	.set = dce_v8_0_set_pageflip_interrupt_state,
+	.process = dce_v8_0_pageflip_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v8_0_hpd_irq_funcs = {
+	.set = dce_v8_0_set_hpd_interrupt_state,
+	.process = dce_v8_0_hpd_irq,
+};
+
+static void dce_v8_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.num_crtc > 0)
+		adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_VLINE1 + adev->mode_info.num_crtc;
+	else
+		adev->crtc_irq.num_types = 0;
+	adev->crtc_irq.funcs = &dce_v8_0_crtc_irq_funcs;
+
+	adev->pageflip_irq.num_types = adev->mode_info.num_crtc;
+	adev->pageflip_irq.funcs = &dce_v8_0_pageflip_irq_funcs;
+
+	adev->hpd_irq.num_types = adev->mode_info.num_hpd;
+	adev->hpd_irq.funcs = &dce_v8_0_hpd_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version dce_v8_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 8,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &dce_v8_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version dce_v8_1_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 8,
+	.minor = 1,
+	.rev = 0,
+	.funcs = &dce_v8_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version dce_v8_2_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 8,
+	.minor = 2,
+	.rev = 0,
+	.funcs = &dce_v8_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version dce_v8_3_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 8,
+	.minor = 3,
+	.rev = 0,
+	.funcs = &dce_v8_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version dce_v8_5_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 8,
+	.minor = 5,
+	.rev = 0,
+	.funcs = &dce_v8_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/dce_v8_0.h b/sys/dev/pci/drm/amd/amdgpu/dce_v8_0.h
new file mode 100644
index 00000000000..13b802dd946
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/dce_v8_0.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DCE_V8_0_H__
+#define __DCE_V8_0_H__
+
+extern const struct amdgpu_ip_block_version dce_v8_0_ip_block;
+extern const struct amdgpu_ip_block_version dce_v8_1_ip_block;
+extern const struct amdgpu_ip_block_version dce_v8_2_ip_block;
+extern const struct amdgpu_ip_block_version dce_v8_3_ip_block;
+extern const struct amdgpu_ip_block_version dce_v8_5_ip_block;
+
+void dce_v8_0_disable_dce(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/dce_virtual.c b/sys/dev/pci/drm/amd/amdgpu/dce_virtual.c
new file mode 100644
index 00000000000..2514769d3e3
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/dce_virtual.c
@@ -0,0 +1,804 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_i2c.h"
+#include "atom.h"
+#include "amdgpu_pll.h"
+#include "amdgpu_connectors.h"
+#ifdef CONFIG_DRM_AMDGPU_SI
+#include "dce_v6_0.h"
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+#include "dce_v8_0.h"
+#endif
+#include "dce_v10_0.h"
+#include "dce_v11_0.h"
+#include "dce_virtual.h"
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+#define DCE_VIRTUAL_VBLANK_PERIOD 16666666
+
+
+static void dce_virtual_set_display_funcs(struct amdgpu_device *adev);
+static void dce_virtual_set_irq_funcs(struct amdgpu_device *adev);
+static int dce_virtual_connector_encoder_init(struct amdgpu_device *adev,
+					      int index);
+static void dce_virtual_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
+							int crtc,
+							enum amdgpu_interrupt_state state);
+
+static u32 dce_virtual_vblank_get_counter(struct amdgpu_device *adev, int crtc)
+{
+	return 0;
+}
+
+static void dce_virtual_page_flip(struct amdgpu_device *adev,
+			      int crtc_id, u64 crtc_base, bool async)
+{
+	return;
+}
+
+static int dce_virtual_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
+					u32 *vbl, u32 *position)
+{
+	*vbl = 0;
+	*position = 0;
+
+	return -EINVAL;
+}
+
+static bool dce_virtual_hpd_sense(struct amdgpu_device *adev,
+			       enum amdgpu_hpd_id hpd)
+{
+	return true;
+}
+
+static void dce_virtual_hpd_set_polarity(struct amdgpu_device *adev,
+				      enum amdgpu_hpd_id hpd)
+{
+	return;
+}
+
+static u32 dce_virtual_hpd_get_gpio_reg(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
+/**
+ * dce_virtual_bandwidth_update - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Calculate and program the display watermarks and line
+ * buffer allocation (CIK).
+ */
+static void dce_virtual_bandwidth_update(struct amdgpu_device *adev)
+{
+	return;
+}
+
+static int dce_virtual_crtc_gamma_set(struct drm_crtc *crtc, u16 *red,
+				      u16 *green, u16 *blue, uint32_t size,
+				      struct drm_modeset_acquire_ctx *ctx)
+{
+	return 0;
+}
+
+static void dce_virtual_crtc_destroy(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	drm_crtc_cleanup(crtc);
+	kfree(amdgpu_crtc);
+}
+
+static const struct drm_crtc_funcs dce_virtual_crtc_funcs = {
+	.cursor_set2 = NULL,
+	.cursor_move = NULL,
+	.gamma_set = dce_virtual_crtc_gamma_set,
+	.set_config = amdgpu_display_crtc_set_config,
+	.destroy = dce_virtual_crtc_destroy,
+	.page_flip_target = amdgpu_display_crtc_page_flip_target,
+};
+
+static void dce_virtual_crtc_dpms(struct drm_crtc *crtc, int mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	unsigned type;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	switch (mode) {
+	case DRM_MODE_DPMS_ON:
+		amdgpu_crtc->enabled = true;
+		/* Make sure VBLANK interrupts are still enabled */
+		type = amdgpu_display_crtc_idx_to_irq_type(adev,
+						amdgpu_crtc->crtc_id);
+		amdgpu_irq_update(adev, &adev->crtc_irq, type);
+		drm_crtc_vblank_on(crtc);
+		break;
+	case DRM_MODE_DPMS_STANDBY:
+	case DRM_MODE_DPMS_SUSPEND:
+	case DRM_MODE_DPMS_OFF:
+		drm_crtc_vblank_off(crtc);
+		amdgpu_crtc->enabled = false;
+		break;
+	}
+}
+
+
+static void dce_virtual_crtc_prepare(struct drm_crtc *crtc)
+{
+	dce_virtual_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+}
+
+static void dce_virtual_crtc_commit(struct drm_crtc *crtc)
+{
+	dce_virtual_crtc_dpms(crtc, DRM_MODE_DPMS_ON);
+}
+
+static void dce_virtual_crtc_disable(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	dce_virtual_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+	if (crtc->primary->fb) {
+		int r;
+		struct amdgpu_bo *abo;
+
+		abo = gem_to_amdgpu_bo(crtc->primary->fb->obj[0]);
+		r = amdgpu_bo_reserve(abo, true);
+		if (unlikely(r))
+			DRM_ERROR("failed to reserve abo before unpin\n");
+		else {
+			amdgpu_bo_unpin(abo);
+			amdgpu_bo_unreserve(abo);
+		}
+	}
+
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+}
+
+static int dce_virtual_crtc_mode_set(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode,
+				  int x, int y, struct drm_framebuffer *old_fb)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	/* update the hw version fpr dpm */
+	amdgpu_crtc->hw_mode = *adjusted_mode;
+
+	return 0;
+}
+
+static bool dce_virtual_crtc_mode_fixup(struct drm_crtc *crtc,
+				     const struct drm_display_mode *mode,
+				     struct drm_display_mode *adjusted_mode)
+{
+	return true;
+}
+
+
+static int dce_virtual_crtc_set_base(struct drm_crtc *crtc, int x, int y,
+				  struct drm_framebuffer *old_fb)
+{
+	return 0;
+}
+
+static int dce_virtual_crtc_set_base_atomic(struct drm_crtc *crtc,
+					 struct drm_framebuffer *fb,
+					 int x, int y, enum mode_set_atomic state)
+{
+	return 0;
+}
+
+static const struct drm_crtc_helper_funcs dce_virtual_crtc_helper_funcs = {
+	.dpms = dce_virtual_crtc_dpms,
+	.mode_fixup = dce_virtual_crtc_mode_fixup,
+	.mode_set = dce_virtual_crtc_mode_set,
+	.mode_set_base = dce_virtual_crtc_set_base,
+	.mode_set_base_atomic = dce_virtual_crtc_set_base_atomic,
+	.prepare = dce_virtual_crtc_prepare,
+	.commit = dce_virtual_crtc_commit,
+	.disable = dce_virtual_crtc_disable,
+};
+
+static int dce_virtual_crtc_init(struct amdgpu_device *adev, int index)
+{
+	struct amdgpu_crtc *amdgpu_crtc;
+
+	amdgpu_crtc = kzalloc(sizeof(struct amdgpu_crtc) +
+			      (AMDGPUFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
+	if (amdgpu_crtc == NULL)
+		return -ENOMEM;
+
+	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_virtual_crtc_funcs);
+
+	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
+	amdgpu_crtc->crtc_id = index;
+	adev->mode_info.crtcs[index] = amdgpu_crtc;
+
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+	amdgpu_crtc->vsync_timer_enabled = AMDGPU_IRQ_STATE_DISABLE;
+	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_virtual_crtc_helper_funcs);
+
+	return 0;
+}
+
+static int dce_virtual_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dce_virtual_set_display_funcs(adev);
+	dce_virtual_set_irq_funcs(adev);
+
+	adev->mode_info.num_hpd = 1;
+	adev->mode_info.num_dig = 1;
+	return 0;
+}
+
+static struct drm_encoder *
+dce_virtual_encoder(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder;
+	int i;
+
+	drm_connector_for_each_possible_encoder(connector, encoder, i) {
+		if (encoder->encoder_type == DRM_MODE_ENCODER_VIRTUAL)
+			return encoder;
+	}
+
+	/* pick the first one */
+	drm_connector_for_each_possible_encoder(connector, encoder, i)
+		return encoder;
+
+	return NULL;
+}
+
+static int dce_virtual_get_modes(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+	struct drm_display_mode *mode = NULL;
+	unsigned i;
+	static const struct mode_size {
+		int w;
+		int h;
+	} common_modes[17] = {
+		{ 640,  480},
+		{ 720,  480},
+		{ 800,  600},
+		{ 848,  480},
+		{1024,  768},
+		{1152,  768},
+		{1280,  720},
+		{1280,  800},
+		{1280,  854},
+		{1280,  960},
+		{1280, 1024},
+		{1440,  900},
+		{1400, 1050},
+		{1680, 1050},
+		{1600, 1200},
+		{1920, 1080},
+		{1920, 1200}
+	};
+
+	for (i = 0; i < 17; i++) {
+		mode = drm_cvt_mode(dev, common_modes[i].w, common_modes[i].h, 60, false, false, false);
+		drm_mode_probed_add(connector, mode);
+	}
+
+	return 0;
+}
+
+static enum drm_mode_status dce_virtual_mode_valid(struct drm_connector *connector,
+				  struct drm_display_mode *mode)
+{
+	return MODE_OK;
+}
+
+static int
+dce_virtual_dpms(struct drm_connector *connector, int mode)
+{
+	return 0;
+}
+
+static int
+dce_virtual_set_property(struct drm_connector *connector,
+			 struct drm_property *property,
+			 uint64_t val)
+{
+	return 0;
+}
+
+static void dce_virtual_destroy(struct drm_connector *connector)
+{
+	drm_connector_unregister(connector);
+	drm_connector_cleanup(connector);
+	kfree(connector);
+}
+
+static void dce_virtual_force(struct drm_connector *connector)
+{
+	return;
+}
+
+static const struct drm_connector_helper_funcs dce_virtual_connector_helper_funcs = {
+	.get_modes = dce_virtual_get_modes,
+	.mode_valid = dce_virtual_mode_valid,
+	.best_encoder = dce_virtual_encoder,
+};
+
+static const struct drm_connector_funcs dce_virtual_connector_funcs = {
+	.dpms = dce_virtual_dpms,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.set_property = dce_virtual_set_property,
+	.destroy = dce_virtual_destroy,
+	.force = dce_virtual_force,
+};
+
+static int dce_virtual_sw_init(void *handle)
+{
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SMU_DISP_TIMER2_TRIGGER, &adev->crtc_irq);
+	if (r)
+		return r;
+
+	adev->ddev->max_vblank_count = 0;
+
+	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	adev->ddev->mode_config.preferred_depth = 24;
+	adev->ddev->mode_config.prefer_shadow = 1;
+
+	adev->ddev->mode_config.fb_base = adev->gmc.aper_base;
+
+	r = amdgpu_display_modeset_create_props(adev);
+	if (r)
+		return r;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	/* allocate crtcs, encoders, connectors */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = dce_virtual_crtc_init(adev, i);
+		if (r)
+			return r;
+		r = dce_virtual_connector_encoder_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	drm_kms_helper_poll_init(adev->ddev);
+
+	adev->mode_info.mode_config_initialized = true;
+	return 0;
+}
+
+static int dce_virtual_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	kfree(adev->mode_info.bios_hardcoded_edid);
+
+	drm_kms_helper_poll_fini(adev->ddev);
+
+	drm_mode_config_cleanup(adev->ddev);
+	/* clear crtcs pointer to avoid dce irq finish routine access freed data */
+	memset(adev->mode_info.crtcs, 0, sizeof(adev->mode_info.crtcs[0]) * AMDGPU_MAX_CRTCS);
+	adev->mode_info.mode_config_initialized = false;
+	return 0;
+}
+
+static int dce_virtual_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_VERDE:
+	case CHIP_OLAND:
+		dce_v6_0_disable_dce(adev);
+		break;
+#endif
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		dce_v8_0_disable_dce(adev);
+		break;
+#endif
+	case CHIP_FIJI:
+	case CHIP_TONGA:
+		dce_v10_0_disable_dce(adev);
+		break;
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_VEGAM:
+		dce_v11_0_disable_dce(adev);
+		break;
+	case CHIP_TOPAZ:
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_HAINAN:
+#endif
+		/* no DCE */
+		break;
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+		break;
+	default:
+		DRM_ERROR("Virtual display unsupported ASIC type: 0x%X\n", adev->asic_type);
+	}
+	return 0;
+}
+
+static int dce_virtual_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i = 0;
+
+	for (i = 0; i<adev->mode_info.num_crtc; i++)
+		if (adev->mode_info.crtcs[i])
+			dce_virtual_set_crtc_vblank_interrupt_state(adev, i, AMDGPU_IRQ_STATE_DISABLE);
+
+	return 0;
+}
+
+static int dce_virtual_suspend(void *handle)
+{
+	return dce_virtual_hw_fini(handle);
+}
+
+static int dce_virtual_resume(void *handle)
+{
+	return dce_virtual_hw_init(handle);
+}
+
+static bool dce_virtual_is_idle(void *handle)
+{
+	return true;
+}
+
+static int dce_virtual_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static int dce_virtual_soft_reset(void *handle)
+{
+	return 0;
+}
+
+static int dce_virtual_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int dce_virtual_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs dce_virtual_ip_funcs = {
+	.name = "dce_virtual",
+	.early_init = dce_virtual_early_init,
+	.late_init = NULL,
+	.sw_init = dce_virtual_sw_init,
+	.sw_fini = dce_virtual_sw_fini,
+	.hw_init = dce_virtual_hw_init,
+	.hw_fini = dce_virtual_hw_fini,
+	.suspend = dce_virtual_suspend,
+	.resume = dce_virtual_resume,
+	.is_idle = dce_virtual_is_idle,
+	.wait_for_idle = dce_virtual_wait_for_idle,
+	.soft_reset = dce_virtual_soft_reset,
+	.set_clockgating_state = dce_virtual_set_clockgating_state,
+	.set_powergating_state = dce_virtual_set_powergating_state,
+};
+
+/* these are handled by the primary encoders */
+static void dce_virtual_encoder_prepare(struct drm_encoder *encoder)
+{
+	return;
+}
+
+static void dce_virtual_encoder_commit(struct drm_encoder *encoder)
+{
+	return;
+}
+
+static void
+dce_virtual_encoder_mode_set(struct drm_encoder *encoder,
+			     struct drm_display_mode *mode,
+			     struct drm_display_mode *adjusted_mode)
+{
+	return;
+}
+
+static void dce_virtual_encoder_disable(struct drm_encoder *encoder)
+{
+	return;
+}
+
+static void
+dce_virtual_encoder_dpms(struct drm_encoder *encoder, int mode)
+{
+	return;
+}
+
+static bool dce_virtual_encoder_mode_fixup(struct drm_encoder *encoder,
+				    const struct drm_display_mode *mode,
+				    struct drm_display_mode *adjusted_mode)
+{
+	return true;
+}
+
+static const struct drm_encoder_helper_funcs dce_virtual_encoder_helper_funcs = {
+	.dpms = dce_virtual_encoder_dpms,
+	.mode_fixup = dce_virtual_encoder_mode_fixup,
+	.prepare = dce_virtual_encoder_prepare,
+	.mode_set = dce_virtual_encoder_mode_set,
+	.commit = dce_virtual_encoder_commit,
+	.disable = dce_virtual_encoder_disable,
+};
+
+static void dce_virtual_encoder_destroy(struct drm_encoder *encoder)
+{
+	drm_encoder_cleanup(encoder);
+	kfree(encoder);
+}
+
+static const struct drm_encoder_funcs dce_virtual_encoder_funcs = {
+	.destroy = dce_virtual_encoder_destroy,
+};
+
+static int dce_virtual_connector_encoder_init(struct amdgpu_device *adev,
+					      int index)
+{
+	struct drm_encoder *encoder;
+	struct drm_connector *connector;
+
+	/* add a new encoder */
+	encoder = kzalloc(sizeof(struct drm_encoder), GFP_KERNEL);
+	if (!encoder)
+		return -ENOMEM;
+	encoder->possible_crtcs = 1 << index;
+	drm_encoder_init(adev->ddev, encoder, &dce_virtual_encoder_funcs,
+			 DRM_MODE_ENCODER_VIRTUAL, NULL);
+	drm_encoder_helper_add(encoder, &dce_virtual_encoder_helper_funcs);
+
+	connector = kzalloc(sizeof(struct drm_connector), GFP_KERNEL);
+	if (!connector) {
+		kfree(encoder);
+		return -ENOMEM;
+	}
+
+	/* add a new connector */
+	drm_connector_init(adev->ddev, connector, &dce_virtual_connector_funcs,
+			   DRM_MODE_CONNECTOR_VIRTUAL);
+	drm_connector_helper_add(connector, &dce_virtual_connector_helper_funcs);
+	connector->display_info.subpixel_order = SubPixelHorizontalRGB;
+	connector->interlace_allowed = false;
+	connector->doublescan_allowed = false;
+	drm_connector_register(connector);
+
+	/* link them */
+	drm_connector_attach_encoder(connector, encoder);
+
+	return 0;
+}
+
+static const struct amdgpu_display_funcs dce_virtual_display_funcs = {
+	.bandwidth_update = &dce_virtual_bandwidth_update,
+	.vblank_get_counter = &dce_virtual_vblank_get_counter,
+	.backlight_set_level = NULL,
+	.backlight_get_level = NULL,
+	.hpd_sense = &dce_virtual_hpd_sense,
+	.hpd_set_polarity = &dce_virtual_hpd_set_polarity,
+	.hpd_get_gpio_reg = &dce_virtual_hpd_get_gpio_reg,
+	.page_flip = &dce_virtual_page_flip,
+	.page_flip_get_scanoutpos = &dce_virtual_crtc_get_scanoutpos,
+	.add_encoder = NULL,
+	.add_connector = NULL,
+};
+
+static void dce_virtual_set_display_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.funcs == NULL)
+		adev->mode_info.funcs = &dce_virtual_display_funcs;
+}
+
+static int dce_virtual_pageflip(struct amdgpu_device *adev,
+				unsigned crtc_id)
+{
+	unsigned long flags;
+	struct amdgpu_crtc *amdgpu_crtc;
+	struct amdgpu_flip_work *works;
+
+	amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	if (crtc_id >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", crtc_id);
+		return -EINVAL;
+	}
+
+	/* IRQ could occur when in initial stage */
+	if (amdgpu_crtc == NULL)
+		return 0;
+
+	spin_lock_irqsave(&adev->ddev->event_lock, flags);
+	works = amdgpu_crtc->pflip_works;
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED) {
+		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
+			"AMDGPU_FLIP_SUBMITTED(%d)\n",
+			amdgpu_crtc->pflip_status,
+			AMDGPU_FLIP_SUBMITTED);
+		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+		return 0;
+	}
+
+	/* page flip completed. clean up */
+	amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE;
+	amdgpu_crtc->pflip_works = NULL;
+
+	/* wakeup usersapce */
+	if (works->event)
+		drm_crtc_send_vblank_event(&amdgpu_crtc->base, works->event);
+
+	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+
+	drm_crtc_vblank_put(&amdgpu_crtc->base);
+	schedule_work(&works->unpin_work);
+
+	return 0;
+}
+
+#ifdef __linux__
+
+static enum hrtimer_restart dce_virtual_vblank_timer_handle(struct hrtimer *vblank_timer)
+{
+	struct amdgpu_crtc *amdgpu_crtc = container_of(vblank_timer,
+						       struct amdgpu_crtc, vblank_timer);
+	struct drm_device *ddev = amdgpu_crtc->base.dev;
+	struct amdgpu_device *adev = ddev->dev_private;
+
+	drm_handle_vblank(ddev, amdgpu_crtc->crtc_id);
+	dce_virtual_pageflip(adev, amdgpu_crtc->crtc_id);
+	hrtimer_start(vblank_timer, DCE_VIRTUAL_VBLANK_PERIOD,
+		      HRTIMER_MODE_REL);
+
+	return HRTIMER_NORESTART;
+}
+
+#else
+
+static void dce_virtual_vblank_timer_handle(void *arg)
+{
+	struct timeout *vblank_timer = arg;
+	struct amdgpu_crtc *amdgpu_crtc = container_of(vblank_timer,
+						       struct amdgpu_crtc, vblank_timer);
+	struct drm_device *ddev = amdgpu_crtc->base.dev;
+	struct amdgpu_device *adev = ddev->dev_private;
+
+	drm_handle_vblank(ddev, amdgpu_crtc->crtc_id);
+	dce_virtual_pageflip(adev, amdgpu_crtc->crtc_id);
+	timeout_add_nsec(vblank_timer, DCE_VIRTUAL_VBLANK_PERIOD);
+
+}
+
+#endif
+
+static void dce_virtual_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
+							int crtc,
+							enum amdgpu_interrupt_state state)
+{
+	if (crtc >= adev->mode_info.num_crtc || !adev->mode_info.crtcs[crtc]) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	if (state && !adev->mode_info.crtcs[crtc]->vsync_timer_enabled) {
+		DRM_DEBUG("Enable software vsync timer\n");
+#ifdef __linux__
+		hrtimer_init(&adev->mode_info.crtcs[crtc]->vblank_timer,
+			     CLOCK_MONOTONIC, HRTIMER_MODE_REL);
+		hrtimer_set_expires(&adev->mode_info.crtcs[crtc]->vblank_timer,
+				    DCE_VIRTUAL_VBLANK_PERIOD);
+		adev->mode_info.crtcs[crtc]->vblank_timer.function =
+			dce_virtual_vblank_timer_handle;
+		hrtimer_start(&adev->mode_info.crtcs[crtc]->vblank_timer,
+			      DCE_VIRTUAL_VBLANK_PERIOD, HRTIMER_MODE_REL);
+#else
+	timeout_set(&adev->mode_info.crtcs[crtc]->vblank_timer,
+	    dce_virtual_vblank_timer_handle,
+	    &adev->mode_info.crtcs[crtc]->vblank_timer);
+	timeout_add_nsec(&adev->mode_info.crtcs[crtc]->vblank_timer,
+	    DCE_VIRTUAL_VBLANK_PERIOD);
+#endif
+	} else if (!state && adev->mode_info.crtcs[crtc]->vsync_timer_enabled) {
+		DRM_DEBUG("Disable software vsync timer\n");
+		hrtimer_cancel(&adev->mode_info.crtcs[crtc]->vblank_timer);
+	}
+
+	adev->mode_info.crtcs[crtc]->vsync_timer_enabled = state;
+	DRM_DEBUG("[FM]set crtc %d vblank interrupt state %d\n", crtc, state);
+}
+
+
+static int dce_virtual_set_crtc_irq_state(struct amdgpu_device *adev,
+					  struct amdgpu_irq_src *source,
+					  unsigned type,
+					  enum amdgpu_interrupt_state state)
+{
+	if (type > AMDGPU_CRTC_IRQ_VBLANK6)
+		return -EINVAL;
+
+	dce_virtual_set_crtc_vblank_interrupt_state(adev, type, state);
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs dce_virtual_crtc_irq_funcs = {
+	.set = dce_virtual_set_crtc_irq_state,
+	.process = NULL,
+};
+
+static void dce_virtual_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_VBLANK6 + 1;
+	adev->crtc_irq.funcs = &dce_virtual_crtc_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version dce_virtual_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_DCE,
+	.major = 1,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &dce_virtual_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/dce_virtual.h b/sys/dev/pci/drm/amd/amdgpu/dce_virtual.h
new file mode 100644
index 00000000000..ed422012c8c
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/dce_virtual.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DCE_VIRTUAL_H__
+#define __DCE_VIRTUAL_H__
+
+extern const struct amdgpu_ip_block_version dce_virtual_ip_block;
+
+#endif
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/df_v1_7.c b/sys/dev/pci/drm/amd/amdgpu/df_v1_7.c
new file mode 100644
index 00000000000..9935371db7c
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/df_v1_7.c
@@ -0,0 +1,120 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "df_v1_7.h"
+
+#include "df/df_1_7_default.h"
+#include "df/df_1_7_offset.h"
+#include "df/df_1_7_sh_mask.h"
+
+static u32 df_v1_7_channel_number[] = {1, 2, 0, 4, 0, 8, 0, 16, 2};
+
+static void df_v1_7_init (struct amdgpu_device *adev)
+{
+}
+
+static void df_v1_7_enable_broadcast_mode(struct amdgpu_device *adev,
+                                          bool enable)
+{
+	u32 tmp;
+
+	if (enable) {
+		tmp = RREG32_SOC15(DF, 0, mmFabricConfigAccessControl);
+		tmp &= ~FabricConfigAccessControl__CfgRegInstAccEn_MASK;
+		WREG32_SOC15(DF, 0, mmFabricConfigAccessControl, tmp);
+	} else
+		WREG32_SOC15(DF, 0, mmFabricConfigAccessControl,
+			     mmFabricConfigAccessControl_DEFAULT);
+}
+
+static u32 df_v1_7_get_fb_channel_number(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(DF, 0, mmDF_CS_AON0_DramBaseAddress0);
+	tmp &= DF_CS_AON0_DramBaseAddress0__IntLvNumChan_MASK;
+	tmp >>= DF_CS_AON0_DramBaseAddress0__IntLvNumChan__SHIFT;
+
+	return tmp;
+}
+
+static u32 df_v1_7_get_hbm_channel_number(struct amdgpu_device *adev)
+{
+	int fb_channel_number;
+
+	fb_channel_number = adev->df_funcs->get_fb_channel_number(adev);
+
+	return df_v1_7_channel_number[fb_channel_number];
+}
+
+static void df_v1_7_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						     bool enable)
+{
+	u32 tmp;
+
+	/* Put DF on broadcast mode */
+	adev->df_funcs->enable_broadcast_mode(adev, true);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_DF_MGCG)) {
+		tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater);
+		tmp &= ~DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK;
+		tmp |= DF_V1_7_MGCG_ENABLE_15_CYCLE_DELAY;
+		WREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater, tmp);
+	} else {
+		tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater);
+		tmp &= ~DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK;
+		tmp |= DF_V1_7_MGCG_DISABLE;
+		WREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater, tmp);
+	}
+
+	/* Exit boradcast mode */
+	adev->df_funcs->enable_broadcast_mode(adev, false);
+}
+
+static void df_v1_7_get_clockgating_state(struct amdgpu_device *adev,
+					  u32 *flags)
+{
+	u32 tmp;
+
+	/* AMD_CG_SUPPORT_DF_MGCG */
+	tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater);
+	if (tmp & DF_V1_7_MGCG_ENABLE_15_CYCLE_DELAY)
+		*flags |= AMD_CG_SUPPORT_DF_MGCG;
+}
+
+static void df_v1_7_enable_ecc_force_par_wr_rmw(struct amdgpu_device *adev,
+						bool enable)
+{
+	WREG32_FIELD15(DF, 0, DF_CS_AON0_CoherentSlaveModeCtrlA0,
+		       ForceParWrRMW, enable);
+}
+
+const struct amdgpu_df_funcs df_v1_7_funcs = {
+	.init = df_v1_7_init,
+	.enable_broadcast_mode = df_v1_7_enable_broadcast_mode,
+	.get_fb_channel_number = df_v1_7_get_fb_channel_number,
+	.get_hbm_channel_number = df_v1_7_get_hbm_channel_number,
+	.update_medium_grain_clock_gating = df_v1_7_update_medium_grain_clock_gating,
+	.get_clockgating_state = df_v1_7_get_clockgating_state,
+	.enable_ecc_force_par_wr_rmw = df_v1_7_enable_ecc_force_par_wr_rmw,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/df_v1_7.h b/sys/dev/pci/drm/amd/amdgpu/df_v1_7.h
new file mode 100644
index 00000000000..74621104c48
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/df_v1_7.h
@@ -0,0 +1,40 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DF_V1_7_H__
+#define __DF_V1_7_H__
+
+#include "soc15_common.h"
+enum DF_V1_7_MGCG
+{
+	DF_V1_7_MGCG_DISABLE = 0,
+	DF_V1_7_MGCG_ENABLE_00_CYCLE_DELAY =1,
+	DF_V1_7_MGCG_ENABLE_01_CYCLE_DELAY =2,
+	DF_V1_7_MGCG_ENABLE_15_CYCLE_DELAY =13,
+	DF_V1_7_MGCG_ENABLE_31_CYCLE_DELAY =14,
+	DF_V1_7_MGCG_ENABLE_63_CYCLE_DELAY =15
+};
+
+extern const struct amdgpu_df_funcs df_v1_7_funcs;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/df_v3_6.c b/sys/dev/pci/drm/amd/amdgpu/df_v3_6.c
new file mode 100644
index 00000000000..d5ebe566809
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/df_v3_6.c
@@ -0,0 +1,116 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "df_v3_6.h"
+
+#include "df/df_3_6_default.h"
+#include "df/df_3_6_offset.h"
+#include "df/df_3_6_sh_mask.h"
+
+static u32 df_v3_6_channel_number[] = {1, 2, 0, 4, 0, 8, 0,
+				       16, 32, 0, 0, 0, 2, 4, 8};
+
+static void df_v3_6_init(struct amdgpu_device *adev)
+{
+}
+
+static void df_v3_6_enable_broadcast_mode(struct amdgpu_device *adev,
+					  bool enable)
+{
+	u32 tmp;
+
+	if (enable) {
+		tmp = RREG32_SOC15(DF, 0, mmFabricConfigAccessControl);
+		tmp &= ~FabricConfigAccessControl__CfgRegInstAccEn_MASK;
+		WREG32_SOC15(DF, 0, mmFabricConfigAccessControl, tmp);
+	} else
+		WREG32_SOC15(DF, 0, mmFabricConfigAccessControl,
+			     mmFabricConfigAccessControl_DEFAULT);
+}
+
+static u32 df_v3_6_get_fb_channel_number(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32_SOC15(DF, 0, mmDF_CS_UMC_AON0_DramBaseAddress0);
+	tmp &= DF_CS_UMC_AON0_DramBaseAddress0__IntLvNumChan_MASK;
+	tmp >>= DF_CS_UMC_AON0_DramBaseAddress0__IntLvNumChan__SHIFT;
+
+	return tmp;
+}
+
+static u32 df_v3_6_get_hbm_channel_number(struct amdgpu_device *adev)
+{
+	int fb_channel_number;
+
+	fb_channel_number = adev->df_funcs->get_fb_channel_number(adev);
+	if (fb_channel_number >= ARRAY_SIZE(df_v3_6_channel_number))
+		fb_channel_number = 0;
+
+	return df_v3_6_channel_number[fb_channel_number];
+}
+
+static void df_v3_6_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						     bool enable)
+{
+	u32 tmp;
+
+	/* Put DF on broadcast mode */
+	adev->df_funcs->enable_broadcast_mode(adev, true);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_DF_MGCG)) {
+		tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater);
+		tmp &= ~DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK;
+		tmp |= DF_V3_6_MGCG_ENABLE_15_CYCLE_DELAY;
+		WREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater, tmp);
+	} else {
+		tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater);
+		tmp &= ~DF_PIE_AON0_DfGlobalClkGater__MGCGMode_MASK;
+		tmp |= DF_V3_6_MGCG_DISABLE;
+		WREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater, tmp);
+	}
+
+	/* Exit broadcast mode */
+	adev->df_funcs->enable_broadcast_mode(adev, false);
+}
+
+static void df_v3_6_get_clockgating_state(struct amdgpu_device *adev,
+					  u32 *flags)
+{
+	u32 tmp;
+
+	/* AMD_CG_SUPPORT_DF_MGCG */
+	tmp = RREG32_SOC15(DF, 0, mmDF_PIE_AON0_DfGlobalClkGater);
+	if (tmp & DF_V3_6_MGCG_ENABLE_15_CYCLE_DELAY)
+		*flags |= AMD_CG_SUPPORT_DF_MGCG;
+}
+
+const struct amdgpu_df_funcs df_v3_6_funcs = {
+	.init = df_v3_6_init,
+	.enable_broadcast_mode = df_v3_6_enable_broadcast_mode,
+	.get_fb_channel_number = df_v3_6_get_fb_channel_number,
+	.get_hbm_channel_number = df_v3_6_get_hbm_channel_number,
+	.update_medium_grain_clock_gating =
+			df_v3_6_update_medium_grain_clock_gating,
+	.get_clockgating_state = df_v3_6_get_clockgating_state,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/df_v3_6.h b/sys/dev/pci/drm/amd/amdgpu/df_v3_6.h
new file mode 100644
index 00000000000..e79c58e5efc
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/df_v3_6.h
@@ -0,0 +1,40 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DF_V3_6_H__
+#define __DF_V3_6_H__
+
+#include "soc15_common.h"
+
+enum DF_V3_6_MGCG {
+	DF_V3_6_MGCG_DISABLE = 0,
+	DF_V3_6_MGCG_ENABLE_00_CYCLE_DELAY = 1,
+	DF_V3_6_MGCG_ENABLE_01_CYCLE_DELAY = 2,
+	DF_V3_6_MGCG_ENABLE_15_CYCLE_DELAY = 13,
+	DF_V3_6_MGCG_ENABLE_31_CYCLE_DELAY = 14,
+	DF_V3_6_MGCG_ENABLE_63_CYCLE_DELAY = 15
+};
+
+extern const struct amdgpu_df_funcs df_v3_6_funcs;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/emu_soc.c b/sys/dev/pci/drm/amd/amdgpu/emu_soc.c
new file mode 100644
index 00000000000..d72c25c1b98
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/emu_soc.c
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "soc15.h"
+
+#include "soc15_common.h"
+#include "soc15_hw_ip.h"
+
+int emu_soc_asic_init(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/gfx_v6_0.c b/sys/dev/pci/drm/amd/amdgpu/gfx_v6_0.c
new file mode 100644
index 00000000000..48d23487d24
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gfx_v6_0.c
@@ -0,0 +1,3623 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_gfx.h"
+#include "amdgpu_ucode.h"
+#include "clearstate_si.h"
+#include "bif/bif_3_0_d.h"
+#include "bif/bif_3_0_sh_mask.h"
+#include "oss/oss_1_0_d.h"
+#include "oss/oss_1_0_sh_mask.h"
+#include "gca/gfx_6_0_d.h"
+#include "gca/gfx_6_0_sh_mask.h"
+#include "gmc/gmc_6_0_d.h"
+#include "gmc/gmc_6_0_sh_mask.h"
+#include "dce/dce_6_0_d.h"
+#include "dce/dce_6_0_sh_mask.h"
+#include "gca/gfx_7_2_enum.h"
+#include "si_enums.h"
+#include "si.h"
+
+static void gfx_v6_0_set_ring_funcs(struct amdgpu_device *adev);
+static void gfx_v6_0_set_irq_funcs(struct amdgpu_device *adev);
+static void gfx_v6_0_get_cu_info(struct amdgpu_device *adev);
+
+MODULE_FIRMWARE("amdgpu/tahiti_pfp.bin");
+MODULE_FIRMWARE("amdgpu/tahiti_me.bin");
+MODULE_FIRMWARE("amdgpu/tahiti_ce.bin");
+MODULE_FIRMWARE("amdgpu/tahiti_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/pitcairn_pfp.bin");
+MODULE_FIRMWARE("amdgpu/pitcairn_me.bin");
+MODULE_FIRMWARE("amdgpu/pitcairn_ce.bin");
+MODULE_FIRMWARE("amdgpu/pitcairn_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/verde_pfp.bin");
+MODULE_FIRMWARE("amdgpu/verde_me.bin");
+MODULE_FIRMWARE("amdgpu/verde_ce.bin");
+MODULE_FIRMWARE("amdgpu/verde_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/oland_pfp.bin");
+MODULE_FIRMWARE("amdgpu/oland_me.bin");
+MODULE_FIRMWARE("amdgpu/oland_ce.bin");
+MODULE_FIRMWARE("amdgpu/oland_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/hainan_pfp.bin");
+MODULE_FIRMWARE("amdgpu/hainan_me.bin");
+MODULE_FIRMWARE("amdgpu/hainan_ce.bin");
+MODULE_FIRMWARE("amdgpu/hainan_rlc.bin");
+
+static u32 gfx_v6_0_get_csb_size(struct amdgpu_device *adev);
+static void gfx_v6_0_get_csb_buffer(struct amdgpu_device *adev, volatile u32 *buffer);
+//static void gfx_v6_0_init_cp_pg_table(struct amdgpu_device *adev);
+static void gfx_v6_0_init_pg(struct amdgpu_device *adev);
+
+#define ARRAY_MODE(x)					((x) << GB_TILE_MODE0__ARRAY_MODE__SHIFT)
+#define PIPE_CONFIG(x)					((x) << GB_TILE_MODE0__PIPE_CONFIG__SHIFT)
+#define TILE_SPLIT(x)					((x) << GB_TILE_MODE0__TILE_SPLIT__SHIFT)
+#define MICRO_TILE_MODE(x)				((x) << 0)
+#define SAMPLE_SPLIT(x)					((x) << GB_TILE_MODE0__SAMPLE_SPLIT__SHIFT)
+#define BANK_WIDTH(x)					((x) << 14)
+#define BANK_HEIGHT(x)					((x) << 16)
+#define MACRO_TILE_ASPECT(x)				((x) << 18)
+#define NUM_BANKS(x)					((x) << 20)
+
+static const u32 verde_rlc_save_restore_register_list[] =
+{
+	(0x8000 << 16) | (0x98f4 >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x98f4 >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0xe80 >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0xe80 >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x89bc >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x89bc >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x8c1c >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x8c1c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x98f0 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0xe7c >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x9148 >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x9148 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9150 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x897c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8d8c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0xac54 >> 2),
+	0X00000000,
+	0x3,
+	(0x9c00 << 16) | (0x98f8 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9910 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9914 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9918 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x991c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9920 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9924 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9928 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x992c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9930 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9934 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9938 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x993c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9940 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9944 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9948 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x994c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9950 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9954 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9958 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x995c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9960 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9964 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9968 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x996c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9970 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9974 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9978 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x997c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9980 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9984 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9988 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x998c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8c00 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8c14 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8c04 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8c08 >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x9b7c >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x9b7c >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0xe84 >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0xe84 >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x89c0 >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x89c0 >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x914c >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x914c >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x8c20 >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x8c20 >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x9354 >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x9354 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9060 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9364 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9100 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x913c >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x90e0 >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x90e4 >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x90e8 >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x90e0 >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x90e4 >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x90e8 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8bcc >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8b24 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x88c4 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8e50 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8c0c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8e58 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8e5c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9508 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x950c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9494 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0xac0c >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0xac10 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0xac14 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0xae00 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0xac08 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x88d4 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x88c8 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x88cc >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x89b0 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8b10 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x8a14 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9830 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9834 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9838 >> 2),
+	0x00000000,
+	(0x9c00 << 16) | (0x9a10 >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x9870 >> 2),
+	0x00000000,
+	(0x8000 << 16) | (0x9874 >> 2),
+	0x00000000,
+	(0x8001 << 16) | (0x9870 >> 2),
+	0x00000000,
+	(0x8001 << 16) | (0x9874 >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x9870 >> 2),
+	0x00000000,
+	(0x8040 << 16) | (0x9874 >> 2),
+	0x00000000,
+	(0x8041 << 16) | (0x9870 >> 2),
+	0x00000000,
+	(0x8041 << 16) | (0x9874 >> 2),
+	0x00000000,
+	0x00000000
+};
+
+static int gfx_v6_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+	const struct gfx_firmware_header_v1_0 *cp_hdr;
+	const struct rlc_firmware_header_v1_0 *rlc_hdr;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_TAHITI:
+		chip_name = "tahiti";
+		break;
+	case CHIP_PITCAIRN:
+		chip_name = "pitcairn";
+		break;
+	case CHIP_VERDE:
+		chip_name = "verde";
+		break;
+	case CHIP_OLAND:
+		chip_name = "oland";
+		break;
+	case CHIP_HAINAN:
+		chip_name = "hainan";
+		break;
+	default: BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_pfp.bin", chip_name);
+	err = request_firmware(&adev->gfx.pfp_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.pfp_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+	adev->gfx.pfp_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.pfp_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_me.bin", chip_name);
+	err = request_firmware(&adev->gfx.me_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.me_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+	adev->gfx.me_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.me_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ce.bin", chip_name);
+	err = request_firmware(&adev->gfx.ce_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.ce_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+	adev->gfx.ce_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.ce_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc.bin", chip_name);
+	err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.rlc_fw);
+	rlc_hdr = (const struct rlc_firmware_header_v1_0 *)adev->gfx.rlc_fw->data;
+	adev->gfx.rlc_fw_version = le32_to_cpu(rlc_hdr->header.ucode_version);
+	adev->gfx.rlc_feature_version = le32_to_cpu(rlc_hdr->ucode_feature_version);
+
+out:
+	if (err) {
+		pr_err("gfx6: Failed to load firmware \"%s\"\n", fw_name);
+		release_firmware(adev->gfx.pfp_fw);
+		adev->gfx.pfp_fw = NULL;
+		release_firmware(adev->gfx.me_fw);
+		adev->gfx.me_fw = NULL;
+		release_firmware(adev->gfx.ce_fw);
+		adev->gfx.ce_fw = NULL;
+		release_firmware(adev->gfx.rlc_fw);
+		adev->gfx.rlc_fw = NULL;
+	}
+	return err;
+}
+
+static void gfx_v6_0_tiling_mode_table_init(struct amdgpu_device *adev)
+{
+	const u32 num_tile_mode_states = ARRAY_SIZE(adev->gfx.config.tile_mode_array);
+	u32 reg_offset, split_equal_to_row_size, *tilemode;
+
+	memset(adev->gfx.config.tile_mode_array, 0, sizeof(adev->gfx.config.tile_mode_array));
+	tilemode = adev->gfx.config.tile_mode_array;
+
+	switch (adev->gfx.config.mem_row_size_in_kb) {
+	case 1:
+		split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_1KB;
+		break;
+	case 2:
+	default:
+		split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_2KB;
+		break;
+	case 4:
+		split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_4KB;
+		break;
+	}
+
+	if (adev->asic_type == CHIP_VERDE) {
+		tilemode[0] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[1] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[2] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[3] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[4] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16);
+		tilemode[5] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		tilemode[6] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		tilemode[7] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[8] =   ARRAY_MODE(ARRAY_LINEAR_ALIGNED);
+		tilemode[9] =   MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16);
+		tilemode[10] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[11] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[12] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[13] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16);
+		tilemode[14] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[15] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[16] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[17] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[18] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16);
+		tilemode[19] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[20] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[21] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK);
+		tilemode[22] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK);
+		tilemode[23] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		tilemode[24] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		tilemode[25] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[26] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[27] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[28] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[29] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[30] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			WREG32(mmGB_TILE_MODE0 + reg_offset, tilemode[reg_offset]);
+	} else if (adev->asic_type == CHIP_OLAND) {
+		tilemode[0] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[1] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[2] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[3] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[4] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[5] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(split_equal_to_row_size) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[6] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(split_equal_to_row_size) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[7] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(split_equal_to_row_size) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[8] =   MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[9] =   MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[10] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[11] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[12] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[13] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[14] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[15] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[16] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[17] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(split_equal_to_row_size) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[21] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[22] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4);
+		tilemode[23] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[24] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2);
+		tilemode[25] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				NUM_BANKS(ADDR_SURF_8_BANK) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1);
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			WREG32(mmGB_TILE_MODE0 + reg_offset, tilemode[reg_offset]);
+	} else if (adev->asic_type == CHIP_HAINAN) {
+		tilemode[0] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[1] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[2] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[3] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[4] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2);
+		tilemode[5] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK);
+		tilemode[6] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK);
+		tilemode[7] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		tilemode[8] =   ARRAY_MODE(ARRAY_LINEAR_ALIGNED);
+		tilemode[9] =   MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2);
+		tilemode[10] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[11] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[12] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[13] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2);
+		tilemode[14] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[15] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[16] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[17] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[18] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P2);
+		tilemode[19] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[20] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[21] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK);
+		tilemode[22] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK);
+		tilemode[23] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK);
+		tilemode[24] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK);
+		tilemode[25] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		tilemode[26] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		tilemode[27] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		tilemode[28] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		tilemode[29] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		tilemode[30] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			WREG32(mmGB_TILE_MODE0 + reg_offset, tilemode[reg_offset]);
+	} else if ((adev->asic_type == CHIP_TAHITI) || (adev->asic_type == CHIP_PITCAIRN)) {
+		tilemode[0] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[1] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[2] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[3] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[4] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16);
+		tilemode[5] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[6] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[7] =   MICRO_TILE_MODE(ADDR_SURF_DEPTH_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[8] =   ARRAY_MODE(ARRAY_LINEAR_ALIGNED);
+		tilemode[9] =   MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16);
+		tilemode[10] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[11] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[12] =  MICRO_TILE_MODE(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[13] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16);
+		tilemode[14] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[15] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[16] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK);
+		tilemode[17] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[18] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16);
+		tilemode[19] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[20] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK) |
+				TILE_SPLIT(split_equal_to_row_size);
+		tilemode[21] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		tilemode[22] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK);
+		tilemode[23] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[24] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[25] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[26] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[27] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[28] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[29] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_1KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		tilemode[30] =  MICRO_TILE_MODE(ADDR_SURF_THIN_MICRO_TILING) |
+				ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_8x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_2_BANK);
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			WREG32(mmGB_TILE_MODE0 + reg_offset, tilemode[reg_offset]);
+	} else {
+		DRM_ERROR("unknown asic: 0x%x\n", adev->asic_type);
+	}
+}
+
+static void gfx_v6_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
+				  u32 sh_num, u32 instance)
+{
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX, instance);
+
+	if ((se_num == 0xffffffff) && (sh_num == 0xffffffff))
+		data |= GRBM_GFX_INDEX__SH_BROADCAST_WRITES_MASK |
+			GRBM_GFX_INDEX__SE_BROADCAST_WRITES_MASK;
+	else if (se_num == 0xffffffff)
+		data |= GRBM_GFX_INDEX__SE_BROADCAST_WRITES_MASK |
+			(sh_num << GRBM_GFX_INDEX__SH_INDEX__SHIFT);
+	else if (sh_num == 0xffffffff)
+		data |= GRBM_GFX_INDEX__SH_BROADCAST_WRITES_MASK |
+			(se_num << GRBM_GFX_INDEX__SE_INDEX__SHIFT);
+	else
+		data |= (sh_num << GRBM_GFX_INDEX__SH_INDEX__SHIFT) |
+			(se_num << GRBM_GFX_INDEX__SE_INDEX__SHIFT);
+	WREG32(mmGRBM_GFX_INDEX, data);
+}
+
+static u32 gfx_v6_0_get_rb_active_bitmap(struct amdgpu_device *adev)
+{
+	u32 data, mask;
+
+	data = RREG32(mmCC_RB_BACKEND_DISABLE) |
+		RREG32(mmGC_USER_RB_BACKEND_DISABLE);
+
+	data = REG_GET_FIELD(data, GC_USER_RB_BACKEND_DISABLE, BACKEND_DISABLE);
+
+	mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_backends_per_se/
+					 adev->gfx.config.max_sh_per_se);
+
+	return ~data & mask;
+}
+
+static void gfx_v6_0_raster_config(struct amdgpu_device *adev, u32 *rconf)
+{
+	switch (adev->asic_type) {
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+		*rconf |=
+			   (2 << PA_SC_RASTER_CONFIG__RB_XSEL2__SHIFT) |
+			   (1 << PA_SC_RASTER_CONFIG__RB_XSEL__SHIFT) |
+			   (2 << PA_SC_RASTER_CONFIG__PKR_MAP__SHIFT) |
+			   (1 << PA_SC_RASTER_CONFIG__PKR_YSEL__SHIFT) |
+			   (2 << PA_SC_RASTER_CONFIG__SE_MAP__SHIFT) |
+			   (2 << PA_SC_RASTER_CONFIG__SE_XSEL__SHIFT) |
+			   (2 << PA_SC_RASTER_CONFIG__SE_YSEL__SHIFT);
+		break;
+	case CHIP_VERDE:
+		*rconf |=
+			   (1 << PA_SC_RASTER_CONFIG__RB_XSEL__SHIFT) |
+			   (2 << PA_SC_RASTER_CONFIG__PKR_MAP__SHIFT) |
+			   (1 << PA_SC_RASTER_CONFIG__PKR_YSEL__SHIFT);
+		break;
+	case CHIP_OLAND:
+		*rconf |= (1 << PA_SC_RASTER_CONFIG__RB_YSEL__SHIFT);
+		break;
+	case CHIP_HAINAN:
+		*rconf |= 0x0;
+		break;
+	default:
+		DRM_ERROR("unknown asic: 0x%x\n", adev->asic_type);
+		break;
+	}
+}
+
+static void gfx_v6_0_write_harvested_raster_configs(struct amdgpu_device *adev,
+						    u32 raster_config, unsigned rb_mask,
+						    unsigned num_rb)
+{
+	unsigned sh_per_se = max_t(unsigned, adev->gfx.config.max_sh_per_se, 1);
+	unsigned num_se = max_t(unsigned, adev->gfx.config.max_shader_engines, 1);
+	unsigned rb_per_pkr = min_t(unsigned, num_rb / num_se / sh_per_se, 2);
+	unsigned rb_per_se = num_rb / num_se;
+	unsigned se_mask[4];
+	unsigned se;
+
+	se_mask[0] = ((1 << rb_per_se) - 1) & rb_mask;
+	se_mask[1] = (se_mask[0] << rb_per_se) & rb_mask;
+	se_mask[2] = (se_mask[1] << rb_per_se) & rb_mask;
+	se_mask[3] = (se_mask[2] << rb_per_se) & rb_mask;
+
+	WARN_ON(!(num_se == 1 || num_se == 2 || num_se == 4));
+	WARN_ON(!(sh_per_se == 1 || sh_per_se == 2));
+	WARN_ON(!(rb_per_pkr == 1 || rb_per_pkr == 2));
+
+	for (se = 0; se < num_se; se++) {
+		unsigned raster_config_se = raster_config;
+		unsigned pkr0_mask = ((1 << rb_per_pkr) - 1) << (se * rb_per_se);
+		unsigned pkr1_mask = pkr0_mask << rb_per_pkr;
+		int idx = (se / 2) * 2;
+
+		if ((num_se > 1) && (!se_mask[idx] || !se_mask[idx + 1])) {
+			raster_config_se &= ~PA_SC_RASTER_CONFIG__SE_MAP_MASK;
+
+			if (!se_mask[idx])
+				raster_config_se |= RASTER_CONFIG_SE_MAP_3 << PA_SC_RASTER_CONFIG__SE_MAP__SHIFT;
+			else
+				raster_config_se |= RASTER_CONFIG_SE_MAP_0 << PA_SC_RASTER_CONFIG__SE_MAP__SHIFT;
+		}
+
+		pkr0_mask &= rb_mask;
+		pkr1_mask &= rb_mask;
+		if (rb_per_se > 2 && (!pkr0_mask || !pkr1_mask)) {
+			raster_config_se &= ~PA_SC_RASTER_CONFIG__PKR_MAP_MASK;
+
+			if (!pkr0_mask)
+				raster_config_se |= RASTER_CONFIG_PKR_MAP_3 << PA_SC_RASTER_CONFIG__PKR_MAP__SHIFT;
+			else
+				raster_config_se |= RASTER_CONFIG_PKR_MAP_0 << PA_SC_RASTER_CONFIG__PKR_MAP__SHIFT;
+		}
+
+		if (rb_per_se >= 2) {
+			unsigned rb0_mask = 1 << (se * rb_per_se);
+			unsigned rb1_mask = rb0_mask << 1;
+
+			rb0_mask &= rb_mask;
+			rb1_mask &= rb_mask;
+			if (!rb0_mask || !rb1_mask) {
+				raster_config_se &= ~PA_SC_RASTER_CONFIG__RB_MAP_PKR0_MASK;
+
+				if (!rb0_mask)
+					raster_config_se |=
+						RASTER_CONFIG_RB_MAP_3 << PA_SC_RASTER_CONFIG__RB_MAP_PKR0__SHIFT;
+				else
+					raster_config_se |=
+						RASTER_CONFIG_RB_MAP_0 << PA_SC_RASTER_CONFIG__RB_MAP_PKR0__SHIFT;
+			}
+
+			if (rb_per_se > 2) {
+				rb0_mask = 1 << (se * rb_per_se + rb_per_pkr);
+				rb1_mask = rb0_mask << 1;
+				rb0_mask &= rb_mask;
+				rb1_mask &= rb_mask;
+				if (!rb0_mask || !rb1_mask) {
+					raster_config_se &= ~PA_SC_RASTER_CONFIG__RB_MAP_PKR1_MASK;
+
+					if (!rb0_mask)
+						raster_config_se |=
+							RASTER_CONFIG_RB_MAP_3 << PA_SC_RASTER_CONFIG__RB_MAP_PKR1__SHIFT;
+					else
+						raster_config_se |=
+							RASTER_CONFIG_RB_MAP_0 << PA_SC_RASTER_CONFIG__RB_MAP_PKR1__SHIFT;
+				}
+			}
+		}
+
+		/* GRBM_GFX_INDEX has a different offset on SI */
+		gfx_v6_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff);
+		WREG32(mmPA_SC_RASTER_CONFIG, raster_config_se);
+	}
+
+	/* GRBM_GFX_INDEX has a different offset on SI */
+	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+}
+
+static void gfx_v6_0_setup_rb(struct amdgpu_device *adev)
+{
+	int i, j;
+	u32 data;
+	u32 raster_config = 0;
+	u32 active_rbs = 0;
+	u32 rb_bitmap_width_per_sh = adev->gfx.config.max_backends_per_se /
+					adev->gfx.config.max_sh_per_se;
+	unsigned num_rb_pipes;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff);
+			data = gfx_v6_0_get_rb_active_bitmap(adev);
+			active_rbs |= data <<
+				((i * adev->gfx.config.max_sh_per_se + j) *
+				 rb_bitmap_width_per_sh);
+		}
+	}
+	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+
+	adev->gfx.config.backend_enable_mask = active_rbs;
+	adev->gfx.config.num_rbs = hweight32(active_rbs);
+
+	num_rb_pipes = min_t(unsigned, adev->gfx.config.max_backends_per_se *
+			     adev->gfx.config.max_shader_engines, 16);
+
+	gfx_v6_0_raster_config(adev, &raster_config);
+
+	if (!adev->gfx.config.backend_enable_mask ||
+	     adev->gfx.config.num_rbs >= num_rb_pipes)
+		WREG32(mmPA_SC_RASTER_CONFIG, raster_config);
+	else
+		gfx_v6_0_write_harvested_raster_configs(adev, raster_config,
+							adev->gfx.config.backend_enable_mask,
+							num_rb_pipes);
+
+	/* cache the values for userspace */
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff);
+			adev->gfx.config.rb_config[i][j].rb_backend_disable =
+				RREG32(mmCC_RB_BACKEND_DISABLE);
+			adev->gfx.config.rb_config[i][j].user_rb_backend_disable =
+				RREG32(mmGC_USER_RB_BACKEND_DISABLE);
+			adev->gfx.config.rb_config[i][j].raster_config =
+				RREG32(mmPA_SC_RASTER_CONFIG);
+		}
+	}
+	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+static void gfx_v6_0_set_user_cu_inactive_bitmap(struct amdgpu_device *adev,
+						 u32 bitmap)
+{
+	u32 data;
+
+	if (!bitmap)
+		return;
+
+	data = bitmap << GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT;
+	data &= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK;
+
+	WREG32(mmGC_USER_SHADER_ARRAY_CONFIG, data);
+}
+
+static u32 gfx_v6_0_get_cu_enabled(struct amdgpu_device *adev)
+{
+	u32 data, mask;
+
+	data = RREG32(mmCC_GC_SHADER_ARRAY_CONFIG) |
+		RREG32(mmGC_USER_SHADER_ARRAY_CONFIG);
+
+	mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_cu_per_sh);
+	return ~REG_GET_FIELD(data, CC_GC_SHADER_ARRAY_CONFIG, INACTIVE_CUS) & mask;
+}
+
+
+static void gfx_v6_0_setup_spi(struct amdgpu_device *adev)
+{
+	int i, j, k;
+	u32 data, mask;
+	u32 active_cu = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff);
+			data = RREG32(mmSPI_STATIC_THREAD_MGMT_3);
+			active_cu = gfx_v6_0_get_cu_enabled(adev);
+
+			mask = 1;
+			for (k = 0; k < 16; k++) {
+				mask <<= k;
+				if (active_cu & mask) {
+					data &= ~mask;
+					WREG32(mmSPI_STATIC_THREAD_MGMT_3, data);
+					break;
+				}
+			}
+		}
+	}
+	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+static void gfx_v6_0_config_init(struct amdgpu_device *adev)
+{
+	adev->gfx.config.double_offchip_lds_buf = 0;
+}
+
+static void gfx_v6_0_gpu_init(struct amdgpu_device *adev)
+{
+	u32 gb_addr_config = 0;
+	u32 mc_shared_chmap, mc_arb_ramcfg;
+	u32 sx_debug_1;
+	u32 hdp_host_path_cntl;
+	u32 tmp;
+
+	switch (adev->asic_type) {
+	case CHIP_TAHITI:
+		adev->gfx.config.max_shader_engines = 2;
+		adev->gfx.config.max_tile_pipes = 12;
+		adev->gfx.config.max_cu_per_sh = 8;
+		adev->gfx.config.max_sh_per_se = 2;
+		adev->gfx.config.max_backends_per_se = 4;
+		adev->gfx.config.max_texture_channel_caches = 12;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = TAHITI_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_PITCAIRN:
+		adev->gfx.config.max_shader_engines = 2;
+		adev->gfx.config.max_tile_pipes = 8;
+		adev->gfx.config.max_cu_per_sh = 5;
+		adev->gfx.config.max_sh_per_se = 2;
+		adev->gfx.config.max_backends_per_se = 4;
+		adev->gfx.config.max_texture_channel_caches = 8;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = TAHITI_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_VERDE:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 4;
+		adev->gfx.config.max_cu_per_sh = 5;
+		adev->gfx.config.max_sh_per_se = 2;
+		adev->gfx.config.max_backends_per_se = 4;
+		adev->gfx.config.max_texture_channel_caches = 4;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x40;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = VERDE_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_OLAND:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 4;
+		adev->gfx.config.max_cu_per_sh = 6;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 2;
+		adev->gfx.config.max_texture_channel_caches = 4;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 16;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x40;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = VERDE_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_HAINAN:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 4;
+		adev->gfx.config.max_cu_per_sh = 5;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 1;
+		adev->gfx.config.max_texture_channel_caches = 2;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 16;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x40;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = HAINAN_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	default:
+		BUG();
+		break;
+	}
+
+	WREG32(mmGRBM_CNTL, (0xff << GRBM_CNTL__READ_TIMEOUT__SHIFT));
+	WREG32(mmSRBM_INT_CNTL, 1);
+	WREG32(mmSRBM_INT_ACK, 1);
+
+	WREG32(mmBIF_FB_EN, BIF_FB_EN__FB_READ_EN_MASK | BIF_FB_EN__FB_WRITE_EN_MASK);
+
+	mc_shared_chmap = RREG32(mmMC_SHARED_CHMAP);
+	adev->gfx.config.mc_arb_ramcfg = RREG32(mmMC_ARB_RAMCFG);
+	mc_arb_ramcfg = adev->gfx.config.mc_arb_ramcfg;
+
+	adev->gfx.config.num_tile_pipes = adev->gfx.config.max_tile_pipes;
+	adev->gfx.config.mem_max_burst_length_bytes = 256;
+	tmp = (mc_arb_ramcfg & MC_ARB_RAMCFG__NOOFCOLS_MASK) >> MC_ARB_RAMCFG__NOOFCOLS__SHIFT;
+	adev->gfx.config.mem_row_size_in_kb = (4 * (1 << (8 + tmp))) / 1024;
+	if (adev->gfx.config.mem_row_size_in_kb > 4)
+		adev->gfx.config.mem_row_size_in_kb = 4;
+	adev->gfx.config.shader_engine_tile_size = 32;
+	adev->gfx.config.num_gpus = 1;
+	adev->gfx.config.multi_gpu_tile_size = 64;
+
+	gb_addr_config &= ~GB_ADDR_CONFIG__ROW_SIZE_MASK;
+	switch (adev->gfx.config.mem_row_size_in_kb) {
+	case 1:
+	default:
+		gb_addr_config |= 0 << GB_ADDR_CONFIG__ROW_SIZE__SHIFT;
+		break;
+	case 2:
+		gb_addr_config |= 1 << GB_ADDR_CONFIG__ROW_SIZE__SHIFT;
+		break;
+	case 4:
+		gb_addr_config |= 2 << GB_ADDR_CONFIG__ROW_SIZE__SHIFT;
+		break;
+	}
+	gb_addr_config &= ~GB_ADDR_CONFIG__NUM_SHADER_ENGINES_MASK;
+	if (adev->gfx.config.max_shader_engines == 2)
+		gb_addr_config |= 1 << GB_ADDR_CONFIG__NUM_SHADER_ENGINES__SHIFT;
+	adev->gfx.config.gb_addr_config = gb_addr_config;
+
+	WREG32(mmGB_ADDR_CONFIG, gb_addr_config);
+	WREG32(mmDMIF_ADDR_CONFIG, gb_addr_config);
+	WREG32(mmDMIF_ADDR_CALC, gb_addr_config);
+	WREG32(mmHDP_ADDR_CONFIG, gb_addr_config);
+	WREG32(mmDMA_TILING_CONFIG + DMA0_REGISTER_OFFSET, gb_addr_config);
+	WREG32(mmDMA_TILING_CONFIG + DMA1_REGISTER_OFFSET, gb_addr_config);
+
+#if 0
+	if (adev->has_uvd) {
+		WREG32(mmUVD_UDEC_ADDR_CONFIG, gb_addr_config);
+		WREG32(mmUVD_UDEC_DB_ADDR_CONFIG, gb_addr_config);
+		WREG32(mmUVD_UDEC_DBW_ADDR_CONFIG, gb_addr_config);
+	}
+#endif
+	gfx_v6_0_tiling_mode_table_init(adev);
+
+	gfx_v6_0_setup_rb(adev);
+
+	gfx_v6_0_setup_spi(adev);
+
+	gfx_v6_0_get_cu_info(adev);
+	gfx_v6_0_config_init(adev);
+
+	WREG32(mmCP_QUEUE_THRESHOLDS, ((0x16 << CP_QUEUE_THRESHOLDS__ROQ_IB1_START__SHIFT) |
+				       (0x2b << CP_QUEUE_THRESHOLDS__ROQ_IB2_START__SHIFT)));
+	WREG32(mmCP_MEQ_THRESHOLDS, (0x30 << CP_MEQ_THRESHOLDS__MEQ1_START__SHIFT) |
+				    (0x60 << CP_MEQ_THRESHOLDS__MEQ2_START__SHIFT));
+
+	sx_debug_1 = RREG32(mmSX_DEBUG_1);
+	WREG32(mmSX_DEBUG_1, sx_debug_1);
+
+	WREG32(mmSPI_CONFIG_CNTL_1, (4 << SPI_CONFIG_CNTL_1__VTX_DONE_DELAY__SHIFT));
+
+	WREG32(mmPA_SC_FIFO_SIZE, ((adev->gfx.config.sc_prim_fifo_size_frontend << PA_SC_FIFO_SIZE__SC_FRONTEND_PRIM_FIFO_SIZE__SHIFT) |
+				   (adev->gfx.config.sc_prim_fifo_size_backend << PA_SC_FIFO_SIZE__SC_BACKEND_PRIM_FIFO_SIZE__SHIFT) |
+				   (adev->gfx.config.sc_hiz_tile_fifo_size << PA_SC_FIFO_SIZE__SC_HIZ_TILE_FIFO_SIZE__SHIFT) |
+				   (adev->gfx.config.sc_earlyz_tile_fifo_size << PA_SC_FIFO_SIZE__SC_EARLYZ_TILE_FIFO_SIZE__SHIFT)));
+
+	WREG32(mmVGT_NUM_INSTANCES, 1);
+	WREG32(mmCP_PERFMON_CNTL, 0);
+	WREG32(mmSQ_CONFIG, 0);
+	WREG32(mmPA_SC_FORCE_EOV_MAX_CNTS, ((4095 << PA_SC_FORCE_EOV_MAX_CNTS__FORCE_EOV_MAX_CLK_CNT__SHIFT) |
+					  (255 << PA_SC_FORCE_EOV_MAX_CNTS__FORCE_EOV_MAX_REZ_CNT__SHIFT)));
+
+	WREG32(mmVGT_CACHE_INVALIDATION,
+		(VC_AND_TC << VGT_CACHE_INVALIDATION__CACHE_INVALIDATION__SHIFT) |
+		(ES_AND_GS_AUTO << VGT_CACHE_INVALIDATION__AUTO_INVLD_EN__SHIFT));
+
+	WREG32(mmVGT_GS_VERTEX_REUSE, 16);
+	WREG32(mmPA_SC_LINE_STIPPLE_STATE, 0);
+
+	WREG32(mmCB_PERFCOUNTER0_SELECT0, 0);
+	WREG32(mmCB_PERFCOUNTER0_SELECT1, 0);
+	WREG32(mmCB_PERFCOUNTER1_SELECT0, 0);
+	WREG32(mmCB_PERFCOUNTER1_SELECT1, 0);
+	WREG32(mmCB_PERFCOUNTER2_SELECT0, 0);
+	WREG32(mmCB_PERFCOUNTER2_SELECT1, 0);
+	WREG32(mmCB_PERFCOUNTER3_SELECT0, 0);
+	WREG32(mmCB_PERFCOUNTER3_SELECT1, 0);
+
+	hdp_host_path_cntl = RREG32(mmHDP_HOST_PATH_CNTL);
+	WREG32(mmHDP_HOST_PATH_CNTL, hdp_host_path_cntl);
+
+	WREG32(mmPA_CL_ENHANCE, PA_CL_ENHANCE__CLIP_VTX_REORDER_ENA_MASK |
+				(3 << PA_CL_ENHANCE__NUM_CLIP_SEQ__SHIFT));
+
+	udelay(50);
+}
+
+
+static void gfx_v6_0_scratch_init(struct amdgpu_device *adev)
+{
+	adev->gfx.scratch.num_reg = 8;
+	adev->gfx.scratch.reg_base = mmSCRATCH_REG0;
+	adev->gfx.scratch.free_mask = (1u << adev->gfx.scratch.num_reg) - 1;
+}
+
+static int gfx_v6_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t scratch;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_gfx_scratch_free(adev, scratch);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
+	amdgpu_ring_write(ring, (scratch - PACKET3_SET_CONFIG_REG_START));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
+			  ring->idx, scratch, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_gfx_scratch_free(adev, scratch);
+	return r;
+}
+
+static void gfx_v6_0_ring_emit_vgt_flush(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE, 0));
+	amdgpu_ring_write(ring, EVENT_TYPE(VGT_FLUSH) |
+		EVENT_INDEX(0));
+}
+
+static void gfx_v6_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+				     u64 seq, unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+	/* flush read cache over gart */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
+	amdgpu_ring_write(ring, (mmCP_COHER_CNTL2 - PACKET3_SET_CONFIG_REG_START));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SURFACE_SYNC, 3));
+	amdgpu_ring_write(ring, PACKET3_TCL1_ACTION_ENA |
+			  PACKET3_TC_ACTION_ENA |
+			  PACKET3_SH_KCACHE_ACTION_ENA |
+			  PACKET3_SH_ICACHE_ACTION_ENA);
+	amdgpu_ring_write(ring, 0xFFFFFFFF);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 10); /* poll interval */
+	/* EVENT_WRITE_EOP - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE_EOP, 4));
+	amdgpu_ring_write(ring, EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) | EVENT_INDEX(5));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) |
+				((write64bit ? 2 : 1) << CP_EOP_DONE_DATA_CNTL__DATA_SEL__SHIFT) |
+				((int_sel ? 2 : 0) << CP_EOP_DONE_DATA_CNTL__INT_SEL__SHIFT));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+}
+
+static void gfx_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib,
+				  unsigned vmid, bool ctx_switch)
+{
+	u32 header, control = 0;
+
+	/* insert SWITCH_BUFFER packet before first IB in the ring frame */
+	if (ctx_switch) {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+
+	if (ib->flags & AMDGPU_IB_FLAG_CE)
+		header = PACKET3(PACKET3_INDIRECT_BUFFER_CONST, 2);
+	else
+		header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
+
+	control |= ib->length_dw | (vmid << 24);
+
+	amdgpu_ring_write(ring, header);
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+			  (2 << 0) |
+#endif
+			  (ib->gpu_addr & 0xFFFFFFFC));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFFFF);
+	amdgpu_ring_write(ring, control);
+}
+
+/**
+ * gfx_v6_0_ring_test_ib - basic ring IB test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Allocate an IB and execute it on the gfx ring (SI).
+ * Provides a basic gfx ring test to verify that IBs are working.
+ * Returns 0 on success, error on failure.
+ */
+static int gfx_v6_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	uint32_t scratch;
+	uint32_t tmp = 0;
+	long r;
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get scratch reg (%ld).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, 256, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err1;
+	}
+	ib.ptr[0] = PACKET3(PACKET3_SET_CONFIG_REG, 1);
+	ib.ptr[1] = ((scratch - PACKET3_SET_CONFIG_REG_START));
+	ib.ptr[2] = 0xDEADBEEF;
+	ib.length_dw = 3;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err2;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
+		goto err2;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto err2;
+	}
+	tmp = RREG32(scratch);
+	if (tmp == 0xDEADBEEF) {
+		DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
+	} else {
+		DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
+			  scratch, tmp);
+		r = -EINVAL;
+	}
+
+err2:
+	amdgpu_ib_free(adev, &ib, NULL);
+	dma_fence_put(f);
+err1:
+	amdgpu_gfx_scratch_free(adev, scratch);
+	return r;
+}
+
+static void gfx_v6_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+	if (enable) {
+		WREG32(mmCP_ME_CNTL, 0);
+	} else {
+		WREG32(mmCP_ME_CNTL, (CP_ME_CNTL__ME_HALT_MASK |
+				      CP_ME_CNTL__PFP_HALT_MASK |
+				      CP_ME_CNTL__CE_HALT_MASK));
+		WREG32(mmSCRATCH_UMSK, 0);
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+			adev->gfx.gfx_ring[i].ready = false;
+		for (i = 0; i < adev->gfx.num_compute_rings; i++)
+			adev->gfx.compute_ring[i].ready = false;
+	}
+	udelay(50);
+}
+
+static int gfx_v6_0_cp_gfx_load_microcode(struct amdgpu_device *adev)
+{
+	unsigned i;
+	const struct gfx_firmware_header_v1_0 *pfp_hdr;
+	const struct gfx_firmware_header_v1_0 *ce_hdr;
+	const struct gfx_firmware_header_v1_0 *me_hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+
+	if (!adev->gfx.me_fw || !adev->gfx.pfp_fw || !adev->gfx.ce_fw)
+		return -EINVAL;
+
+	gfx_v6_0_cp_gfx_enable(adev, false);
+	pfp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+	ce_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+	me_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&pfp_hdr->header);
+	amdgpu_ucode_print_gfx_hdr(&ce_hdr->header);
+	amdgpu_ucode_print_gfx_hdr(&me_hdr->header);
+
+	/* PFP */
+	fw_data = (const __le32 *)
+		(adev->gfx.pfp_fw->data + le32_to_cpu(pfp_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(pfp_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_PFP_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_PFP_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_PFP_UCODE_ADDR, 0);
+
+	/* CE */
+	fw_data = (const __le32 *)
+		(adev->gfx.ce_fw->data + le32_to_cpu(ce_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(ce_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_CE_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_CE_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_CE_UCODE_ADDR, 0);
+
+	/* ME */
+	fw_data = (const __be32 *)
+		(adev->gfx.me_fw->data + le32_to_cpu(me_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(me_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_ME_RAM_WADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_ME_RAM_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_ME_RAM_WADDR, 0);
+
+	WREG32(mmCP_PFP_UCODE_ADDR, 0);
+	WREG32(mmCP_CE_UCODE_ADDR, 0);
+	WREG32(mmCP_ME_RAM_WADDR, 0);
+	WREG32(mmCP_ME_RAM_RADDR, 0);
+	return 0;
+}
+
+static int gfx_v6_0_cp_gfx_start(struct amdgpu_device *adev)
+{
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+	struct amdgpu_ring *ring = &adev->gfx.gfx_ring[0];
+	int r, i;
+
+	r = amdgpu_ring_alloc(ring, 7 + 4);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET3(PACKET3_ME_INITIALIZE, 5));
+	amdgpu_ring_write(ring, 0x1);
+	amdgpu_ring_write(ring, 0x0);
+	amdgpu_ring_write(ring, adev->gfx.config.max_hw_contexts - 1);
+	amdgpu_ring_write(ring, PACKET3_ME_INITIALIZE_DEVICE_ID(1));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_BASE, 2));
+	amdgpu_ring_write(ring, PACKET3_BASE_INDEX(CE_PARTITION_BASE));
+	amdgpu_ring_write(ring, 0xc000);
+	amdgpu_ring_write(ring, 0xe000);
+	amdgpu_ring_commit(ring);
+
+	gfx_v6_0_cp_gfx_enable(adev, true);
+
+	r = amdgpu_ring_alloc(ring, gfx_v6_0_get_csb_size(adev) + 10);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				amdgpu_ring_write(ring,
+						  PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
+				amdgpu_ring_write(ring, ext->reg_index - PACKET3_SET_CONTEXT_REG_START);
+				for (i = 0; i < ext->reg_count; i++)
+					amdgpu_ring_write(ring, ext->extent[i]);
+			}
+		}
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CLEAR_STATE, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONTEXT_REG, 2));
+	amdgpu_ring_write(ring, 0x00000316);
+	amdgpu_ring_write(ring, 0x0000000e);
+	amdgpu_ring_write(ring, 0x00000010);
+
+	amdgpu_ring_commit(ring);
+
+	return 0;
+}
+
+static int gfx_v6_0_cp_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 tmp;
+	u32 rb_bufsz;
+	int r;
+	u64 rptr_addr;
+
+	WREG32(mmCP_SEM_WAIT_TIMER, 0x0);
+	WREG32(mmCP_SEM_INCOMPLETE_TIMER_CNTL, 0x0);
+
+	/* Set the write pointer delay */
+	WREG32(mmCP_RB_WPTR_DELAY, 0);
+
+	WREG32(mmCP_DEBUG, 0);
+	WREG32(mmSCRATCH_ADDR, 0);
+
+	/* ring 0 - compute and gfx */
+	/* Set ring buffer size */
+	ring = &adev->gfx.gfx_ring[0];
+	rb_bufsz = order_base_2(ring->ring_size / 8);
+	tmp = (order_base_2(AMDGPU_GPU_PAGE_SIZE/8) << 8) | rb_bufsz;
+
+#ifdef __BIG_ENDIAN
+	tmp |= BUF_SWAP_32BIT;
+#endif
+	WREG32(mmCP_RB0_CNTL, tmp);
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(mmCP_RB0_CNTL, tmp | CP_RB0_CNTL__RB_RPTR_WR_ENA_MASK);
+	ring->wptr = 0;
+	WREG32(mmCP_RB0_WPTR, ring->wptr);
+
+	/* set the wb address whether it's enabled or not */
+	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	WREG32(mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
+	WREG32(mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF);
+
+	WREG32(mmSCRATCH_UMSK, 0);
+
+	mdelay(1);
+	WREG32(mmCP_RB0_CNTL, tmp);
+
+	WREG32(mmCP_RB0_BASE, ring->gpu_addr >> 8);
+
+	/* start the rings */
+	gfx_v6_0_cp_gfx_start(adev);
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		return r;
+	}
+
+	return 0;
+}
+
+static u64 gfx_v6_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	return ring->adev->wb.wb[ring->rptr_offs];
+}
+
+static u64 gfx_v6_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->gfx.gfx_ring[0])
+		return RREG32(mmCP_RB0_WPTR);
+	else if (ring == &adev->gfx.compute_ring[0])
+		return RREG32(mmCP_RB1_WPTR);
+	else if (ring == &adev->gfx.compute_ring[1])
+		return RREG32(mmCP_RB2_WPTR);
+	else
+		BUG();
+}
+
+static void gfx_v6_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32(mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
+	(void)RREG32(mmCP_RB0_WPTR);
+}
+
+static void gfx_v6_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->gfx.compute_ring[0]) {
+		WREG32(mmCP_RB1_WPTR, lower_32_bits(ring->wptr));
+		(void)RREG32(mmCP_RB1_WPTR);
+	} else if (ring == &adev->gfx.compute_ring[1]) {
+		WREG32(mmCP_RB2_WPTR, lower_32_bits(ring->wptr));
+		(void)RREG32(mmCP_RB2_WPTR);
+	} else {
+		BUG();
+	}
+
+}
+
+static int gfx_v6_0_cp_compute_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 tmp;
+	u32 rb_bufsz;
+	int i, r;
+	u64 rptr_addr;
+
+	/* ring1  - compute only */
+	/* Set ring buffer size */
+
+	ring = &adev->gfx.compute_ring[0];
+	rb_bufsz = order_base_2(ring->ring_size / 8);
+	tmp = (order_base_2(AMDGPU_GPU_PAGE_SIZE/8) << 8) | rb_bufsz;
+#ifdef __BIG_ENDIAN
+	tmp |= BUF_SWAP_32BIT;
+#endif
+	WREG32(mmCP_RB1_CNTL, tmp);
+
+	WREG32(mmCP_RB1_CNTL, tmp | CP_RB1_CNTL__RB_RPTR_WR_ENA_MASK);
+	ring->wptr = 0;
+	WREG32(mmCP_RB1_WPTR, ring->wptr);
+
+	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	WREG32(mmCP_RB1_RPTR_ADDR, lower_32_bits(rptr_addr));
+	WREG32(mmCP_RB1_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF);
+
+	mdelay(1);
+	WREG32(mmCP_RB1_CNTL, tmp);
+	WREG32(mmCP_RB1_BASE, ring->gpu_addr >> 8);
+
+	ring = &adev->gfx.compute_ring[1];
+	rb_bufsz = order_base_2(ring->ring_size / 8);
+	tmp = (order_base_2(AMDGPU_GPU_PAGE_SIZE/8) << 8) | rb_bufsz;
+#ifdef __BIG_ENDIAN
+	tmp |= BUF_SWAP_32BIT;
+#endif
+	WREG32(mmCP_RB2_CNTL, tmp);
+
+	WREG32(mmCP_RB2_CNTL, tmp | CP_RB2_CNTL__RB_RPTR_WR_ENA_MASK);
+	ring->wptr = 0;
+	WREG32(mmCP_RB2_WPTR, ring->wptr);
+	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	WREG32(mmCP_RB2_RPTR_ADDR, lower_32_bits(rptr_addr));
+	WREG32(mmCP_RB2_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF);
+
+	mdelay(1);
+	WREG32(mmCP_RB2_CNTL, tmp);
+	WREG32(mmCP_RB2_BASE, ring->gpu_addr >> 8);
+
+	adev->gfx.compute_ring[0].ready = false;
+	adev->gfx.compute_ring[1].ready = false;
+
+	for (i = 0; i < 2; i++) {
+		r = amdgpu_ring_test_ring(&adev->gfx.compute_ring[i]);
+		if (r)
+			return r;
+		adev->gfx.compute_ring[i].ready = true;
+	}
+
+	return 0;
+}
+
+static void gfx_v6_0_cp_enable(struct amdgpu_device *adev, bool enable)
+{
+	gfx_v6_0_cp_gfx_enable(adev, enable);
+}
+
+static int gfx_v6_0_cp_load_microcode(struct amdgpu_device *adev)
+{
+	return gfx_v6_0_cp_gfx_load_microcode(adev);
+}
+
+static void gfx_v6_0_enable_gui_idle_interrupt(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 tmp = RREG32(mmCP_INT_CNTL_RING0);
+	u32 mask;
+	int i;
+
+	if (enable)
+		tmp |= (CP_INT_CNTL__CNTX_BUSY_INT_ENABLE_MASK |
+			CP_INT_CNTL__CNTX_EMPTY_INT_ENABLE_MASK);
+	else
+		tmp &= ~(CP_INT_CNTL__CNTX_BUSY_INT_ENABLE_MASK |
+			 CP_INT_CNTL__CNTX_EMPTY_INT_ENABLE_MASK);
+	WREG32(mmCP_INT_CNTL_RING0, tmp);
+
+	if (!enable) {
+		/* read a gfx register */
+		tmp = RREG32(mmDB_DEPTH_INFO);
+
+		mask = RLC_BUSY_STATUS | GFX_POWER_STATUS | GFX_CLOCK_STATUS | GFX_LS_STATUS;
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if ((RREG32(mmRLC_STAT) & mask) == (GFX_CLOCK_STATUS | GFX_POWER_STATUS))
+				break;
+			udelay(1);
+		}
+	}
+}
+
+static int gfx_v6_0_cp_resume(struct amdgpu_device *adev)
+{
+	int r;
+
+	gfx_v6_0_enable_gui_idle_interrupt(adev, false);
+
+	r = gfx_v6_0_cp_load_microcode(adev);
+	if (r)
+		return r;
+
+	r = gfx_v6_0_cp_gfx_resume(adev);
+	if (r)
+		return r;
+	r = gfx_v6_0_cp_compute_resume(adev);
+	if (r)
+		return r;
+
+	gfx_v6_0_enable_gui_idle_interrupt(adev, true);
+
+	return 0;
+}
+
+static void gfx_v6_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_MEM_SPACE(1) | /* memory */
+				 WAIT_REG_MEM_FUNCTION(3) | /* equal */
+				 WAIT_REG_MEM_ENGINE(usepfp)));   /* pfp or me */
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, 0xffffffff);
+	amdgpu_ring_write(ring, 4); /* poll interval */
+
+	if (usepfp) {
+		/* synce CE with ME to prevent CE fetch CEIB before context switch done */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static void gfx_v6_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned vmid, uint64_t pd_addr)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for the invalidate to complete */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_FUNCTION(0) |  /* always */
+				 WAIT_REG_MEM_ENGINE(0))); /* me */
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0); /* ref */
+	amdgpu_ring_write(ring, 0); /* mask */
+	amdgpu_ring_write(ring, 0x20); /* poll interval */
+
+	if (usepfp) {
+		/* sync PFP to ME, otherwise we might get invalid PFP reads */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
+		amdgpu_ring_write(ring, 0x0);
+
+		/* synce CE with ME to prevent CE fetch CEIB before context switch done */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static void gfx_v6_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				    uint32_t reg, uint32_t val)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(usepfp) |
+				 WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
+static void gfx_v6_0_rlc_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.save_restore_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL);
+}
+
+static int gfx_v6_0_rlc_init(struct amdgpu_device *adev)
+{
+	const u32 *src_ptr;
+	volatile u32 *dst_ptr;
+	u32 dws, i;
+	u64 reg_list_mc_addr;
+	const struct cs_section_def *cs_data;
+	int r;
+
+	adev->gfx.rlc.reg_list = verde_rlc_save_restore_register_list;
+	adev->gfx.rlc.reg_list_size =
+			(u32)ARRAY_SIZE(verde_rlc_save_restore_register_list);
+
+	adev->gfx.rlc.cs_data = si_cs_data;
+	src_ptr = adev->gfx.rlc.reg_list;
+	dws = adev->gfx.rlc.reg_list_size;
+	cs_data = adev->gfx.rlc.cs_data;
+
+	if (src_ptr) {
+		/* save restore block */
+		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.save_restore_obj,
+					      &adev->gfx.rlc.save_restore_gpu_addr,
+					      (void **)&adev->gfx.rlc.sr_ptr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create RLC sr bo failed\n",
+				 r);
+			gfx_v6_0_rlc_fini(adev);
+			return r;
+		}
+
+		/* write the sr buffer */
+		dst_ptr = adev->gfx.rlc.sr_ptr;
+		for (i = 0; i < adev->gfx.rlc.reg_list_size; i++)
+			dst_ptr[i] = cpu_to_le32(src_ptr[i]);
+
+		amdgpu_bo_kunmap(adev->gfx.rlc.save_restore_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
+	}
+
+	if (cs_data) {
+		/* clear state block */
+		adev->gfx.rlc.clear_state_size = gfx_v6_0_get_csb_size(adev);
+		dws = adev->gfx.rlc.clear_state_size + (256 / 4);
+
+		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.clear_state_obj,
+					      &adev->gfx.rlc.clear_state_gpu_addr,
+					      (void **)&adev->gfx.rlc.cs_ptr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
+			gfx_v6_0_rlc_fini(adev);
+			return r;
+		}
+
+		/* set up the cs buffer */
+		dst_ptr = adev->gfx.rlc.cs_ptr;
+		reg_list_mc_addr = adev->gfx.rlc.clear_state_gpu_addr + 256;
+		dst_ptr[0] = cpu_to_le32(upper_32_bits(reg_list_mc_addr));
+		dst_ptr[1] = cpu_to_le32(lower_32_bits(reg_list_mc_addr));
+		dst_ptr[2] = cpu_to_le32(adev->gfx.rlc.clear_state_size);
+		gfx_v6_0_get_csb_buffer(adev, &dst_ptr[(256/4)]);
+		amdgpu_bo_kunmap(adev->gfx.rlc.clear_state_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
+	}
+
+	return 0;
+}
+
+static void gfx_v6_0_enable_lbpw(struct amdgpu_device *adev, bool enable)
+{
+	WREG32_FIELD(RLC_LB_CNTL, LOAD_BALANCE_ENABLE, enable ? 1 : 0);
+
+	if (!enable) {
+		gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		WREG32(mmSPI_LB_CU_MASK, 0x00ff);
+	}
+}
+
+static void gfx_v6_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmRLC_SERDES_MASTER_BUSY_0) == 0)
+			break;
+		udelay(1);
+	}
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmRLC_SERDES_MASTER_BUSY_1) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v6_0_update_rlc(struct amdgpu_device *adev, u32 rlc)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmRLC_CNTL);
+	if (tmp != rlc)
+		WREG32(mmRLC_CNTL, rlc);
+}
+
+static u32 gfx_v6_0_halt_rlc(struct amdgpu_device *adev)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_CNTL);
+
+	if (data & RLC_CNTL__RLC_ENABLE_F32_MASK) {
+		data &= ~RLC_CNTL__RLC_ENABLE_F32_MASK;
+		WREG32(mmRLC_CNTL, data);
+
+		gfx_v6_0_wait_for_rlc_serdes(adev);
+	}
+
+	return orig;
+}
+
+static void gfx_v6_0_rlc_stop(struct amdgpu_device *adev)
+{
+	WREG32(mmRLC_CNTL, 0);
+
+	gfx_v6_0_enable_gui_idle_interrupt(adev, false);
+	gfx_v6_0_wait_for_rlc_serdes(adev);
+}
+
+static void gfx_v6_0_rlc_start(struct amdgpu_device *adev)
+{
+	WREG32(mmRLC_CNTL, RLC_CNTL__RLC_ENABLE_F32_MASK);
+
+	gfx_v6_0_enable_gui_idle_interrupt(adev, true);
+
+	udelay(50);
+}
+
+static void gfx_v6_0_rlc_reset(struct amdgpu_device *adev)
+{
+	WREG32_FIELD(GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+	udelay(50);
+	WREG32_FIELD(GRBM_SOFT_RESET, SOFT_RESET_RLC, 0);
+	udelay(50);
+}
+
+static bool gfx_v6_0_lbpw_supported(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	/* Enable LBPW only for DDR3 */
+	tmp = RREG32(mmMC_SEQ_MISC0);
+	if ((tmp & 0xF0000000) == 0xB0000000)
+		return true;
+	return false;
+}
+
+static void gfx_v6_0_init_cg(struct amdgpu_device *adev)
+{
+}
+
+static int gfx_v6_0_rlc_resume(struct amdgpu_device *adev)
+{
+	u32 i;
+	const struct rlc_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+
+
+	if (!adev->gfx.rlc_fw)
+		return -EINVAL;
+
+	gfx_v6_0_rlc_stop(adev);
+	gfx_v6_0_rlc_reset(adev);
+	gfx_v6_0_init_pg(adev);
+	gfx_v6_0_init_cg(adev);
+
+	WREG32(mmRLC_RL_BASE, 0);
+	WREG32(mmRLC_RL_SIZE, 0);
+	WREG32(mmRLC_LB_CNTL, 0);
+	WREG32(mmRLC_LB_CNTR_MAX, 0xffffffff);
+	WREG32(mmRLC_LB_CNTR_INIT, 0);
+	WREG32(mmRLC_LB_INIT_CU_MASK, 0xffffffff);
+
+	WREG32(mmRLC_MC_CNTL, 0);
+	WREG32(mmRLC_UCODE_CNTL, 0);
+
+	hdr = (const struct rlc_firmware_header_v1_0 *)adev->gfx.rlc_fw->data;
+	fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+	fw_data = (const __le32 *)
+		(adev->gfx.rlc_fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	amdgpu_ucode_print_rlc_hdr(&hdr->header);
+
+	for (i = 0; i < fw_size; i++) {
+		WREG32(mmRLC_UCODE_ADDR, i);
+		WREG32(mmRLC_UCODE_DATA, le32_to_cpup(fw_data++));
+	}
+	WREG32(mmRLC_UCODE_ADDR, 0);
+
+	gfx_v6_0_enable_lbpw(adev, gfx_v6_0_lbpw_supported(adev));
+	gfx_v6_0_rlc_start(adev);
+
+	return 0;
+}
+
+static void gfx_v6_0_enable_cgcg(struct amdgpu_device *adev, bool enable)
+{
+	u32 data, orig, tmp;
+
+	orig = data = RREG32(mmRLC_CGCG_CGLS_CTRL);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)) {
+		gfx_v6_0_enable_gui_idle_interrupt(adev, true);
+
+		WREG32(mmRLC_GCPM_GENERAL_3, 0x00000080);
+
+		tmp = gfx_v6_0_halt_rlc(adev);
+
+		WREG32(mmRLC_SERDES_WR_MASTER_MASK_0, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_MASTER_MASK_1, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_CTRL, 0x00b000ff);
+
+		gfx_v6_0_wait_for_rlc_serdes(adev);
+		gfx_v6_0_update_rlc(adev, tmp);
+
+		WREG32(mmRLC_SERDES_WR_CTRL, 0x007000ff);
+
+		data |= RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK | RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+	} else {
+		gfx_v6_0_enable_gui_idle_interrupt(adev, false);
+
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+
+		data &= ~(RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK | RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK);
+	}
+
+	if (orig != data)
+		WREG32(mmRLC_CGCG_CGLS_CTRL, data);
+
+}
+
+static void gfx_v6_0_enable_mgcg(struct amdgpu_device *adev, bool enable)
+{
+
+	u32 data, orig, tmp = 0;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG)) {
+		orig = data = RREG32(mmCGTS_SM_CTRL_REG);
+		data = 0x96940200;
+		if (orig != data)
+			WREG32(mmCGTS_SM_CTRL_REG, data);
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CP_LS) {
+			orig = data = RREG32(mmCP_MEM_SLP_CNTL);
+			data |= CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+			if (orig != data)
+				WREG32(mmCP_MEM_SLP_CNTL, data);
+		}
+
+		orig = data = RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
+		data &= 0xffffffc0;
+		if (orig != data)
+			WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		tmp = gfx_v6_0_halt_rlc(adev);
+
+		WREG32(mmRLC_SERDES_WR_MASTER_MASK_0, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_MASTER_MASK_1, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_CTRL, 0x00d000ff);
+
+		gfx_v6_0_update_rlc(adev, tmp);
+	} else {
+		orig = data = RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
+		data |= 0x00000003;
+		if (orig != data)
+			WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		data = RREG32(mmCP_MEM_SLP_CNTL);
+		if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK) {
+			data &= ~CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+			WREG32(mmCP_MEM_SLP_CNTL, data);
+		}
+		orig = data = RREG32(mmCGTS_SM_CTRL_REG);
+		data |= CGTS_SM_CTRL_REG__LS_OVERRIDE_MASK | CGTS_SM_CTRL_REG__OVERRIDE_MASK;
+		if (orig != data)
+			WREG32(mmCGTS_SM_CTRL_REG, data);
+
+		tmp = gfx_v6_0_halt_rlc(adev);
+
+		WREG32(mmRLC_SERDES_WR_MASTER_MASK_0, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_MASTER_MASK_1, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_CTRL, 0x00e000ff);
+
+		gfx_v6_0_update_rlc(adev, tmp);
+	}
+}
+/*
+static void gfx_v6_0_update_cg(struct amdgpu_device *adev,
+			       bool enable)
+{
+	gfx_v6_0_enable_gui_idle_interrupt(adev, false);
+	if (enable) {
+		gfx_v6_0_enable_mgcg(adev, true);
+		gfx_v6_0_enable_cgcg(adev, true);
+	} else {
+		gfx_v6_0_enable_cgcg(adev, false);
+		gfx_v6_0_enable_mgcg(adev, false);
+	}
+	gfx_v6_0_enable_gui_idle_interrupt(adev, true);
+}
+*/
+
+static void gfx_v6_0_enable_sclk_slowdown_on_pu(struct amdgpu_device *adev,
+						bool enable)
+{
+}
+
+static void gfx_v6_0_enable_sclk_slowdown_on_pd(struct amdgpu_device *adev,
+						bool enable)
+{
+}
+
+static void gfx_v6_0_enable_cp_pg(struct amdgpu_device *adev, bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_CP))
+		data &= ~0x8000;
+	else
+		data |= 0x8000;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v6_0_enable_gds_pg(struct amdgpu_device *adev, bool enable)
+{
+}
+/*
+static void gfx_v6_0_init_cp_pg_table(struct amdgpu_device *adev)
+{
+	const __le32 *fw_data;
+	volatile u32 *dst_ptr;
+	int me, i, max_me = 4;
+	u32 bo_offset = 0;
+	u32 table_offset, table_size;
+
+	if (adev->asic_type == CHIP_KAVERI)
+		max_me = 5;
+
+	if (adev->gfx.rlc.cp_table_ptr == NULL)
+		return;
+
+	dst_ptr = adev->gfx.rlc.cp_table_ptr;
+	for (me = 0; me < max_me; me++) {
+		if (me == 0) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.ce_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 1) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.pfp_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 2) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.me_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 3) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec2_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		}
+
+		for (i = 0; i < table_size; i ++) {
+			dst_ptr[bo_offset + i] =
+				cpu_to_le32(le32_to_cpu(fw_data[table_offset + i]));
+		}
+
+		bo_offset += table_size;
+	}
+}
+*/
+static void gfx_v6_0_enable_gfx_cgpg(struct amdgpu_device *adev,
+				     bool enable)
+{
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG)) {
+		WREG32(mmRLC_TTOP_D, RLC_PUD(0x10) | RLC_PDD(0x10) | RLC_TTPD(0x10) | RLC_MSD(0x10));
+		WREG32_FIELD(RLC_PG_CNTL, GFX_POWER_GATING_ENABLE, 1);
+		WREG32_FIELD(RLC_AUTO_PG_CTRL, AUTO_PG_EN, 1);
+	} else {
+		WREG32_FIELD(RLC_AUTO_PG_CTRL, AUTO_PG_EN, 0);
+		(void)RREG32(mmDB_RENDER_CONTROL);
+	}
+}
+
+static void gfx_v6_0_init_ao_cu_mask(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	WREG32(mmRLC_PG_ALWAYS_ON_CU_MASK, adev->gfx.cu_info.ao_cu_mask);
+
+	tmp = RREG32(mmRLC_MAX_PG_CU);
+	tmp &= ~RLC_MAX_PG_CU__MAX_POWERED_UP_CU_MASK;
+	tmp |= (adev->gfx.cu_info.number << RLC_MAX_PG_CU__MAX_POWERED_UP_CU__SHIFT);
+	WREG32(mmRLC_MAX_PG_CU, tmp);
+}
+
+static void gfx_v6_0_enable_gfx_static_mgpg(struct amdgpu_device *adev,
+					    bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_SMG))
+		data |= RLC_PG_CNTL__STATIC_PER_CU_PG_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__STATIC_PER_CU_PG_ENABLE_MASK;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v6_0_enable_gfx_dynamic_mgpg(struct amdgpu_device *adev,
+					     bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_DMG))
+		data |= RLC_PG_CNTL__DYN_PER_CU_PG_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__DYN_PER_CU_PG_ENABLE_MASK;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v6_0_init_gfx_cgpg(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	WREG32(mmRLC_SAVE_AND_RESTORE_BASE, adev->gfx.rlc.save_restore_gpu_addr >> 8);
+	WREG32_FIELD(RLC_PG_CNTL, GFX_POWER_GATING_SRC, 1);
+	WREG32(mmRLC_CLEAR_STATE_RESTORE_BASE, adev->gfx.rlc.clear_state_gpu_addr >> 8);
+
+	tmp = RREG32(mmRLC_AUTO_PG_CTRL);
+	tmp &= ~RLC_AUTO_PG_CTRL__GRBM_REG_SAVE_GFX_IDLE_THRESHOLD_MASK;
+	tmp |= (0x700 << RLC_AUTO_PG_CTRL__GRBM_REG_SAVE_GFX_IDLE_THRESHOLD__SHIFT);
+	tmp &= ~RLC_AUTO_PG_CTRL__PG_AFTER_GRBM_REG_SAVE_THRESHOLD_MASK;
+	WREG32(mmRLC_AUTO_PG_CTRL, tmp);
+}
+
+static void gfx_v6_0_update_gfx_pg(struct amdgpu_device *adev, bool enable)
+{
+	gfx_v6_0_enable_gfx_cgpg(adev, enable);
+	gfx_v6_0_enable_gfx_static_mgpg(adev, enable);
+	gfx_v6_0_enable_gfx_dynamic_mgpg(adev, enable);
+}
+
+static u32 gfx_v6_0_get_csb_size(struct amdgpu_device *adev)
+{
+	u32 count = 0;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	if (adev->gfx.rlc.cs_data == NULL)
+		return 0;
+
+	/* begin clear state */
+	count += 2;
+	/* context control state */
+	count += 3;
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT)
+				count += 2 + ext->reg_count;
+			else
+				return 0;
+		}
+	}
+	/* pa_sc_raster_config */
+	count += 3;
+	/* end clear state */
+	count += 2;
+	/* clear state */
+	count += 2;
+
+	return count;
+}
+
+static void gfx_v6_0_get_csb_buffer(struct amdgpu_device *adev,
+				    volatile u32 *buffer)
+{
+	u32 count = 0, i;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	if (adev->gfx.rlc.cs_data == NULL)
+		return;
+	if (buffer == NULL)
+		return;
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	buffer[count++] = cpu_to_le32(0x80000000);
+	buffer[count++] = cpu_to_le32(0x80000000);
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				buffer[count++] =
+					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
+				buffer[count++] = cpu_to_le32(ext->reg_index - 0xa000);
+				for (i = 0; i < ext->reg_count; i++)
+					buffer[count++] = cpu_to_le32(ext->extent[i]);
+			} else {
+				return;
+			}
+		}
+	}
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 1));
+	buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
+	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
+	buffer[count++] = cpu_to_le32(0);
+}
+
+static void gfx_v6_0_init_pg(struct amdgpu_device *adev)
+{
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_PG |
+			      AMD_PG_SUPPORT_GFX_SMG |
+			      AMD_PG_SUPPORT_GFX_DMG |
+			      AMD_PG_SUPPORT_CP |
+			      AMD_PG_SUPPORT_GDS |
+			      AMD_PG_SUPPORT_RLC_SMU_HS)) {
+		gfx_v6_0_enable_sclk_slowdown_on_pu(adev, true);
+		gfx_v6_0_enable_sclk_slowdown_on_pd(adev, true);
+		if (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG) {
+			gfx_v6_0_init_gfx_cgpg(adev);
+			gfx_v6_0_enable_cp_pg(adev, true);
+			gfx_v6_0_enable_gds_pg(adev, true);
+		} else {
+			WREG32(mmRLC_SAVE_AND_RESTORE_BASE, adev->gfx.rlc.save_restore_gpu_addr >> 8);
+			WREG32(mmRLC_CLEAR_STATE_RESTORE_BASE, adev->gfx.rlc.clear_state_gpu_addr >> 8);
+
+		}
+		gfx_v6_0_init_ao_cu_mask(adev);
+		gfx_v6_0_update_gfx_pg(adev, true);
+	} else {
+
+		WREG32(mmRLC_SAVE_AND_RESTORE_BASE, adev->gfx.rlc.save_restore_gpu_addr >> 8);
+		WREG32(mmRLC_CLEAR_STATE_RESTORE_BASE, adev->gfx.rlc.clear_state_gpu_addr >> 8);
+	}
+}
+
+static void gfx_v6_0_fini_pg(struct amdgpu_device *adev)
+{
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_PG |
+			      AMD_PG_SUPPORT_GFX_SMG |
+			      AMD_PG_SUPPORT_GFX_DMG |
+			      AMD_PG_SUPPORT_CP |
+			      AMD_PG_SUPPORT_GDS |
+			      AMD_PG_SUPPORT_RLC_SMU_HS)) {
+		gfx_v6_0_update_gfx_pg(adev, false);
+		if (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG) {
+			gfx_v6_0_enable_cp_pg(adev, false);
+			gfx_v6_0_enable_gds_pg(adev, false);
+		}
+	}
+}
+
+static uint64_t gfx_v6_0_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
+	uint64_t clock;
+
+	mutex_lock(&adev->gfx.gpu_clock_mutex);
+	WREG32(mmRLC_CAPTURE_GPU_CLOCK_COUNT, 1);
+	clock = (uint64_t)RREG32(mmRLC_GPU_CLOCK_COUNT_LSB) |
+	        ((uint64_t)RREG32(mmRLC_GPU_CLOCK_COUNT_MSB) << 32ULL);
+	mutex_unlock(&adev->gfx.gpu_clock_mutex);
+	return clock;
+}
+
+static void gfx_v6_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags)
+{
+	if (flags & AMDGPU_HAVE_CTX_SWITCH)
+		gfx_v6_0_ring_emit_vgt_flush(ring);
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, 0x80000000);
+	amdgpu_ring_write(ring, 0);
+}
+
+
+static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t address)
+{
+	WREG32(mmSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(address << SQ_IND_INDEX__INDEX__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK));
+	return RREG32(mmSQ_IND_DATA);
+}
+
+static void wave_read_regs(struct amdgpu_device *adev, uint32_t simd,
+			   uint32_t wave, uint32_t thread,
+			   uint32_t regno, uint32_t num, uint32_t *out)
+{
+	WREG32(mmSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(regno << SQ_IND_INDEX__INDEX__SHIFT) |
+		(thread << SQ_IND_INDEX__THREAD_ID__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK) |
+		(SQ_IND_INDEX__AUTO_INCR_MASK));
+	while (num--)
+		*(out++) = RREG32(mmSQ_IND_DATA);
+}
+
+static void gfx_v6_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
+{
+	/* type 0 wave data */
+	dst[(*no_fields)++] = 0;
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_STATUS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_HW_ID);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW1);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_GPR_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_LDS_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TRAPSTS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_STS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TBA_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TBA_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TMA_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TMA_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_DBG0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_M0);
+}
+
+static void gfx_v6_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd,
+				     uint32_t wave, uint32_t start,
+				     uint32_t size, uint32_t *dst)
+{
+	wave_read_regs(
+		adev, simd, wave, 0,
+		start + SQIND_WAVE_SGPRS_OFFSET, size, dst);
+}
+
+static void gfx_v6_0_select_me_pipe_q(struct amdgpu_device *adev,
+				  u32 me, u32 pipe, u32 q)
+{
+	DRM_INFO("Not implemented\n");
+}
+
+static const struct amdgpu_gfx_funcs gfx_v6_0_gfx_funcs = {
+	.get_gpu_clock_counter = &gfx_v6_0_get_gpu_clock_counter,
+	.select_se_sh = &gfx_v6_0_select_se_sh,
+	.read_wave_data = &gfx_v6_0_read_wave_data,
+	.read_wave_sgprs = &gfx_v6_0_read_wave_sgprs,
+	.select_me_pipe_q = &gfx_v6_0_select_me_pipe_q
+};
+
+static int gfx_v6_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->gfx.num_gfx_rings = GFX6_NUM_GFX_RINGS;
+	adev->gfx.num_compute_rings = GFX6_NUM_COMPUTE_RINGS;
+	adev->gfx.funcs = &gfx_v6_0_gfx_funcs;
+	gfx_v6_0_set_ring_funcs(adev);
+	gfx_v6_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int gfx_v6_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i, r;
+
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 181, &adev->gfx.eop_irq);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 184, &adev->gfx.priv_reg_irq);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 185, &adev->gfx.priv_inst_irq);
+	if (r)
+		return r;
+
+	gfx_v6_0_scratch_init(adev);
+
+	r = gfx_v6_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load gfx firmware!\n");
+		return r;
+	}
+
+	r = gfx_v6_0_rlc_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init rlc BOs!\n");
+		return r;
+	}
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		ring = &adev->gfx.gfx_ring[i];
+		ring->ring_obj = NULL;
+		snprintf(ring->name, sizeof(ring->name), "gfx");
+		r = amdgpu_ring_init(adev, ring, 1024,
+				     &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP);
+		if (r)
+			return r;
+	}
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		unsigned irq_type;
+
+		if ((i >= 32) || (i >= AMDGPU_MAX_COMPUTE_RINGS)) {
+			DRM_ERROR("Too many (%d) compute rings!\n", i);
+			break;
+		}
+		ring = &adev->gfx.compute_ring[i];
+		ring->ring_obj = NULL;
+		ring->use_doorbell = false;
+		ring->doorbell_index = 0;
+		ring->me = 1;
+		ring->pipe = i;
+		ring->queue = i;
+		snprintf(ring->name, sizeof(ring->name), "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+		irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP + ring->pipe;
+		r = amdgpu_ring_init(adev, ring, 1024,
+				     &adev->gfx.eop_irq, irq_type);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+static int gfx_v6_0_sw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
+
+	gfx_v6_0_rlc_fini(adev);
+
+	return 0;
+}
+
+static int gfx_v6_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gfx_v6_0_gpu_init(adev);
+
+	r = gfx_v6_0_rlc_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v6_0_cp_resume(adev);
+	if (r)
+		return r;
+
+	adev->gfx.ce_ram_size = 0x8000;
+
+	return r;
+}
+
+static int gfx_v6_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gfx_v6_0_cp_enable(adev, false);
+	gfx_v6_0_rlc_stop(adev);
+	gfx_v6_0_fini_pg(adev);
+
+	return 0;
+}
+
+static int gfx_v6_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return gfx_v6_0_hw_fini(adev);
+}
+
+static int gfx_v6_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return gfx_v6_0_hw_init(adev);
+}
+
+static bool gfx_v6_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (RREG32(mmGRBM_STATUS) & GRBM_STATUS__GUI_ACTIVE_MASK)
+		return false;
+	else
+		return true;
+}
+
+static int gfx_v6_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (gfx_v6_0_is_idle(handle))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int gfx_v6_0_soft_reset(void *handle)
+{
+	return 0;
+}
+
+static void gfx_v6_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
+						 enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl &= ~CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl |= CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v6_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev,
+						     int ring,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+	switch (state){
+	case AMDGPU_IRQ_STATE_DISABLE:
+		if (ring == 0) {
+			cp_int_cntl = RREG32(mmCP_INT_CNTL_RING1);
+			cp_int_cntl &= ~CP_INT_CNTL_RING1__TIME_STAMP_INT_ENABLE_MASK;
+			WREG32(mmCP_INT_CNTL_RING1, cp_int_cntl);
+			break;
+		} else {
+			cp_int_cntl = RREG32(mmCP_INT_CNTL_RING2);
+			cp_int_cntl &= ~CP_INT_CNTL_RING2__TIME_STAMP_INT_ENABLE_MASK;
+			WREG32(mmCP_INT_CNTL_RING2, cp_int_cntl);
+			break;
+
+		}
+	case AMDGPU_IRQ_STATE_ENABLE:
+		if (ring == 0) {
+			cp_int_cntl = RREG32(mmCP_INT_CNTL_RING1);
+			cp_int_cntl |= CP_INT_CNTL_RING1__TIME_STAMP_INT_ENABLE_MASK;
+			WREG32(mmCP_INT_CNTL_RING1, cp_int_cntl);
+			break;
+		} else {
+			cp_int_cntl = RREG32(mmCP_INT_CNTL_RING2);
+			cp_int_cntl |= CP_INT_CNTL_RING2__TIME_STAMP_INT_ENABLE_MASK;
+			WREG32(mmCP_INT_CNTL_RING2, cp_int_cntl);
+			break;
+
+		}
+
+	default:
+		BUG();
+		break;
+
+	}
+}
+
+static int gfx_v6_0_set_priv_reg_fault_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl &= ~CP_INT_CNTL_RING0__PRIV_REG_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl |= CP_INT_CNTL_RING0__PRIV_REG_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v6_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *src,
+					      unsigned type,
+					      enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl &= ~CP_INT_CNTL_RING0__PRIV_INSTR_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl |= CP_INT_CNTL_RING0__PRIV_INSTR_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v6_0_set_eop_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CP_IRQ_GFX_EOP:
+		gfx_v6_0_set_gfx_eop_interrupt_state(adev, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
+		gfx_v6_0_set_compute_eop_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP:
+		gfx_v6_0_set_compute_eop_interrupt_state(adev, 1, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v6_0_eop_irq(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *source,
+			    struct amdgpu_iv_entry *entry)
+{
+	switch (entry->ring_id) {
+	case 0:
+		amdgpu_fence_process(&adev->gfx.gfx_ring[0]);
+		break;
+	case 1:
+	case 2:
+		amdgpu_fence_process(&adev->gfx.compute_ring[entry->ring_id - 1]);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v6_0_priv_reg_irq(struct amdgpu_device *adev,
+				 struct amdgpu_irq_src *source,
+				 struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal register access in command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int gfx_v6_0_priv_inst_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int gfx_v6_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	bool gate = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_CG_STATE_GATE)
+		gate = true;
+
+	gfx_v6_0_enable_gui_idle_interrupt(adev, false);
+	if (gate) {
+		gfx_v6_0_enable_mgcg(adev, true);
+		gfx_v6_0_enable_cgcg(adev, true);
+	} else {
+		gfx_v6_0_enable_cgcg(adev, false);
+		gfx_v6_0_enable_mgcg(adev, false);
+	}
+	gfx_v6_0_enable_gui_idle_interrupt(adev, true);
+
+	return 0;
+}
+
+static int gfx_v6_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	bool gate = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_PG_STATE_GATE)
+		gate = true;
+
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_PG |
+			      AMD_PG_SUPPORT_GFX_SMG |
+			      AMD_PG_SUPPORT_GFX_DMG |
+			      AMD_PG_SUPPORT_CP |
+			      AMD_PG_SUPPORT_GDS |
+			      AMD_PG_SUPPORT_RLC_SMU_HS)) {
+		gfx_v6_0_update_gfx_pg(adev, gate);
+		if (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG) {
+			gfx_v6_0_enable_cp_pg(adev, gate);
+			gfx_v6_0_enable_gds_pg(adev, gate);
+		}
+	}
+
+	return 0;
+}
+
+static const struct amd_ip_funcs gfx_v6_0_ip_funcs = {
+	.name = "gfx_v6_0",
+	.early_init = gfx_v6_0_early_init,
+	.late_init = NULL,
+	.sw_init = gfx_v6_0_sw_init,
+	.sw_fini = gfx_v6_0_sw_fini,
+	.hw_init = gfx_v6_0_hw_init,
+	.hw_fini = gfx_v6_0_hw_fini,
+	.suspend = gfx_v6_0_suspend,
+	.resume = gfx_v6_0_resume,
+	.is_idle = gfx_v6_0_is_idle,
+	.wait_for_idle = gfx_v6_0_wait_for_idle,
+	.soft_reset = gfx_v6_0_soft_reset,
+	.set_clockgating_state = gfx_v6_0_set_clockgating_state,
+	.set_powergating_state = gfx_v6_0_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs gfx_v6_0_ring_funcs_gfx = {
+	.type = AMDGPU_RING_TYPE_GFX,
+	.align_mask = 0xff,
+	.nop = 0x80000000,
+	.support_64bit_ptrs = false,
+	.get_rptr = gfx_v6_0_ring_get_rptr,
+	.get_wptr = gfx_v6_0_ring_get_wptr,
+	.set_wptr = gfx_v6_0_ring_set_wptr_gfx,
+	.emit_frame_size =
+		5 + 5 + /* hdp flush / invalidate */
+		14 + 14 + 14 + /* gfx_v6_0_ring_emit_fence x3 for user fence, vm fence */
+		7 + 4 + /* gfx_v6_0_ring_emit_pipeline_sync */
+		SI_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + 6 + /* gfx_v6_0_ring_emit_vm_flush */
+		3 + 2, /* gfx_v6_ring_emit_cntxcntl including vgt flush */
+	.emit_ib_size = 6, /* gfx_v6_0_ring_emit_ib */
+	.emit_ib = gfx_v6_0_ring_emit_ib,
+	.emit_fence = gfx_v6_0_ring_emit_fence,
+	.emit_pipeline_sync = gfx_v6_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v6_0_ring_emit_vm_flush,
+	.test_ring = gfx_v6_0_ring_test_ring,
+	.test_ib = gfx_v6_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.emit_cntxcntl = gfx_v6_ring_emit_cntxcntl,
+	.emit_wreg = gfx_v6_0_ring_emit_wreg,
+};
+
+static const struct amdgpu_ring_funcs gfx_v6_0_ring_funcs_compute = {
+	.type = AMDGPU_RING_TYPE_COMPUTE,
+	.align_mask = 0xff,
+	.nop = 0x80000000,
+	.get_rptr = gfx_v6_0_ring_get_rptr,
+	.get_wptr = gfx_v6_0_ring_get_wptr,
+	.set_wptr = gfx_v6_0_ring_set_wptr_compute,
+	.emit_frame_size =
+		5 + 5 + /* hdp flush / invalidate */
+		7 + /* gfx_v6_0_ring_emit_pipeline_sync */
+		SI_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + /* gfx_v6_0_ring_emit_vm_flush */
+		14 + 14 + 14, /* gfx_v6_0_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 6, /* gfx_v6_0_ring_emit_ib */
+	.emit_ib = gfx_v6_0_ring_emit_ib,
+	.emit_fence = gfx_v6_0_ring_emit_fence,
+	.emit_pipeline_sync = gfx_v6_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v6_0_ring_emit_vm_flush,
+	.test_ring = gfx_v6_0_ring_test_ring,
+	.test_ib = gfx_v6_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.emit_wreg = gfx_v6_0_ring_emit_wreg,
+};
+
+static void gfx_v6_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		adev->gfx.gfx_ring[i].funcs = &gfx_v6_0_ring_funcs_gfx;
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		adev->gfx.compute_ring[i].funcs = &gfx_v6_0_ring_funcs_compute;
+}
+
+static const struct amdgpu_irq_src_funcs gfx_v6_0_eop_irq_funcs = {
+	.set = gfx_v6_0_set_eop_interrupt_state,
+	.process = gfx_v6_0_eop_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v6_0_priv_reg_irq_funcs = {
+	.set = gfx_v6_0_set_priv_reg_fault_state,
+	.process = gfx_v6_0_priv_reg_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v6_0_priv_inst_irq_funcs = {
+	.set = gfx_v6_0_set_priv_inst_fault_state,
+	.process = gfx_v6_0_priv_inst_irq,
+};
+
+static void gfx_v6_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.eop_irq.num_types = AMDGPU_CP_IRQ_LAST;
+	adev->gfx.eop_irq.funcs = &gfx_v6_0_eop_irq_funcs;
+
+	adev->gfx.priv_reg_irq.num_types = 1;
+	adev->gfx.priv_reg_irq.funcs = &gfx_v6_0_priv_reg_irq_funcs;
+
+	adev->gfx.priv_inst_irq.num_types = 1;
+	adev->gfx.priv_inst_irq.funcs = &gfx_v6_0_priv_inst_irq_funcs;
+}
+
+static void gfx_v6_0_get_cu_info(struct amdgpu_device *adev)
+{
+	int i, j, k, counter, active_cu_number = 0;
+	u32 mask, bitmap, ao_bitmap, ao_cu_mask = 0;
+	struct amdgpu_cu_info *cu_info = &adev->gfx.cu_info;
+	unsigned disable_masks[4 * 2];
+	u32 ao_cu_num;
+
+	if (adev->flags & AMD_IS_APU)
+		ao_cu_num = 2;
+	else
+		ao_cu_num = adev->gfx.config.max_cu_per_sh;
+
+	memset(cu_info, 0, sizeof(*cu_info));
+
+	amdgpu_gfx_parse_disable_cu(disable_masks, 4, 2);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			mask = 1;
+			ao_bitmap = 0;
+			counter = 0;
+			gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff);
+			if (i < 4 && j < 2)
+				gfx_v6_0_set_user_cu_inactive_bitmap(
+					adev, disable_masks[i * 2 + j]);
+			bitmap = gfx_v6_0_get_cu_enabled(adev);
+			cu_info->bitmap[i][j] = bitmap;
+
+			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) {
+				if (bitmap & mask) {
+					if (counter < ao_cu_num)
+						ao_bitmap |= mask;
+					counter ++;
+				}
+				mask <<= 1;
+			}
+			active_cu_number += counter;
+			if (i < 2 && j < 2)
+				ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8));
+			cu_info->ao_cu_bitmap[i][j] = ao_bitmap;
+		}
+	}
+
+	gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	cu_info->number = active_cu_number;
+	cu_info->ao_cu_mask = ao_cu_mask;
+}
+
+const struct amdgpu_ip_block_version gfx_v6_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GFX,
+	.major = 6,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gfx_v6_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/gfx_v6_0.h b/sys/dev/pci/drm/amd/amdgpu/gfx_v6_0.h
new file mode 100644
index 00000000000..ced6fc42f68
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gfx_v6_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V6_0_H__
+#define __GFX_V6_0_H__
+
+extern const struct amdgpu_ip_block_version gfx_v6_0_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/gfx_v7_0.c b/sys/dev/pci/drm/amd/amdgpu/gfx_v7_0.c
new file mode 100644
index 00000000000..6768f81061a
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gfx_v7_0.c
@@ -0,0 +1,5278 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_gfx.h"
+#include "cikd.h"
+#include "cik.h"
+#include "cik_structs.h"
+#include "atom.h"
+#include "amdgpu_ucode.h"
+#include "clearstate_ci.h"
+
+#include "dce/dce_8_0_d.h"
+#include "dce/dce_8_0_sh_mask.h"
+
+#include "bif/bif_4_1_d.h"
+#include "bif/bif_4_1_sh_mask.h"
+
+#include "gca/gfx_7_0_d.h"
+#include "gca/gfx_7_2_enum.h"
+#include "gca/gfx_7_2_sh_mask.h"
+
+#include "gmc/gmc_7_0_d.h"
+#include "gmc/gmc_7_0_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+#define NUM_SIMD_PER_CU 0x4 /* missing from the gfx_7 IP headers */
+
+#define GFX7_NUM_GFX_RINGS     1
+#define GFX7_MEC_HPD_SIZE      2048
+
+static void gfx_v7_0_set_ring_funcs(struct amdgpu_device *adev);
+static void gfx_v7_0_set_irq_funcs(struct amdgpu_device *adev);
+static void gfx_v7_0_set_gds_init(struct amdgpu_device *adev);
+
+MODULE_FIRMWARE("amdgpu/bonaire_pfp.bin");
+MODULE_FIRMWARE("amdgpu/bonaire_me.bin");
+MODULE_FIRMWARE("amdgpu/bonaire_ce.bin");
+MODULE_FIRMWARE("amdgpu/bonaire_rlc.bin");
+MODULE_FIRMWARE("amdgpu/bonaire_mec.bin");
+
+MODULE_FIRMWARE("amdgpu/hawaii_pfp.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_me.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_ce.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_rlc.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_mec.bin");
+
+MODULE_FIRMWARE("amdgpu/kaveri_pfp.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_me.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_ce.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_rlc.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_mec.bin");
+MODULE_FIRMWARE("amdgpu/kaveri_mec2.bin");
+
+MODULE_FIRMWARE("amdgpu/kabini_pfp.bin");
+MODULE_FIRMWARE("amdgpu/kabini_me.bin");
+MODULE_FIRMWARE("amdgpu/kabini_ce.bin");
+MODULE_FIRMWARE("amdgpu/kabini_rlc.bin");
+MODULE_FIRMWARE("amdgpu/kabini_mec.bin");
+
+MODULE_FIRMWARE("amdgpu/mullins_pfp.bin");
+MODULE_FIRMWARE("amdgpu/mullins_me.bin");
+MODULE_FIRMWARE("amdgpu/mullins_ce.bin");
+MODULE_FIRMWARE("amdgpu/mullins_rlc.bin");
+MODULE_FIRMWARE("amdgpu/mullins_mec.bin");
+
+static const struct amdgpu_gds_reg_offset amdgpu_gds_reg_offset[] =
+{
+	{mmGDS_VMID0_BASE, mmGDS_VMID0_SIZE, mmGDS_GWS_VMID0, mmGDS_OA_VMID0},
+	{mmGDS_VMID1_BASE, mmGDS_VMID1_SIZE, mmGDS_GWS_VMID1, mmGDS_OA_VMID1},
+	{mmGDS_VMID2_BASE, mmGDS_VMID2_SIZE, mmGDS_GWS_VMID2, mmGDS_OA_VMID2},
+	{mmGDS_VMID3_BASE, mmGDS_VMID3_SIZE, mmGDS_GWS_VMID3, mmGDS_OA_VMID3},
+	{mmGDS_VMID4_BASE, mmGDS_VMID4_SIZE, mmGDS_GWS_VMID4, mmGDS_OA_VMID4},
+	{mmGDS_VMID5_BASE, mmGDS_VMID5_SIZE, mmGDS_GWS_VMID5, mmGDS_OA_VMID5},
+	{mmGDS_VMID6_BASE, mmGDS_VMID6_SIZE, mmGDS_GWS_VMID6, mmGDS_OA_VMID6},
+	{mmGDS_VMID7_BASE, mmGDS_VMID7_SIZE, mmGDS_GWS_VMID7, mmGDS_OA_VMID7},
+	{mmGDS_VMID8_BASE, mmGDS_VMID8_SIZE, mmGDS_GWS_VMID8, mmGDS_OA_VMID8},
+	{mmGDS_VMID9_BASE, mmGDS_VMID9_SIZE, mmGDS_GWS_VMID9, mmGDS_OA_VMID9},
+	{mmGDS_VMID10_BASE, mmGDS_VMID10_SIZE, mmGDS_GWS_VMID10, mmGDS_OA_VMID10},
+	{mmGDS_VMID11_BASE, mmGDS_VMID11_SIZE, mmGDS_GWS_VMID11, mmGDS_OA_VMID11},
+	{mmGDS_VMID12_BASE, mmGDS_VMID12_SIZE, mmGDS_GWS_VMID12, mmGDS_OA_VMID12},
+	{mmGDS_VMID13_BASE, mmGDS_VMID13_SIZE, mmGDS_GWS_VMID13, mmGDS_OA_VMID13},
+	{mmGDS_VMID14_BASE, mmGDS_VMID14_SIZE, mmGDS_GWS_VMID14, mmGDS_OA_VMID14},
+	{mmGDS_VMID15_BASE, mmGDS_VMID15_SIZE, mmGDS_GWS_VMID15, mmGDS_OA_VMID15}
+};
+
+static const u32 spectre_rlc_save_restore_register_list[] =
+{
+	(0x0e00 << 16) | (0xc12c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc140 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc150 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc15c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc168 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc170 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc178 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc204 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2b8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2bc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2c0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8228 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x829c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x869c >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x98f4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x98f8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9900 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc260 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x90e8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c000 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c00c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c1c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9700 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x89bc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8900 >> 2),
+	0x00000000,
+	0x3,
+	(0x0e00 << 16) | (0xc130 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc134 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc1fc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc208 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc264 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc268 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc26c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc270 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc274 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc278 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc27c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc280 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc284 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc288 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc28c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc290 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc294 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc298 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc29c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2a0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2a4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2a8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2ac  >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2b0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x301d0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30238 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30250 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30254 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30258 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3025c >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc99c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9834 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f00 >> 2),
+	0x00000000,
+	(0x0001 << 16) | (0x30f00 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f04 >> 2),
+	0x00000000,
+	(0x0001 << 16) | (0x30f04 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f08 >> 2),
+	0x00000000,
+	(0x0001 << 16) | (0x30f08 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f0c >> 2),
+	0x00000000,
+	(0x0001 << 16) | (0x30f0c >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x9b7c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8a14 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8a18 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a00 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8bf0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8bcc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8b24 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30a04 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a10 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a14 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a18 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a2c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc700 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc704 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc708 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc768 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc770 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc774 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc778 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc77c >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc780 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc784 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc788 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc78c >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc798 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc79c >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc7a0 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc7a4 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc7a8 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc7ac >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc7b0 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc7b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9100 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c010 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92a8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92ac >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92b8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92bc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92c0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92c4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92c8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92cc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92d0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c00 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c04 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c20 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c38 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c3c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xae00 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9604 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac08 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac0c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac10 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac14 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac58 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac68 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac6c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac70 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac74 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac78 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac7c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac80 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac84 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac88 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac8c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x970c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9714 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9718 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x971c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xcd10 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xcd14 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88b0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88b8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88bc >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0x89c0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88c4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88c8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88d0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88d4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88d8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8980 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30938 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3093c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30940 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x89a0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30900 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30904 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x89b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c210 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c214 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c218 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8904 >> 2),
+	0x00000000,
+	0x5,
+	(0x0e00 << 16) | (0x8c28 >> 2),
+	(0x0e00 << 16) | (0x8c2c >> 2),
+	(0x0e00 << 16) | (0x8c30 >> 2),
+	(0x0e00 << 16) | (0x8c34 >> 2),
+	(0x0e00 << 16) | (0x9600 >> 2),
+};
+
+static const u32 kalindi_rlc_save_restore_register_list[] =
+{
+	(0x0e00 << 16) | (0xc12c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc140 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc150 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc15c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc168 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc170 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc204 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2b8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2bc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2c0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8228 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x829c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x869c >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x98f4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x98f8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9900 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc260 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x90e8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c000 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c00c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c1c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9700 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x89bc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8900 >> 2),
+	0x00000000,
+	0x3,
+	(0x0e00 << 16) | (0xc130 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc134 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc1fc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc208 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc264 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc268 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc26c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc270 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc274 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc28c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc290 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc294 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc298 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2a0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2a4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2a8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2ac >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x301d0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30238 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30250 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30254 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30258 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3025c >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc99c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9834 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f00 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f04 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f08 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f0c >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x9b7c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8a14 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8a18 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a00 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8bf0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8bcc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8b24 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30a04 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a10 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a14 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a18 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a2c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc700 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc704 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc708 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc768 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc770 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc774 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc798 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc79c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9100 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c010 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c00 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c04 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c20 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c38 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c3c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xae00 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9604 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac08 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac0c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac10 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac14 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac58 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac68 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac6c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac70 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac74 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac78 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac7c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac80 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac84 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac88 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac8c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x970c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9714 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9718 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x971c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xcd10 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xcd14 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88b0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88b8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88bc >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0x89c0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88c4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88c8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88d0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88d4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88d8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8980 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30938 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3093c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30940 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x89a0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30900 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30904 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x89b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3e1fc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c210 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c214 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c218 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8904 >> 2),
+	0x00000000,
+	0x5,
+	(0x0e00 << 16) | (0x8c28 >> 2),
+	(0x0e00 << 16) | (0x8c2c >> 2),
+	(0x0e00 << 16) | (0x8c30 >> 2),
+	(0x0e00 << 16) | (0x8c34 >> 2),
+	(0x0e00 << 16) | (0x9600 >> 2),
+};
+
+static u32 gfx_v7_0_get_csb_size(struct amdgpu_device *adev);
+static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, volatile u32 *buffer);
+static void gfx_v7_0_init_cp_pg_table(struct amdgpu_device *adev);
+static void gfx_v7_0_init_pg(struct amdgpu_device *adev);
+static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev);
+
+/*
+ * Core functions
+ */
+/**
+ * gfx_v7_0_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int gfx_v7_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		chip_name = "bonaire";
+		break;
+	case CHIP_HAWAII:
+		chip_name = "hawaii";
+		break;
+	case CHIP_KAVERI:
+		chip_name = "kaveri";
+		break;
+	case CHIP_KABINI:
+		chip_name = "kabini";
+		break;
+	case CHIP_MULLINS:
+		chip_name = "mullins";
+		break;
+	default: BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_pfp.bin", chip_name);
+	err = request_firmware(&adev->gfx.pfp_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.pfp_fw);
+	if (err)
+		goto out;
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_me.bin", chip_name);
+	err = request_firmware(&adev->gfx.me_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.me_fw);
+	if (err)
+		goto out;
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ce.bin", chip_name);
+	err = request_firmware(&adev->gfx.ce_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.ce_fw);
+	if (err)
+		goto out;
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec.bin", chip_name);
+	err = request_firmware(&adev->gfx.mec_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.mec_fw);
+	if (err)
+		goto out;
+
+	if (adev->asic_type == CHIP_KAVERI) {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec2.bin", chip_name);
+		err = request_firmware(&adev->gfx.mec2_fw, fw_name, adev->dev);
+		if (err)
+			goto out;
+		err = amdgpu_ucode_validate(adev->gfx.mec2_fw);
+		if (err)
+			goto out;
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc.bin", chip_name);
+	err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.rlc_fw);
+
+out:
+	if (err) {
+		pr_err("gfx7: Failed to load firmware \"%s\"\n", fw_name);
+		release_firmware(adev->gfx.pfp_fw);
+		adev->gfx.pfp_fw = NULL;
+		release_firmware(adev->gfx.me_fw);
+		adev->gfx.me_fw = NULL;
+		release_firmware(adev->gfx.ce_fw);
+		adev->gfx.ce_fw = NULL;
+		release_firmware(adev->gfx.mec_fw);
+		adev->gfx.mec_fw = NULL;
+		release_firmware(adev->gfx.mec2_fw);
+		adev->gfx.mec2_fw = NULL;
+		release_firmware(adev->gfx.rlc_fw);
+		adev->gfx.rlc_fw = NULL;
+	}
+	return err;
+}
+
+static void gfx_v7_0_free_microcode(struct amdgpu_device *adev)
+{
+	release_firmware(adev->gfx.pfp_fw);
+	adev->gfx.pfp_fw = NULL;
+	release_firmware(adev->gfx.me_fw);
+	adev->gfx.me_fw = NULL;
+	release_firmware(adev->gfx.ce_fw);
+	adev->gfx.ce_fw = NULL;
+	release_firmware(adev->gfx.mec_fw);
+	adev->gfx.mec_fw = NULL;
+	release_firmware(adev->gfx.mec2_fw);
+	adev->gfx.mec2_fw = NULL;
+	release_firmware(adev->gfx.rlc_fw);
+	adev->gfx.rlc_fw = NULL;
+}
+
+/**
+ * gfx_v7_0_tiling_mode_table_init - init the hw tiling table
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Starting with SI, the tiling setup is done globally in a
+ * set of 32 tiling modes.  Rather than selecting each set of
+ * parameters per surface as on older asics, we just select
+ * which index in the tiling table we want to use, and the
+ * surface uses those parameters (CIK).
+ */
+static void gfx_v7_0_tiling_mode_table_init(struct amdgpu_device *adev)
+{
+	const u32 num_tile_mode_states =
+			ARRAY_SIZE(adev->gfx.config.tile_mode_array);
+	const u32 num_secondary_tile_mode_states =
+			ARRAY_SIZE(adev->gfx.config.macrotile_mode_array);
+	u32 reg_offset, split_equal_to_row_size;
+	uint32_t *tile, *macrotile;
+
+	tile = adev->gfx.config.tile_mode_array;
+	macrotile = adev->gfx.config.macrotile_mode_array;
+
+	switch (adev->gfx.config.mem_row_size_in_kb) {
+	case 1:
+		split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_1KB;
+		break;
+	case 2:
+	default:
+		split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_2KB;
+		break;
+	case 4:
+		split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_4KB;
+		break;
+	}
+
+	for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+		tile[reg_offset] = 0;
+	for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++)
+		macrotile[reg_offset] = 0;
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		tile[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			   TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			   TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			   TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			   TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+			   TILE_SPLIT(split_equal_to_row_size));
+		tile[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+			   TILE_SPLIT(split_equal_to_row_size));
+		tile[7] = (TILE_SPLIT(split_equal_to_row_size));
+		tile[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+			   PIPE_CONFIG(ADDR_SURF_P4_16x16));
+		tile[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING));
+		tile[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		tile[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		tile[12] = (TILE_SPLIT(split_equal_to_row_size));
+		tile[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING));
+		tile[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		tile[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		tile[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		tile[17] = (TILE_SPLIT(split_equal_to_row_size));
+		tile[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING));
+		tile[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[21] =  (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[23] = (TILE_SPLIT(split_equal_to_row_size));
+		tile[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING));
+		tile[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		tile[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		tile[30] = (TILE_SPLIT(split_equal_to_row_size));
+
+		macrotile[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		macrotile[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK));
+		macrotile[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		macrotile[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK));
+
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			WREG32(mmGB_TILE_MODE0 + reg_offset, tile[reg_offset]);
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++)
+			if (reg_offset != 7)
+				WREG32(mmGB_MACROTILE_MODE0 + reg_offset, macrotile[reg_offset]);
+		break;
+	case CHIP_HAWAII:
+		tile[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			   TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			   TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			   TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			   TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+			   TILE_SPLIT(split_equal_to_row_size));
+		tile[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+			   TILE_SPLIT(split_equal_to_row_size));
+		tile[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+			   TILE_SPLIT(split_equal_to_row_size));
+		tile[7] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+			   TILE_SPLIT(split_equal_to_row_size));
+		tile[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+			   PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16));
+		tile[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING));
+		tile[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		tile[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		tile[12] = (ARRAY_MODE(ARRAY_PRT_2D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		tile[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING));
+		tile[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		tile[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		tile[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		tile[17] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		tile[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING));
+		tile[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[23] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING));
+		tile[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		tile[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		tile[30] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+
+		macrotile[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		macrotile[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK));
+		macrotile[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK));
+		macrotile[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		macrotile[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		macrotile[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK));
+
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			WREG32(mmGB_TILE_MODE0 + reg_offset, tile[reg_offset]);
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++)
+			if (reg_offset != 7)
+				WREG32(mmGB_MACROTILE_MODE0 + reg_offset, macrotile[reg_offset]);
+		break;
+	case CHIP_KABINI:
+	case CHIP_KAVERI:
+	case CHIP_MULLINS:
+	default:
+		tile[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P2) |
+			   TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P2) |
+			   TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P2) |
+			   TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P2) |
+			   TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P2) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+			   TILE_SPLIT(split_equal_to_row_size));
+		tile[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P2) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		tile[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P2) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+			   TILE_SPLIT(split_equal_to_row_size));
+		tile[7] = (TILE_SPLIT(split_equal_to_row_size));
+		tile[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+			   PIPE_CONFIG(ADDR_SURF_P2));
+		tile[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+			   PIPE_CONFIG(ADDR_SURF_P2) |
+			   MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING));
+		tile[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		tile[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		tile[12] = (TILE_SPLIT(split_equal_to_row_size));
+		tile[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING));
+		tile[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		tile[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		tile[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		tile[17] = (TILE_SPLIT(split_equal_to_row_size));
+		tile[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING));
+		tile[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[23] = (TILE_SPLIT(split_equal_to_row_size));
+		tile[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		tile[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING));
+		tile[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		tile[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+			    PIPE_CONFIG(ADDR_SURF_P2) |
+			    MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+			    SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		tile[30] = (TILE_SPLIT(split_equal_to_row_size));
+
+		macrotile[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		macrotile[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		macrotile[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		macrotile[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		macrotile[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		macrotile[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		macrotile[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		macrotile[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		macrotile[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			WREG32(mmGB_TILE_MODE0 + reg_offset, tile[reg_offset]);
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++)
+			if (reg_offset != 7)
+				WREG32(mmGB_MACROTILE_MODE0 + reg_offset, macrotile[reg_offset]);
+		break;
+	}
+}
+
+/**
+ * gfx_v7_0_select_se_sh - select which SE, SH to address
+ *
+ * @adev: amdgpu_device pointer
+ * @se_num: shader engine to address
+ * @sh_num: sh block to address
+ *
+ * Select which SE, SH combinations to address. Certain
+ * registers are instanced per SE or SH.  0xffffffff means
+ * broadcast to all SEs or SHs (CIK).
+ */
+static void gfx_v7_0_select_se_sh(struct amdgpu_device *adev,
+				  u32 se_num, u32 sh_num, u32 instance)
+{
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX, instance);
+
+	if ((se_num == 0xffffffff) && (sh_num == 0xffffffff))
+		data |= GRBM_GFX_INDEX__SH_BROADCAST_WRITES_MASK |
+			GRBM_GFX_INDEX__SE_BROADCAST_WRITES_MASK;
+	else if (se_num == 0xffffffff)
+		data |= GRBM_GFX_INDEX__SE_BROADCAST_WRITES_MASK |
+			(sh_num << GRBM_GFX_INDEX__SH_INDEX__SHIFT);
+	else if (sh_num == 0xffffffff)
+		data |= GRBM_GFX_INDEX__SH_BROADCAST_WRITES_MASK |
+			(se_num << GRBM_GFX_INDEX__SE_INDEX__SHIFT);
+	else
+		data |= (sh_num << GRBM_GFX_INDEX__SH_INDEX__SHIFT) |
+			(se_num << GRBM_GFX_INDEX__SE_INDEX__SHIFT);
+	WREG32(mmGRBM_GFX_INDEX, data);
+}
+
+/**
+ * gfx_v7_0_get_rb_active_bitmap - computes the mask of enabled RBs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Calculates the bitmask of enabled RBs (CIK).
+ * Returns the enabled RB bitmask.
+ */
+static u32 gfx_v7_0_get_rb_active_bitmap(struct amdgpu_device *adev)
+{
+	u32 data, mask;
+
+	data = RREG32(mmCC_RB_BACKEND_DISABLE);
+	data |= RREG32(mmGC_USER_RB_BACKEND_DISABLE);
+
+	data &= CC_RB_BACKEND_DISABLE__BACKEND_DISABLE_MASK;
+	data >>= GC_USER_RB_BACKEND_DISABLE__BACKEND_DISABLE__SHIFT;
+
+	mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_backends_per_se /
+					 adev->gfx.config.max_sh_per_se);
+
+	return (~data) & mask;
+}
+
+static void
+gfx_v7_0_raster_config(struct amdgpu_device *adev, u32 *rconf, u32 *rconf1)
+{
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		*rconf |= RB_MAP_PKR0(2) | RB_XSEL2(1) | SE_MAP(2) |
+			  SE_XSEL(1) | SE_YSEL(1);
+		*rconf1 |= 0x0;
+		break;
+	case CHIP_HAWAII:
+		*rconf |= RB_MAP_PKR0(2) | RB_MAP_PKR1(2) |
+			  RB_XSEL2(1) | PKR_MAP(2) | PKR_XSEL(1) |
+			  PKR_YSEL(1) | SE_MAP(2) | SE_XSEL(2) |
+			  SE_YSEL(3);
+		*rconf1 |= SE_PAIR_MAP(2) | SE_PAIR_XSEL(3) |
+			   SE_PAIR_YSEL(2);
+		break;
+	case CHIP_KAVERI:
+		*rconf |= RB_MAP_PKR0(2);
+		*rconf1 |= 0x0;
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		*rconf |= 0x0;
+		*rconf1 |= 0x0;
+		break;
+	default:
+		DRM_ERROR("unknown asic: 0x%x\n", adev->asic_type);
+		break;
+	}
+}
+
+static void
+gfx_v7_0_write_harvested_raster_configs(struct amdgpu_device *adev,
+					u32 raster_config, u32 raster_config_1,
+					unsigned rb_mask, unsigned num_rb)
+{
+	unsigned sh_per_se = max_t(unsigned, adev->gfx.config.max_sh_per_se, 1);
+	unsigned num_se = max_t(unsigned, adev->gfx.config.max_shader_engines, 1);
+	unsigned rb_per_pkr = min_t(unsigned, num_rb / num_se / sh_per_se, 2);
+	unsigned rb_per_se = num_rb / num_se;
+	unsigned se_mask[4];
+	unsigned se;
+
+	se_mask[0] = ((1 << rb_per_se) - 1) & rb_mask;
+	se_mask[1] = (se_mask[0] << rb_per_se) & rb_mask;
+	se_mask[2] = (se_mask[1] << rb_per_se) & rb_mask;
+	se_mask[3] = (se_mask[2] << rb_per_se) & rb_mask;
+
+	WARN_ON(!(num_se == 1 || num_se == 2 || num_se == 4));
+	WARN_ON(!(sh_per_se == 1 || sh_per_se == 2));
+	WARN_ON(!(rb_per_pkr == 1 || rb_per_pkr == 2));
+
+	if ((num_se > 2) && ((!se_mask[0] && !se_mask[1]) ||
+			     (!se_mask[2] && !se_mask[3]))) {
+		raster_config_1 &= ~SE_PAIR_MAP_MASK;
+
+		if (!se_mask[0] && !se_mask[1]) {
+			raster_config_1 |=
+				SE_PAIR_MAP(RASTER_CONFIG_SE_PAIR_MAP_3);
+		} else {
+			raster_config_1 |=
+				SE_PAIR_MAP(RASTER_CONFIG_SE_PAIR_MAP_0);
+		}
+	}
+
+	for (se = 0; se < num_se; se++) {
+		unsigned raster_config_se = raster_config;
+		unsigned pkr0_mask = ((1 << rb_per_pkr) - 1) << (se * rb_per_se);
+		unsigned pkr1_mask = pkr0_mask << rb_per_pkr;
+		int idx = (se / 2) * 2;
+
+		if ((num_se > 1) && (!se_mask[idx] || !se_mask[idx + 1])) {
+			raster_config_se &= ~SE_MAP_MASK;
+
+			if (!se_mask[idx]) {
+				raster_config_se |= SE_MAP(RASTER_CONFIG_SE_MAP_3);
+			} else {
+				raster_config_se |= SE_MAP(RASTER_CONFIG_SE_MAP_0);
+			}
+		}
+
+		pkr0_mask &= rb_mask;
+		pkr1_mask &= rb_mask;
+		if (rb_per_se > 2 && (!pkr0_mask || !pkr1_mask)) {
+			raster_config_se &= ~PKR_MAP_MASK;
+
+			if (!pkr0_mask) {
+				raster_config_se |= PKR_MAP(RASTER_CONFIG_PKR_MAP_3);
+			} else {
+				raster_config_se |= PKR_MAP(RASTER_CONFIG_PKR_MAP_0);
+			}
+		}
+
+		if (rb_per_se >= 2) {
+			unsigned rb0_mask = 1 << (se * rb_per_se);
+			unsigned rb1_mask = rb0_mask << 1;
+
+			rb0_mask &= rb_mask;
+			rb1_mask &= rb_mask;
+			if (!rb0_mask || !rb1_mask) {
+				raster_config_se &= ~RB_MAP_PKR0_MASK;
+
+				if (!rb0_mask) {
+					raster_config_se |=
+						RB_MAP_PKR0(RASTER_CONFIG_RB_MAP_3);
+				} else {
+					raster_config_se |=
+						RB_MAP_PKR0(RASTER_CONFIG_RB_MAP_0);
+				}
+			}
+
+			if (rb_per_se > 2) {
+				rb0_mask = 1 << (se * rb_per_se + rb_per_pkr);
+				rb1_mask = rb0_mask << 1;
+				rb0_mask &= rb_mask;
+				rb1_mask &= rb_mask;
+				if (!rb0_mask || !rb1_mask) {
+					raster_config_se &= ~RB_MAP_PKR1_MASK;
+
+					if (!rb0_mask) {
+						raster_config_se |=
+							RB_MAP_PKR1(RASTER_CONFIG_RB_MAP_3);
+					} else {
+						raster_config_se |=
+							RB_MAP_PKR1(RASTER_CONFIG_RB_MAP_0);
+					}
+				}
+			}
+		}
+
+		/* GRBM_GFX_INDEX has a different offset on CI+ */
+		gfx_v7_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff);
+		WREG32(mmPA_SC_RASTER_CONFIG, raster_config_se);
+		WREG32(mmPA_SC_RASTER_CONFIG_1, raster_config_1);
+	}
+
+	/* GRBM_GFX_INDEX has a different offset on CI+ */
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+}
+
+/**
+ * gfx_v7_0_setup_rb - setup the RBs on the asic
+ *
+ * @adev: amdgpu_device pointer
+ * @se_num: number of SEs (shader engines) for the asic
+ * @sh_per_se: number of SH blocks per SE for the asic
+ *
+ * Configures per-SE/SH RB registers (CIK).
+ */
+static void gfx_v7_0_setup_rb(struct amdgpu_device *adev)
+{
+	int i, j;
+	u32 data;
+	u32 raster_config = 0, raster_config_1 = 0;
+	u32 active_rbs = 0;
+	u32 rb_bitmap_width_per_sh = adev->gfx.config.max_backends_per_se /
+					adev->gfx.config.max_sh_per_se;
+	unsigned num_rb_pipes;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
+			data = gfx_v7_0_get_rb_active_bitmap(adev);
+			active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
+					       rb_bitmap_width_per_sh);
+		}
+	}
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+
+	adev->gfx.config.backend_enable_mask = active_rbs;
+	adev->gfx.config.num_rbs = hweight32(active_rbs);
+
+	num_rb_pipes = min_t(unsigned, adev->gfx.config.max_backends_per_se *
+			     adev->gfx.config.max_shader_engines, 16);
+
+	gfx_v7_0_raster_config(adev, &raster_config, &raster_config_1);
+
+	if (!adev->gfx.config.backend_enable_mask ||
+			adev->gfx.config.num_rbs >= num_rb_pipes) {
+		WREG32(mmPA_SC_RASTER_CONFIG, raster_config);
+		WREG32(mmPA_SC_RASTER_CONFIG_1, raster_config_1);
+	} else {
+		gfx_v7_0_write_harvested_raster_configs(adev, raster_config, raster_config_1,
+							adev->gfx.config.backend_enable_mask,
+							num_rb_pipes);
+	}
+
+	/* cache the values for userspace */
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
+			adev->gfx.config.rb_config[i][j].rb_backend_disable =
+				RREG32(mmCC_RB_BACKEND_DISABLE);
+			adev->gfx.config.rb_config[i][j].user_rb_backend_disable =
+				RREG32(mmGC_USER_RB_BACKEND_DISABLE);
+			adev->gfx.config.rb_config[i][j].raster_config =
+				RREG32(mmPA_SC_RASTER_CONFIG);
+			adev->gfx.config.rb_config[i][j].raster_config_1 =
+				RREG32(mmPA_SC_RASTER_CONFIG_1);
+		}
+	}
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+/**
+ * gfx_v7_0_init_compute_vmid - gart enable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize compute vmid sh_mem registers
+ *
+ */
+#define DEFAULT_SH_MEM_BASES	(0x6000)
+#define FIRST_COMPUTE_VMID	(8)
+#define LAST_COMPUTE_VMID	(16)
+static void gfx_v7_0_init_compute_vmid(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t sh_mem_config;
+	uint32_t sh_mem_bases;
+
+	/*
+	 * Configure apertures:
+	 * LDS:         0x60000000'00000000 - 0x60000001'00000000 (4GB)
+	 * Scratch:     0x60000001'00000000 - 0x60000002'00000000 (4GB)
+	 * GPUVM:       0x60010000'00000000 - 0x60020000'00000000 (1TB)
+	*/
+	sh_mem_bases = DEFAULT_SH_MEM_BASES | (DEFAULT_SH_MEM_BASES << 16);
+	sh_mem_config = SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
+			SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT;
+	sh_mem_config |= MTYPE_NONCACHED << SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT;
+	mutex_lock(&adev->srbm_mutex);
+	for (i = FIRST_COMPUTE_VMID; i < LAST_COMPUTE_VMID; i++) {
+		cik_srbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32(mmSH_MEM_CONFIG, sh_mem_config);
+		WREG32(mmSH_MEM_APE1_BASE, 1);
+		WREG32(mmSH_MEM_APE1_LIMIT, 0);
+		WREG32(mmSH_MEM_BASES, sh_mem_bases);
+	}
+	cik_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void gfx_v7_0_config_init(struct amdgpu_device *adev)
+{
+	adev->gfx.config.double_offchip_lds_buf = 1;
+}
+
+/**
+ * gfx_v7_0_gpu_init - setup the 3D engine
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Configures the 3D engine and tiling configuration
+ * registers so that the 3D engine is usable.
+ */
+static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
+{
+	u32 sh_mem_cfg, sh_static_mem_cfg, sh_mem_base;
+	u32 tmp;
+	int i;
+
+	WREG32(mmGRBM_CNTL, (0xff << GRBM_CNTL__READ_TIMEOUT__SHIFT));
+
+	WREG32(mmGB_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmHDP_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmDMIF_ADDR_CALC, adev->gfx.config.gb_addr_config);
+
+	gfx_v7_0_tiling_mode_table_init(adev);
+
+	gfx_v7_0_setup_rb(adev);
+	gfx_v7_0_get_cu_info(adev);
+	gfx_v7_0_config_init(adev);
+
+	/* set HW defaults for 3D engine */
+	WREG32(mmCP_MEQ_THRESHOLDS,
+	       (0x30 << CP_MEQ_THRESHOLDS__MEQ1_START__SHIFT) |
+	       (0x60 << CP_MEQ_THRESHOLDS__MEQ2_START__SHIFT));
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	/*
+	 * making sure that the following register writes will be broadcasted
+	 * to all the shaders
+	 */
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+
+	/* XXX SH_MEM regs */
+	/* where to put LDS, scratch, GPUVM in FSA64 space */
+	sh_mem_cfg = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE,
+				   SH_MEM_ALIGNMENT_MODE_UNALIGNED);
+	sh_mem_cfg = REG_SET_FIELD(sh_mem_cfg, SH_MEM_CONFIG, DEFAULT_MTYPE,
+				   MTYPE_NC);
+	sh_mem_cfg = REG_SET_FIELD(sh_mem_cfg, SH_MEM_CONFIG, APE1_MTYPE,
+				   MTYPE_UC);
+	sh_mem_cfg = REG_SET_FIELD(sh_mem_cfg, SH_MEM_CONFIG, PRIVATE_ATC, 0);
+
+	sh_static_mem_cfg = REG_SET_FIELD(0, SH_STATIC_MEM_CONFIG,
+				   SWIZZLE_ENABLE, 1);
+	sh_static_mem_cfg = REG_SET_FIELD(sh_static_mem_cfg, SH_STATIC_MEM_CONFIG,
+				   ELEMENT_SIZE, 1);
+	sh_static_mem_cfg = REG_SET_FIELD(sh_static_mem_cfg, SH_STATIC_MEM_CONFIG,
+				   INDEX_STRIDE, 3);
+	WREG32(mmSH_STATIC_MEM_CONFIG, sh_static_mem_cfg);
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->vm_manager.id_mgr[0].num_ids; i++) {
+		if (i == 0)
+			sh_mem_base = 0;
+		else
+			sh_mem_base = adev->gmc.shared_aperture_start >> 48;
+		cik_srbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32(mmSH_MEM_CONFIG, sh_mem_cfg);
+		WREG32(mmSH_MEM_APE1_BASE, 1);
+		WREG32(mmSH_MEM_APE1_LIMIT, 0);
+		WREG32(mmSH_MEM_BASES, sh_mem_base);
+	}
+	cik_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	gfx_v7_0_init_compute_vmid(adev);
+
+	WREG32(mmSX_DEBUG_1, 0x20);
+
+	WREG32(mmTA_CNTL_AUX, 0x00010000);
+
+	tmp = RREG32(mmSPI_CONFIG_CNTL);
+	tmp |= 0x03000000;
+	WREG32(mmSPI_CONFIG_CNTL, tmp);
+
+	WREG32(mmSQ_CONFIG, 1);
+
+	WREG32(mmDB_DEBUG, 0);
+
+	tmp = RREG32(mmDB_DEBUG2) & ~0xf00fffff;
+	tmp |= 0x00000400;
+	WREG32(mmDB_DEBUG2, tmp);
+
+	tmp = RREG32(mmDB_DEBUG3) & ~0x0002021c;
+	tmp |= 0x00020200;
+	WREG32(mmDB_DEBUG3, tmp);
+
+	tmp = RREG32(mmCB_HW_CONTROL) & ~0x00010000;
+	tmp |= 0x00018208;
+	WREG32(mmCB_HW_CONTROL, tmp);
+
+	WREG32(mmSPI_CONFIG_CNTL_1, (4 << SPI_CONFIG_CNTL_1__VTX_DONE_DELAY__SHIFT));
+
+	WREG32(mmPA_SC_FIFO_SIZE,
+		((adev->gfx.config.sc_prim_fifo_size_frontend << PA_SC_FIFO_SIZE__SC_FRONTEND_PRIM_FIFO_SIZE__SHIFT) |
+		(adev->gfx.config.sc_prim_fifo_size_backend << PA_SC_FIFO_SIZE__SC_BACKEND_PRIM_FIFO_SIZE__SHIFT) |
+		(adev->gfx.config.sc_hiz_tile_fifo_size << PA_SC_FIFO_SIZE__SC_HIZ_TILE_FIFO_SIZE__SHIFT) |
+		(adev->gfx.config.sc_earlyz_tile_fifo_size << PA_SC_FIFO_SIZE__SC_EARLYZ_TILE_FIFO_SIZE__SHIFT)));
+
+	WREG32(mmVGT_NUM_INSTANCES, 1);
+
+	WREG32(mmCP_PERFMON_CNTL, 0);
+
+	WREG32(mmSQ_CONFIG, 0);
+
+	WREG32(mmPA_SC_FORCE_EOV_MAX_CNTS,
+		((4095 << PA_SC_FORCE_EOV_MAX_CNTS__FORCE_EOV_MAX_CLK_CNT__SHIFT) |
+		(255 << PA_SC_FORCE_EOV_MAX_CNTS__FORCE_EOV_MAX_REZ_CNT__SHIFT)));
+
+	WREG32(mmVGT_CACHE_INVALIDATION,
+		(VC_AND_TC << VGT_CACHE_INVALIDATION__CACHE_INVALIDATION__SHIFT) |
+		(ES_AND_GS_AUTO << VGT_CACHE_INVALIDATION__AUTO_INVLD_EN__SHIFT));
+
+	WREG32(mmVGT_GS_VERTEX_REUSE, 16);
+	WREG32(mmPA_SC_LINE_STIPPLE_STATE, 0);
+
+	WREG32(mmPA_CL_ENHANCE, PA_CL_ENHANCE__CLIP_VTX_REORDER_ENA_MASK |
+			(3 << PA_CL_ENHANCE__NUM_CLIP_SEQ__SHIFT));
+	WREG32(mmPA_SC_ENHANCE, PA_SC_ENHANCE__ENABLE_PA_SC_OUT_OF_ORDER_MASK);
+
+	tmp = RREG32(mmSPI_ARB_PRIORITY);
+	tmp = REG_SET_FIELD(tmp, SPI_ARB_PRIORITY, PIPE_ORDER_TS0, 2);
+	tmp = REG_SET_FIELD(tmp, SPI_ARB_PRIORITY, PIPE_ORDER_TS1, 2);
+	tmp = REG_SET_FIELD(tmp, SPI_ARB_PRIORITY, PIPE_ORDER_TS2, 2);
+	tmp = REG_SET_FIELD(tmp, SPI_ARB_PRIORITY, PIPE_ORDER_TS3, 2);
+	WREG32(mmSPI_ARB_PRIORITY, tmp);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	udelay(50);
+}
+
+/*
+ * GPU scratch registers helpers function.
+ */
+/**
+ * gfx_v7_0_scratch_init - setup driver info for CP scratch regs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the number and offset of the CP scratch registers.
+ * NOTE: use of CP scratch registers is a legacy inferface and
+ * is not used by default on newer asics (r6xx+).  On newer asics,
+ * memory buffers are used for fences rather than scratch regs.
+ */
+static void gfx_v7_0_scratch_init(struct amdgpu_device *adev)
+{
+	adev->gfx.scratch.num_reg = 8;
+	adev->gfx.scratch.reg_base = mmSCRATCH_REG0;
+	adev->gfx.scratch.free_mask = (1u << adev->gfx.scratch.num_reg) - 1;
+}
+
+/**
+ * gfx_v7_0_ring_test_ring - basic gfx ring test
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Allocate a scratch register and write to it using the gfx ring (CIK).
+ * Provides a basic gfx ring test to verify that the ring is working.
+ * Used by gfx_v7_0_cp_gfx_resume();
+ * Returns 0 on success, error on failure.
+ */
+static int gfx_v7_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t scratch;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_gfx_scratch_free(adev, scratch);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+	amdgpu_ring_write(ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
+			  ring->idx, scratch, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_gfx_scratch_free(adev, scratch);
+	return r;
+}
+
+/**
+ * gfx_v7_0_ring_emit_hdp - emit an hdp flush on the cp
+ *
+ * @adev: amdgpu_device pointer
+ * @ridx: amdgpu ring index
+ *
+ * Emits an hdp flush on the cp.
+ */
+static void gfx_v7_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	u32 ref_and_mask;
+	int usepfp = ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE ? 0 : 1;
+
+	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) {
+		switch (ring->me) {
+		case 1:
+			ref_and_mask = GPU_HDP_FLUSH_DONE__CP2_MASK << ring->pipe;
+			break;
+		case 2:
+			ref_and_mask = GPU_HDP_FLUSH_DONE__CP6_MASK << ring->pipe;
+			break;
+		default:
+			return;
+		}
+	} else {
+		ref_and_mask = GPU_HDP_FLUSH_DONE__CP0_MASK;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_OPERATION(1) | /* write, wait, write */
+				 WAIT_REG_MEM_FUNCTION(3) |  /* == */
+				 WAIT_REG_MEM_ENGINE(usepfp)));   /* pfp or me */
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_REQ);
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_DONE);
+	amdgpu_ring_write(ring, ref_and_mask);
+	amdgpu_ring_write(ring, ref_and_mask);
+	amdgpu_ring_write(ring, 0x20); /* poll interval */
+}
+
+static void gfx_v7_0_ring_emit_vgt_flush(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE, 0));
+	amdgpu_ring_write(ring, EVENT_TYPE(VS_PARTIAL_FLUSH) |
+		EVENT_INDEX(4));
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE, 0));
+	amdgpu_ring_write(ring, EVENT_TYPE(VGT_FLUSH) |
+		EVENT_INDEX(0));
+}
+
+/**
+ * gfx_v7_0_ring_emit_fence_gfx - emit a fence on the gfx ring
+ *
+ * @adev: amdgpu_device pointer
+ * @fence: amdgpu fence object
+ *
+ * Emits a fence sequnce number on the gfx ring and flushes
+ * GPU caches.
+ */
+static void gfx_v7_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr,
+					 u64 seq, unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+	/* Workaround for cache flush problems. First send a dummy EOP
+	 * event down the pipe with seq one below.
+	 */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE_EOP, 4));
+	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
+				 EOP_TC_ACTION_EN |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) |
+				DATA_SEL(1) | INT_SEL(0));
+	amdgpu_ring_write(ring, lower_32_bits(seq - 1));
+	amdgpu_ring_write(ring, upper_32_bits(seq - 1));
+
+	/* Then send the real EOP event down the pipe. */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE_EOP, 4));
+	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
+				 EOP_TC_ACTION_EN |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) |
+				DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+}
+
+/**
+ * gfx_v7_0_ring_emit_fence_compute - emit a fence on the compute ring
+ *
+ * @adev: amdgpu_device pointer
+ * @fence: amdgpu fence object
+ *
+ * Emits a fence sequnce number on the compute ring and flushes
+ * GPU caches.
+ */
+static void gfx_v7_0_ring_emit_fence_compute(struct amdgpu_ring *ring,
+					     u64 addr, u64 seq,
+					     unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+
+	/* RELEASE_MEM - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RELEASE_MEM, 5));
+	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
+				 EOP_TC_ACTION_EN |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+}
+
+/*
+ * IB stuff
+ */
+/**
+ * gfx_v7_0_ring_emit_ib - emit an IB (Indirect Buffer) on the ring
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @ib: amdgpu indirect buffer object
+ *
+ * Emits an DE (drawing engine) or CE (constant engine) IB
+ * on the gfx ring.  IBs are usually generated by userspace
+ * acceleration drivers and submitted to the kernel for
+ * sheduling on the ring.  This function schedules the IB
+ * on the gfx ring for execution by the GPU.
+ */
+static void gfx_v7_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
+				      struct amdgpu_ib *ib,
+				      unsigned vmid, bool ctx_switch)
+{
+	u32 header, control = 0;
+
+	/* insert SWITCH_BUFFER packet before first IB in the ring frame */
+	if (ctx_switch) {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+
+	if (ib->flags & AMDGPU_IB_FLAG_CE)
+		header = PACKET3(PACKET3_INDIRECT_BUFFER_CONST, 2);
+	else
+		header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
+
+	control |= ib->length_dw | (vmid << 24);
+
+	amdgpu_ring_write(ring, header);
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+			  (2 << 0) |
+#endif
+			  (ib->gpu_addr & 0xFFFFFFFC));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFFFF);
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v7_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
+					  struct amdgpu_ib *ib,
+					  unsigned vmid, bool ctx_switch)
+{
+	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+					  (2 << 0) |
+#endif
+					  (ib->gpu_addr & 0xFFFFFFFC));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFFFF);
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v7_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags)
+{
+	uint32_t dw2 = 0;
+
+	dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */
+	if (flags & AMDGPU_HAVE_CTX_SWITCH) {
+		gfx_v7_0_ring_emit_vgt_flush(ring);
+		/* set load_global_config & load_global_uconfig */
+		dw2 |= 0x8001;
+		/* set load_cs_sh_regs */
+		dw2 |= 0x01000000;
+		/* set load_per_context_state & load_gfx_sh_regs */
+		dw2 |= 0x10002;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, dw2);
+	amdgpu_ring_write(ring, 0);
+}
+
+/**
+ * gfx_v7_0_ring_test_ib - basic ring IB test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Allocate an IB and execute it on the gfx ring (CIK).
+ * Provides a basic gfx ring test to verify that IBs are working.
+ * Returns 0 on success, error on failure.
+ */
+static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	uint32_t scratch;
+	uint32_t tmp = 0;
+	long r;
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get scratch reg (%ld).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, 256, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err1;
+	}
+	ib.ptr[0] = PACKET3(PACKET3_SET_UCONFIG_REG, 1);
+	ib.ptr[1] = ((scratch - PACKET3_SET_UCONFIG_REG_START));
+	ib.ptr[2] = 0xDEADBEEF;
+	ib.length_dw = 3;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err2;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
+		goto err2;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto err2;
+	}
+	tmp = RREG32(scratch);
+	if (tmp == 0xDEADBEEF) {
+		DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
+	} else {
+		DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
+			  scratch, tmp);
+		r = -EINVAL;
+	}
+
+err2:
+	amdgpu_ib_free(adev, &ib, NULL);
+	dma_fence_put(f);
+err1:
+	amdgpu_gfx_scratch_free(adev, scratch);
+	return r;
+}
+
+/*
+ * CP.
+ * On CIK, gfx and compute now have independant command processors.
+ *
+ * GFX
+ * Gfx consists of a single ring and can process both gfx jobs and
+ * compute jobs.  The gfx CP consists of three microengines (ME):
+ * PFP - Pre-Fetch Parser
+ * ME - Micro Engine
+ * CE - Constant Engine
+ * The PFP and ME make up what is considered the Drawing Engine (DE).
+ * The CE is an asynchronous engine used for updating buffer desciptors
+ * used by the DE so that they can be loaded into cache in parallel
+ * while the DE is processing state update packets.
+ *
+ * Compute
+ * The compute CP consists of two microengines (ME):
+ * MEC1 - Compute MicroEngine 1
+ * MEC2 - Compute MicroEngine 2
+ * Each MEC supports 4 compute pipes and each pipe supports 8 queues.
+ * The queues are exposed to userspace and are programmed directly
+ * by the compute runtime.
+ */
+/**
+ * gfx_v7_0_cp_gfx_enable - enable/disable the gfx CP MEs
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable or disable the MEs
+ *
+ * Halts or unhalts the gfx MEs.
+ */
+static void gfx_v7_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+
+	if (enable) {
+		WREG32(mmCP_ME_CNTL, 0);
+	} else {
+		WREG32(mmCP_ME_CNTL, (CP_ME_CNTL__ME_HALT_MASK | CP_ME_CNTL__PFP_HALT_MASK | CP_ME_CNTL__CE_HALT_MASK));
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+			adev->gfx.gfx_ring[i].ready = false;
+	}
+	udelay(50);
+}
+
+/**
+ * gfx_v7_0_cp_gfx_load_microcode - load the gfx CP ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the gfx PFP, ME, and CE ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int gfx_v7_0_cp_gfx_load_microcode(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v1_0 *pfp_hdr;
+	const struct gfx_firmware_header_v1_0 *ce_hdr;
+	const struct gfx_firmware_header_v1_0 *me_hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.me_fw || !adev->gfx.pfp_fw || !adev->gfx.ce_fw)
+		return -EINVAL;
+
+	pfp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+	ce_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+	me_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&pfp_hdr->header);
+	amdgpu_ucode_print_gfx_hdr(&ce_hdr->header);
+	amdgpu_ucode_print_gfx_hdr(&me_hdr->header);
+	adev->gfx.pfp_fw_version = le32_to_cpu(pfp_hdr->header.ucode_version);
+	adev->gfx.ce_fw_version = le32_to_cpu(ce_hdr->header.ucode_version);
+	adev->gfx.me_fw_version = le32_to_cpu(me_hdr->header.ucode_version);
+	adev->gfx.me_feature_version = le32_to_cpu(me_hdr->ucode_feature_version);
+	adev->gfx.ce_feature_version = le32_to_cpu(ce_hdr->ucode_feature_version);
+	adev->gfx.pfp_feature_version = le32_to_cpu(pfp_hdr->ucode_feature_version);
+
+	gfx_v7_0_cp_gfx_enable(adev, false);
+
+	/* PFP */
+	fw_data = (const __le32 *)
+		(adev->gfx.pfp_fw->data +
+		 le32_to_cpu(pfp_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(pfp_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_PFP_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_PFP_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_PFP_UCODE_ADDR, adev->gfx.pfp_fw_version);
+
+	/* CE */
+	fw_data = (const __le32 *)
+		(adev->gfx.ce_fw->data +
+		 le32_to_cpu(ce_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(ce_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_CE_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_CE_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_CE_UCODE_ADDR, adev->gfx.ce_fw_version);
+
+	/* ME */
+	fw_data = (const __le32 *)
+		(adev->gfx.me_fw->data +
+		 le32_to_cpu(me_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(me_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_ME_RAM_WADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_ME_RAM_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_ME_RAM_WADDR, adev->gfx.me_fw_version);
+
+	return 0;
+}
+
+/**
+ * gfx_v7_0_cp_gfx_start - start the gfx ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enables the ring and loads the clear state context and other
+ * packets required to init the ring.
+ * Returns 0 for success, error for failure.
+ */
+static int gfx_v7_0_cp_gfx_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->gfx.gfx_ring[0];
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+	int r, i;
+
+	/* init the CP */
+	WREG32(mmCP_MAX_CONTEXT, adev->gfx.config.max_hw_contexts - 1);
+	WREG32(mmCP_ENDIAN_SWAP, 0);
+	WREG32(mmCP_DEVICE_ID, 1);
+
+	gfx_v7_0_cp_gfx_enable(adev, true);
+
+	r = amdgpu_ring_alloc(ring, gfx_v7_0_get_csb_size(adev) + 8);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
+		return r;
+	}
+
+	/* init the CE partitions.  CE only used for gfx on CIK */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_BASE, 2));
+	amdgpu_ring_write(ring, PACKET3_BASE_INDEX(CE_PARTITION_BASE));
+	amdgpu_ring_write(ring, 0x8000);
+	amdgpu_ring_write(ring, 0x8000);
+
+	/* clear state buffer */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, 0x80000000);
+	amdgpu_ring_write(ring, 0x80000000);
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				amdgpu_ring_write(ring,
+						  PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
+				amdgpu_ring_write(ring, ext->reg_index - PACKET3_SET_CONTEXT_REG_START);
+				for (i = 0; i < ext->reg_count; i++)
+					amdgpu_ring_write(ring, ext->extent[i]);
+			}
+		}
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONTEXT_REG, 2));
+	amdgpu_ring_write(ring, mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
+	amdgpu_ring_write(ring, adev->gfx.config.rb_config[0][0].raster_config);
+	amdgpu_ring_write(ring, adev->gfx.config.rb_config[0][0].raster_config_1);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CLEAR_STATE, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONTEXT_REG, 2));
+	amdgpu_ring_write(ring, 0x00000316);
+	amdgpu_ring_write(ring, 0x0000000e); /* VGT_VERTEX_REUSE_BLOCK_CNTL */
+	amdgpu_ring_write(ring, 0x00000010); /* VGT_OUT_DEALLOC_CNTL */
+
+	amdgpu_ring_commit(ring);
+
+	return 0;
+}
+
+/**
+ * gfx_v7_0_cp_gfx_resume - setup the gfx ring buffer registers
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Program the location and size of the gfx ring buffer
+ * and test it to make sure it's working.
+ * Returns 0 for success, error for failure.
+ */
+static int gfx_v7_0_cp_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 tmp;
+	u32 rb_bufsz;
+	u64 rb_addr, rptr_addr;
+	int r;
+
+	WREG32(mmCP_SEM_WAIT_TIMER, 0x0);
+	if (adev->asic_type != CHIP_HAWAII)
+		WREG32(mmCP_SEM_INCOMPLETE_TIMER_CNTL, 0x0);
+
+	/* Set the write pointer delay */
+	WREG32(mmCP_RB_WPTR_DELAY, 0);
+
+	/* set the RB to use vmid 0 */
+	WREG32(mmCP_RB_VMID, 0);
+
+	WREG32(mmSCRATCH_ADDR, 0);
+
+	/* ring 0 - compute and gfx */
+	/* Set ring buffer size */
+	ring = &adev->gfx.gfx_ring[0];
+	rb_bufsz = order_base_2(ring->ring_size / 8);
+	tmp = (order_base_2(AMDGPU_GPU_PAGE_SIZE/8) << 8) | rb_bufsz;
+#ifdef __BIG_ENDIAN
+	tmp |= 2 << CP_RB0_CNTL__BUF_SWAP__SHIFT;
+#endif
+	WREG32(mmCP_RB0_CNTL, tmp);
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(mmCP_RB0_CNTL, tmp | CP_RB0_CNTL__RB_RPTR_WR_ENA_MASK);
+	ring->wptr = 0;
+	WREG32(mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
+
+	/* set the wb address wether it's enabled or not */
+	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	WREG32(mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
+	WREG32(mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF);
+
+	/* scratch register shadowing is no longer supported */
+	WREG32(mmSCRATCH_UMSK, 0);
+
+	mdelay(1);
+	WREG32(mmCP_RB0_CNTL, tmp);
+
+	rb_addr = ring->gpu_addr >> 8;
+	WREG32(mmCP_RB0_BASE, rb_addr);
+	WREG32(mmCP_RB0_BASE_HI, upper_32_bits(rb_addr));
+
+	/* start the ring */
+	gfx_v7_0_cp_gfx_start(adev);
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		return r;
+	}
+
+	return 0;
+}
+
+static u64 gfx_v7_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	return ring->adev->wb.wb[ring->rptr_offs];
+}
+
+static u64 gfx_v7_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmCP_RB0_WPTR);
+}
+
+static void gfx_v7_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32(mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
+	(void)RREG32(mmCP_RB0_WPTR);
+}
+
+static u64 gfx_v7_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
+{
+	/* XXX check if swapping is necessary on BE */
+	return ring->adev->wb.wb[ring->wptr_offs];
+}
+
+static void gfx_v7_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* XXX check if swapping is necessary on BE */
+	adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
+	WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+}
+
+/**
+ * gfx_v7_0_cp_compute_enable - enable/disable the compute CP MEs
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable or disable the MEs
+ *
+ * Halts or unhalts the compute MEs.
+ */
+static void gfx_v7_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+
+	if (enable) {
+		WREG32(mmCP_MEC_CNTL, 0);
+	} else {
+		WREG32(mmCP_MEC_CNTL, (CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK));
+		for (i = 0; i < adev->gfx.num_compute_rings; i++)
+			adev->gfx.compute_ring[i].ready = false;
+	}
+	udelay(50);
+}
+
+/**
+ * gfx_v7_0_cp_compute_load_microcode - load the compute CP ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the compute MEC1&2 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int gfx_v7_0_cp_compute_load_microcode(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v1_0 *mec_hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.mec_fw)
+		return -EINVAL;
+
+	mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+	amdgpu_ucode_print_gfx_hdr(&mec_hdr->header);
+	adev->gfx.mec_fw_version = le32_to_cpu(mec_hdr->header.ucode_version);
+	adev->gfx.mec_feature_version = le32_to_cpu(
+					mec_hdr->ucode_feature_version);
+
+	gfx_v7_0_cp_compute_enable(adev, false);
+
+	/* MEC1 */
+	fw_data = (const __le32 *)
+		(adev->gfx.mec_fw->data +
+		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_MEC_ME1_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_MEC_ME1_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_MEC_ME1_UCODE_ADDR, 0);
+
+	if (adev->asic_type == CHIP_KAVERI) {
+		const struct gfx_firmware_header_v1_0 *mec2_hdr;
+
+		if (!adev->gfx.mec2_fw)
+			return -EINVAL;
+
+		mec2_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
+		amdgpu_ucode_print_gfx_hdr(&mec2_hdr->header);
+		adev->gfx.mec2_fw_version = le32_to_cpu(mec2_hdr->header.ucode_version);
+		adev->gfx.mec2_feature_version = le32_to_cpu(
+				mec2_hdr->ucode_feature_version);
+
+		/* MEC2 */
+		fw_data = (const __le32 *)
+			(adev->gfx.mec2_fw->data +
+			 le32_to_cpu(mec2_hdr->header.ucode_array_offset_bytes));
+		fw_size = le32_to_cpu(mec2_hdr->header.ucode_size_bytes) / 4;
+		WREG32(mmCP_MEC_ME2_UCODE_ADDR, 0);
+		for (i = 0; i < fw_size; i++)
+			WREG32(mmCP_MEC_ME2_UCODE_DATA, le32_to_cpup(fw_data++));
+		WREG32(mmCP_MEC_ME2_UCODE_ADDR, 0);
+	}
+
+	return 0;
+}
+
+/**
+ * gfx_v7_0_cp_compute_fini - stop the compute queues
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute queues and tear down the driver queue
+ * info.
+ */
+static void gfx_v7_0_cp_compute_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
+
+		amdgpu_bo_free_kernel(&ring->mqd_obj, NULL, NULL);
+	}
+}
+
+static void gfx_v7_0_mec_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
+}
+
+static int gfx_v7_0_mec_init(struct amdgpu_device *adev)
+{
+	int r;
+	u32 *hpd;
+	size_t mec_hpd_size;
+
+	bitmap_zero(adev->gfx.mec.queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+
+	/* take ownership of the relevant compute queues */
+	amdgpu_gfx_compute_queue_acquire(adev);
+
+	/* allocate space for ALL pipes (even the ones we don't own) */
+	mec_hpd_size = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe_per_mec
+		* GFX7_MEC_HPD_SIZE * 2;
+
+	r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.mec.hpd_eop_obj,
+				      &adev->gfx.mec.hpd_eop_gpu_addr,
+				      (void **)&hpd);
+	if (r) {
+		dev_warn(adev->dev, "(%d) create, pin or map of HDP EOP bo failed\n", r);
+		gfx_v7_0_mec_fini(adev);
+		return r;
+	}
+
+	/* clear memory.  Not sure if this is required or not */
+	memset(hpd, 0, mec_hpd_size);
+
+	amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
+	amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+
+	return 0;
+}
+
+struct hqd_registers
+{
+	u32 cp_mqd_base_addr;
+	u32 cp_mqd_base_addr_hi;
+	u32 cp_hqd_active;
+	u32 cp_hqd_vmid;
+	u32 cp_hqd_persistent_state;
+	u32 cp_hqd_pipe_priority;
+	u32 cp_hqd_queue_priority;
+	u32 cp_hqd_quantum;
+	u32 cp_hqd_pq_base;
+	u32 cp_hqd_pq_base_hi;
+	u32 cp_hqd_pq_rptr;
+	u32 cp_hqd_pq_rptr_report_addr;
+	u32 cp_hqd_pq_rptr_report_addr_hi;
+	u32 cp_hqd_pq_wptr_poll_addr;
+	u32 cp_hqd_pq_wptr_poll_addr_hi;
+	u32 cp_hqd_pq_doorbell_control;
+	u32 cp_hqd_pq_wptr;
+	u32 cp_hqd_pq_control;
+	u32 cp_hqd_ib_base_addr;
+	u32 cp_hqd_ib_base_addr_hi;
+	u32 cp_hqd_ib_rptr;
+	u32 cp_hqd_ib_control;
+	u32 cp_hqd_iq_timer;
+	u32 cp_hqd_iq_rptr;
+	u32 cp_hqd_dequeue_request;
+	u32 cp_hqd_dma_offload;
+	u32 cp_hqd_sema_cmd;
+	u32 cp_hqd_msg_type;
+	u32 cp_hqd_atomic0_preop_lo;
+	u32 cp_hqd_atomic0_preop_hi;
+	u32 cp_hqd_atomic1_preop_lo;
+	u32 cp_hqd_atomic1_preop_hi;
+	u32 cp_hqd_hq_scheduler0;
+	u32 cp_hqd_hq_scheduler1;
+	u32 cp_mqd_control;
+};
+
+static void gfx_v7_0_compute_pipe_init(struct amdgpu_device *adev,
+				       int mec, int pipe)
+{
+	u64 eop_gpu_addr;
+	u32 tmp;
+	size_t eop_offset = (mec * adev->gfx.mec.num_pipe_per_mec + pipe)
+			    * GFX7_MEC_HPD_SIZE * 2;
+
+	mutex_lock(&adev->srbm_mutex);
+	eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr + eop_offset;
+
+	cik_srbm_select(adev, mec + 1, pipe, 0, 0);
+
+	/* write the EOP addr */
+	WREG32(mmCP_HPD_EOP_BASE_ADDR, eop_gpu_addr >> 8);
+	WREG32(mmCP_HPD_EOP_BASE_ADDR_HI, upper_32_bits(eop_gpu_addr) >> 8);
+
+	/* set the VMID assigned */
+	WREG32(mmCP_HPD_EOP_VMID, 0);
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	tmp = RREG32(mmCP_HPD_EOP_CONTROL);
+	tmp &= ~CP_HPD_EOP_CONTROL__EOP_SIZE_MASK;
+	tmp |= order_base_2(GFX7_MEC_HPD_SIZE / 8);
+	WREG32(mmCP_HPD_EOP_CONTROL, tmp);
+
+	cik_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static int gfx_v7_0_mqd_deactivate(struct amdgpu_device *adev)
+{
+	int i;
+
+	/* disable the queue if it's active */
+	if (RREG32(mmCP_HQD_ACTIVE) & 1) {
+		WREG32(mmCP_HQD_DEQUEUE_REQUEST, 1);
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32(mmCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+
+		if (i == adev->usec_timeout)
+			return -ETIMEDOUT;
+
+		WREG32(mmCP_HQD_DEQUEUE_REQUEST, 0);
+		WREG32(mmCP_HQD_PQ_RPTR, 0);
+		WREG32(mmCP_HQD_PQ_WPTR, 0);
+	}
+
+	return 0;
+}
+
+static void gfx_v7_0_mqd_init(struct amdgpu_device *adev,
+			     struct cik_mqd *mqd,
+			     uint64_t mqd_gpu_addr,
+			     struct amdgpu_ring *ring)
+{
+	u64 hqd_gpu_addr;
+	u64 wb_gpu_addr;
+
+	/* init the mqd struct */
+	memset(mqd, 0, sizeof(struct cik_mqd));
+
+	mqd->header = 0xC0310800;
+	mqd->compute_static_thread_mgmt_se0 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se1 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se2 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se3 = 0xffffffff;
+
+	/* enable doorbell? */
+	mqd->cp_hqd_pq_doorbell_control =
+		RREG32(mmCP_HQD_PQ_DOORBELL_CONTROL);
+	if (ring->use_doorbell)
+		mqd->cp_hqd_pq_doorbell_control |= CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_EN_MASK;
+	else
+		mqd->cp_hqd_pq_doorbell_control &= ~CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_EN_MASK;
+
+	/* set the pointer to the MQD */
+	mqd->cp_mqd_base_addr_lo = mqd_gpu_addr & 0xfffffffc;
+	mqd->cp_mqd_base_addr_hi = upper_32_bits(mqd_gpu_addr);
+
+	/* set MQD vmid to 0 */
+	mqd->cp_mqd_control = RREG32(mmCP_MQD_CONTROL);
+	mqd->cp_mqd_control &= ~CP_MQD_CONTROL__VMID_MASK;
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	hqd_gpu_addr = ring->gpu_addr >> 8;
+	mqd->cp_hqd_pq_base_lo = hqd_gpu_addr;
+	mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	mqd->cp_hqd_pq_control = RREG32(mmCP_HQD_PQ_CONTROL);
+	mqd->cp_hqd_pq_control &=
+		~(CP_HQD_PQ_CONTROL__QUEUE_SIZE_MASK |
+				CP_HQD_PQ_CONTROL__RPTR_BLOCK_SIZE_MASK);
+
+	mqd->cp_hqd_pq_control |=
+		order_base_2(ring->ring_size / 8);
+	mqd->cp_hqd_pq_control |=
+		(order_base_2(AMDGPU_GPU_PAGE_SIZE/8) << 8);
+#ifdef __BIG_ENDIAN
+	mqd->cp_hqd_pq_control |=
+		2 << CP_HQD_PQ_CONTROL__ENDIAN_SWAP__SHIFT;
+#endif
+	mqd->cp_hqd_pq_control &=
+		~(CP_HQD_PQ_CONTROL__UNORD_DISPATCH_MASK |
+				CP_HQD_PQ_CONTROL__ROQ_PQ_IB_FLIP_MASK |
+				CP_HQD_PQ_CONTROL__PQ_VOLATILE_MASK);
+	mqd->cp_hqd_pq_control |=
+		CP_HQD_PQ_CONTROL__PRIV_STATE_MASK |
+		CP_HQD_PQ_CONTROL__KMD_QUEUE_MASK; /* assuming kernel queue control */
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	wb_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* set the wb address wether it's enabled or not */
+	wb_gpu_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_rptr_report_addr_hi =
+		upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* enable the doorbell if requested */
+	if (ring->use_doorbell) {
+		mqd->cp_hqd_pq_doorbell_control =
+			RREG32(mmCP_HQD_PQ_DOORBELL_CONTROL);
+		mqd->cp_hqd_pq_doorbell_control &=
+			~CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET_MASK;
+		mqd->cp_hqd_pq_doorbell_control |=
+			(ring->doorbell_index <<
+			 CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT);
+		mqd->cp_hqd_pq_doorbell_control |=
+			CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_EN_MASK;
+		mqd->cp_hqd_pq_doorbell_control &=
+			~(CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_SOURCE_MASK |
+					CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_HIT_MASK);
+
+	} else {
+		mqd->cp_hqd_pq_doorbell_control = 0;
+	}
+
+	/* read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	ring->wptr = 0;
+	mqd->cp_hqd_pq_wptr = lower_32_bits(ring->wptr);
+	mqd->cp_hqd_pq_rptr = RREG32(mmCP_HQD_PQ_RPTR);
+
+	/* set the vmid for the queue */
+	mqd->cp_hqd_vmid = 0;
+
+	/* defaults */
+	mqd->cp_hqd_ib_control = RREG32(mmCP_HQD_IB_CONTROL);
+	mqd->cp_hqd_ib_base_addr_lo = RREG32(mmCP_HQD_IB_BASE_ADDR);
+	mqd->cp_hqd_ib_base_addr_hi = RREG32(mmCP_HQD_IB_BASE_ADDR_HI);
+	mqd->cp_hqd_ib_rptr = RREG32(mmCP_HQD_IB_RPTR);
+	mqd->cp_hqd_persistent_state = RREG32(mmCP_HQD_PERSISTENT_STATE);
+	mqd->cp_hqd_sema_cmd = RREG32(mmCP_HQD_SEMA_CMD);
+	mqd->cp_hqd_msg_type = RREG32(mmCP_HQD_MSG_TYPE);
+	mqd->cp_hqd_atomic0_preop_lo = RREG32(mmCP_HQD_ATOMIC0_PREOP_LO);
+	mqd->cp_hqd_atomic0_preop_hi = RREG32(mmCP_HQD_ATOMIC0_PREOP_HI);
+	mqd->cp_hqd_atomic1_preop_lo = RREG32(mmCP_HQD_ATOMIC1_PREOP_LO);
+	mqd->cp_hqd_atomic1_preop_hi = RREG32(mmCP_HQD_ATOMIC1_PREOP_HI);
+	mqd->cp_hqd_pq_rptr = RREG32(mmCP_HQD_PQ_RPTR);
+	mqd->cp_hqd_quantum = RREG32(mmCP_HQD_QUANTUM);
+	mqd->cp_hqd_pipe_priority = RREG32(mmCP_HQD_PIPE_PRIORITY);
+	mqd->cp_hqd_queue_priority = RREG32(mmCP_HQD_QUEUE_PRIORITY);
+	mqd->cp_hqd_iq_rptr = RREG32(mmCP_HQD_IQ_RPTR);
+
+	/* activate the queue */
+	mqd->cp_hqd_active = 1;
+}
+
+int gfx_v7_0_mqd_commit(struct amdgpu_device *adev, struct cik_mqd *mqd)
+{
+	uint32_t tmp;
+	uint32_t mqd_reg;
+	uint32_t *mqd_data;
+
+	/* HQD registers extend from mmCP_MQD_BASE_ADDR to mmCP_MQD_CONTROL */
+	mqd_data = &mqd->cp_mqd_base_addr_lo;
+
+	/* disable wptr polling */
+	tmp = RREG32(mmCP_PQ_WPTR_POLL_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_PQ_WPTR_POLL_CNTL, EN, 0);
+	WREG32(mmCP_PQ_WPTR_POLL_CNTL, tmp);
+
+	/* program all HQD registers */
+	for (mqd_reg = mmCP_HQD_VMID; mqd_reg <= mmCP_MQD_CONTROL; mqd_reg++)
+		WREG32(mqd_reg, mqd_data[mqd_reg - mmCP_MQD_BASE_ADDR]);
+
+	/* activate the HQD */
+	for (mqd_reg = mmCP_MQD_BASE_ADDR; mqd_reg <= mmCP_HQD_ACTIVE; mqd_reg++)
+		WREG32(mqd_reg, mqd_data[mqd_reg - mmCP_MQD_BASE_ADDR]);
+
+	return 0;
+}
+
+static int gfx_v7_0_compute_queue_init(struct amdgpu_device *adev, int ring_id)
+{
+	int r;
+	u64 mqd_gpu_addr;
+	struct cik_mqd *mqd;
+	struct amdgpu_ring *ring = &adev->gfx.compute_ring[ring_id];
+
+	r = amdgpu_bo_create_reserved(adev, sizeof(struct cik_mqd), PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT, &ring->mqd_obj,
+				      &mqd_gpu_addr, (void **)&mqd);
+	if (r) {
+		dev_warn(adev->dev, "(%d) create MQD bo failed\n", r);
+		return r;
+	}
+
+	mutex_lock(&adev->srbm_mutex);
+	cik_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+
+	gfx_v7_0_mqd_init(adev, mqd, mqd_gpu_addr, ring);
+	gfx_v7_0_mqd_deactivate(adev);
+	gfx_v7_0_mqd_commit(adev, mqd);
+
+	cik_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	amdgpu_bo_kunmap(ring->mqd_obj);
+	amdgpu_bo_unreserve(ring->mqd_obj);
+	return 0;
+}
+
+/**
+ * gfx_v7_0_cp_compute_resume - setup the compute queue registers
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Program the compute queues and test them to make sure they
+ * are working.
+ * Returns 0 for success, error for failure.
+ */
+static int gfx_v7_0_cp_compute_resume(struct amdgpu_device *adev)
+{
+	int r, i, j;
+	u32 tmp;
+	struct amdgpu_ring *ring;
+
+	/* fix up chicken bits */
+	tmp = RREG32(mmCP_CPF_DEBUG);
+	tmp |= (1 << 23);
+	WREG32(mmCP_CPF_DEBUG, tmp);
+
+	/* init all pipes (even the ones we don't own) */
+	for (i = 0; i < adev->gfx.mec.num_mec; i++)
+		for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++)
+			gfx_v7_0_compute_pipe_init(adev, i, j);
+
+	/* init the queues */
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		r = gfx_v7_0_compute_queue_init(adev, i);
+		if (r) {
+			gfx_v7_0_cp_compute_fini(adev);
+			return r;
+		}
+	}
+
+	gfx_v7_0_cp_compute_enable(adev, true);
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+		ring->ready = true;
+		r = amdgpu_ring_test_ring(ring);
+		if (r)
+			ring->ready = false;
+	}
+
+	return 0;
+}
+
+static void gfx_v7_0_cp_enable(struct amdgpu_device *adev, bool enable)
+{
+	gfx_v7_0_cp_gfx_enable(adev, enable);
+	gfx_v7_0_cp_compute_enable(adev, enable);
+}
+
+static int gfx_v7_0_cp_load_microcode(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = gfx_v7_0_cp_gfx_load_microcode(adev);
+	if (r)
+		return r;
+	r = gfx_v7_0_cp_compute_load_microcode(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static void gfx_v7_0_enable_gui_idle_interrupt(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 tmp = RREG32(mmCP_INT_CNTL_RING0);
+
+	if (enable)
+		tmp |= (CP_INT_CNTL_RING0__CNTX_BUSY_INT_ENABLE_MASK |
+				CP_INT_CNTL_RING0__CNTX_EMPTY_INT_ENABLE_MASK);
+	else
+		tmp &= ~(CP_INT_CNTL_RING0__CNTX_BUSY_INT_ENABLE_MASK |
+				CP_INT_CNTL_RING0__CNTX_EMPTY_INT_ENABLE_MASK);
+	WREG32(mmCP_INT_CNTL_RING0, tmp);
+}
+
+static int gfx_v7_0_cp_resume(struct amdgpu_device *adev)
+{
+	int r;
+
+	gfx_v7_0_enable_gui_idle_interrupt(adev, false);
+
+	r = gfx_v7_0_cp_load_microcode(adev);
+	if (r)
+		return r;
+
+	r = gfx_v7_0_cp_gfx_resume(adev);
+	if (r)
+		return r;
+	r = gfx_v7_0_cp_compute_resume(adev);
+	if (r)
+		return r;
+
+	gfx_v7_0_enable_gui_idle_interrupt(adev, true);
+
+	return 0;
+}
+
+/**
+ * gfx_v7_0_ring_emit_vm_flush - cik vm flush using the CP
+ *
+ * @ring: the ring to emmit the commands to
+ *
+ * Sync the command pipeline with the PFP. E.g. wait for everything
+ * to be completed.
+ */
+static void gfx_v7_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_MEM_SPACE(1) | /* memory */
+				 WAIT_REG_MEM_FUNCTION(3) | /* equal */
+				 WAIT_REG_MEM_ENGINE(usepfp)));   /* pfp or me */
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, 0xffffffff);
+	amdgpu_ring_write(ring, 4); /* poll interval */
+
+	if (usepfp) {
+		/* synce CE with ME to prevent CE fetch CEIB before context switch done */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+/*
+ * vm
+ * VMID 0 is the physical GPU addresses as used by the kernel.
+ * VMIDs 1-15 are used for userspace clients and are handled
+ * by the amdgpu vm/hsa code.
+ */
+/**
+ * gfx_v7_0_ring_emit_vm_flush - cik vm flush using the CP
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Update the page table base and flush the VM TLB
+ * using the CP (CIK).
+ */
+static void gfx_v7_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned vmid, uint64_t pd_addr)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for the invalidate to complete */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_OPERATION(0) | /* wait */
+				 WAIT_REG_MEM_FUNCTION(0) |  /* always */
+				 WAIT_REG_MEM_ENGINE(0))); /* me */
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0); /* ref */
+	amdgpu_ring_write(ring, 0); /* mask */
+	amdgpu_ring_write(ring, 0x20); /* poll interval */
+
+	/* compute doesn't have PFP */
+	if (usepfp) {
+		/* sync PFP to ME, otherwise we might get invalid PFP reads */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
+		amdgpu_ring_write(ring, 0x0);
+
+		/* synce CE with ME to prevent CE fetch CEIB before context switch done */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static void gfx_v7_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				    uint32_t reg, uint32_t val)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(usepfp) |
+				 WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
+/*
+ * RLC
+ * The RLC is a multi-purpose microengine that handles a
+ * variety of functions.
+ */
+static void gfx_v7_0_rlc_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.save_restore_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL);
+}
+
+static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
+{
+	const u32 *src_ptr;
+	volatile u32 *dst_ptr;
+	u32 dws, i;
+	const struct cs_section_def *cs_data;
+	int r;
+
+	/* allocate rlc buffers */
+	if (adev->flags & AMD_IS_APU) {
+		if (adev->asic_type == CHIP_KAVERI) {
+			adev->gfx.rlc.reg_list = spectre_rlc_save_restore_register_list;
+			adev->gfx.rlc.reg_list_size =
+				(u32)ARRAY_SIZE(spectre_rlc_save_restore_register_list);
+		} else {
+			adev->gfx.rlc.reg_list = kalindi_rlc_save_restore_register_list;
+			adev->gfx.rlc.reg_list_size =
+				(u32)ARRAY_SIZE(kalindi_rlc_save_restore_register_list);
+		}
+	}
+	adev->gfx.rlc.cs_data = ci_cs_data;
+	adev->gfx.rlc.cp_table_size = roundup2(CP_ME_TABLE_SIZE * 5 * 4, 2048); /* CP JT */
+	adev->gfx.rlc.cp_table_size += 64 * 1024; /* GDS */
+
+	src_ptr = adev->gfx.rlc.reg_list;
+	dws = adev->gfx.rlc.reg_list_size;
+	dws += (5 * 16) + 48 + 48 + 64;
+
+	cs_data = adev->gfx.rlc.cs_data;
+
+	if (src_ptr) {
+		/* save restore block */
+		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.save_restore_obj,
+					      &adev->gfx.rlc.save_restore_gpu_addr,
+					      (void **)&adev->gfx.rlc.sr_ptr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create, pin or map of RLC sr bo failed\n", r);
+			gfx_v7_0_rlc_fini(adev);
+			return r;
+		}
+
+		/* write the sr buffer */
+		dst_ptr = adev->gfx.rlc.sr_ptr;
+		for (i = 0; i < adev->gfx.rlc.reg_list_size; i++)
+			dst_ptr[i] = cpu_to_le32(src_ptr[i]);
+		amdgpu_bo_kunmap(adev->gfx.rlc.save_restore_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
+	}
+
+	if (cs_data) {
+		/* clear state block */
+		adev->gfx.rlc.clear_state_size = dws = gfx_v7_0_get_csb_size(adev);
+
+		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.clear_state_obj,
+					      &adev->gfx.rlc.clear_state_gpu_addr,
+					      (void **)&adev->gfx.rlc.cs_ptr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
+			gfx_v7_0_rlc_fini(adev);
+			return r;
+		}
+
+		/* set up the cs buffer */
+		dst_ptr = adev->gfx.rlc.cs_ptr;
+		gfx_v7_0_get_csb_buffer(adev, dst_ptr);
+		amdgpu_bo_kunmap(adev->gfx.rlc.clear_state_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
+	}
+
+	if (adev->gfx.rlc.cp_table_size) {
+
+		r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
+					      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.cp_table_obj,
+					      &adev->gfx.rlc.cp_table_gpu_addr,
+					      (void **)&adev->gfx.rlc.cp_table_ptr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
+			gfx_v7_0_rlc_fini(adev);
+			return r;
+		}
+
+		gfx_v7_0_init_cp_pg_table(adev);
+
+		amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
+
+	}
+
+	return 0;
+}
+
+static void gfx_v7_0_enable_lbpw(struct amdgpu_device *adev, bool enable)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmRLC_LB_CNTL);
+	if (enable)
+		tmp |= RLC_LB_CNTL__LOAD_BALANCE_ENABLE_MASK;
+	else
+		tmp &= ~RLC_LB_CNTL__LOAD_BALANCE_ENABLE_MASK;
+	WREG32(mmRLC_LB_CNTL, tmp);
+}
+
+static void gfx_v7_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
+{
+	u32 i, j, k;
+	u32 mask;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
+			for (k = 0; k < adev->usec_timeout; k++) {
+				if (RREG32(mmRLC_SERDES_CU_MASTER_BUSY) == 0)
+					break;
+				udelay(1);
+			}
+		}
+	}
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__GC_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC0_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC1_MASTER_BUSY_MASK;
+	for (k = 0; k < adev->usec_timeout; k++) {
+		if ((RREG32(mmRLC_SERDES_NONCU_MASTER_BUSY) & mask) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v7_0_update_rlc(struct amdgpu_device *adev, u32 rlc)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmRLC_CNTL);
+	if (tmp != rlc)
+		WREG32(mmRLC_CNTL, rlc);
+}
+
+static u32 gfx_v7_0_halt_rlc(struct amdgpu_device *adev)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_CNTL);
+
+	if (data & RLC_CNTL__RLC_ENABLE_F32_MASK) {
+		u32 i;
+
+		data &= ~RLC_CNTL__RLC_ENABLE_F32_MASK;
+		WREG32(mmRLC_CNTL, data);
+
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if ((RREG32(mmRLC_GPM_STAT) & RLC_GPM_STAT__RLC_BUSY_MASK) == 0)
+				break;
+			udelay(1);
+		}
+
+		gfx_v7_0_wait_for_rlc_serdes(adev);
+	}
+
+	return orig;
+}
+
+static void gfx_v7_0_enter_rlc_safe_mode(struct amdgpu_device *adev)
+{
+	u32 tmp, i, mask;
+
+	tmp = 0x1 | (1 << 1);
+	WREG32(mmRLC_GPR_REG2, tmp);
+
+	mask = RLC_GPM_STAT__GFX_POWER_STATUS_MASK |
+		RLC_GPM_STAT__GFX_CLOCK_STATUS_MASK;
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if ((RREG32(mmRLC_GPM_STAT) & mask) == mask)
+			break;
+		udelay(1);
+	}
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if ((RREG32(mmRLC_GPR_REG2) & 0x1) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v7_0_exit_rlc_safe_mode(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = 0x1 | (0 << 1);
+	WREG32(mmRLC_GPR_REG2, tmp);
+}
+
+/**
+ * gfx_v7_0_rlc_stop - stop the RLC ME
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Halt the RLC ME (MicroEngine) (CIK).
+ */
+static void gfx_v7_0_rlc_stop(struct amdgpu_device *adev)
+{
+	WREG32(mmRLC_CNTL, 0);
+
+	gfx_v7_0_enable_gui_idle_interrupt(adev, false);
+
+	gfx_v7_0_wait_for_rlc_serdes(adev);
+}
+
+/**
+ * gfx_v7_0_rlc_start - start the RLC ME
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Unhalt the RLC ME (MicroEngine) (CIK).
+ */
+static void gfx_v7_0_rlc_start(struct amdgpu_device *adev)
+{
+	WREG32(mmRLC_CNTL, RLC_CNTL__RLC_ENABLE_F32_MASK);
+
+	gfx_v7_0_enable_gui_idle_interrupt(adev, true);
+
+	udelay(50);
+}
+
+static void gfx_v7_0_rlc_reset(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(mmGRBM_SOFT_RESET);
+
+	tmp |= GRBM_SOFT_RESET__SOFT_RESET_RLC_MASK;
+	WREG32(mmGRBM_SOFT_RESET, tmp);
+	udelay(50);
+	tmp &= ~GRBM_SOFT_RESET__SOFT_RESET_RLC_MASK;
+	WREG32(mmGRBM_SOFT_RESET, tmp);
+	udelay(50);
+}
+
+/**
+ * gfx_v7_0_rlc_resume - setup the RLC hw
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the RLC registers, load the ucode,
+ * and start the RLC (CIK).
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int gfx_v7_0_rlc_resume(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+	u32 tmp;
+
+	if (!adev->gfx.rlc_fw)
+		return -EINVAL;
+
+	hdr = (const struct rlc_firmware_header_v1_0 *)adev->gfx.rlc_fw->data;
+	amdgpu_ucode_print_rlc_hdr(&hdr->header);
+	adev->gfx.rlc_fw_version = le32_to_cpu(hdr->header.ucode_version);
+	adev->gfx.rlc_feature_version = le32_to_cpu(
+					hdr->ucode_feature_version);
+
+	gfx_v7_0_rlc_stop(adev);
+
+	/* disable CG */
+	tmp = RREG32(mmRLC_CGCG_CGLS_CTRL) & 0xfffffffc;
+	WREG32(mmRLC_CGCG_CGLS_CTRL, tmp);
+
+	gfx_v7_0_rlc_reset(adev);
+
+	gfx_v7_0_init_pg(adev);
+
+	WREG32(mmRLC_LB_CNTR_INIT, 0);
+	WREG32(mmRLC_LB_CNTR_MAX, 0x00008000);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	WREG32(mmRLC_LB_INIT_CU_MASK, 0xffffffff);
+	WREG32(mmRLC_LB_PARAMS, 0x00600408);
+	WREG32(mmRLC_LB_CNTL, 0x80000004);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	WREG32(mmRLC_MC_CNTL, 0);
+	WREG32(mmRLC_UCODE_CNTL, 0);
+
+	fw_data = (const __le32 *)
+		(adev->gfx.rlc_fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmRLC_GPM_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmRLC_GPM_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmRLC_GPM_UCODE_ADDR, adev->gfx.rlc_fw_version);
+
+	/* XXX - find out what chips support lbpw */
+	gfx_v7_0_enable_lbpw(adev, false);
+
+	if (adev->asic_type == CHIP_BONAIRE)
+		WREG32(mmRLC_DRIVER_CPDMA_STATUS, 0);
+
+	gfx_v7_0_rlc_start(adev);
+
+	return 0;
+}
+
+static void gfx_v7_0_enable_cgcg(struct amdgpu_device *adev, bool enable)
+{
+	u32 data, orig, tmp, tmp2;
+
+	orig = data = RREG32(mmRLC_CGCG_CGLS_CTRL);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)) {
+		gfx_v7_0_enable_gui_idle_interrupt(adev, true);
+
+		tmp = gfx_v7_0_halt_rlc(adev);
+
+		mutex_lock(&adev->grbm_idx_mutex);
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
+		tmp2 = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK |
+			RLC_SERDES_WR_CTRL__CGCG_OVERRIDE_0_MASK |
+			RLC_SERDES_WR_CTRL__CGLS_ENABLE_MASK;
+		WREG32(mmRLC_SERDES_WR_CTRL, tmp2);
+		mutex_unlock(&adev->grbm_idx_mutex);
+
+		gfx_v7_0_update_rlc(adev, tmp);
+
+		data |= RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK | RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+		if (orig != data)
+			WREG32(mmRLC_CGCG_CGLS_CTRL, data);
+
+	} else {
+		gfx_v7_0_enable_gui_idle_interrupt(adev, false);
+
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+
+		data &= ~(RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK | RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK);
+		if (orig != data)
+			WREG32(mmRLC_CGCG_CGLS_CTRL, data);
+
+		gfx_v7_0_enable_gui_idle_interrupt(adev, true);
+	}
+}
+
+static void gfx_v7_0_enable_mgcg(struct amdgpu_device *adev, bool enable)
+{
+	u32 data, orig, tmp = 0;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG)) {
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGLS) {
+			if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CP_LS) {
+				orig = data = RREG32(mmCP_MEM_SLP_CNTL);
+				data |= CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+				if (orig != data)
+					WREG32(mmCP_MEM_SLP_CNTL, data);
+			}
+		}
+
+		orig = data = RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
+		data |= 0x00000001;
+		data &= 0xfffffffd;
+		if (orig != data)
+			WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		tmp = gfx_v7_0_halt_rlc(adev);
+
+		mutex_lock(&adev->grbm_idx_mutex);
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
+		data = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK |
+			RLC_SERDES_WR_CTRL__MGCG_OVERRIDE_0_MASK;
+		WREG32(mmRLC_SERDES_WR_CTRL, data);
+		mutex_unlock(&adev->grbm_idx_mutex);
+
+		gfx_v7_0_update_rlc(adev, tmp);
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGTS) {
+			orig = data = RREG32(mmCGTS_SM_CTRL_REG);
+			data &= ~CGTS_SM_CTRL_REG__SM_MODE_MASK;
+			data |= (0x2 << CGTS_SM_CTRL_REG__SM_MODE__SHIFT);
+			data |= CGTS_SM_CTRL_REG__SM_MODE_ENABLE_MASK;
+			data &= ~CGTS_SM_CTRL_REG__OVERRIDE_MASK;
+			if ((adev->cg_flags & AMD_CG_SUPPORT_GFX_MGLS) &&
+			    (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGTS_LS))
+				data &= ~CGTS_SM_CTRL_REG__LS_OVERRIDE_MASK;
+			data &= ~CGTS_SM_CTRL_REG__ON_MONITOR_ADD_MASK;
+			data |= CGTS_SM_CTRL_REG__ON_MONITOR_ADD_EN_MASK;
+			data |= (0x96 << CGTS_SM_CTRL_REG__ON_MONITOR_ADD__SHIFT);
+			if (orig != data)
+				WREG32(mmCGTS_SM_CTRL_REG, data);
+		}
+	} else {
+		orig = data = RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
+		data |= 0x00000003;
+		if (orig != data)
+			WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		data = RREG32(mmRLC_MEM_SLP_CNTL);
+		if (data & RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK) {
+			data &= ~RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK;
+			WREG32(mmRLC_MEM_SLP_CNTL, data);
+		}
+
+		data = RREG32(mmCP_MEM_SLP_CNTL);
+		if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK) {
+			data &= ~CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+			WREG32(mmCP_MEM_SLP_CNTL, data);
+		}
+
+		orig = data = RREG32(mmCGTS_SM_CTRL_REG);
+		data |= CGTS_SM_CTRL_REG__OVERRIDE_MASK | CGTS_SM_CTRL_REG__LS_OVERRIDE_MASK;
+		if (orig != data)
+			WREG32(mmCGTS_SM_CTRL_REG, data);
+
+		tmp = gfx_v7_0_halt_rlc(adev);
+
+		mutex_lock(&adev->grbm_idx_mutex);
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
+		data = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK | RLC_SERDES_WR_CTRL__MGCG_OVERRIDE_1_MASK;
+		WREG32(mmRLC_SERDES_WR_CTRL, data);
+		mutex_unlock(&adev->grbm_idx_mutex);
+
+		gfx_v7_0_update_rlc(adev, tmp);
+	}
+}
+
+static void gfx_v7_0_update_cg(struct amdgpu_device *adev,
+			       bool enable)
+{
+	gfx_v7_0_enable_gui_idle_interrupt(adev, false);
+	/* order matters! */
+	if (enable) {
+		gfx_v7_0_enable_mgcg(adev, true);
+		gfx_v7_0_enable_cgcg(adev, true);
+	} else {
+		gfx_v7_0_enable_cgcg(adev, false);
+		gfx_v7_0_enable_mgcg(adev, false);
+	}
+	gfx_v7_0_enable_gui_idle_interrupt(adev, true);
+}
+
+static void gfx_v7_0_enable_sclk_slowdown_on_pu(struct amdgpu_device *adev,
+						bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_RLC_SMU_HS))
+		data |= RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PU_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PU_ENABLE_MASK;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v7_0_enable_sclk_slowdown_on_pd(struct amdgpu_device *adev,
+						bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_RLC_SMU_HS))
+		data |= RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PD_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PD_ENABLE_MASK;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v7_0_enable_cp_pg(struct amdgpu_device *adev, bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_CP))
+		data &= ~0x8000;
+	else
+		data |= 0x8000;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v7_0_enable_gds_pg(struct amdgpu_device *adev, bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GDS))
+		data &= ~0x2000;
+	else
+		data |= 0x2000;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v7_0_init_cp_pg_table(struct amdgpu_device *adev)
+{
+	const __le32 *fw_data;
+	volatile u32 *dst_ptr;
+	int me, i, max_me = 4;
+	u32 bo_offset = 0;
+	u32 table_offset, table_size;
+
+	if (adev->asic_type == CHIP_KAVERI)
+		max_me = 5;
+
+	if (adev->gfx.rlc.cp_table_ptr == NULL)
+		return;
+
+	/* write the cp table buffer */
+	dst_ptr = adev->gfx.rlc.cp_table_ptr;
+	for (me = 0; me < max_me; me++) {
+		if (me == 0) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.ce_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 1) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.pfp_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 2) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.me_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 3) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec2_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		}
+
+		for (i = 0; i < table_size; i ++) {
+			dst_ptr[bo_offset + i] =
+				cpu_to_le32(le32_to_cpu(fw_data[table_offset + i]));
+		}
+
+		bo_offset += table_size;
+	}
+}
+
+static void gfx_v7_0_enable_gfx_cgpg(struct amdgpu_device *adev,
+				     bool enable)
+{
+	u32 data, orig;
+
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG)) {
+		orig = data = RREG32(mmRLC_PG_CNTL);
+		data |= RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+		if (orig != data)
+			WREG32(mmRLC_PG_CNTL, data);
+
+		orig = data = RREG32(mmRLC_AUTO_PG_CTRL);
+		data |= RLC_AUTO_PG_CTRL__AUTO_PG_EN_MASK;
+		if (orig != data)
+			WREG32(mmRLC_AUTO_PG_CTRL, data);
+	} else {
+		orig = data = RREG32(mmRLC_PG_CNTL);
+		data &= ~RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+		if (orig != data)
+			WREG32(mmRLC_PG_CNTL, data);
+
+		orig = data = RREG32(mmRLC_AUTO_PG_CTRL);
+		data &= ~RLC_AUTO_PG_CTRL__AUTO_PG_EN_MASK;
+		if (orig != data)
+			WREG32(mmRLC_AUTO_PG_CTRL, data);
+
+		data = RREG32(mmDB_RENDER_CONTROL);
+	}
+}
+
+static void gfx_v7_0_set_user_cu_inactive_bitmap(struct amdgpu_device *adev,
+						 u32 bitmap)
+{
+	u32 data;
+
+	if (!bitmap)
+		return;
+
+	data = bitmap << GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT;
+	data &= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK;
+
+	WREG32(mmGC_USER_SHADER_ARRAY_CONFIG, data);
+}
+
+static u32 gfx_v7_0_get_cu_active_bitmap(struct amdgpu_device *adev)
+{
+	u32 data, mask;
+
+	data = RREG32(mmCC_GC_SHADER_ARRAY_CONFIG);
+	data |= RREG32(mmGC_USER_SHADER_ARRAY_CONFIG);
+
+	data &= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK;
+	data >>= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT;
+
+	mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_cu_per_sh);
+
+	return (~data) & mask;
+}
+
+static void gfx_v7_0_init_ao_cu_mask(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	WREG32(mmRLC_PG_ALWAYS_ON_CU_MASK, adev->gfx.cu_info.ao_cu_mask);
+
+	tmp = RREG32(mmRLC_MAX_PG_CU);
+	tmp &= ~RLC_MAX_PG_CU__MAX_POWERED_UP_CU_MASK;
+	tmp |= (adev->gfx.cu_info.number << RLC_MAX_PG_CU__MAX_POWERED_UP_CU__SHIFT);
+	WREG32(mmRLC_MAX_PG_CU, tmp);
+}
+
+static void gfx_v7_0_enable_gfx_static_mgpg(struct amdgpu_device *adev,
+					    bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_SMG))
+		data |= RLC_PG_CNTL__STATIC_PER_CU_PG_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__STATIC_PER_CU_PG_ENABLE_MASK;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v7_0_enable_gfx_dynamic_mgpg(struct amdgpu_device *adev,
+					     bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_DMG))
+		data |= RLC_PG_CNTL__DYN_PER_CU_PG_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__DYN_PER_CU_PG_ENABLE_MASK;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+#define RLC_SAVE_AND_RESTORE_STARTING_OFFSET 0x90
+#define RLC_CLEAR_STATE_DESCRIPTOR_OFFSET    0x3D
+
+static void gfx_v7_0_init_gfx_cgpg(struct amdgpu_device *adev)
+{
+	u32 data, orig;
+	u32 i;
+
+	if (adev->gfx.rlc.cs_data) {
+		WREG32(mmRLC_GPM_SCRATCH_ADDR, RLC_CLEAR_STATE_DESCRIPTOR_OFFSET);
+		WREG32(mmRLC_GPM_SCRATCH_DATA, upper_32_bits(adev->gfx.rlc.clear_state_gpu_addr));
+		WREG32(mmRLC_GPM_SCRATCH_DATA, lower_32_bits(adev->gfx.rlc.clear_state_gpu_addr));
+		WREG32(mmRLC_GPM_SCRATCH_DATA, adev->gfx.rlc.clear_state_size);
+	} else {
+		WREG32(mmRLC_GPM_SCRATCH_ADDR, RLC_CLEAR_STATE_DESCRIPTOR_OFFSET);
+		for (i = 0; i < 3; i++)
+			WREG32(mmRLC_GPM_SCRATCH_DATA, 0);
+	}
+	if (adev->gfx.rlc.reg_list) {
+		WREG32(mmRLC_GPM_SCRATCH_ADDR, RLC_SAVE_AND_RESTORE_STARTING_OFFSET);
+		for (i = 0; i < adev->gfx.rlc.reg_list_size; i++)
+			WREG32(mmRLC_GPM_SCRATCH_DATA, adev->gfx.rlc.reg_list[i]);
+	}
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	data |= RLC_PG_CNTL__GFX_POWER_GATING_SRC_MASK;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+
+	WREG32(mmRLC_SAVE_AND_RESTORE_BASE, adev->gfx.rlc.save_restore_gpu_addr >> 8);
+	WREG32(mmRLC_JUMP_TABLE_RESTORE, adev->gfx.rlc.cp_table_gpu_addr >> 8);
+
+	data = RREG32(mmCP_RB_WPTR_POLL_CNTL);
+	data &= ~CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT_MASK;
+	data |= (0x60 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT);
+	WREG32(mmCP_RB_WPTR_POLL_CNTL, data);
+
+	data = 0x10101010;
+	WREG32(mmRLC_PG_DELAY, data);
+
+	data = RREG32(mmRLC_PG_DELAY_2);
+	data &= ~0xff;
+	data |= 0x3;
+	WREG32(mmRLC_PG_DELAY_2, data);
+
+	data = RREG32(mmRLC_AUTO_PG_CTRL);
+	data &= ~RLC_AUTO_PG_CTRL__GRBM_REG_SAVE_GFX_IDLE_THRESHOLD_MASK;
+	data |= (0x700 << RLC_AUTO_PG_CTRL__GRBM_REG_SAVE_GFX_IDLE_THRESHOLD__SHIFT);
+	WREG32(mmRLC_AUTO_PG_CTRL, data);
+
+}
+
+static void gfx_v7_0_update_gfx_pg(struct amdgpu_device *adev, bool enable)
+{
+	gfx_v7_0_enable_gfx_cgpg(adev, enable);
+	gfx_v7_0_enable_gfx_static_mgpg(adev, enable);
+	gfx_v7_0_enable_gfx_dynamic_mgpg(adev, enable);
+}
+
+static u32 gfx_v7_0_get_csb_size(struct amdgpu_device *adev)
+{
+	u32 count = 0;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	if (adev->gfx.rlc.cs_data == NULL)
+		return 0;
+
+	/* begin clear state */
+	count += 2;
+	/* context control state */
+	count += 3;
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT)
+				count += 2 + ext->reg_count;
+			else
+				return 0;
+		}
+	}
+	/* pa_sc_raster_config/pa_sc_raster_config1 */
+	count += 4;
+	/* end clear state */
+	count += 2;
+	/* clear state */
+	count += 2;
+
+	return count;
+}
+
+static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev,
+				    volatile u32 *buffer)
+{
+	u32 count = 0, i;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	if (adev->gfx.rlc.cs_data == NULL)
+		return;
+	if (buffer == NULL)
+		return;
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	buffer[count++] = cpu_to_le32(0x80000000);
+	buffer[count++] = cpu_to_le32(0x80000000);
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				buffer[count++] =
+					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
+				buffer[count++] = cpu_to_le32(ext->reg_index - PACKET3_SET_CONTEXT_REG_START);
+				for (i = 0; i < ext->reg_count; i++)
+					buffer[count++] = cpu_to_le32(ext->extent[i]);
+			} else {
+				return;
+			}
+		}
+	}
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 2));
+	buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		buffer[count++] = cpu_to_le32(0x16000012);
+		buffer[count++] = cpu_to_le32(0x00000000);
+		break;
+	case CHIP_KAVERI:
+		buffer[count++] = cpu_to_le32(0x00000000); /* XXX */
+		buffer[count++] = cpu_to_le32(0x00000000);
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		buffer[count++] = cpu_to_le32(0x00000000); /* XXX */
+		buffer[count++] = cpu_to_le32(0x00000000);
+		break;
+	case CHIP_HAWAII:
+		buffer[count++] = cpu_to_le32(0x3a00161a);
+		buffer[count++] = cpu_to_le32(0x0000002e);
+		break;
+	default:
+		buffer[count++] = cpu_to_le32(0x00000000);
+		buffer[count++] = cpu_to_le32(0x00000000);
+		break;
+	}
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
+	buffer[count++] = cpu_to_le32(0);
+}
+
+static void gfx_v7_0_init_pg(struct amdgpu_device *adev)
+{
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_PG |
+			      AMD_PG_SUPPORT_GFX_SMG |
+			      AMD_PG_SUPPORT_GFX_DMG |
+			      AMD_PG_SUPPORT_CP |
+			      AMD_PG_SUPPORT_GDS |
+			      AMD_PG_SUPPORT_RLC_SMU_HS)) {
+		gfx_v7_0_enable_sclk_slowdown_on_pu(adev, true);
+		gfx_v7_0_enable_sclk_slowdown_on_pd(adev, true);
+		if (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG) {
+			gfx_v7_0_init_gfx_cgpg(adev);
+			gfx_v7_0_enable_cp_pg(adev, true);
+			gfx_v7_0_enable_gds_pg(adev, true);
+		}
+		gfx_v7_0_init_ao_cu_mask(adev);
+		gfx_v7_0_update_gfx_pg(adev, true);
+	}
+}
+
+static void gfx_v7_0_fini_pg(struct amdgpu_device *adev)
+{
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_PG |
+			      AMD_PG_SUPPORT_GFX_SMG |
+			      AMD_PG_SUPPORT_GFX_DMG |
+			      AMD_PG_SUPPORT_CP |
+			      AMD_PG_SUPPORT_GDS |
+			      AMD_PG_SUPPORT_RLC_SMU_HS)) {
+		gfx_v7_0_update_gfx_pg(adev, false);
+		if (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG) {
+			gfx_v7_0_enable_cp_pg(adev, false);
+			gfx_v7_0_enable_gds_pg(adev, false);
+		}
+	}
+}
+
+/**
+ * gfx_v7_0_get_gpu_clock_counter - return GPU clock counter snapshot
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Fetches a GPU clock counter snapshot (SI).
+ * Returns the 64 bit clock counter snapshot.
+ */
+static uint64_t gfx_v7_0_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
+	uint64_t clock;
+
+	mutex_lock(&adev->gfx.gpu_clock_mutex);
+	WREG32(mmRLC_CAPTURE_GPU_CLOCK_COUNT, 1);
+	clock = (uint64_t)RREG32(mmRLC_GPU_CLOCK_COUNT_LSB) |
+		((uint64_t)RREG32(mmRLC_GPU_CLOCK_COUNT_MSB) << 32ULL);
+	mutex_unlock(&adev->gfx.gpu_clock_mutex);
+	return clock;
+}
+
+static void gfx_v7_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
+					  uint32_t vmid,
+					  uint32_t gds_base, uint32_t gds_size,
+					  uint32_t gws_base, uint32_t gws_size,
+					  uint32_t oa_base, uint32_t oa_size)
+{
+	gds_base = gds_base >> AMDGPU_GDS_SHIFT;
+	gds_size = gds_size >> AMDGPU_GDS_SHIFT;
+
+	gws_base = gws_base >> AMDGPU_GWS_SHIFT;
+	gws_size = gws_size >> AMDGPU_GWS_SHIFT;
+
+	oa_base = oa_base >> AMDGPU_OA_SHIFT;
+	oa_size = oa_size >> AMDGPU_OA_SHIFT;
+
+	/* GDS Base */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].mem_base);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, gds_base);
+
+	/* GDS Size */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].mem_size);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, gds_size);
+
+	/* GWS */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].gws);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, gws_size << GDS_GWS_VMID0__SIZE__SHIFT | gws_base);
+
+	/* OA */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].oa);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, (1 << (oa_size + oa_base)) - (1 << oa_base));
+}
+
+static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t address)
+{
+	WREG32(mmSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(address << SQ_IND_INDEX__INDEX__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK));
+	return RREG32(mmSQ_IND_DATA);
+}
+
+static void wave_read_regs(struct amdgpu_device *adev, uint32_t simd,
+			   uint32_t wave, uint32_t thread,
+			   uint32_t regno, uint32_t num, uint32_t *out)
+{
+	WREG32(mmSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(regno << SQ_IND_INDEX__INDEX__SHIFT) |
+		(thread << SQ_IND_INDEX__THREAD_ID__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK) |
+		(SQ_IND_INDEX__AUTO_INCR_MASK));
+	while (num--)
+		*(out++) = RREG32(mmSQ_IND_DATA);
+}
+
+static void gfx_v7_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
+{
+	/* type 0 wave data */
+	dst[(*no_fields)++] = 0;
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_STATUS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_HW_ID);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW1);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_GPR_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_LDS_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TRAPSTS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_STS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TBA_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TBA_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TMA_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TMA_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_DBG0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_M0);
+}
+
+static void gfx_v7_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd,
+				     uint32_t wave, uint32_t start,
+				     uint32_t size, uint32_t *dst)
+{
+	wave_read_regs(
+		adev, simd, wave, 0,
+		start + SQIND_WAVE_SGPRS_OFFSET, size, dst);
+}
+
+static void gfx_v7_0_select_me_pipe_q(struct amdgpu_device *adev,
+				  u32 me, u32 pipe, u32 q)
+{
+	cik_srbm_select(adev, me, pipe, q, 0);
+}
+
+static const struct amdgpu_gfx_funcs gfx_v7_0_gfx_funcs = {
+	.get_gpu_clock_counter = &gfx_v7_0_get_gpu_clock_counter,
+	.select_se_sh = &gfx_v7_0_select_se_sh,
+	.read_wave_data = &gfx_v7_0_read_wave_data,
+	.read_wave_sgprs = &gfx_v7_0_read_wave_sgprs,
+	.select_me_pipe_q = &gfx_v7_0_select_me_pipe_q
+};
+
+static const struct amdgpu_rlc_funcs gfx_v7_0_rlc_funcs = {
+	.enter_safe_mode = gfx_v7_0_enter_rlc_safe_mode,
+	.exit_safe_mode = gfx_v7_0_exit_rlc_safe_mode
+};
+
+static int gfx_v7_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->gfx.num_gfx_rings = GFX7_NUM_GFX_RINGS;
+	adev->gfx.num_compute_rings = AMDGPU_MAX_COMPUTE_RINGS;
+	adev->gfx.funcs = &gfx_v7_0_gfx_funcs;
+	adev->gfx.rlc.funcs = &gfx_v7_0_rlc_funcs;
+	gfx_v7_0_set_ring_funcs(adev);
+	gfx_v7_0_set_irq_funcs(adev);
+	gfx_v7_0_set_gds_init(adev);
+
+	return 0;
+}
+
+static int gfx_v7_0_late_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_reg_irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_inst_irq, 0);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static void gfx_v7_0_gpu_early_init(struct amdgpu_device *adev)
+{
+	u32 gb_addr_config;
+	u32 mc_shared_chmap, mc_arb_ramcfg;
+	u32 dimm00_addr_map, dimm01_addr_map, dimm10_addr_map, dimm11_addr_map;
+	u32 tmp;
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		adev->gfx.config.max_shader_engines = 2;
+		adev->gfx.config.max_tile_pipes = 4;
+		adev->gfx.config.max_cu_per_sh = 7;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 2;
+		adev->gfx.config.max_texture_channel_caches = 4;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = BONAIRE_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_HAWAII:
+		adev->gfx.config.max_shader_engines = 4;
+		adev->gfx.config.max_tile_pipes = 16;
+		adev->gfx.config.max_cu_per_sh = 11;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 4;
+		adev->gfx.config.max_texture_channel_caches = 16;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = HAWAII_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_KAVERI:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 4;
+		adev->gfx.config.max_cu_per_sh = 8;
+		adev->gfx.config.max_backends_per_se = 2;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_texture_channel_caches = 4;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 16;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = BONAIRE_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+	default:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 2;
+		adev->gfx.config.max_cu_per_sh = 2;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 1;
+		adev->gfx.config.max_texture_channel_caches = 2;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 16;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = BONAIRE_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	}
+
+	mc_shared_chmap = RREG32(mmMC_SHARED_CHMAP);
+	adev->gfx.config.mc_arb_ramcfg = RREG32(mmMC_ARB_RAMCFG);
+	mc_arb_ramcfg = adev->gfx.config.mc_arb_ramcfg;
+
+	adev->gfx.config.num_tile_pipes = adev->gfx.config.max_tile_pipes;
+	adev->gfx.config.mem_max_burst_length_bytes = 256;
+	if (adev->flags & AMD_IS_APU) {
+		/* Get memory bank mapping mode. */
+		tmp = RREG32(mmMC_FUS_DRAM0_BANK_ADDR_MAPPING);
+		dimm00_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM0_BANK_ADDR_MAPPING, DIMM0ADDRMAP);
+		dimm01_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM0_BANK_ADDR_MAPPING, DIMM1ADDRMAP);
+
+		tmp = RREG32(mmMC_FUS_DRAM1_BANK_ADDR_MAPPING);
+		dimm10_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM1_BANK_ADDR_MAPPING, DIMM0ADDRMAP);
+		dimm11_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM1_BANK_ADDR_MAPPING, DIMM1ADDRMAP);
+
+		/* Validate settings in case only one DIMM installed. */
+		if ((dimm00_addr_map == 0) || (dimm00_addr_map == 3) || (dimm00_addr_map == 4) || (dimm00_addr_map > 12))
+			dimm00_addr_map = 0;
+		if ((dimm01_addr_map == 0) || (dimm01_addr_map == 3) || (dimm01_addr_map == 4) || (dimm01_addr_map > 12))
+			dimm01_addr_map = 0;
+		if ((dimm10_addr_map == 0) || (dimm10_addr_map == 3) || (dimm10_addr_map == 4) || (dimm10_addr_map > 12))
+			dimm10_addr_map = 0;
+		if ((dimm11_addr_map == 0) || (dimm11_addr_map == 3) || (dimm11_addr_map == 4) || (dimm11_addr_map > 12))
+			dimm11_addr_map = 0;
+
+		/* If DIMM Addr map is 8GB, ROW size should be 2KB. Otherwise 1KB. */
+		/* If ROW size(DIMM1) != ROW size(DMIMM0), ROW size should be larger one. */
+		if ((dimm00_addr_map == 11) || (dimm01_addr_map == 11) || (dimm10_addr_map == 11) || (dimm11_addr_map == 11))
+			adev->gfx.config.mem_row_size_in_kb = 2;
+		else
+			adev->gfx.config.mem_row_size_in_kb = 1;
+	} else {
+		tmp = (mc_arb_ramcfg & MC_ARB_RAMCFG__NOOFCOLS_MASK) >> MC_ARB_RAMCFG__NOOFCOLS__SHIFT;
+		adev->gfx.config.mem_row_size_in_kb = (4 * (1 << (8 + tmp))) / 1024;
+		if (adev->gfx.config.mem_row_size_in_kb > 4)
+			adev->gfx.config.mem_row_size_in_kb = 4;
+	}
+	/* XXX use MC settings? */
+	adev->gfx.config.shader_engine_tile_size = 32;
+	adev->gfx.config.num_gpus = 1;
+	adev->gfx.config.multi_gpu_tile_size = 64;
+
+	/* fix up row size */
+	gb_addr_config &= ~GB_ADDR_CONFIG__ROW_SIZE_MASK;
+	switch (adev->gfx.config.mem_row_size_in_kb) {
+	case 1:
+	default:
+		gb_addr_config |= (0 << GB_ADDR_CONFIG__ROW_SIZE__SHIFT);
+		break;
+	case 2:
+		gb_addr_config |= (1 << GB_ADDR_CONFIG__ROW_SIZE__SHIFT);
+		break;
+	case 4:
+		gb_addr_config |= (2 << GB_ADDR_CONFIG__ROW_SIZE__SHIFT);
+		break;
+	}
+	adev->gfx.config.gb_addr_config = gb_addr_config;
+}
+
+static int gfx_v7_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
+					int mec, int pipe, int queue)
+{
+	int r;
+	unsigned irq_type;
+	struct amdgpu_ring *ring = &adev->gfx.compute_ring[ring_id];
+
+	/* mec0 is me1 */
+	ring->me = mec + 1;
+	ring->pipe = pipe;
+	ring->queue = queue;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->doorbell_index = AMDGPU_DOORBELL_MEC_RING0 + ring_id;
+	snprintf(ring->name, sizeof(ring->name), "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+
+	irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
+		+ ((ring->me - 1) * adev->gfx.mec.num_pipe_per_mec)
+		+ ring->pipe;
+
+	/* type-2 packets are deprecated on MEC, use type-3 instead */
+	r = amdgpu_ring_init(adev, ring, 1024,
+			&adev->gfx.eop_irq, irq_type);
+	if (r)
+		return r;
+
+
+	return 0;
+}
+
+static int gfx_v7_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i, j, k, r, ring_id;
+
+	switch (adev->asic_type) {
+	case CHIP_KAVERI:
+		adev->gfx.mec.num_mec = 2;
+		break;
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+	default:
+		adev->gfx.mec.num_mec = 1;
+		break;
+	}
+	adev->gfx.mec.num_pipe_per_mec = 4;
+	adev->gfx.mec.num_queue_per_pipe = 8;
+
+	/* EOP Event */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 181, &adev->gfx.eop_irq);
+	if (r)
+		return r;
+
+	/* Privileged reg */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 184,
+			      &adev->gfx.priv_reg_irq);
+	if (r)
+		return r;
+
+	/* Privileged inst */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 185,
+			      &adev->gfx.priv_inst_irq);
+	if (r)
+		return r;
+
+	gfx_v7_0_scratch_init(adev);
+
+	r = gfx_v7_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load gfx firmware!\n");
+		return r;
+	}
+
+	r = gfx_v7_0_rlc_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init rlc BOs!\n");
+		return r;
+	}
+
+	/* allocate mec buffers */
+	r = gfx_v7_0_mec_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init MEC BOs!\n");
+		return r;
+	}
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		ring = &adev->gfx.gfx_ring[i];
+		ring->ring_obj = NULL;
+		snprintf(ring->name, sizeof(ring->name), "gfx");
+		r = amdgpu_ring_init(adev, ring, 1024,
+				     &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP);
+		if (r)
+			return r;
+	}
+
+	/* set up the compute queues - allocate horizontally across pipes */
+	ring_id = 0;
+	for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
+		for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
+			for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
+				if (!amdgpu_gfx_is_mec_queue_enabled(adev, i, k, j))
+					continue;
+
+				r = gfx_v7_0_compute_ring_init(adev,
+								ring_id,
+								i, k, j);
+				if (r)
+					return r;
+
+				ring_id++;
+			}
+		}
+	}
+
+	/* reserve GDS, GWS and OA resource for gfx */
+	r = amdgpu_bo_create_kernel(adev, adev->gds.mem.gfx_partition_size,
+				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GDS,
+				    &adev->gds.gds_gfx_bo, NULL, NULL);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_create_kernel(adev, adev->gds.gws.gfx_partition_size,
+				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GWS,
+				    &adev->gds.gws_gfx_bo, NULL, NULL);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_create_kernel(adev, adev->gds.oa.gfx_partition_size,
+				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_OA,
+				    &adev->gds.oa_gfx_bo, NULL, NULL);
+	if (r)
+		return r;
+
+	adev->gfx.ce_ram_size = 0x8000;
+
+	gfx_v7_0_gpu_early_init(adev);
+
+	return r;
+}
+
+static int gfx_v7_0_sw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_bo_free_kernel(&adev->gds.oa_gfx_bo, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gds.gws_gfx_bo, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gds.gds_gfx_bo, NULL, NULL);
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
+
+	gfx_v7_0_cp_compute_fini(adev);
+	gfx_v7_0_rlc_fini(adev);
+	gfx_v7_0_mec_fini(adev);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj,
+				&adev->gfx.rlc.clear_state_gpu_addr,
+				(void **)&adev->gfx.rlc.cs_ptr);
+	if (adev->gfx.rlc.cp_table_size) {
+		amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj,
+				&adev->gfx.rlc.cp_table_gpu_addr,
+				(void **)&adev->gfx.rlc.cp_table_ptr);
+	}
+	gfx_v7_0_free_microcode(adev);
+
+	return 0;
+}
+
+static int gfx_v7_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gfx_v7_0_gpu_init(adev);
+
+	/* init rlc */
+	r = gfx_v7_0_rlc_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v7_0_cp_resume(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int gfx_v7_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
+	gfx_v7_0_cp_enable(adev, false);
+	gfx_v7_0_rlc_stop(adev);
+	gfx_v7_0_fini_pg(adev);
+
+	return 0;
+}
+
+static int gfx_v7_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return gfx_v7_0_hw_fini(adev);
+}
+
+static int gfx_v7_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return gfx_v7_0_hw_init(adev);
+}
+
+static bool gfx_v7_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (RREG32(mmGRBM_STATUS) & GRBM_STATUS__GUI_ACTIVE_MASK)
+		return false;
+	else
+		return true;
+}
+
+static int gfx_v7_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmGRBM_STATUS) & GRBM_STATUS__GUI_ACTIVE_MASK;
+
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int gfx_v7_0_soft_reset(void *handle)
+{
+	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* GRBM_STATUS */
+	tmp = RREG32(mmGRBM_STATUS);
+	if (tmp & (GRBM_STATUS__PA_BUSY_MASK | GRBM_STATUS__SC_BUSY_MASK |
+		   GRBM_STATUS__BCI_BUSY_MASK | GRBM_STATUS__SX_BUSY_MASK |
+		   GRBM_STATUS__TA_BUSY_MASK | GRBM_STATUS__VGT_BUSY_MASK |
+		   GRBM_STATUS__DB_BUSY_MASK | GRBM_STATUS__CB_BUSY_MASK |
+		   GRBM_STATUS__GDS_BUSY_MASK | GRBM_STATUS__SPI_BUSY_MASK |
+		   GRBM_STATUS__IA_BUSY_MASK | GRBM_STATUS__IA_BUSY_NO_DMA_MASK))
+		grbm_soft_reset |= GRBM_SOFT_RESET__SOFT_RESET_CP_MASK |
+			GRBM_SOFT_RESET__SOFT_RESET_GFX_MASK;
+
+	if (tmp & (GRBM_STATUS__CP_BUSY_MASK | GRBM_STATUS__CP_COHERENCY_BUSY_MASK)) {
+		grbm_soft_reset |= GRBM_SOFT_RESET__SOFT_RESET_CP_MASK;
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_GRBM_MASK;
+	}
+
+	/* GRBM_STATUS2 */
+	tmp = RREG32(mmGRBM_STATUS2);
+	if (tmp & GRBM_STATUS2__RLC_BUSY_MASK)
+		grbm_soft_reset |= GRBM_SOFT_RESET__SOFT_RESET_RLC_MASK;
+
+	/* SRBM_STATUS */
+	tmp = RREG32(mmSRBM_STATUS);
+	if (tmp & SRBM_STATUS__GRBM_RQ_PENDING_MASK)
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_GRBM_MASK;
+
+	if (grbm_soft_reset || srbm_soft_reset) {
+		/* disable CG/PG */
+		gfx_v7_0_fini_pg(adev);
+		gfx_v7_0_update_cg(adev, false);
+
+		/* stop the rlc */
+		gfx_v7_0_rlc_stop(adev);
+
+		/* Disable GFX parsing/prefetching */
+		WREG32(mmCP_ME_CNTL, CP_ME_CNTL__ME_HALT_MASK | CP_ME_CNTL__PFP_HALT_MASK | CP_ME_CNTL__CE_HALT_MASK);
+
+		/* Disable MEC parsing/prefetching */
+		WREG32(mmCP_MEC_CNTL, CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK);
+
+		if (grbm_soft_reset) {
+			tmp = RREG32(mmGRBM_SOFT_RESET);
+			tmp |= grbm_soft_reset;
+			dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+			WREG32(mmGRBM_SOFT_RESET, tmp);
+			tmp = RREG32(mmGRBM_SOFT_RESET);
+
+			udelay(50);
+
+			tmp &= ~grbm_soft_reset;
+			WREG32(mmGRBM_SOFT_RESET, tmp);
+			tmp = RREG32(mmGRBM_SOFT_RESET);
+		}
+
+		if (srbm_soft_reset) {
+			tmp = RREG32(mmSRBM_SOFT_RESET);
+			tmp |= srbm_soft_reset;
+			dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+			WREG32(mmSRBM_SOFT_RESET, tmp);
+			tmp = RREG32(mmSRBM_SOFT_RESET);
+
+			udelay(50);
+
+			tmp &= ~srbm_soft_reset;
+			WREG32(mmSRBM_SOFT_RESET, tmp);
+			tmp = RREG32(mmSRBM_SOFT_RESET);
+		}
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+	return 0;
+}
+
+static void gfx_v7_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
+						 enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl &= ~CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl |= CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v7_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev,
+						     int me, int pipe,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 mec_int_cntl, mec_int_cntl_reg;
+
+	/*
+	 * amdgpu controls only the first MEC. That's why this function only
+	 * handles the setting of interrupts for this specific MEC. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+
+	if (me == 1) {
+		switch (pipe) {
+		case 0:
+			mec_int_cntl_reg = mmCP_ME1_PIPE0_INT_CNTL;
+			break;
+		case 1:
+			mec_int_cntl_reg = mmCP_ME1_PIPE1_INT_CNTL;
+			break;
+		case 2:
+			mec_int_cntl_reg = mmCP_ME1_PIPE2_INT_CNTL;
+			break;
+		case 3:
+			mec_int_cntl_reg = mmCP_ME1_PIPE3_INT_CNTL;
+			break;
+		default:
+			DRM_DEBUG("invalid pipe %d\n", pipe);
+			return;
+		}
+	} else {
+		DRM_DEBUG("invalid me %d\n", me);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		mec_int_cntl = RREG32(mec_int_cntl_reg);
+		mec_int_cntl &= ~CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK;
+		WREG32(mec_int_cntl_reg, mec_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		mec_int_cntl = RREG32(mec_int_cntl_reg);
+		mec_int_cntl |= CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK;
+		WREG32(mec_int_cntl_reg, mec_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static int gfx_v7_0_set_priv_reg_fault_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl &= ~CP_INT_CNTL_RING0__PRIV_REG_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl |= CP_INT_CNTL_RING0__PRIV_REG_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v7_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *src,
+					      unsigned type,
+					      enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl &= ~CP_INT_CNTL_RING0__PRIV_INSTR_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl |= CP_INT_CNTL_RING0__PRIV_INSTR_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v7_0_set_eop_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CP_IRQ_GFX_EOP:
+		gfx_v7_0_set_gfx_eop_interrupt_state(adev, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 1, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 1, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 1, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE3_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 1, 3, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE0_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 2, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE1_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 2, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE2_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 2, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE3_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 2, 3, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v7_0_eop_irq(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *source,
+			    struct amdgpu_iv_entry *entry)
+{
+	u8 me_id, pipe_id;
+	struct amdgpu_ring *ring;
+	int i;
+
+	DRM_DEBUG("IH: CP EOP\n");
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	switch (me_id) {
+	case 0:
+		amdgpu_fence_process(&adev->gfx.gfx_ring[0]);
+		break;
+	case 1:
+	case 2:
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring[i];
+			if ((ring->me == me_id) && (ring->pipe == pipe_id))
+				amdgpu_fence_process(ring);
+		}
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v7_0_priv_reg_irq(struct amdgpu_device *adev,
+				 struct amdgpu_irq_src *source,
+				 struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal register access in command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int gfx_v7_0_priv_inst_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in command stream\n");
+	// XXX soft reset the gfx block only
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int gfx_v7_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	bool gate = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_CG_STATE_GATE)
+		gate = true;
+
+	gfx_v7_0_enable_gui_idle_interrupt(adev, false);
+	/* order matters! */
+	if (gate) {
+		gfx_v7_0_enable_mgcg(adev, true);
+		gfx_v7_0_enable_cgcg(adev, true);
+	} else {
+		gfx_v7_0_enable_cgcg(adev, false);
+		gfx_v7_0_enable_mgcg(adev, false);
+	}
+	gfx_v7_0_enable_gui_idle_interrupt(adev, true);
+
+	return 0;
+}
+
+static int gfx_v7_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	bool gate = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_PG_STATE_GATE)
+		gate = true;
+
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_PG |
+			      AMD_PG_SUPPORT_GFX_SMG |
+			      AMD_PG_SUPPORT_GFX_DMG |
+			      AMD_PG_SUPPORT_CP |
+			      AMD_PG_SUPPORT_GDS |
+			      AMD_PG_SUPPORT_RLC_SMU_HS)) {
+		gfx_v7_0_update_gfx_pg(adev, gate);
+		if (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG) {
+			gfx_v7_0_enable_cp_pg(adev, gate);
+			gfx_v7_0_enable_gds_pg(adev, gate);
+		}
+	}
+
+	return 0;
+}
+
+static const struct amd_ip_funcs gfx_v7_0_ip_funcs = {
+	.name = "gfx_v7_0",
+	.early_init = gfx_v7_0_early_init,
+	.late_init = gfx_v7_0_late_init,
+	.sw_init = gfx_v7_0_sw_init,
+	.sw_fini = gfx_v7_0_sw_fini,
+	.hw_init = gfx_v7_0_hw_init,
+	.hw_fini = gfx_v7_0_hw_fini,
+	.suspend = gfx_v7_0_suspend,
+	.resume = gfx_v7_0_resume,
+	.is_idle = gfx_v7_0_is_idle,
+	.wait_for_idle = gfx_v7_0_wait_for_idle,
+	.soft_reset = gfx_v7_0_soft_reset,
+	.set_clockgating_state = gfx_v7_0_set_clockgating_state,
+	.set_powergating_state = gfx_v7_0_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = {
+	.type = AMDGPU_RING_TYPE_GFX,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = false,
+	.get_rptr = gfx_v7_0_ring_get_rptr,
+	.get_wptr = gfx_v7_0_ring_get_wptr_gfx,
+	.set_wptr = gfx_v7_0_ring_set_wptr_gfx,
+	.emit_frame_size =
+		20 + /* gfx_v7_0_ring_emit_gds_switch */
+		7 + /* gfx_v7_0_ring_emit_hdp_flush */
+		5 + /* hdp invalidate */
+		12 + 12 + 12 + /* gfx_v7_0_ring_emit_fence_gfx x3 for user fence, vm fence */
+		7 + 4 + /* gfx_v7_0_ring_emit_pipeline_sync */
+		CIK_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + 6 + /* gfx_v7_0_ring_emit_vm_flush */
+		3 + 4, /* gfx_v7_ring_emit_cntxcntl including vgt flush*/
+	.emit_ib_size = 4, /* gfx_v7_0_ring_emit_ib_gfx */
+	.emit_ib = gfx_v7_0_ring_emit_ib_gfx,
+	.emit_fence = gfx_v7_0_ring_emit_fence_gfx,
+	.emit_pipeline_sync = gfx_v7_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v7_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v7_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v7_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v7_0_ring_test_ring,
+	.test_ib = gfx_v7_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_cntxcntl = gfx_v7_ring_emit_cntxcntl,
+	.emit_wreg = gfx_v7_0_ring_emit_wreg,
+};
+
+static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = {
+	.type = AMDGPU_RING_TYPE_COMPUTE,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = false,
+	.get_rptr = gfx_v7_0_ring_get_rptr,
+	.get_wptr = gfx_v7_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v7_0_ring_set_wptr_compute,
+	.emit_frame_size =
+		20 + /* gfx_v7_0_ring_emit_gds_switch */
+		7 + /* gfx_v7_0_ring_emit_hdp_flush */
+		5 + /* hdp invalidate */
+		7 + /* gfx_v7_0_ring_emit_pipeline_sync */
+		CIK_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + /* gfx_v7_0_ring_emit_vm_flush */
+		7 + 7 + 7, /* gfx_v7_0_ring_emit_fence_compute x3 for user fence, vm fence */
+	.emit_ib_size =	4, /* gfx_v7_0_ring_emit_ib_compute */
+	.emit_ib = gfx_v7_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v7_0_ring_emit_fence_compute,
+	.emit_pipeline_sync = gfx_v7_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v7_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v7_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v7_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v7_0_ring_test_ring,
+	.test_ib = gfx_v7_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_wreg = gfx_v7_0_ring_emit_wreg,
+};
+
+static void gfx_v7_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		adev->gfx.gfx_ring[i].funcs = &gfx_v7_0_ring_funcs_gfx;
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		adev->gfx.compute_ring[i].funcs = &gfx_v7_0_ring_funcs_compute;
+}
+
+static const struct amdgpu_irq_src_funcs gfx_v7_0_eop_irq_funcs = {
+	.set = gfx_v7_0_set_eop_interrupt_state,
+	.process = gfx_v7_0_eop_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v7_0_priv_reg_irq_funcs = {
+	.set = gfx_v7_0_set_priv_reg_fault_state,
+	.process = gfx_v7_0_priv_reg_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v7_0_priv_inst_irq_funcs = {
+	.set = gfx_v7_0_set_priv_inst_fault_state,
+	.process = gfx_v7_0_priv_inst_irq,
+};
+
+static void gfx_v7_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.eop_irq.num_types = AMDGPU_CP_IRQ_LAST;
+	adev->gfx.eop_irq.funcs = &gfx_v7_0_eop_irq_funcs;
+
+	adev->gfx.priv_reg_irq.num_types = 1;
+	adev->gfx.priv_reg_irq.funcs = &gfx_v7_0_priv_reg_irq_funcs;
+
+	adev->gfx.priv_inst_irq.num_types = 1;
+	adev->gfx.priv_inst_irq.funcs = &gfx_v7_0_priv_inst_irq_funcs;
+}
+
+static void gfx_v7_0_set_gds_init(struct amdgpu_device *adev)
+{
+	/* init asci gds info */
+	adev->gds.mem.total_size = RREG32(mmGDS_VMID0_SIZE);
+	adev->gds.gws.total_size = 64;
+	adev->gds.oa.total_size = 16;
+
+	if (adev->gds.mem.total_size == 64 * 1024) {
+		adev->gds.mem.gfx_partition_size = 4096;
+		adev->gds.mem.cs_partition_size = 4096;
+
+		adev->gds.gws.gfx_partition_size = 4;
+		adev->gds.gws.cs_partition_size = 4;
+
+		adev->gds.oa.gfx_partition_size = 4;
+		adev->gds.oa.cs_partition_size = 1;
+	} else {
+		adev->gds.mem.gfx_partition_size = 1024;
+		adev->gds.mem.cs_partition_size = 1024;
+
+		adev->gds.gws.gfx_partition_size = 16;
+		adev->gds.gws.cs_partition_size = 16;
+
+		adev->gds.oa.gfx_partition_size = 4;
+		adev->gds.oa.cs_partition_size = 4;
+	}
+}
+
+
+static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev)
+{
+	int i, j, k, counter, active_cu_number = 0;
+	u32 mask, bitmap, ao_bitmap, ao_cu_mask = 0;
+	struct amdgpu_cu_info *cu_info = &adev->gfx.cu_info;
+	unsigned disable_masks[4 * 2];
+	u32 ao_cu_num;
+
+	if (adev->flags & AMD_IS_APU)
+		ao_cu_num = 2;
+	else
+		ao_cu_num = adev->gfx.config.max_cu_per_sh;
+
+	memset(cu_info, 0, sizeof(*cu_info));
+
+	amdgpu_gfx_parse_disable_cu(disable_masks, 4, 2);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			mask = 1;
+			ao_bitmap = 0;
+			counter = 0;
+			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
+			if (i < 4 && j < 2)
+				gfx_v7_0_set_user_cu_inactive_bitmap(
+					adev, disable_masks[i * 2 + j]);
+			bitmap = gfx_v7_0_get_cu_active_bitmap(adev);
+			cu_info->bitmap[i][j] = bitmap;
+
+			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) {
+				if (bitmap & mask) {
+					if (counter < ao_cu_num)
+						ao_bitmap |= mask;
+					counter ++;
+				}
+				mask <<= 1;
+			}
+			active_cu_number += counter;
+			if (i < 2 && j < 2)
+				ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8));
+			cu_info->ao_cu_bitmap[i][j] = ao_bitmap;
+		}
+	}
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	cu_info->number = active_cu_number;
+	cu_info->ao_cu_mask = ao_cu_mask;
+	cu_info->simd_per_cu = NUM_SIMD_PER_CU;
+	cu_info->max_waves_per_simd = 10;
+	cu_info->max_scratch_slots_per_cu = 32;
+	cu_info->wave_front_size = 64;
+	cu_info->lds_size = 64;
+}
+
+const struct amdgpu_ip_block_version gfx_v7_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GFX,
+	.major = 7,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gfx_v7_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version gfx_v7_1_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GFX,
+	.major = 7,
+	.minor = 1,
+	.rev = 0,
+	.funcs = &gfx_v7_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version gfx_v7_2_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GFX,
+	.major = 7,
+	.minor = 2,
+	.rev = 0,
+	.funcs = &gfx_v7_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version gfx_v7_3_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GFX,
+	.major = 7,
+	.minor = 3,
+	.rev = 0,
+	.funcs = &gfx_v7_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/gfx_v7_0.h b/sys/dev/pci/drm/amd/amdgpu/gfx_v7_0.h
new file mode 100644
index 00000000000..6fb9c152469
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gfx_v7_0.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V7_0_H__
+#define __GFX_V7_0_H__
+
+extern const struct amdgpu_ip_block_version gfx_v7_0_ip_block;
+extern const struct amdgpu_ip_block_version gfx_v7_1_ip_block;
+extern const struct amdgpu_ip_block_version gfx_v7_2_ip_block;
+extern const struct amdgpu_ip_block_version gfx_v7_3_ip_block;
+
+struct amdgpu_device;
+struct cik_mqd;
+
+int gfx_v7_0_mqd_commit(struct amdgpu_device *adev, struct cik_mqd *mqd);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/gfx_v8_0.c b/sys/dev/pci/drm/amd/amdgpu/gfx_v8_0.c
new file mode 100644
index 00000000000..d358a5736b7
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gfx_v8_0.c
@@ -0,0 +1,7502 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include "vi.h"
+#include "vi_structs.h"
+#include "vid.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_atombios.h"
+#include "atombios_i2c.h"
+#include "clearstate_vi.h"
+
+#include "gmc/gmc_8_2_d.h"
+#include "gmc/gmc_8_2_sh_mask.h"
+
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+
+#include "bif/bif_5_0_d.h"
+#include "bif/bif_5_0_sh_mask.h"
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_enum.h"
+#include "gca/gfx_8_0_sh_mask.h"
+#include "gca/gfx_8_0_enum.h"
+
+#include "dce/dce_10_0_d.h"
+#include "dce/dce_10_0_sh_mask.h"
+
+#include "smu/smu_7_1_3_d.h"
+
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+#define GFX8_NUM_GFX_RINGS     1
+#define GFX8_MEC_HPD_SIZE 2048
+
+#define TOPAZ_GB_ADDR_CONFIG_GOLDEN 0x22010001
+#define CARRIZO_GB_ADDR_CONFIG_GOLDEN 0x22010001
+#define POLARIS11_GB_ADDR_CONFIG_GOLDEN 0x22011002
+#define TONGA_GB_ADDR_CONFIG_GOLDEN 0x22011003
+
+#define ARRAY_MODE(x)					((x) << GB_TILE_MODE0__ARRAY_MODE__SHIFT)
+#define PIPE_CONFIG(x)					((x) << GB_TILE_MODE0__PIPE_CONFIG__SHIFT)
+#define TILE_SPLIT(x)					((x) << GB_TILE_MODE0__TILE_SPLIT__SHIFT)
+#define MICRO_TILE_MODE_NEW(x)				((x) << GB_TILE_MODE0__MICRO_TILE_MODE_NEW__SHIFT)
+#define SAMPLE_SPLIT(x)					((x) << GB_TILE_MODE0__SAMPLE_SPLIT__SHIFT)
+#define BANK_WIDTH(x)					((x) << GB_MACROTILE_MODE0__BANK_WIDTH__SHIFT)
+#define BANK_HEIGHT(x)					((x) << GB_MACROTILE_MODE0__BANK_HEIGHT__SHIFT)
+#define MACRO_TILE_ASPECT(x)				((x) << GB_MACROTILE_MODE0__MACRO_TILE_ASPECT__SHIFT)
+#define NUM_BANKS(x)					((x) << GB_MACROTILE_MODE0__NUM_BANKS__SHIFT)
+
+#define RLC_CGTT_MGCG_OVERRIDE__CPF_MASK            0x00000001L
+#define RLC_CGTT_MGCG_OVERRIDE__RLC_MASK            0x00000002L
+#define RLC_CGTT_MGCG_OVERRIDE__MGCG_MASK           0x00000004L
+#define RLC_CGTT_MGCG_OVERRIDE__CGCG_MASK           0x00000008L
+#define RLC_CGTT_MGCG_OVERRIDE__CGLS_MASK           0x00000010L
+#define RLC_CGTT_MGCG_OVERRIDE__GRBM_MASK           0x00000020L
+
+/* BPM SERDES CMD */
+#define SET_BPM_SERDES_CMD    1
+#define CLE_BPM_SERDES_CMD    0
+
+/* BPM Register Address*/
+enum {
+	BPM_REG_CGLS_EN = 0,        /* Enable/Disable CGLS */
+	BPM_REG_CGLS_ON,            /* ON/OFF CGLS: shall be controlled by RLC FW */
+	BPM_REG_CGCG_OVERRIDE,      /* Set/Clear CGCG Override */
+	BPM_REG_MGCG_OVERRIDE,      /* Set/Clear MGCG Override */
+	BPM_REG_FGCG_OVERRIDE,      /* Set/Clear FGCG Override */
+	BPM_REG_FGCG_MAX
+};
+
+#define RLC_FormatDirectRegListLength        14
+
+MODULE_FIRMWARE("amdgpu/carrizo_ce.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_pfp.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_me.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_mec.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_mec2.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/stoney_ce.bin");
+MODULE_FIRMWARE("amdgpu/stoney_pfp.bin");
+MODULE_FIRMWARE("amdgpu/stoney_me.bin");
+MODULE_FIRMWARE("amdgpu/stoney_mec.bin");
+MODULE_FIRMWARE("amdgpu/stoney_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/tonga_ce.bin");
+MODULE_FIRMWARE("amdgpu/tonga_pfp.bin");
+MODULE_FIRMWARE("amdgpu/tonga_me.bin");
+MODULE_FIRMWARE("amdgpu/tonga_mec.bin");
+MODULE_FIRMWARE("amdgpu/tonga_mec2.bin");
+MODULE_FIRMWARE("amdgpu/tonga_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/topaz_ce.bin");
+MODULE_FIRMWARE("amdgpu/topaz_pfp.bin");
+MODULE_FIRMWARE("amdgpu/topaz_me.bin");
+MODULE_FIRMWARE("amdgpu/topaz_mec.bin");
+MODULE_FIRMWARE("amdgpu/topaz_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/fiji_ce.bin");
+MODULE_FIRMWARE("amdgpu/fiji_pfp.bin");
+MODULE_FIRMWARE("amdgpu/fiji_me.bin");
+MODULE_FIRMWARE("amdgpu/fiji_mec.bin");
+MODULE_FIRMWARE("amdgpu/fiji_mec2.bin");
+MODULE_FIRMWARE("amdgpu/fiji_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/polaris10_ce.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_ce_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_pfp.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_pfp_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_me.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_me_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_mec.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_mec_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_mec2.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_mec2_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/polaris11_ce.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_ce_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_pfp.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_pfp_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_me.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_me_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_mec.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_mec_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_mec2.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_mec2_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/polaris12_ce.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_ce_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_pfp.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_pfp_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_me.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_me_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_mec.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_mec_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_mec2.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_mec2_2.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/vegam_ce.bin");
+MODULE_FIRMWARE("amdgpu/vegam_pfp.bin");
+MODULE_FIRMWARE("amdgpu/vegam_me.bin");
+MODULE_FIRMWARE("amdgpu/vegam_mec.bin");
+MODULE_FIRMWARE("amdgpu/vegam_mec2.bin");
+MODULE_FIRMWARE("amdgpu/vegam_rlc.bin");
+
+static const struct amdgpu_gds_reg_offset amdgpu_gds_reg_offset[] =
+{
+	{mmGDS_VMID0_BASE, mmGDS_VMID0_SIZE, mmGDS_GWS_VMID0, mmGDS_OA_VMID0},
+	{mmGDS_VMID1_BASE, mmGDS_VMID1_SIZE, mmGDS_GWS_VMID1, mmGDS_OA_VMID1},
+	{mmGDS_VMID2_BASE, mmGDS_VMID2_SIZE, mmGDS_GWS_VMID2, mmGDS_OA_VMID2},
+	{mmGDS_VMID3_BASE, mmGDS_VMID3_SIZE, mmGDS_GWS_VMID3, mmGDS_OA_VMID3},
+	{mmGDS_VMID4_BASE, mmGDS_VMID4_SIZE, mmGDS_GWS_VMID4, mmGDS_OA_VMID4},
+	{mmGDS_VMID5_BASE, mmGDS_VMID5_SIZE, mmGDS_GWS_VMID5, mmGDS_OA_VMID5},
+	{mmGDS_VMID6_BASE, mmGDS_VMID6_SIZE, mmGDS_GWS_VMID6, mmGDS_OA_VMID6},
+	{mmGDS_VMID7_BASE, mmGDS_VMID7_SIZE, mmGDS_GWS_VMID7, mmGDS_OA_VMID7},
+	{mmGDS_VMID8_BASE, mmGDS_VMID8_SIZE, mmGDS_GWS_VMID8, mmGDS_OA_VMID8},
+	{mmGDS_VMID9_BASE, mmGDS_VMID9_SIZE, mmGDS_GWS_VMID9, mmGDS_OA_VMID9},
+	{mmGDS_VMID10_BASE, mmGDS_VMID10_SIZE, mmGDS_GWS_VMID10, mmGDS_OA_VMID10},
+	{mmGDS_VMID11_BASE, mmGDS_VMID11_SIZE, mmGDS_GWS_VMID11, mmGDS_OA_VMID11},
+	{mmGDS_VMID12_BASE, mmGDS_VMID12_SIZE, mmGDS_GWS_VMID12, mmGDS_OA_VMID12},
+	{mmGDS_VMID13_BASE, mmGDS_VMID13_SIZE, mmGDS_GWS_VMID13, mmGDS_OA_VMID13},
+	{mmGDS_VMID14_BASE, mmGDS_VMID14_SIZE, mmGDS_GWS_VMID14, mmGDS_OA_VMID14},
+	{mmGDS_VMID15_BASE, mmGDS_VMID15_SIZE, mmGDS_GWS_VMID15, mmGDS_OA_VMID15}
+};
+
+static const u32 golden_settings_tonga_a11[] =
+{
+	mmCB_HW_CONTROL, 0xfffdf3cf, 0x00007208,
+	mmCB_HW_CONTROL_3, 0x00000040, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmGB_GPU_ID, 0x0000000f, 0x00000000,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_FIFO_DEPTH_CNTL, 0x000003ff, 0x000000fc,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0000003c,
+	mmSQ_RANDOM_WAVE_PRI, 0x001fffff, 0x000006fd,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000002fb,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x0000543b,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0xa9210876,
+	mmVGT_RESET_DEBUG, 0x00000004, 0x00000004,
+};
+
+static const u32 tonga_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG, 0xffffffff, 0x16000012,
+	mmPA_SC_RASTER_CONFIG_1, 0xffffffff, 0x0000002A,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22011003,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF
+};
+
+static const u32 tonga_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xffffffff,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPF_CLK_CTRL, 0xffffffff, 0x40000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_WD_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL4, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCGTS_CU0_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU0_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU0_TA_SQC_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU0_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU0_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU1_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU1_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU1_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU1_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU1_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU2_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU2_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU2_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU2_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU2_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU3_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU3_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU3_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU3_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU3_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU4_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU4_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU4_TA_SQC_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU4_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU4_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU5_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU5_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU5_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU5_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU5_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU6_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU6_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU6_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU6_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU6_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU7_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU7_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU7_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU7_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU7_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96e00200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003c,
+	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
+};
+
+static const u32 golden_settings_vegam_a11[] =
+{
+	mmCB_HW_CONTROL, 0x0001f3cf, 0x00007208,
+	mmCB_HW_CONTROL_2, 0x0f000000, 0x0d000000,
+	mmCB_HW_CONTROL_3, 0x000001ff, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmPA_SC_RASTER_CONFIG, 0x3f3fffff, 0x3a00161a,
+	mmPA_SC_RASTER_CONFIG_1, 0x0000003f, 0x0000002e,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0001003c,
+	mmRLC_CGCG_CGLS_CTRL_3D, 0xffffffff, 0x0001003c,
+	mmSQ_CONFIG, 0x07f80000, 0x01180000,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000f7,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x32761054,
+	mmVGT_RESET_DEBUG, 0x00000004, 0x00000004,
+};
+
+static const u32 vegam_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22011003,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF,
+};
+
+static const u32 golden_settings_polaris11_a11[] =
+{
+	mmCB_HW_CONTROL, 0x0000f3cf, 0x00007208,
+	mmCB_HW_CONTROL_2, 0x0f000000, 0x0f000000,
+	mmCB_HW_CONTROL_3, 0x000001ff, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmPA_SC_RASTER_CONFIG, 0x3f3fffff, 0x16000012,
+	mmPA_SC_RASTER_CONFIG_1, 0x0000003f, 0x00000000,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0001003c,
+	mmRLC_CGCG_CGLS_CTRL_3D, 0xffffffff, 0x0001003c,
+	mmSQ_CONFIG, 0x07f80000, 0x01180000,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000f3,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00003210,
+	mmVGT_RESET_DEBUG, 0x00000004, 0x00000004,
+};
+
+static const u32 polaris11_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22011002,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF,
+};
+
+static const u32 golden_settings_polaris10_a11[] =
+{
+	mmATC_MISC_CG, 0x000c0fc0, 0x000c0200,
+	mmCB_HW_CONTROL, 0x0001f3cf, 0x00007208,
+	mmCB_HW_CONTROL_2, 0x0f000000, 0x0f000000,
+	mmCB_HW_CONTROL_3, 0x000001ff, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmPA_SC_RASTER_CONFIG, 0x3f3fffff, 0x16000012,
+	mmPA_SC_RASTER_CONFIG_1, 0x0000003f, 0x0000002a,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0001003c,
+	mmRLC_CGCG_CGLS_CTRL_3D, 0xffffffff, 0x0001003c,
+	mmSQ_CONFIG, 0x07f80000, 0x07180000,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000f7,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000,
+	mmVGT_RESET_DEBUG, 0x00000004, 0x00000004,
+};
+
+static const u32 polaris10_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG, 0xffffffff, 0x16000012,
+	mmPA_SC_RASTER_CONFIG_1, 0xffffffff, 0x0000002A,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22011003,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF,
+};
+
+static const u32 fiji_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG, 0xffffffff, 0x3a00161a,
+	mmPA_SC_RASTER_CONFIG_1, 0xffffffff, 0x0000002e,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22011003,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmSPI_CONFIG_CNTL_1, 0x0000000f, 0x00000009,
+};
+
+static const u32 golden_settings_fiji_a10[] =
+{
+	mmCB_HW_CONTROL_3, 0x000001ff, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0001003c,
+	mmSQ_RANDOM_WAVE_PRI, 0x001fffff, 0x000006fd,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000ff,
+	mmVGT_RESET_DEBUG, 0x00000004, 0x00000004,
+};
+
+static const u32 fiji_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xffffffff,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPF_CLK_CTRL, 0xffffffff, 0x40000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_WD_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL4, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96e00200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003c,
+	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
+};
+
+static const u32 golden_settings_iceland_a11[] =
+{
+	mmCB_HW_CONTROL_3, 0x00000040, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmDB_DEBUG3, 0xc0000000, 0xc0000000,
+	mmGB_GPU_ID, 0x0000000f, 0x00000000,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmPA_SC_RASTER_CONFIG, 0x3f3fffff, 0x00000002,
+	mmPA_SC_RASTER_CONFIG_1, 0x0000003f, 0x00000000,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0000003c,
+	mmSQ_RANDOM_WAVE_PRI, 0x001fffff, 0x000006fd,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000f1,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00000010,
+};
+
+static const u32 iceland_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG, 0xffffffff, 0x00000002,
+	mmPA_SC_RASTER_CONFIG_1, 0xffffffff, 0x00000000,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22010001,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF
+};
+
+static const u32 iceland_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xffffffff,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0xc0000100,
+	mmCGTT_CPC_CLK_CTRL, 0xffffffff, 0xc0000100,
+	mmCGTT_CPF_CLK_CTRL, 0xffffffff, 0xc0000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_WD_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL4, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0xff000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCGTS_CU0_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU0_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU0_TA_SQC_CTRL_REG, 0xffffffff, 0x0f840f87,
+	mmCGTS_CU0_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU0_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU1_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU1_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU1_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU1_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU1_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU2_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU2_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU2_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU2_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU2_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU3_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU3_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU3_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU3_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU3_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU4_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU4_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU4_TA_SQC_CTRL_REG, 0xffffffff, 0x0f840f87,
+	mmCGTS_CU4_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU4_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU5_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU5_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU5_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU5_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU5_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96e00200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003c,
+};
+
+static const u32 cz_golden_settings_a11[] =
+{
+	mmCB_HW_CONTROL_3, 0x00000040, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmGB_GPU_ID, 0x0000000f, 0x00000000,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x00000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0000003c,
+	mmSQ_RANDOM_WAVE_PRI, 0x001fffff, 0x000006fd,
+	mmTA_CNTL_AUX, 0x000f000f, 0x00010000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
+	mmTCP_ADDR_CONFIG, 0x0000000f, 0x000000f3,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00001302
+};
+
+static const u32 cz_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG, 0xffffffff, 0x00000002,
+	mmPA_SC_RASTER_CONFIG_1, 0xffffffff, 0x00000000,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22010001,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF
+};
+
+static const u32 cz_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xffffffff,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPF_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_WD_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL4, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCGTS_CU0_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU0_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU0_TA_SQC_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU0_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU0_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU1_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU1_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU1_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU1_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU1_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU2_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU2_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU2_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU2_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU2_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU3_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU3_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU3_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU3_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU3_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU4_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU4_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU4_TA_SQC_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU4_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU4_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU5_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU5_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU5_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU5_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU5_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU6_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU6_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU6_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU6_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU6_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU7_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU7_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU7_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU7_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU7_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96e00200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003f,
+	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
+};
+
+static const u32 stoney_golden_settings_a11[] =
+{
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmGB_GPU_ID, 0x0000000f, 0x00000000,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0001003c,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
+	mmTCP_ADDR_CONFIG, 0x0000000f, 0x000000f1,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x10101010,
+};
+
+static const u32 stoney_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG, 0xffffffff, 0x00000000,
+	mmPA_SC_RASTER_CONFIG_1, 0xffffffff, 0x00000000,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x12010001,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF,
+};
+
+static const u32 stoney_mgcg_cgcg_init[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003f,
+	mmCP_MEM_SLP_CNTL, 0xffffffff, 0x00020201,
+	mmRLC_MEM_SLP_CNTL, 0xffffffff, 0x00020201,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96940200,
+};
+
+
+static const char * const sq_edc_source_names[] = {
+	"SQ_EDC_INFO_SOURCE_INVALID: No EDC error has occurred",
+	"SQ_EDC_INFO_SOURCE_INST: EDC source is Instruction Fetch",
+	"SQ_EDC_INFO_SOURCE_SGPR: EDC source is SGPR or SQC data return",
+	"SQ_EDC_INFO_SOURCE_VGPR: EDC source is VGPR",
+	"SQ_EDC_INFO_SOURCE_LDS: EDC source is LDS",
+	"SQ_EDC_INFO_SOURCE_GDS: EDC source is GDS",
+	"SQ_EDC_INFO_SOURCE_TA: EDC source is TA",
+};
+
+static void gfx_v8_0_set_ring_funcs(struct amdgpu_device *adev);
+static void gfx_v8_0_set_irq_funcs(struct amdgpu_device *adev);
+static void gfx_v8_0_set_gds_init(struct amdgpu_device *adev);
+static void gfx_v8_0_set_rlc_funcs(struct amdgpu_device *adev);
+static u32 gfx_v8_0_get_csb_size(struct amdgpu_device *adev);
+static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev);
+static void gfx_v8_0_ring_emit_ce_meta(struct amdgpu_ring *ring);
+static void gfx_v8_0_ring_emit_de_meta(struct amdgpu_ring *ring);
+
+static void gfx_v8_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		amdgpu_device_program_register_sequence(adev,
+							iceland_mgcg_cgcg_init,
+							ARRAY_SIZE(iceland_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_iceland_a11,
+							ARRAY_SIZE(golden_settings_iceland_a11));
+		amdgpu_device_program_register_sequence(adev,
+							iceland_golden_common_all,
+							ARRAY_SIZE(iceland_golden_common_all));
+		break;
+	case CHIP_FIJI:
+		amdgpu_device_program_register_sequence(adev,
+							fiji_mgcg_cgcg_init,
+							ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_fiji_a10,
+							ARRAY_SIZE(golden_settings_fiji_a10));
+		amdgpu_device_program_register_sequence(adev,
+							fiji_golden_common_all,
+							ARRAY_SIZE(fiji_golden_common_all));
+		break;
+
+	case CHIP_TONGA:
+		amdgpu_device_program_register_sequence(adev,
+							tonga_mgcg_cgcg_init,
+							ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_tonga_a11,
+							ARRAY_SIZE(golden_settings_tonga_a11));
+		amdgpu_device_program_register_sequence(adev,
+							tonga_golden_common_all,
+							ARRAY_SIZE(tonga_golden_common_all));
+		break;
+	case CHIP_VEGAM:
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_vegam_a11,
+							ARRAY_SIZE(golden_settings_vegam_a11));
+		amdgpu_device_program_register_sequence(adev,
+							vegam_golden_common_all,
+							ARRAY_SIZE(vegam_golden_common_all));
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_polaris11_a11,
+							ARRAY_SIZE(golden_settings_polaris11_a11));
+		amdgpu_device_program_register_sequence(adev,
+							polaris11_golden_common_all,
+							ARRAY_SIZE(polaris11_golden_common_all));
+		break;
+	case CHIP_POLARIS10:
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_polaris10_a11,
+							ARRAY_SIZE(golden_settings_polaris10_a11));
+		amdgpu_device_program_register_sequence(adev,
+							polaris10_golden_common_all,
+							ARRAY_SIZE(polaris10_golden_common_all));
+		WREG32_SMC(ixCG_ACLK_CNTL, 0x0000001C);
+		if (adev->pdev->revision == 0xc7 &&
+		    ((adev->pdev->subsystem_device == 0xb37 && adev->pdev->subsystem_vendor == 0x1002) ||
+		     (adev->pdev->subsystem_device == 0x4a8 && adev->pdev->subsystem_vendor == 0x1043) ||
+		     (adev->pdev->subsystem_device == 0x9480 && adev->pdev->subsystem_vendor == 0x1682))) {
+			amdgpu_atombios_i2c_channel_trans(adev, 0x10, 0x96, 0x1E, 0xDD);
+			amdgpu_atombios_i2c_channel_trans(adev, 0x10, 0x96, 0x1F, 0xD0);
+		}
+		break;
+	case CHIP_CARRIZO:
+		amdgpu_device_program_register_sequence(adev,
+							cz_mgcg_cgcg_init,
+							ARRAY_SIZE(cz_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							cz_golden_settings_a11,
+							ARRAY_SIZE(cz_golden_settings_a11));
+		amdgpu_device_program_register_sequence(adev,
+							cz_golden_common_all,
+							ARRAY_SIZE(cz_golden_common_all));
+		break;
+	case CHIP_STONEY:
+		amdgpu_device_program_register_sequence(adev,
+							stoney_mgcg_cgcg_init,
+							ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							stoney_golden_settings_a11,
+							ARRAY_SIZE(stoney_golden_settings_a11));
+		amdgpu_device_program_register_sequence(adev,
+							stoney_golden_common_all,
+							ARRAY_SIZE(stoney_golden_common_all));
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v8_0_scratch_init(struct amdgpu_device *adev)
+{
+	adev->gfx.scratch.num_reg = 8;
+	adev->gfx.scratch.reg_base = mmSCRATCH_REG0;
+	adev->gfx.scratch.free_mask = (1u << adev->gfx.scratch.num_reg) - 1;
+}
+
+static int gfx_v8_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t scratch;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		amdgpu_gfx_scratch_free(adev, scratch);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+	amdgpu_ring_write(ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
+			  ring->idx, scratch, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_gfx_scratch_free(adev, scratch);
+	return r;
+}
+
+static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+
+	unsigned int index;
+	uint64_t gpu_addr;
+	uint32_t tmp;
+	long r;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, 16, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err1;
+	}
+	ib.ptr[0] = PACKET3(PACKET3_WRITE_DATA, 3);
+	ib.ptr[1] = WRITE_DATA_DST_SEL(5) | WR_CONFIRM;
+	ib.ptr[2] = lower_32_bits(gpu_addr);
+	ib.ptr[3] = upper_32_bits(gpu_addr);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.length_dw = 5;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err2;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out.\n");
+		r = -ETIMEDOUT;
+		goto err2;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto err2;
+	}
+
+	tmp = adev->wb.wb[index];
+	if (tmp == 0xDEADBEEF) {
+		DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
+	} else {
+		DRM_ERROR("ib test on ring %d failed\n", ring->idx);
+		r = -EINVAL;
+	}
+
+err2:
+	amdgpu_ib_free(adev, &ib, NULL);
+	dma_fence_put(f);
+err1:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+
+static void gfx_v8_0_free_microcode(struct amdgpu_device *adev)
+{
+	release_firmware(adev->gfx.pfp_fw);
+	adev->gfx.pfp_fw = NULL;
+	release_firmware(adev->gfx.me_fw);
+	adev->gfx.me_fw = NULL;
+	release_firmware(adev->gfx.ce_fw);
+	adev->gfx.ce_fw = NULL;
+	release_firmware(adev->gfx.rlc_fw);
+	adev->gfx.rlc_fw = NULL;
+	release_firmware(adev->gfx.mec_fw);
+	adev->gfx.mec_fw = NULL;
+	if ((adev->asic_type != CHIP_STONEY) &&
+	    (adev->asic_type != CHIP_TOPAZ))
+		release_firmware(adev->gfx.mec2_fw);
+	adev->gfx.mec2_fw = NULL;
+
+	kfree(adev->gfx.rlc.register_list_format);
+}
+
+static int gfx_v8_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+	struct amdgpu_firmware_info *info = NULL;
+	const struct common_firmware_header *header = NULL;
+	const struct gfx_firmware_header_v1_0 *cp_hdr;
+	const struct rlc_firmware_header_v2_0 *rlc_hdr;
+	unsigned int *tmp = NULL, i;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		chip_name = "topaz";
+		break;
+	case CHIP_TONGA:
+		chip_name = "tonga";
+		break;
+	case CHIP_CARRIZO:
+		chip_name = "carrizo";
+		break;
+	case CHIP_FIJI:
+		chip_name = "fiji";
+		break;
+	case CHIP_STONEY:
+		chip_name = "stoney";
+		break;
+	case CHIP_POLARIS10:
+		chip_name = "polaris10";
+		break;
+	case CHIP_POLARIS11:
+		chip_name = "polaris11";
+		break;
+	case CHIP_POLARIS12:
+		chip_name = "polaris12";
+		break;
+	case CHIP_VEGAM:
+		chip_name = "vegam";
+		break;
+	default:
+		BUG();
+	}
+
+	if (adev->asic_type >= CHIP_POLARIS10 && adev->asic_type <= CHIP_POLARIS12) {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_pfp_2.bin", chip_name);
+		err = request_firmware(&adev->gfx.pfp_fw, fw_name, adev->dev);
+		if (err == -ENOENT) {
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_pfp.bin", chip_name);
+			err = request_firmware(&adev->gfx.pfp_fw, fw_name, adev->dev);
+		}
+	} else {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_pfp.bin", chip_name);
+		err = request_firmware(&adev->gfx.pfp_fw, fw_name, adev->dev);
+	}
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.pfp_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+	adev->gfx.pfp_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.pfp_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	if (adev->asic_type >= CHIP_POLARIS10 && adev->asic_type <= CHIP_POLARIS12) {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_me_2.bin", chip_name);
+		err = request_firmware(&adev->gfx.me_fw, fw_name, adev->dev);
+		if (err == -ENOENT) {
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_me.bin", chip_name);
+			err = request_firmware(&adev->gfx.me_fw, fw_name, adev->dev);
+		}
+	} else {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_me.bin", chip_name);
+		err = request_firmware(&adev->gfx.me_fw, fw_name, adev->dev);
+	}
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.me_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+	adev->gfx.me_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+
+	adev->gfx.me_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	if (adev->asic_type >= CHIP_POLARIS10 && adev->asic_type <= CHIP_POLARIS12) {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ce_2.bin", chip_name);
+		err = request_firmware(&adev->gfx.ce_fw, fw_name, adev->dev);
+		if (err == -ENOENT) {
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ce.bin", chip_name);
+			err = request_firmware(&adev->gfx.ce_fw, fw_name, adev->dev);
+		}
+	} else {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ce.bin", chip_name);
+		err = request_firmware(&adev->gfx.ce_fw, fw_name, adev->dev);
+	}
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.ce_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+	adev->gfx.ce_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.ce_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	/*
+	 * Support for MCBP/Virtualization in combination with chained IBs is
+	 * formal released on feature version #46
+	 */
+	if (adev->gfx.ce_feature_version >= 46 &&
+	    adev->gfx.pfp_feature_version >= 46) {
+		adev->virt.chained_ib_support = true;
+		DRM_INFO("Chained IB support enabled!\n");
+	} else
+		adev->virt.chained_ib_support = false;
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc.bin", chip_name);
+	err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.rlc_fw);
+	rlc_hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+	adev->gfx.rlc_fw_version = le32_to_cpu(rlc_hdr->header.ucode_version);
+	adev->gfx.rlc_feature_version = le32_to_cpu(rlc_hdr->ucode_feature_version);
+
+	adev->gfx.rlc.save_and_restore_offset =
+			le32_to_cpu(rlc_hdr->save_and_restore_offset);
+	adev->gfx.rlc.clear_state_descriptor_offset =
+			le32_to_cpu(rlc_hdr->clear_state_descriptor_offset);
+	adev->gfx.rlc.avail_scratch_ram_locations =
+			le32_to_cpu(rlc_hdr->avail_scratch_ram_locations);
+	adev->gfx.rlc.reg_restore_list_size =
+			le32_to_cpu(rlc_hdr->reg_restore_list_size);
+	adev->gfx.rlc.reg_list_format_start =
+			le32_to_cpu(rlc_hdr->reg_list_format_start);
+	adev->gfx.rlc.reg_list_format_separate_start =
+			le32_to_cpu(rlc_hdr->reg_list_format_separate_start);
+	adev->gfx.rlc.starting_offsets_start =
+			le32_to_cpu(rlc_hdr->starting_offsets_start);
+	adev->gfx.rlc.reg_list_format_size_bytes =
+			le32_to_cpu(rlc_hdr->reg_list_format_size_bytes);
+	adev->gfx.rlc.reg_list_size_bytes =
+			le32_to_cpu(rlc_hdr->reg_list_size_bytes);
+
+	adev->gfx.rlc.register_list_format =
+			kmalloc(adev->gfx.rlc.reg_list_format_size_bytes +
+					adev->gfx.rlc.reg_list_size_bytes, GFP_KERNEL);
+
+	if (!adev->gfx.rlc.register_list_format) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	tmp = (unsigned int *)((uintptr_t)rlc_hdr +
+			le32_to_cpu(rlc_hdr->reg_list_format_array_offset_bytes));
+	for (i = 0 ; i < (rlc_hdr->reg_list_format_size_bytes >> 2); i++)
+		adev->gfx.rlc.register_list_format[i] =	le32_to_cpu(tmp[i]);
+
+	adev->gfx.rlc.register_restore = adev->gfx.rlc.register_list_format + i;
+
+	tmp = (unsigned int *)((uintptr_t)rlc_hdr +
+			le32_to_cpu(rlc_hdr->reg_list_array_offset_bytes));
+	for (i = 0 ; i < (rlc_hdr->reg_list_size_bytes >> 2); i++)
+		adev->gfx.rlc.register_restore[i] = le32_to_cpu(tmp[i]);
+
+	if (adev->asic_type >= CHIP_POLARIS10 && adev->asic_type <= CHIP_POLARIS12) {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec_2.bin", chip_name);
+		err = request_firmware(&adev->gfx.mec_fw, fw_name, adev->dev);
+		if (err == -ENOENT) {
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec.bin", chip_name);
+			err = request_firmware(&adev->gfx.mec_fw, fw_name, adev->dev);
+		}
+	} else {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec.bin", chip_name);
+		err = request_firmware(&adev->gfx.mec_fw, fw_name, adev->dev);
+	}
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.mec_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+	adev->gfx.mec_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.mec_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	if ((adev->asic_type != CHIP_STONEY) &&
+	    (adev->asic_type != CHIP_TOPAZ)) {
+		if (adev->asic_type >= CHIP_POLARIS10 && adev->asic_type <= CHIP_POLARIS12) {
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec2_2.bin", chip_name);
+			err = request_firmware(&adev->gfx.mec2_fw, fw_name, adev->dev);
+			if (err == -ENOENT) {
+				snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec2.bin", chip_name);
+				err = request_firmware(&adev->gfx.mec2_fw, fw_name, adev->dev);
+			}
+		} else {
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec2.bin", chip_name);
+			err = request_firmware(&adev->gfx.mec2_fw, fw_name, adev->dev);
+		}
+		if (!err) {
+			err = amdgpu_ucode_validate(adev->gfx.mec2_fw);
+			if (err)
+				goto out;
+			cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+				adev->gfx.mec2_fw->data;
+			adev->gfx.mec2_fw_version =
+				le32_to_cpu(cp_hdr->header.ucode_version);
+			adev->gfx.mec2_feature_version =
+				le32_to_cpu(cp_hdr->ucode_feature_version);
+		} else {
+			err = 0;
+			adev->gfx.mec2_fw = NULL;
+		}
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_SMU) {
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_PFP];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_PFP;
+		info->fw = adev->gfx.pfp_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_ME];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_ME;
+		info->fw = adev->gfx.me_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_CE];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_CE;
+		info->fw = adev->gfx.ce_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_G];
+		info->ucode_id = AMDGPU_UCODE_ID_RLC_G;
+		info->fw = adev->gfx.rlc_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC1];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_MEC1;
+		info->fw = adev->gfx.mec_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		/* we need account JT in */
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(cp_hdr->jt_size) << 2, PAGE_SIZE);
+
+		if (amdgpu_sriov_vf(adev)) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_STORAGE];
+			info->ucode_id = AMDGPU_UCODE_ID_STORAGE;
+			info->fw = adev->gfx.mec_fw;
+			adev->firmware.fw_size +=
+				roundup2(le32_to_cpu(64 * PAGE_SIZE), PAGE_SIZE);
+		}
+
+		if (adev->gfx.mec2_fw) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC2];
+			info->ucode_id = AMDGPU_UCODE_ID_CP_MEC2;
+			info->fw = adev->gfx.mec2_fw;
+			header = (const struct common_firmware_header *)info->fw->data;
+			adev->firmware.fw_size +=
+				roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+		}
+
+	}
+
+out:
+	if (err) {
+		dev_err(adev->dev,
+			"gfx8: Failed to load firmware \"%s\"\n",
+			fw_name);
+		release_firmware(adev->gfx.pfp_fw);
+		adev->gfx.pfp_fw = NULL;
+		release_firmware(adev->gfx.me_fw);
+		adev->gfx.me_fw = NULL;
+		release_firmware(adev->gfx.ce_fw);
+		adev->gfx.ce_fw = NULL;
+		release_firmware(adev->gfx.rlc_fw);
+		adev->gfx.rlc_fw = NULL;
+		release_firmware(adev->gfx.mec_fw);
+		adev->gfx.mec_fw = NULL;
+		release_firmware(adev->gfx.mec2_fw);
+		adev->gfx.mec2_fw = NULL;
+	}
+	return err;
+}
+
+static void gfx_v8_0_get_csb_buffer(struct amdgpu_device *adev,
+				    volatile u32 *buffer)
+{
+	u32 count = 0, i;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	if (adev->gfx.rlc.cs_data == NULL)
+		return;
+	if (buffer == NULL)
+		return;
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	buffer[count++] = cpu_to_le32(0x80000000);
+	buffer[count++] = cpu_to_le32(0x80000000);
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				buffer[count++] =
+					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
+				buffer[count++] = cpu_to_le32(ext->reg_index -
+						PACKET3_SET_CONTEXT_REG_START);
+				for (i = 0; i < ext->reg_count; i++)
+					buffer[count++] = cpu_to_le32(ext->extent[i]);
+			} else {
+				return;
+			}
+		}
+	}
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 2));
+	buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG -
+			PACKET3_SET_CONTEXT_REG_START);
+	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config);
+	buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config_1);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
+	buffer[count++] = cpu_to_le32(0);
+}
+
+static void cz_init_cp_jump_table(struct amdgpu_device *adev)
+{
+	const __le32 *fw_data;
+	volatile u32 *dst_ptr;
+	int me, i, max_me = 4;
+	u32 bo_offset = 0;
+	u32 table_offset, table_size;
+
+	if (adev->asic_type == CHIP_CARRIZO)
+		max_me = 5;
+
+	/* write the cp table buffer */
+	dst_ptr = adev->gfx.rlc.cp_table_ptr;
+	for (me = 0; me < max_me; me++) {
+		if (me == 0) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.ce_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 1) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.pfp_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 2) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.me_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 3) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else  if (me == 4) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec2_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		}
+
+		for (i = 0; i < table_size; i ++) {
+			dst_ptr[bo_offset + i] =
+				cpu_to_le32(le32_to_cpu(fw_data[table_offset + i]));
+		}
+
+		bo_offset += table_size;
+	}
+}
+
+static void gfx_v8_0_rlc_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL);
+}
+
+static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
+{
+	volatile u32 *dst_ptr;
+	u32 dws;
+	const struct cs_section_def *cs_data;
+	int r;
+
+	adev->gfx.rlc.cs_data = vi_cs_data;
+
+	cs_data = adev->gfx.rlc.cs_data;
+
+	if (cs_data) {
+		/* clear state block */
+		adev->gfx.rlc.clear_state_size = dws = gfx_v8_0_get_csb_size(adev);
+
+		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.clear_state_obj,
+					      &adev->gfx.rlc.clear_state_gpu_addr,
+					      (void **)&adev->gfx.rlc.cs_ptr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
+			gfx_v8_0_rlc_fini(adev);
+			return r;
+		}
+
+		/* set up the cs buffer */
+		dst_ptr = adev->gfx.rlc.cs_ptr;
+		gfx_v8_0_get_csb_buffer(adev, dst_ptr);
+		amdgpu_bo_kunmap(adev->gfx.rlc.clear_state_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
+	}
+
+	if ((adev->asic_type == CHIP_CARRIZO) ||
+	    (adev->asic_type == CHIP_STONEY)) {
+		adev->gfx.rlc.cp_table_size = roundup2(96 * 5 * 4, 2048) + (64 * 1024); /* JT + GDS */
+		r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
+					      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.cp_table_obj,
+					      &adev->gfx.rlc.cp_table_gpu_addr,
+					      (void **)&adev->gfx.rlc.cp_table_ptr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
+			return r;
+		}
+
+		cz_init_cp_jump_table(adev);
+
+		amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
+	}
+
+	return 0;
+}
+
+static void gfx_v8_0_mec_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
+}
+
+static int gfx_v8_0_mec_init(struct amdgpu_device *adev)
+{
+	int r;
+	u32 *hpd;
+	size_t mec_hpd_size;
+
+	bitmap_zero(adev->gfx.mec.queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+
+	/* take ownership of the relevant compute queues */
+	amdgpu_gfx_compute_queue_acquire(adev);
+
+	mec_hpd_size = adev->gfx.num_compute_rings * GFX8_MEC_HPD_SIZE;
+
+	r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.mec.hpd_eop_obj,
+				      &adev->gfx.mec.hpd_eop_gpu_addr,
+				      (void **)&hpd);
+	if (r) {
+		dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
+		return r;
+	}
+
+	memset(hpd, 0, mec_hpd_size);
+
+	amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
+	amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+
+	return 0;
+}
+
+static const u32 vgpr_init_compute_shader[] =
+{
+	0x7e000209, 0x7e020208,
+	0x7e040207, 0x7e060206,
+	0x7e080205, 0x7e0a0204,
+	0x7e0c0203, 0x7e0e0202,
+	0x7e100201, 0x7e120200,
+	0x7e140209, 0x7e160208,
+	0x7e180207, 0x7e1a0206,
+	0x7e1c0205, 0x7e1e0204,
+	0x7e200203, 0x7e220202,
+	0x7e240201, 0x7e260200,
+	0x7e280209, 0x7e2a0208,
+	0x7e2c0207, 0x7e2e0206,
+	0x7e300205, 0x7e320204,
+	0x7e340203, 0x7e360202,
+	0x7e380201, 0x7e3a0200,
+	0x7e3c0209, 0x7e3e0208,
+	0x7e400207, 0x7e420206,
+	0x7e440205, 0x7e460204,
+	0x7e480203, 0x7e4a0202,
+	0x7e4c0201, 0x7e4e0200,
+	0x7e500209, 0x7e520208,
+	0x7e540207, 0x7e560206,
+	0x7e580205, 0x7e5a0204,
+	0x7e5c0203, 0x7e5e0202,
+	0x7e600201, 0x7e620200,
+	0x7e640209, 0x7e660208,
+	0x7e680207, 0x7e6a0206,
+	0x7e6c0205, 0x7e6e0204,
+	0x7e700203, 0x7e720202,
+	0x7e740201, 0x7e760200,
+	0x7e780209, 0x7e7a0208,
+	0x7e7c0207, 0x7e7e0206,
+	0xbf8a0000, 0xbf810000,
+};
+
+static const u32 sgpr_init_compute_shader[] =
+{
+	0xbe8a0100, 0xbe8c0102,
+	0xbe8e0104, 0xbe900106,
+	0xbe920108, 0xbe940100,
+	0xbe960102, 0xbe980104,
+	0xbe9a0106, 0xbe9c0108,
+	0xbe9e0100, 0xbea00102,
+	0xbea20104, 0xbea40106,
+	0xbea60108, 0xbea80100,
+	0xbeaa0102, 0xbeac0104,
+	0xbeae0106, 0xbeb00108,
+	0xbeb20100, 0xbeb40102,
+	0xbeb60104, 0xbeb80106,
+	0xbeba0108, 0xbebc0100,
+	0xbebe0102, 0xbec00104,
+	0xbec20106, 0xbec40108,
+	0xbec60100, 0xbec80102,
+	0xbee60004, 0xbee70005,
+	0xbeea0006, 0xbeeb0007,
+	0xbee80008, 0xbee90009,
+	0xbefc0000, 0xbf8a0000,
+	0xbf810000, 0x00000000,
+};
+
+static const u32 vgpr_init_regs[] =
+{
+	mmCOMPUTE_STATIC_THREAD_MGMT_SE0, 0xffffffff,
+	mmCOMPUTE_RESOURCE_LIMITS, 0x1000000, /* CU_GROUP_COUNT=1 */
+	mmCOMPUTE_NUM_THREAD_X, 256*4,
+	mmCOMPUTE_NUM_THREAD_Y, 1,
+	mmCOMPUTE_NUM_THREAD_Z, 1,
+	mmCOMPUTE_PGM_RSRC1, 0x100004f, /* VGPRS=15 (64 logical VGPRs), SGPRS=1 (16 SGPRs), BULKY=1 */
+	mmCOMPUTE_PGM_RSRC2, 20,
+	mmCOMPUTE_USER_DATA_0, 0xedcedc00,
+	mmCOMPUTE_USER_DATA_1, 0xedcedc01,
+	mmCOMPUTE_USER_DATA_2, 0xedcedc02,
+	mmCOMPUTE_USER_DATA_3, 0xedcedc03,
+	mmCOMPUTE_USER_DATA_4, 0xedcedc04,
+	mmCOMPUTE_USER_DATA_5, 0xedcedc05,
+	mmCOMPUTE_USER_DATA_6, 0xedcedc06,
+	mmCOMPUTE_USER_DATA_7, 0xedcedc07,
+	mmCOMPUTE_USER_DATA_8, 0xedcedc08,
+	mmCOMPUTE_USER_DATA_9, 0xedcedc09,
+};
+
+static const u32 sgpr1_init_regs[] =
+{
+	mmCOMPUTE_STATIC_THREAD_MGMT_SE0, 0x0f,
+	mmCOMPUTE_RESOURCE_LIMITS, 0x1000000, /* CU_GROUP_COUNT=1 */
+	mmCOMPUTE_NUM_THREAD_X, 256*5,
+	mmCOMPUTE_NUM_THREAD_Y, 1,
+	mmCOMPUTE_NUM_THREAD_Z, 1,
+	mmCOMPUTE_PGM_RSRC1, 0x240, /* SGPRS=9 (80 GPRS) */
+	mmCOMPUTE_PGM_RSRC2, 20,
+	mmCOMPUTE_USER_DATA_0, 0xedcedc00,
+	mmCOMPUTE_USER_DATA_1, 0xedcedc01,
+	mmCOMPUTE_USER_DATA_2, 0xedcedc02,
+	mmCOMPUTE_USER_DATA_3, 0xedcedc03,
+	mmCOMPUTE_USER_DATA_4, 0xedcedc04,
+	mmCOMPUTE_USER_DATA_5, 0xedcedc05,
+	mmCOMPUTE_USER_DATA_6, 0xedcedc06,
+	mmCOMPUTE_USER_DATA_7, 0xedcedc07,
+	mmCOMPUTE_USER_DATA_8, 0xedcedc08,
+	mmCOMPUTE_USER_DATA_9, 0xedcedc09,
+};
+
+static const u32 sgpr2_init_regs[] =
+{
+	mmCOMPUTE_STATIC_THREAD_MGMT_SE0, 0xf0,
+	mmCOMPUTE_RESOURCE_LIMITS, 0x1000000,
+	mmCOMPUTE_NUM_THREAD_X, 256*5,
+	mmCOMPUTE_NUM_THREAD_Y, 1,
+	mmCOMPUTE_NUM_THREAD_Z, 1,
+	mmCOMPUTE_PGM_RSRC1, 0x240, /* SGPRS=9 (80 GPRS) */
+	mmCOMPUTE_PGM_RSRC2, 20,
+	mmCOMPUTE_USER_DATA_0, 0xedcedc00,
+	mmCOMPUTE_USER_DATA_1, 0xedcedc01,
+	mmCOMPUTE_USER_DATA_2, 0xedcedc02,
+	mmCOMPUTE_USER_DATA_3, 0xedcedc03,
+	mmCOMPUTE_USER_DATA_4, 0xedcedc04,
+	mmCOMPUTE_USER_DATA_5, 0xedcedc05,
+	mmCOMPUTE_USER_DATA_6, 0xedcedc06,
+	mmCOMPUTE_USER_DATA_7, 0xedcedc07,
+	mmCOMPUTE_USER_DATA_8, 0xedcedc08,
+	mmCOMPUTE_USER_DATA_9, 0xedcedc09,
+};
+
+static const u32 sec_ded_counter_registers[] =
+{
+	mmCPC_EDC_ATC_CNT,
+	mmCPC_EDC_SCRATCH_CNT,
+	mmCPC_EDC_UCODE_CNT,
+	mmCPF_EDC_ATC_CNT,
+	mmCPF_EDC_ROQ_CNT,
+	mmCPF_EDC_TAG_CNT,
+	mmCPG_EDC_ATC_CNT,
+	mmCPG_EDC_DMA_CNT,
+	mmCPG_EDC_TAG_CNT,
+	mmDC_EDC_CSINVOC_CNT,
+	mmDC_EDC_RESTORE_CNT,
+	mmDC_EDC_STATE_CNT,
+	mmGDS_EDC_CNT,
+	mmGDS_EDC_GRBM_CNT,
+	mmGDS_EDC_OA_DED,
+	mmSPI_EDC_CNT,
+	mmSQC_ATC_EDC_GATCL1_CNT,
+	mmSQC_EDC_CNT,
+	mmSQ_EDC_DED_CNT,
+	mmSQ_EDC_INFO,
+	mmSQ_EDC_SEC_CNT,
+	mmTCC_EDC_CNT,
+	mmTCP_ATC_EDC_GATCL1_CNT,
+	mmTCP_EDC_CNT,
+	mmTD_EDC_CNT
+};
+
+static int gfx_v8_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->gfx.compute_ring[0];
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	int r, i;
+	u32 tmp;
+	unsigned total_size, vgpr_offset, sgpr_offset;
+	u64 gpu_addr;
+
+	/* only supported on CZ */
+	if (adev->asic_type != CHIP_CARRIZO)
+		return 0;
+
+	/* bail if the compute ring is not ready */
+	if (!ring->ready)
+		return 0;
+
+	tmp = RREG32(mmGB_EDC_MODE);
+	WREG32(mmGB_EDC_MODE, 0);
+
+	total_size =
+		(((ARRAY_SIZE(vgpr_init_regs) / 2) * 3) + 4 + 5 + 2) * 4;
+	total_size +=
+		(((ARRAY_SIZE(sgpr1_init_regs) / 2) * 3) + 4 + 5 + 2) * 4;
+	total_size +=
+		(((ARRAY_SIZE(sgpr2_init_regs) / 2) * 3) + 4 + 5 + 2) * 4;
+	total_size = roundup2(total_size, 256);
+	vgpr_offset = total_size;
+	total_size += roundup2(sizeof(vgpr_init_compute_shader), 256);
+	sgpr_offset = total_size;
+	total_size += sizeof(sgpr_init_compute_shader);
+
+	/* allocate an indirect buffer to put the commands in */
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, total_size, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		return r;
+	}
+
+	/* load the compute shaders */
+	for (i = 0; i < ARRAY_SIZE(vgpr_init_compute_shader); i++)
+		ib.ptr[i + (vgpr_offset / 4)] = vgpr_init_compute_shader[i];
+
+	for (i = 0; i < ARRAY_SIZE(sgpr_init_compute_shader); i++)
+		ib.ptr[i + (sgpr_offset / 4)] = sgpr_init_compute_shader[i];
+
+	/* init the ib length to 0 */
+	ib.length_dw = 0;
+
+	/* VGPR */
+	/* write the register state for the compute dispatch */
+	for (i = 0; i < ARRAY_SIZE(vgpr_init_regs); i += 2) {
+		ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1);
+		ib.ptr[ib.length_dw++] = vgpr_init_regs[i] - PACKET3_SET_SH_REG_START;
+		ib.ptr[ib.length_dw++] = vgpr_init_regs[i + 1];
+	}
+	/* write the shader start address: mmCOMPUTE_PGM_LO, mmCOMPUTE_PGM_HI */
+	gpu_addr = (ib.gpu_addr + (u64)vgpr_offset) >> 8;
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 2);
+	ib.ptr[ib.length_dw++] = mmCOMPUTE_PGM_LO - PACKET3_SET_SH_REG_START;
+	ib.ptr[ib.length_dw++] = lower_32_bits(gpu_addr);
+	ib.ptr[ib.length_dw++] = upper_32_bits(gpu_addr);
+
+	/* write dispatch packet */
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3);
+	ib.ptr[ib.length_dw++] = 8; /* x */
+	ib.ptr[ib.length_dw++] = 1; /* y */
+	ib.ptr[ib.length_dw++] = 1; /* z */
+	ib.ptr[ib.length_dw++] =
+		REG_SET_FIELD(0, COMPUTE_DISPATCH_INITIATOR, COMPUTE_SHADER_EN, 1);
+
+	/* write CS partial flush packet */
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_EVENT_WRITE, 0);
+	ib.ptr[ib.length_dw++] = EVENT_TYPE(7) | EVENT_INDEX(4);
+
+	/* SGPR1 */
+	/* write the register state for the compute dispatch */
+	for (i = 0; i < ARRAY_SIZE(sgpr1_init_regs); i += 2) {
+		ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1);
+		ib.ptr[ib.length_dw++] = sgpr1_init_regs[i] - PACKET3_SET_SH_REG_START;
+		ib.ptr[ib.length_dw++] = sgpr1_init_regs[i + 1];
+	}
+	/* write the shader start address: mmCOMPUTE_PGM_LO, mmCOMPUTE_PGM_HI */
+	gpu_addr = (ib.gpu_addr + (u64)sgpr_offset) >> 8;
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 2);
+	ib.ptr[ib.length_dw++] = mmCOMPUTE_PGM_LO - PACKET3_SET_SH_REG_START;
+	ib.ptr[ib.length_dw++] = lower_32_bits(gpu_addr);
+	ib.ptr[ib.length_dw++] = upper_32_bits(gpu_addr);
+
+	/* write dispatch packet */
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3);
+	ib.ptr[ib.length_dw++] = 8; /* x */
+	ib.ptr[ib.length_dw++] = 1; /* y */
+	ib.ptr[ib.length_dw++] = 1; /* z */
+	ib.ptr[ib.length_dw++] =
+		REG_SET_FIELD(0, COMPUTE_DISPATCH_INITIATOR, COMPUTE_SHADER_EN, 1);
+
+	/* write CS partial flush packet */
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_EVENT_WRITE, 0);
+	ib.ptr[ib.length_dw++] = EVENT_TYPE(7) | EVENT_INDEX(4);
+
+	/* SGPR2 */
+	/* write the register state for the compute dispatch */
+	for (i = 0; i < ARRAY_SIZE(sgpr2_init_regs); i += 2) {
+		ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1);
+		ib.ptr[ib.length_dw++] = sgpr2_init_regs[i] - PACKET3_SET_SH_REG_START;
+		ib.ptr[ib.length_dw++] = sgpr2_init_regs[i + 1];
+	}
+	/* write the shader start address: mmCOMPUTE_PGM_LO, mmCOMPUTE_PGM_HI */
+	gpu_addr = (ib.gpu_addr + (u64)sgpr_offset) >> 8;
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 2);
+	ib.ptr[ib.length_dw++] = mmCOMPUTE_PGM_LO - PACKET3_SET_SH_REG_START;
+	ib.ptr[ib.length_dw++] = lower_32_bits(gpu_addr);
+	ib.ptr[ib.length_dw++] = upper_32_bits(gpu_addr);
+
+	/* write dispatch packet */
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3);
+	ib.ptr[ib.length_dw++] = 8; /* x */
+	ib.ptr[ib.length_dw++] = 1; /* y */
+	ib.ptr[ib.length_dw++] = 1; /* z */
+	ib.ptr[ib.length_dw++] =
+		REG_SET_FIELD(0, COMPUTE_DISPATCH_INITIATOR, COMPUTE_SHADER_EN, 1);
+
+	/* write CS partial flush packet */
+	ib.ptr[ib.length_dw++] = PACKET3(PACKET3_EVENT_WRITE, 0);
+	ib.ptr[ib.length_dw++] = EVENT_TYPE(7) | EVENT_INDEX(4);
+
+	/* shedule the ib on the ring */
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r) {
+		DRM_ERROR("amdgpu: ib submit failed (%d).\n", r);
+		goto fail;
+	}
+
+	/* wait for the GPU to finish processing the IB */
+	r = dma_fence_wait(f, false);
+	if (r) {
+		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+		goto fail;
+	}
+
+	tmp = REG_SET_FIELD(tmp, GB_EDC_MODE, DED_MODE, 2);
+	tmp = REG_SET_FIELD(tmp, GB_EDC_MODE, PROP_FED, 1);
+	WREG32(mmGB_EDC_MODE, tmp);
+
+	tmp = RREG32(mmCC_GC_EDC_CONFIG);
+	tmp = REG_SET_FIELD(tmp, CC_GC_EDC_CONFIG, DIS_EDC, 0) | 1;
+	WREG32(mmCC_GC_EDC_CONFIG, tmp);
+
+
+	/* read back registers to clear the counters */
+	for (i = 0; i < ARRAY_SIZE(sec_ded_counter_registers); i++)
+		RREG32(sec_ded_counter_registers[i]);
+
+fail:
+	amdgpu_ib_free(adev, &ib, NULL);
+	dma_fence_put(f);
+
+	return r;
+}
+
+static int gfx_v8_0_gpu_early_init(struct amdgpu_device *adev)
+{
+	u32 gb_addr_config;
+	u32 mc_shared_chmap, mc_arb_ramcfg;
+	u32 dimm00_addr_map, dimm01_addr_map, dimm10_addr_map, dimm11_addr_map;
+	u32 tmp;
+	int ret;
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 2;
+		adev->gfx.config.max_cu_per_sh = 6;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 2;
+		adev->gfx.config.max_texture_channel_caches = 2;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = TOPAZ_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_FIJI:
+		adev->gfx.config.max_shader_engines = 4;
+		adev->gfx.config.max_tile_pipes = 16;
+		adev->gfx.config.max_cu_per_sh = 16;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 4;
+		adev->gfx.config.max_texture_channel_caches = 16;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = TONGA_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+		ret = amdgpu_atombios_get_gfx_info(adev);
+		if (ret)
+			return ret;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = POLARIS11_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_POLARIS10:
+	case CHIP_VEGAM:
+		ret = amdgpu_atombios_get_gfx_info(adev);
+		if (ret)
+			return ret;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = TONGA_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_TONGA:
+		adev->gfx.config.max_shader_engines = 4;
+		adev->gfx.config.max_tile_pipes = 8;
+		adev->gfx.config.max_cu_per_sh = 8;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 2;
+		adev->gfx.config.max_texture_channel_caches = 8;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = TONGA_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_CARRIZO:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 2;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 2;
+		adev->gfx.config.max_cu_per_sh = 8;
+		adev->gfx.config.max_texture_channel_caches = 2;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = CARRIZO_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_STONEY:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 2;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 1;
+		adev->gfx.config.max_cu_per_sh = 3;
+		adev->gfx.config.max_texture_channel_caches = 2;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 16;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = CARRIZO_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	default:
+		adev->gfx.config.max_shader_engines = 2;
+		adev->gfx.config.max_tile_pipes = 4;
+		adev->gfx.config.max_cu_per_sh = 2;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 2;
+		adev->gfx.config.max_texture_channel_caches = 4;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = TONGA_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	}
+
+	mc_shared_chmap = RREG32(mmMC_SHARED_CHMAP);
+	adev->gfx.config.mc_arb_ramcfg = RREG32(mmMC_ARB_RAMCFG);
+	mc_arb_ramcfg = adev->gfx.config.mc_arb_ramcfg;
+
+	adev->gfx.config.num_tile_pipes = adev->gfx.config.max_tile_pipes;
+	adev->gfx.config.mem_max_burst_length_bytes = 256;
+	if (adev->flags & AMD_IS_APU) {
+		/* Get memory bank mapping mode. */
+		tmp = RREG32(mmMC_FUS_DRAM0_BANK_ADDR_MAPPING);
+		dimm00_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM0_BANK_ADDR_MAPPING, DIMM0ADDRMAP);
+		dimm01_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM0_BANK_ADDR_MAPPING, DIMM1ADDRMAP);
+
+		tmp = RREG32(mmMC_FUS_DRAM1_BANK_ADDR_MAPPING);
+		dimm10_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM1_BANK_ADDR_MAPPING, DIMM0ADDRMAP);
+		dimm11_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM1_BANK_ADDR_MAPPING, DIMM1ADDRMAP);
+
+		/* Validate settings in case only one DIMM installed. */
+		if ((dimm00_addr_map == 0) || (dimm00_addr_map == 3) || (dimm00_addr_map == 4) || (dimm00_addr_map > 12))
+			dimm00_addr_map = 0;
+		if ((dimm01_addr_map == 0) || (dimm01_addr_map == 3) || (dimm01_addr_map == 4) || (dimm01_addr_map > 12))
+			dimm01_addr_map = 0;
+		if ((dimm10_addr_map == 0) || (dimm10_addr_map == 3) || (dimm10_addr_map == 4) || (dimm10_addr_map > 12))
+			dimm10_addr_map = 0;
+		if ((dimm11_addr_map == 0) || (dimm11_addr_map == 3) || (dimm11_addr_map == 4) || (dimm11_addr_map > 12))
+			dimm11_addr_map = 0;
+
+		/* If DIMM Addr map is 8GB, ROW size should be 2KB. Otherwise 1KB. */
+		/* If ROW size(DIMM1) != ROW size(DMIMM0), ROW size should be larger one. */
+		if ((dimm00_addr_map == 11) || (dimm01_addr_map == 11) || (dimm10_addr_map == 11) || (dimm11_addr_map == 11))
+			adev->gfx.config.mem_row_size_in_kb = 2;
+		else
+			adev->gfx.config.mem_row_size_in_kb = 1;
+	} else {
+		tmp = REG_GET_FIELD(mc_arb_ramcfg, MC_ARB_RAMCFG, NOOFCOLS);
+		adev->gfx.config.mem_row_size_in_kb = (4 * (1 << (8 + tmp))) / 1024;
+		if (adev->gfx.config.mem_row_size_in_kb > 4)
+			adev->gfx.config.mem_row_size_in_kb = 4;
+	}
+
+	adev->gfx.config.shader_engine_tile_size = 32;
+	adev->gfx.config.num_gpus = 1;
+	adev->gfx.config.multi_gpu_tile_size = 64;
+
+	/* fix up row size */
+	switch (adev->gfx.config.mem_row_size_in_kb) {
+	case 1:
+	default:
+		gb_addr_config = REG_SET_FIELD(gb_addr_config, GB_ADDR_CONFIG, ROW_SIZE, 0);
+		break;
+	case 2:
+		gb_addr_config = REG_SET_FIELD(gb_addr_config, GB_ADDR_CONFIG, ROW_SIZE, 1);
+		break;
+	case 4:
+		gb_addr_config = REG_SET_FIELD(gb_addr_config, GB_ADDR_CONFIG, ROW_SIZE, 2);
+		break;
+	}
+	adev->gfx.config.gb_addr_config = gb_addr_config;
+
+	return 0;
+}
+
+static int gfx_v8_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
+					int mec, int pipe, int queue)
+{
+	int r;
+	unsigned irq_type;
+	struct amdgpu_ring *ring = &adev->gfx.compute_ring[ring_id];
+
+	ring = &adev->gfx.compute_ring[ring_id];
+
+	/* mec0 is me1 */
+	ring->me = mec + 1;
+	ring->pipe = pipe;
+	ring->queue = queue;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->doorbell_index = AMDGPU_DOORBELL_MEC_RING0 + ring_id;
+	ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr
+				+ (ring_id * GFX8_MEC_HPD_SIZE);
+	snprintf(ring->name, sizeof(ring->name), "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+
+	irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
+		+ ((ring->me - 1) * adev->gfx.mec.num_pipe_per_mec)
+		+ ring->pipe;
+
+	/* type-2 packets are deprecated on MEC, use type-3 instead */
+	r = amdgpu_ring_init(adev, ring, 1024,
+			&adev->gfx.eop_irq, irq_type);
+	if (r)
+		return r;
+
+
+	return 0;
+}
+
+static void gfx_v8_0_sq_irq_work_func(struct work_struct *work);
+
+static int gfx_v8_0_sw_init(void *handle)
+{
+	int i, j, k, r, ring_id;
+	struct amdgpu_ring *ring;
+	struct amdgpu_kiq *kiq;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	switch (adev->asic_type) {
+	case CHIP_TONGA:
+	case CHIP_CARRIZO:
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		adev->gfx.mec.num_mec = 2;
+		break;
+	case CHIP_TOPAZ:
+	case CHIP_STONEY:
+	default:
+		adev->gfx.mec.num_mec = 1;
+		break;
+	}
+
+	adev->gfx.mec.num_pipe_per_mec = 4;
+	adev->gfx.mec.num_queue_per_pipe = 8;
+
+	/* KIQ event */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_INT_IB2, &adev->gfx.kiq.irq);
+	if (r)
+		return r;
+
+	/* EOP Event */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_END_OF_PIPE, &adev->gfx.eop_irq);
+	if (r)
+		return r;
+
+	/* Privileged reg */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_PRIV_REG_FAULT,
+			      &adev->gfx.priv_reg_irq);
+	if (r)
+		return r;
+
+	/* Privileged inst */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_PRIV_INSTR_FAULT,
+			      &adev->gfx.priv_inst_irq);
+	if (r)
+		return r;
+
+	/* Add CP EDC/ECC irq  */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_CP_ECC_ERROR,
+			      &adev->gfx.cp_ecc_error_irq);
+	if (r)
+		return r;
+
+	/* SQ interrupts. */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SQ_INTERRUPT_MSG,
+			      &adev->gfx.sq_irq);
+	if (r) {
+		DRM_ERROR("amdgpu_irq_add() for SQ failed: %d\n", r);
+		return r;
+	}
+
+	INIT_WORK(&adev->gfx.sq_work.work, gfx_v8_0_sq_irq_work_func);
+
+	adev->gfx.gfx_current_status = AMDGPU_GFX_NORMAL_MODE;
+
+	gfx_v8_0_scratch_init(adev);
+
+	r = gfx_v8_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load gfx firmware!\n");
+		return r;
+	}
+
+	r = gfx_v8_0_rlc_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init rlc BOs!\n");
+		return r;
+	}
+
+	r = gfx_v8_0_mec_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init MEC BOs!\n");
+		return r;
+	}
+
+	/* set up the gfx ring */
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		ring = &adev->gfx.gfx_ring[i];
+		ring->ring_obj = NULL;
+		snprintf(ring->name, sizeof(ring->name), "gfx");
+		/* no gfx doorbells on iceland */
+		if (adev->asic_type != CHIP_TOPAZ) {
+			ring->use_doorbell = true;
+			ring->doorbell_index = AMDGPU_DOORBELL_GFX_RING0;
+		}
+
+		r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq,
+				     AMDGPU_CP_IRQ_GFX_EOP);
+		if (r)
+			return r;
+	}
+
+
+	/* set up the compute queues - allocate horizontally across pipes */
+	ring_id = 0;
+	for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
+		for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
+			for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
+				if (!amdgpu_gfx_is_mec_queue_enabled(adev, i, k, j))
+					continue;
+
+				r = gfx_v8_0_compute_ring_init(adev,
+								ring_id,
+								i, k, j);
+				if (r)
+					return r;
+
+				ring_id++;
+			}
+		}
+	}
+
+	r = amdgpu_gfx_kiq_init(adev, GFX8_MEC_HPD_SIZE);
+	if (r) {
+		DRM_ERROR("Failed to init KIQ BOs!\n");
+		return r;
+	}
+
+	kiq = &adev->gfx.kiq;
+	r = amdgpu_gfx_kiq_init_ring(adev, &kiq->ring, &kiq->irq);
+	if (r)
+		return r;
+
+	/* create MQD for all compute queues as well as KIQ for SRIOV case */
+	r = amdgpu_gfx_compute_mqd_sw_init(adev, sizeof(struct vi_mqd_allocation));
+	if (r)
+		return r;
+
+	/* reserve GDS, GWS and OA resource for gfx */
+	r = amdgpu_bo_create_kernel(adev, adev->gds.mem.gfx_partition_size,
+				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GDS,
+				    &adev->gds.gds_gfx_bo, NULL, NULL);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_create_kernel(adev, adev->gds.gws.gfx_partition_size,
+				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GWS,
+				    &adev->gds.gws_gfx_bo, NULL, NULL);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_create_kernel(adev, adev->gds.oa.gfx_partition_size,
+				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_OA,
+				    &adev->gds.oa_gfx_bo, NULL, NULL);
+	if (r)
+		return r;
+
+	adev->gfx.ce_ram_size = 0x8000;
+
+	r = gfx_v8_0_gpu_early_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int gfx_v8_0_sw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_bo_free_kernel(&adev->gds.oa_gfx_bo, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gds.gws_gfx_bo, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gds.gds_gfx_bo, NULL, NULL);
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
+
+	amdgpu_gfx_compute_mqd_sw_fini(adev);
+	amdgpu_gfx_kiq_free_ring(&adev->gfx.kiq.ring, &adev->gfx.kiq.irq);
+	amdgpu_gfx_kiq_fini(adev);
+
+	gfx_v8_0_mec_fini(adev);
+	gfx_v8_0_rlc_fini(adev);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj,
+				&adev->gfx.rlc.clear_state_gpu_addr,
+				(void **)&adev->gfx.rlc.cs_ptr);
+	if ((adev->asic_type == CHIP_CARRIZO) ||
+	    (adev->asic_type == CHIP_STONEY)) {
+		amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj,
+				&adev->gfx.rlc.cp_table_gpu_addr,
+				(void **)&adev->gfx.rlc.cp_table_ptr);
+	}
+	gfx_v8_0_free_microcode(adev);
+
+	return 0;
+}
+
+static void gfx_v8_0_tiling_mode_table_init(struct amdgpu_device *adev)
+{
+	uint32_t *modearray, *mod2array;
+	const u32 num_tile_mode_states = ARRAY_SIZE(adev->gfx.config.tile_mode_array);
+	const u32 num_secondary_tile_mode_states = ARRAY_SIZE(adev->gfx.config.macrotile_mode_array);
+	u32 reg_offset;
+
+	modearray = adev->gfx.config.tile_mode_array;
+	mod2array = adev->gfx.config.macrotile_mode_array;
+
+	for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+		modearray[reg_offset] = 0;
+
+	for (reg_offset = 0; reg_offset <  num_secondary_tile_mode_states; reg_offset++)
+		mod2array[reg_offset] = 0;
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+				PIPE_CONFIG(ADDR_SURF_P2));
+		modearray[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+
+		mod2array[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				 NUM_BANKS(ADDR_SURF_8_BANK));
+
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			if (reg_offset != 7 && reg_offset != 12 && reg_offset != 17 &&
+			    reg_offset != 23)
+				WREG32(mmGB_TILE_MODE0 + reg_offset, modearray[reg_offset]);
+
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++)
+			if (reg_offset != 7)
+				WREG32(mmGB_MACROTILE_MODE0 + reg_offset, mod2array[reg_offset]);
+
+		break;
+	case CHIP_FIJI:
+	case CHIP_VEGAM:
+		modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[7] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+				PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16));
+		modearray[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[12] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[17] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[23] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[30] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+
+		mod2array[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				 NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				 NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				 NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				 NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				 NUM_BANKS(ADDR_SURF_4_BANK));
+
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			WREG32(mmGB_TILE_MODE0 + reg_offset, modearray[reg_offset]);
+
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++)
+			if (reg_offset != 7)
+				WREG32(mmGB_MACROTILE_MODE0 + reg_offset, mod2array[reg_offset]);
+
+		break;
+	case CHIP_TONGA:
+		modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[7] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16));
+		modearray[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[12] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[17] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[23] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[30] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+
+		mod2array[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				 NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				 NUM_BANKS(ADDR_SURF_4_BANK));
+		mod2array[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				 NUM_BANKS(ADDR_SURF_4_BANK));
+
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			WREG32(mmGB_TILE_MODE0 + reg_offset, modearray[reg_offset]);
+
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++)
+			if (reg_offset != 7)
+				WREG32(mmGB_MACROTILE_MODE0 + reg_offset, mod2array[reg_offset]);
+
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+		modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[7] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16));
+		modearray[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[12] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[17] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[23] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[30] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+
+		mod2array[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+
+		mod2array[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK));
+
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			WREG32(mmGB_TILE_MODE0 + reg_offset, modearray[reg_offset]);
+
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++)
+			if (reg_offset != 7)
+				WREG32(mmGB_MACROTILE_MODE0 + reg_offset, mod2array[reg_offset]);
+
+		break;
+	case CHIP_POLARIS10:
+		modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[7] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16));
+		modearray[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[12] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[17] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[23] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[30] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+
+		mod2array[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+
+		mod2array[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+
+		mod2array[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK));
+
+		mod2array[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+				NUM_BANKS(ADDR_SURF_4_BANK));
+
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			WREG32(mmGB_TILE_MODE0 + reg_offset, modearray[reg_offset]);
+
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++)
+			if (reg_offset != 7)
+				WREG32(mmGB_MACROTILE_MODE0 + reg_offset, mod2array[reg_offset]);
+
+		break;
+	case CHIP_STONEY:
+		modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+				PIPE_CONFIG(ADDR_SURF_P2));
+		modearray[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+
+		mod2array[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				 NUM_BANKS(ADDR_SURF_8_BANK));
+
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			if (reg_offset != 7 && reg_offset != 12 && reg_offset != 17 &&
+			    reg_offset != 23)
+				WREG32(mmGB_TILE_MODE0 + reg_offset, modearray[reg_offset]);
+
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++)
+			if (reg_offset != 7)
+				WREG32(mmGB_MACROTILE_MODE0 + reg_offset, mod2array[reg_offset]);
+
+		break;
+	default:
+		dev_warn(adev->dev,
+			 "Unknown chip type (%d) in function gfx_v8_0_tiling_mode_table_init() falling through to CHIP_CARRIZO\n",
+			 adev->asic_type);
+
+	case CHIP_CARRIZO:
+		modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+		modearray[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+				PIPE_CONFIG(ADDR_SURF_P2));
+		modearray[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				PIPE_CONFIG(ADDR_SURF_P2) |
+				MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+		modearray[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+		modearray[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+		modearray[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+				 PIPE_CONFIG(ADDR_SURF_P2) |
+				 MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+				 SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+
+		mod2array[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				NUM_BANKS(ADDR_SURF_8_BANK));
+		mod2array[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+				BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+				 NUM_BANKS(ADDR_SURF_16_BANK));
+		mod2array[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+				 BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+				 MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+				 NUM_BANKS(ADDR_SURF_8_BANK));
+
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++)
+			if (reg_offset != 7 && reg_offset != 12 && reg_offset != 17 &&
+			    reg_offset != 23)
+				WREG32(mmGB_TILE_MODE0 + reg_offset, modearray[reg_offset]);
+
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++)
+			if (reg_offset != 7)
+				WREG32(mmGB_MACROTILE_MODE0 + reg_offset, mod2array[reg_offset]);
+
+		break;
+	}
+}
+
+static void gfx_v8_0_select_se_sh(struct amdgpu_device *adev,
+				  u32 se_num, u32 sh_num, u32 instance)
+{
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX, instance);
+
+	if (se_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
+
+	if (sh_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
+
+	WREG32(mmGRBM_GFX_INDEX, data);
+}
+
+static void gfx_v8_0_select_me_pipe_q(struct amdgpu_device *adev,
+				  u32 me, u32 pipe, u32 q)
+{
+	vi_srbm_select(adev, me, pipe, q, 0);
+}
+
+static u32 gfx_v8_0_get_rb_active_bitmap(struct amdgpu_device *adev)
+{
+	u32 data, mask;
+
+	data =  RREG32(mmCC_RB_BACKEND_DISABLE) |
+		RREG32(mmGC_USER_RB_BACKEND_DISABLE);
+
+	data = REG_GET_FIELD(data, GC_USER_RB_BACKEND_DISABLE, BACKEND_DISABLE);
+
+	mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_backends_per_se /
+					 adev->gfx.config.max_sh_per_se);
+
+	return (~data) & mask;
+}
+
+static void
+gfx_v8_0_raster_config(struct amdgpu_device *adev, u32 *rconf, u32 *rconf1)
+{
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+	case CHIP_VEGAM:
+		*rconf |= RB_MAP_PKR0(2) | RB_MAP_PKR1(2) |
+			  RB_XSEL2(1) | PKR_MAP(2) |
+			  PKR_XSEL(1) | PKR_YSEL(1) |
+			  SE_MAP(2) | SE_XSEL(2) | SE_YSEL(3);
+		*rconf1 |= SE_PAIR_MAP(2) | SE_PAIR_XSEL(3) |
+			   SE_PAIR_YSEL(2);
+		break;
+	case CHIP_TONGA:
+	case CHIP_POLARIS10:
+		*rconf |= RB_MAP_PKR0(2) | RB_XSEL2(1) | SE_MAP(2) |
+			  SE_XSEL(1) | SE_YSEL(1);
+		*rconf1 |= SE_PAIR_MAP(2) | SE_PAIR_XSEL(2) |
+			   SE_PAIR_YSEL(2);
+		break;
+	case CHIP_TOPAZ:
+	case CHIP_CARRIZO:
+		*rconf |= RB_MAP_PKR0(2);
+		*rconf1 |= 0x0;
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+		*rconf |= RB_MAP_PKR0(2) | RB_XSEL2(1) | SE_MAP(2) |
+			  SE_XSEL(1) | SE_YSEL(1);
+		*rconf1 |= 0x0;
+		break;
+	case CHIP_STONEY:
+		*rconf |= 0x0;
+		*rconf1 |= 0x0;
+		break;
+	default:
+		DRM_ERROR("unknown asic: 0x%x\n", adev->asic_type);
+		break;
+	}
+}
+
+static void
+gfx_v8_0_write_harvested_raster_configs(struct amdgpu_device *adev,
+					u32 raster_config, u32 raster_config_1,
+					unsigned rb_mask, unsigned num_rb)
+{
+	unsigned sh_per_se = max_t(unsigned, adev->gfx.config.max_sh_per_se, 1);
+	unsigned num_se = max_t(unsigned, adev->gfx.config.max_shader_engines, 1);
+	unsigned rb_per_pkr = min_t(unsigned, num_rb / num_se / sh_per_se, 2);
+	unsigned rb_per_se = num_rb / num_se;
+	unsigned se_mask[4];
+	unsigned se;
+
+	se_mask[0] = ((1 << rb_per_se) - 1) & rb_mask;
+	se_mask[1] = (se_mask[0] << rb_per_se) & rb_mask;
+	se_mask[2] = (se_mask[1] << rb_per_se) & rb_mask;
+	se_mask[3] = (se_mask[2] << rb_per_se) & rb_mask;
+
+	WARN_ON(!(num_se == 1 || num_se == 2 || num_se == 4));
+	WARN_ON(!(sh_per_se == 1 || sh_per_se == 2));
+	WARN_ON(!(rb_per_pkr == 1 || rb_per_pkr == 2));
+
+	if ((num_se > 2) && ((!se_mask[0] && !se_mask[1]) ||
+			     (!se_mask[2] && !se_mask[3]))) {
+		raster_config_1 &= ~SE_PAIR_MAP_MASK;
+
+		if (!se_mask[0] && !se_mask[1]) {
+			raster_config_1 |=
+				SE_PAIR_MAP(RASTER_CONFIG_SE_PAIR_MAP_3);
+		} else {
+			raster_config_1 |=
+				SE_PAIR_MAP(RASTER_CONFIG_SE_PAIR_MAP_0);
+		}
+	}
+
+	for (se = 0; se < num_se; se++) {
+		unsigned raster_config_se = raster_config;
+		unsigned pkr0_mask = ((1 << rb_per_pkr) - 1) << (se * rb_per_se);
+		unsigned pkr1_mask = pkr0_mask << rb_per_pkr;
+		int idx = (se / 2) * 2;
+
+		if ((num_se > 1) && (!se_mask[idx] || !se_mask[idx + 1])) {
+			raster_config_se &= ~SE_MAP_MASK;
+
+			if (!se_mask[idx]) {
+				raster_config_se |= SE_MAP(RASTER_CONFIG_SE_MAP_3);
+			} else {
+				raster_config_se |= SE_MAP(RASTER_CONFIG_SE_MAP_0);
+			}
+		}
+
+		pkr0_mask &= rb_mask;
+		pkr1_mask &= rb_mask;
+		if (rb_per_se > 2 && (!pkr0_mask || !pkr1_mask)) {
+			raster_config_se &= ~PKR_MAP_MASK;
+
+			if (!pkr0_mask) {
+				raster_config_se |= PKR_MAP(RASTER_CONFIG_PKR_MAP_3);
+			} else {
+				raster_config_se |= PKR_MAP(RASTER_CONFIG_PKR_MAP_0);
+			}
+		}
+
+		if (rb_per_se >= 2) {
+			unsigned rb0_mask = 1 << (se * rb_per_se);
+			unsigned rb1_mask = rb0_mask << 1;
+
+			rb0_mask &= rb_mask;
+			rb1_mask &= rb_mask;
+			if (!rb0_mask || !rb1_mask) {
+				raster_config_se &= ~RB_MAP_PKR0_MASK;
+
+				if (!rb0_mask) {
+					raster_config_se |=
+						RB_MAP_PKR0(RASTER_CONFIG_RB_MAP_3);
+				} else {
+					raster_config_se |=
+						RB_MAP_PKR0(RASTER_CONFIG_RB_MAP_0);
+				}
+			}
+
+			if (rb_per_se > 2) {
+				rb0_mask = 1 << (se * rb_per_se + rb_per_pkr);
+				rb1_mask = rb0_mask << 1;
+				rb0_mask &= rb_mask;
+				rb1_mask &= rb_mask;
+				if (!rb0_mask || !rb1_mask) {
+					raster_config_se &= ~RB_MAP_PKR1_MASK;
+
+					if (!rb0_mask) {
+						raster_config_se |=
+							RB_MAP_PKR1(RASTER_CONFIG_RB_MAP_3);
+					} else {
+						raster_config_se |=
+							RB_MAP_PKR1(RASTER_CONFIG_RB_MAP_0);
+					}
+				}
+			}
+		}
+
+		/* GRBM_GFX_INDEX has a different offset on VI */
+		gfx_v8_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff);
+		WREG32(mmPA_SC_RASTER_CONFIG, raster_config_se);
+		WREG32(mmPA_SC_RASTER_CONFIG_1, raster_config_1);
+	}
+
+	/* GRBM_GFX_INDEX has a different offset on VI */
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+}
+
+static void gfx_v8_0_setup_rb(struct amdgpu_device *adev)
+{
+	int i, j;
+	u32 data;
+	u32 raster_config = 0, raster_config_1 = 0;
+	u32 active_rbs = 0;
+	u32 rb_bitmap_width_per_sh = adev->gfx.config.max_backends_per_se /
+					adev->gfx.config.max_sh_per_se;
+	unsigned num_rb_pipes;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+			data = gfx_v8_0_get_rb_active_bitmap(adev);
+			active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
+					       rb_bitmap_width_per_sh);
+		}
+	}
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+
+	adev->gfx.config.backend_enable_mask = active_rbs;
+	adev->gfx.config.num_rbs = hweight32(active_rbs);
+
+	num_rb_pipes = min_t(unsigned, adev->gfx.config.max_backends_per_se *
+			     adev->gfx.config.max_shader_engines, 16);
+
+	gfx_v8_0_raster_config(adev, &raster_config, &raster_config_1);
+
+	if (!adev->gfx.config.backend_enable_mask ||
+			adev->gfx.config.num_rbs >= num_rb_pipes) {
+		WREG32(mmPA_SC_RASTER_CONFIG, raster_config);
+		WREG32(mmPA_SC_RASTER_CONFIG_1, raster_config_1);
+	} else {
+		gfx_v8_0_write_harvested_raster_configs(adev, raster_config, raster_config_1,
+							adev->gfx.config.backend_enable_mask,
+							num_rb_pipes);
+	}
+
+	/* cache the values for userspace */
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+			adev->gfx.config.rb_config[i][j].rb_backend_disable =
+				RREG32(mmCC_RB_BACKEND_DISABLE);
+			adev->gfx.config.rb_config[i][j].user_rb_backend_disable =
+				RREG32(mmGC_USER_RB_BACKEND_DISABLE);
+			adev->gfx.config.rb_config[i][j].raster_config =
+				RREG32(mmPA_SC_RASTER_CONFIG);
+			adev->gfx.config.rb_config[i][j].raster_config_1 =
+				RREG32(mmPA_SC_RASTER_CONFIG_1);
+		}
+	}
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+/**
+ * gfx_v8_0_init_compute_vmid - gart enable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize compute vmid sh_mem registers
+ *
+ */
+#define DEFAULT_SH_MEM_BASES	(0x6000)
+#define FIRST_COMPUTE_VMID	(8)
+#define LAST_COMPUTE_VMID	(16)
+static void gfx_v8_0_init_compute_vmid(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t sh_mem_config;
+	uint32_t sh_mem_bases;
+
+	/*
+	 * Configure apertures:
+	 * LDS:         0x60000000'00000000 - 0x60000001'00000000 (4GB)
+	 * Scratch:     0x60000001'00000000 - 0x60000002'00000000 (4GB)
+	 * GPUVM:       0x60010000'00000000 - 0x60020000'00000000 (1TB)
+	 */
+	sh_mem_bases = DEFAULT_SH_MEM_BASES | (DEFAULT_SH_MEM_BASES << 16);
+
+	sh_mem_config = SH_MEM_ADDRESS_MODE_HSA64 <<
+			SH_MEM_CONFIG__ADDRESS_MODE__SHIFT |
+			SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
+			SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT |
+			MTYPE_CC << SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT |
+			SH_MEM_CONFIG__PRIVATE_ATC_MASK;
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = FIRST_COMPUTE_VMID; i < LAST_COMPUTE_VMID; i++) {
+		vi_srbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32(mmSH_MEM_CONFIG, sh_mem_config);
+		WREG32(mmSH_MEM_APE1_BASE, 1);
+		WREG32(mmSH_MEM_APE1_LIMIT, 0);
+		WREG32(mmSH_MEM_BASES, sh_mem_bases);
+	}
+	vi_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void gfx_v8_0_config_init(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	default:
+		adev->gfx.config.double_offchip_lds_buf = 1;
+		break;
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		adev->gfx.config.double_offchip_lds_buf = 0;
+		break;
+	}
+}
+
+static void gfx_v8_0_gpu_init(struct amdgpu_device *adev)
+{
+	u32 tmp, sh_static_mem_cfg;
+	int i;
+
+	WREG32_FIELD(GRBM_CNTL, READ_TIMEOUT, 0xFF);
+	WREG32(mmGB_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmHDP_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmDMIF_ADDR_CALC, adev->gfx.config.gb_addr_config);
+
+	gfx_v8_0_tiling_mode_table_init(adev);
+	gfx_v8_0_setup_rb(adev);
+	gfx_v8_0_get_cu_info(adev);
+	gfx_v8_0_config_init(adev);
+
+	/* XXX SH_MEM regs */
+	/* where to put LDS, scratch, GPUVM in FSA64 space */
+	sh_static_mem_cfg = REG_SET_FIELD(0, SH_STATIC_MEM_CONFIG,
+				   SWIZZLE_ENABLE, 1);
+	sh_static_mem_cfg = REG_SET_FIELD(sh_static_mem_cfg, SH_STATIC_MEM_CONFIG,
+				   ELEMENT_SIZE, 1);
+	sh_static_mem_cfg = REG_SET_FIELD(sh_static_mem_cfg, SH_STATIC_MEM_CONFIG,
+				   INDEX_STRIDE, 3);
+	WREG32(mmSH_STATIC_MEM_CONFIG, sh_static_mem_cfg);
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->vm_manager.id_mgr[0].num_ids; i++) {
+		vi_srbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		if (i == 0) {
+			tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, DEFAULT_MTYPE, MTYPE_UC);
+			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, APE1_MTYPE, MTYPE_UC);
+			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, ALIGNMENT_MODE,
+					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
+			WREG32(mmSH_MEM_CONFIG, tmp);
+			WREG32(mmSH_MEM_BASES, 0);
+		} else {
+			tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, DEFAULT_MTYPE, MTYPE_NC);
+			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, APE1_MTYPE, MTYPE_UC);
+			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, ALIGNMENT_MODE,
+					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
+			WREG32(mmSH_MEM_CONFIG, tmp);
+			tmp = adev->gmc.shared_aperture_start >> 48;
+			WREG32(mmSH_MEM_BASES, tmp);
+		}
+
+		WREG32(mmSH_MEM_APE1_BASE, 1);
+		WREG32(mmSH_MEM_APE1_LIMIT, 0);
+	}
+	vi_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	gfx_v8_0_init_compute_vmid(adev);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	/*
+	 * making sure that the following register writes will be broadcasted
+	 * to all the shaders
+	 */
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+
+	WREG32(mmPA_SC_FIFO_SIZE,
+		   (adev->gfx.config.sc_prim_fifo_size_frontend <<
+			PA_SC_FIFO_SIZE__SC_FRONTEND_PRIM_FIFO_SIZE__SHIFT) |
+		   (adev->gfx.config.sc_prim_fifo_size_backend <<
+			PA_SC_FIFO_SIZE__SC_BACKEND_PRIM_FIFO_SIZE__SHIFT) |
+		   (adev->gfx.config.sc_hiz_tile_fifo_size <<
+			PA_SC_FIFO_SIZE__SC_HIZ_TILE_FIFO_SIZE__SHIFT) |
+		   (adev->gfx.config.sc_earlyz_tile_fifo_size <<
+			PA_SC_FIFO_SIZE__SC_EARLYZ_TILE_FIFO_SIZE__SHIFT));
+
+	tmp = RREG32(mmSPI_ARB_PRIORITY);
+	tmp = REG_SET_FIELD(tmp, SPI_ARB_PRIORITY, PIPE_ORDER_TS0, 2);
+	tmp = REG_SET_FIELD(tmp, SPI_ARB_PRIORITY, PIPE_ORDER_TS1, 2);
+	tmp = REG_SET_FIELD(tmp, SPI_ARB_PRIORITY, PIPE_ORDER_TS2, 2);
+	tmp = REG_SET_FIELD(tmp, SPI_ARB_PRIORITY, PIPE_ORDER_TS3, 2);
+	WREG32(mmSPI_ARB_PRIORITY, tmp);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+}
+
+static void gfx_v8_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
+{
+	u32 i, j, k;
+	u32 mask;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+			for (k = 0; k < adev->usec_timeout; k++) {
+				if (RREG32(mmRLC_SERDES_CU_MASTER_BUSY) == 0)
+					break;
+				udelay(1);
+			}
+			if (k == adev->usec_timeout) {
+				gfx_v8_0_select_se_sh(adev, 0xffffffff,
+						      0xffffffff, 0xffffffff);
+				mutex_unlock(&adev->grbm_idx_mutex);
+				DRM_INFO("Timeout wait for RLC serdes %u,%u\n",
+					 i, j);
+				return;
+			}
+		}
+	}
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__GC_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC0_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC1_MASTER_BUSY_MASK;
+	for (k = 0; k < adev->usec_timeout; k++) {
+		if ((RREG32(mmRLC_SERDES_NONCU_MASTER_BUSY) & mask) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v8_0_enable_gui_idle_interrupt(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 tmp = RREG32(mmCP_INT_CNTL_RING0);
+
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE, enable ? 1 : 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE, enable ? 1 : 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE, enable ? 1 : 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE, enable ? 1 : 0);
+
+	WREG32(mmCP_INT_CNTL_RING0, tmp);
+}
+
+static void gfx_v8_0_init_csb(struct amdgpu_device *adev)
+{
+	/* csib */
+	WREG32(mmRLC_CSIB_ADDR_HI,
+			adev->gfx.rlc.clear_state_gpu_addr >> 32);
+	WREG32(mmRLC_CSIB_ADDR_LO,
+			adev->gfx.rlc.clear_state_gpu_addr & 0xfffffffc);
+	WREG32(mmRLC_CSIB_LENGTH,
+			adev->gfx.rlc.clear_state_size);
+}
+
+static void gfx_v8_0_parse_ind_reg_list(int *register_list_format,
+				int ind_offset,
+				int list_size,
+				int *unique_indices,
+				int *indices_count,
+				int max_indices,
+				int *ind_start_offsets,
+				int *offset_count,
+				int max_offset)
+{
+	int indices;
+	bool new_entry = true;
+
+	for (; ind_offset < list_size; ind_offset++) {
+
+		if (new_entry) {
+			new_entry = false;
+			ind_start_offsets[*offset_count] = ind_offset;
+			*offset_count = *offset_count + 1;
+			BUG_ON(*offset_count >= max_offset);
+		}
+
+		if (register_list_format[ind_offset] == 0xFFFFFFFF) {
+			new_entry = true;
+			continue;
+		}
+
+		ind_offset += 2;
+
+		/* look for the matching indice */
+		for (indices = 0;
+			indices < *indices_count;
+			indices++) {
+			if (unique_indices[indices] ==
+				register_list_format[ind_offset])
+				break;
+		}
+
+		if (indices >= *indices_count) {
+			unique_indices[*indices_count] =
+				register_list_format[ind_offset];
+			indices = *indices_count;
+			*indices_count = *indices_count + 1;
+			BUG_ON(*indices_count >= max_indices);
+		}
+
+		register_list_format[ind_offset] = indices;
+	}
+}
+
+static int gfx_v8_0_init_save_restore_list(struct amdgpu_device *adev)
+{
+	int i, temp, data;
+	int unique_indices[] = {0, 0, 0, 0, 0, 0, 0, 0};
+	int indices_count = 0;
+	int indirect_start_offsets[] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+	int offset_count = 0;
+
+	int list_size;
+	unsigned int *register_list_format =
+		kmalloc(adev->gfx.rlc.reg_list_format_size_bytes, GFP_KERNEL);
+	if (!register_list_format)
+		return -ENOMEM;
+	memcpy(register_list_format, adev->gfx.rlc.register_list_format,
+			adev->gfx.rlc.reg_list_format_size_bytes);
+
+	gfx_v8_0_parse_ind_reg_list(register_list_format,
+				RLC_FormatDirectRegListLength,
+				adev->gfx.rlc.reg_list_format_size_bytes >> 2,
+				unique_indices,
+				&indices_count,
+				ARRAY_SIZE(unique_indices),
+				indirect_start_offsets,
+				&offset_count,
+				ARRAY_SIZE(indirect_start_offsets));
+
+	/* save and restore list */
+	WREG32_FIELD(RLC_SRM_CNTL, AUTO_INCR_ADDR, 1);
+
+	WREG32(mmRLC_SRM_ARAM_ADDR, 0);
+	for (i = 0; i < adev->gfx.rlc.reg_list_size_bytes >> 2; i++)
+		WREG32(mmRLC_SRM_ARAM_DATA, adev->gfx.rlc.register_restore[i]);
+
+	/* indirect list */
+	WREG32(mmRLC_GPM_SCRATCH_ADDR, adev->gfx.rlc.reg_list_format_start);
+	for (i = 0; i < adev->gfx.rlc.reg_list_format_size_bytes >> 2; i++)
+		WREG32(mmRLC_GPM_SCRATCH_DATA, register_list_format[i]);
+
+	list_size = adev->gfx.rlc.reg_list_size_bytes >> 2;
+	list_size = list_size >> 1;
+	WREG32(mmRLC_GPM_SCRATCH_ADDR, adev->gfx.rlc.reg_restore_list_size);
+	WREG32(mmRLC_GPM_SCRATCH_DATA, list_size);
+
+	/* starting offsets starts */
+	WREG32(mmRLC_GPM_SCRATCH_ADDR,
+		adev->gfx.rlc.starting_offsets_start);
+	for (i = 0; i < ARRAY_SIZE(indirect_start_offsets); i++)
+		WREG32(mmRLC_GPM_SCRATCH_DATA,
+				indirect_start_offsets[i]);
+
+	/* unique indices */
+	temp = mmRLC_SRM_INDEX_CNTL_ADDR_0;
+	data = mmRLC_SRM_INDEX_CNTL_DATA_0;
+	for (i = 0; i < ARRAY_SIZE(unique_indices); i++) {
+		if (unique_indices[i] != 0) {
+			WREG32(temp + i, unique_indices[i] & 0x3FFFF);
+			WREG32(data + i, unique_indices[i] >> 20);
+		}
+	}
+	kfree(register_list_format);
+
+	return 0;
+}
+
+static void gfx_v8_0_enable_save_restore_machine(struct amdgpu_device *adev)
+{
+	WREG32_FIELD(RLC_SRM_CNTL, SRM_ENABLE, 1);
+}
+
+static void gfx_v8_0_init_power_gating(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	WREG32_FIELD(CP_RB_WPTR_POLL_CNTL, IDLE_POLL_COUNT, 0x60);
+
+	data = REG_SET_FIELD(0, RLC_PG_DELAY, POWER_UP_DELAY, 0x10);
+	data = REG_SET_FIELD(data, RLC_PG_DELAY, POWER_DOWN_DELAY, 0x10);
+	data = REG_SET_FIELD(data, RLC_PG_DELAY, CMD_PROPAGATE_DELAY, 0x10);
+	data = REG_SET_FIELD(data, RLC_PG_DELAY, MEM_SLEEP_DELAY, 0x10);
+	WREG32(mmRLC_PG_DELAY, data);
+
+	WREG32_FIELD(RLC_PG_DELAY_2, SERDES_CMD_DELAY, 0x3);
+	WREG32_FIELD(RLC_AUTO_PG_CTRL, GRBM_REG_SAVE_GFX_IDLE_THRESHOLD, 0x55f0);
+
+}
+
+static void cz_enable_sck_slow_down_on_power_up(struct amdgpu_device *adev,
+						bool enable)
+{
+	WREG32_FIELD(RLC_PG_CNTL, SMU_CLK_SLOWDOWN_ON_PU_ENABLE, enable ? 1 : 0);
+}
+
+static void cz_enable_sck_slow_down_on_power_down(struct amdgpu_device *adev,
+						  bool enable)
+{
+	WREG32_FIELD(RLC_PG_CNTL, SMU_CLK_SLOWDOWN_ON_PD_ENABLE, enable ? 1 : 0);
+}
+
+static void cz_enable_cp_power_gating(struct amdgpu_device *adev, bool enable)
+{
+	WREG32_FIELD(RLC_PG_CNTL, CP_PG_DISABLE, enable ? 0 : 1);
+}
+
+static void gfx_v8_0_init_pg(struct amdgpu_device *adev)
+{
+	if ((adev->asic_type == CHIP_CARRIZO) ||
+	    (adev->asic_type == CHIP_STONEY)) {
+		gfx_v8_0_init_csb(adev);
+		gfx_v8_0_init_save_restore_list(adev);
+		gfx_v8_0_enable_save_restore_machine(adev);
+		WREG32(mmRLC_JUMP_TABLE_RESTORE, adev->gfx.rlc.cp_table_gpu_addr >> 8);
+		gfx_v8_0_init_power_gating(adev);
+		WREG32(mmRLC_PG_ALWAYS_ON_CU_MASK, adev->gfx.cu_info.ao_cu_mask);
+	} else if ((adev->asic_type == CHIP_POLARIS11) ||
+		   (adev->asic_type == CHIP_POLARIS12) ||
+		   (adev->asic_type == CHIP_VEGAM)) {
+		gfx_v8_0_init_csb(adev);
+		gfx_v8_0_init_save_restore_list(adev);
+		gfx_v8_0_enable_save_restore_machine(adev);
+		gfx_v8_0_init_power_gating(adev);
+	}
+
+}
+
+static void gfx_v8_0_rlc_stop(struct amdgpu_device *adev)
+{
+	WREG32_FIELD(RLC_CNTL, RLC_ENABLE_F32, 0);
+
+	gfx_v8_0_enable_gui_idle_interrupt(adev, false);
+	gfx_v8_0_wait_for_rlc_serdes(adev);
+}
+
+static void gfx_v8_0_rlc_reset(struct amdgpu_device *adev)
+{
+	WREG32_FIELD(GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+	udelay(50);
+
+	WREG32_FIELD(GRBM_SOFT_RESET, SOFT_RESET_RLC, 0);
+	udelay(50);
+}
+
+static void gfx_v8_0_rlc_start(struct amdgpu_device *adev)
+{
+	WREG32_FIELD(RLC_CNTL, RLC_ENABLE_F32, 1);
+
+	/* carrizo do enable cp interrupt after cp inited */
+	if (!(adev->flags & AMD_IS_APU))
+		gfx_v8_0_enable_gui_idle_interrupt(adev, true);
+
+	udelay(50);
+}
+
+static int gfx_v8_0_rlc_load_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_0 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.rlc_fw)
+		return -EINVAL;
+
+	hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+	amdgpu_ucode_print_rlc_hdr(&hdr->header);
+
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			   le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+
+	WREG32(mmRLC_GPM_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmRLC_GPM_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmRLC_GPM_UCODE_ADDR, adev->gfx.rlc_fw_version);
+
+	return 0;
+}
+
+static int gfx_v8_0_rlc_resume(struct amdgpu_device *adev)
+{
+	int r;
+	u32 tmp;
+
+	gfx_v8_0_rlc_stop(adev);
+
+	/* disable CG */
+	tmp = RREG32(mmRLC_CGCG_CGLS_CTRL);
+	tmp &= ~(RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK |
+		 RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK);
+	WREG32(mmRLC_CGCG_CGLS_CTRL, tmp);
+	if (adev->asic_type == CHIP_POLARIS11 ||
+	    adev->asic_type == CHIP_POLARIS10 ||
+	    adev->asic_type == CHIP_POLARIS12 ||
+	    adev->asic_type == CHIP_VEGAM) {
+		tmp = RREG32(mmRLC_CGCG_CGLS_CTRL_3D);
+		tmp &= ~0x3;
+		WREG32(mmRLC_CGCG_CGLS_CTRL_3D, tmp);
+	}
+
+	/* disable PG */
+	WREG32(mmRLC_PG_CNTL, 0);
+
+	gfx_v8_0_rlc_reset(adev);
+	gfx_v8_0_init_pg(adev);
+
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		/* legacy rlc firmware loading */
+		r = gfx_v8_0_rlc_load_microcode(adev);
+		if (r)
+			return r;
+	}
+
+	gfx_v8_0_rlc_start(adev);
+
+	return 0;
+}
+
+static void gfx_v8_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+	u32 tmp = RREG32(mmCP_ME_CNTL);
+
+	if (enable) {
+		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_HALT, 0);
+		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, 0);
+		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, 0);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_HALT, 1);
+		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, 1);
+		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, 1);
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+			adev->gfx.gfx_ring[i].ready = false;
+	}
+	WREG32(mmCP_ME_CNTL, tmp);
+	udelay(50);
+}
+
+static int gfx_v8_0_cp_gfx_load_microcode(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v1_0 *pfp_hdr;
+	const struct gfx_firmware_header_v1_0 *ce_hdr;
+	const struct gfx_firmware_header_v1_0 *me_hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.me_fw || !adev->gfx.pfp_fw || !adev->gfx.ce_fw)
+		return -EINVAL;
+
+	pfp_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.pfp_fw->data;
+	ce_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.ce_fw->data;
+	me_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.me_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&pfp_hdr->header);
+	amdgpu_ucode_print_gfx_hdr(&ce_hdr->header);
+	amdgpu_ucode_print_gfx_hdr(&me_hdr->header);
+
+	gfx_v8_0_cp_gfx_enable(adev, false);
+
+	/* PFP */
+	fw_data = (const __le32 *)
+		(adev->gfx.pfp_fw->data +
+		 le32_to_cpu(pfp_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(pfp_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_PFP_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_PFP_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_PFP_UCODE_ADDR, adev->gfx.pfp_fw_version);
+
+	/* CE */
+	fw_data = (const __le32 *)
+		(adev->gfx.ce_fw->data +
+		 le32_to_cpu(ce_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(ce_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_CE_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_CE_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_CE_UCODE_ADDR, adev->gfx.ce_fw_version);
+
+	/* ME */
+	fw_data = (const __le32 *)
+		(adev->gfx.me_fw->data +
+		 le32_to_cpu(me_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(me_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_ME_RAM_WADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_ME_RAM_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_ME_RAM_WADDR, adev->gfx.me_fw_version);
+
+	return 0;
+}
+
+static u32 gfx_v8_0_get_csb_size(struct amdgpu_device *adev)
+{
+	u32 count = 0;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	/* begin clear state */
+	count += 2;
+	/* context control state */
+	count += 3;
+
+	for (sect = vi_cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT)
+				count += 2 + ext->reg_count;
+			else
+				return 0;
+		}
+	}
+	/* pa_sc_raster_config/pa_sc_raster_config1 */
+	count += 4;
+	/* end clear state */
+	count += 2;
+	/* clear state */
+	count += 2;
+
+	return count;
+}
+
+static int gfx_v8_0_cp_gfx_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->gfx.gfx_ring[0];
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+	int r, i;
+
+	/* init the CP */
+	WREG32(mmCP_MAX_CONTEXT, adev->gfx.config.max_hw_contexts - 1);
+	WREG32(mmCP_ENDIAN_SWAP, 0);
+	WREG32(mmCP_DEVICE_ID, 1);
+
+	gfx_v8_0_cp_gfx_enable(adev, true);
+
+	r = amdgpu_ring_alloc(ring, gfx_v8_0_get_csb_size(adev) + 4);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
+		return r;
+	}
+
+	/* clear state buffer */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, 0x80000000);
+	amdgpu_ring_write(ring, 0x80000000);
+
+	for (sect = vi_cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				amdgpu_ring_write(ring,
+				       PACKET3(PACKET3_SET_CONTEXT_REG,
+					       ext->reg_count));
+				amdgpu_ring_write(ring,
+				       ext->reg_index - PACKET3_SET_CONTEXT_REG_START);
+				for (i = 0; i < ext->reg_count; i++)
+					amdgpu_ring_write(ring, ext->extent[i]);
+			}
+		}
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONTEXT_REG, 2));
+	amdgpu_ring_write(ring, mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
+	amdgpu_ring_write(ring, adev->gfx.config.rb_config[0][0].raster_config);
+	amdgpu_ring_write(ring, adev->gfx.config.rb_config[0][0].raster_config_1);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CLEAR_STATE, 0));
+	amdgpu_ring_write(ring, 0);
+
+	/* init the CE partitions */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_BASE, 2));
+	amdgpu_ring_write(ring, PACKET3_BASE_INDEX(CE_PARTITION_BASE));
+	amdgpu_ring_write(ring, 0x8000);
+	amdgpu_ring_write(ring, 0x8000);
+
+	amdgpu_ring_commit(ring);
+
+	return 0;
+}
+static void gfx_v8_0_set_cpg_door_bell(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	u32 tmp;
+	/* no gfx doorbells on iceland */
+	if (adev->asic_type == CHIP_TOPAZ)
+		return;
+
+	tmp = RREG32(mmCP_RB_DOORBELL_CONTROL);
+
+	if (ring->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				DOORBELL_OFFSET, ring->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+						DOORBELL_HIT, 0);
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+					    DOORBELL_EN, 1);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL, DOORBELL_EN, 0);
+	}
+
+	WREG32(mmCP_RB_DOORBELL_CONTROL, tmp);
+
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	tmp = REG_SET_FIELD(0, CP_RB_DOORBELL_RANGE_LOWER,
+					DOORBELL_RANGE_LOWER,
+					AMDGPU_DOORBELL_GFX_RING0);
+	WREG32(mmCP_RB_DOORBELL_RANGE_LOWER, tmp);
+
+	WREG32(mmCP_RB_DOORBELL_RANGE_UPPER,
+		CP_RB_DOORBELL_RANGE_UPPER__DOORBELL_RANGE_UPPER_MASK);
+}
+
+static int gfx_v8_0_cp_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 tmp;
+	u32 rb_bufsz;
+	u64 rb_addr, rptr_addr, wptr_gpu_addr;
+	int r;
+
+	/* Set the write pointer delay */
+	WREG32(mmCP_RB_WPTR_DELAY, 0);
+
+	/* set the RB to use vmid 0 */
+	WREG32(mmCP_RB_VMID, 0);
+
+	/* Set ring buffer size */
+	ring = &adev->gfx.gfx_ring[0];
+	rb_bufsz = order_base_2(ring->ring_size / 8);
+	tmp = REG_SET_FIELD(0, CP_RB0_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, RB_BLKSZ, rb_bufsz - 2);
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, MTYPE, 3);
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, MIN_IB_AVAILSZ, 1);
+#ifdef __BIG_ENDIAN
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, BUF_SWAP, 1);
+#endif
+	WREG32(mmCP_RB0_CNTL, tmp);
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(mmCP_RB0_CNTL, tmp | CP_RB0_CNTL__RB_RPTR_WR_ENA_MASK);
+	ring->wptr = 0;
+	WREG32(mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
+
+	/* set the wb address wether it's enabled or not */
+	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	WREG32(mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
+	WREG32(mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF);
+
+	wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	WREG32(mmCP_RB_WPTR_POLL_ADDR_LO, lower_32_bits(wptr_gpu_addr));
+	WREG32(mmCP_RB_WPTR_POLL_ADDR_HI, upper_32_bits(wptr_gpu_addr));
+	mdelay(1);
+	WREG32(mmCP_RB0_CNTL, tmp);
+
+	rb_addr = ring->gpu_addr >> 8;
+	WREG32(mmCP_RB0_BASE, rb_addr);
+	WREG32(mmCP_RB0_BASE_HI, upper_32_bits(rb_addr));
+
+	gfx_v8_0_set_cpg_door_bell(adev, ring);
+	/* start the ring */
+	amdgpu_ring_clear_ring(ring);
+	gfx_v8_0_cp_gfx_start(adev);
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r)
+		ring->ready = false;
+
+	return r;
+}
+
+static void gfx_v8_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+
+	if (enable) {
+		WREG32(mmCP_MEC_CNTL, 0);
+	} else {
+		WREG32(mmCP_MEC_CNTL, (CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK));
+		for (i = 0; i < adev->gfx.num_compute_rings; i++)
+			adev->gfx.compute_ring[i].ready = false;
+		adev->gfx.kiq.ring.ready = false;
+	}
+	udelay(50);
+}
+
+static int gfx_v8_0_cp_compute_load_microcode(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v1_0 *mec_hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.mec_fw)
+		return -EINVAL;
+
+	gfx_v8_0_cp_compute_enable(adev, false);
+
+	mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+	amdgpu_ucode_print_gfx_hdr(&mec_hdr->header);
+
+	fw_data = (const __le32 *)
+		(adev->gfx.mec_fw->data +
+		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes) / 4;
+
+	/* MEC1 */
+	WREG32(mmCP_MEC_ME1_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_MEC_ME1_UCODE_DATA, le32_to_cpup(fw_data+i));
+	WREG32(mmCP_MEC_ME1_UCODE_ADDR, adev->gfx.mec_fw_version);
+
+	/* Loading MEC2 firmware is only necessary if MEC2 should run different microcode than MEC1. */
+	if (adev->gfx.mec2_fw) {
+		const struct gfx_firmware_header_v1_0 *mec2_hdr;
+
+		mec2_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
+		amdgpu_ucode_print_gfx_hdr(&mec2_hdr->header);
+
+		fw_data = (const __le32 *)
+			(adev->gfx.mec2_fw->data +
+			 le32_to_cpu(mec2_hdr->header.ucode_array_offset_bytes));
+		fw_size = le32_to_cpu(mec2_hdr->header.ucode_size_bytes) / 4;
+
+		WREG32(mmCP_MEC_ME2_UCODE_ADDR, 0);
+		for (i = 0; i < fw_size; i++)
+			WREG32(mmCP_MEC_ME2_UCODE_DATA, le32_to_cpup(fw_data+i));
+		WREG32(mmCP_MEC_ME2_UCODE_ADDR, adev->gfx.mec2_fw_version);
+	}
+
+	return 0;
+}
+
+/* KIQ functions */
+static void gfx_v8_0_kiq_setting(struct amdgpu_ring *ring)
+{
+	uint32_t tmp;
+	struct amdgpu_device *adev = ring->adev;
+
+	/* tell RLC which is KIQ queue */
+	tmp = RREG32(mmRLC_CP_SCHEDULERS);
+	tmp &= 0xffffff00;
+	tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue);
+	WREG32(mmRLC_CP_SCHEDULERS, tmp);
+	tmp |= 0x80;
+	WREG32(mmRLC_CP_SCHEDULERS, tmp);
+}
+
+static int gfx_v8_0_kiq_kcq_enable(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq.ring;
+	uint32_t scratch, tmp = 0;
+	uint64_t queue_mask = 0;
+	int r, i;
+
+	for (i = 0; i < AMDGPU_MAX_COMPUTE_QUEUES; ++i) {
+		if (!test_bit(i, adev->gfx.mec.queue_bitmap))
+			continue;
+
+		/* This situation may be hit in the future if a new HW
+		 * generation exposes more than 64 queues. If so, the
+		 * definition of queue_mask needs updating */
+		if (WARN_ON(i >= (sizeof(queue_mask)*8))) {
+			DRM_ERROR("Invalid KCQ enabled: %d\n", i);
+			break;
+		}
+
+		queue_mask |= (1ull << i);
+	}
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("Failed to get scratch reg (%d).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+
+	r = amdgpu_ring_alloc(kiq_ring, (8 * adev->gfx.num_compute_rings) + 11);
+	if (r) {
+		DRM_ERROR("Failed to lock KIQ (%d).\n", r);
+		amdgpu_gfx_scratch_free(adev, scratch);
+		return r;
+	}
+	/* set resources */
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_RESOURCES, 6));
+	amdgpu_ring_write(kiq_ring, 0);	/* vmid_mask:0 queue_type:0 (KIQ) */
+	amdgpu_ring_write(kiq_ring, lower_32_bits(queue_mask));	/* queue mask lo */
+	amdgpu_ring_write(kiq_ring, upper_32_bits(queue_mask));	/* queue mask hi */
+	amdgpu_ring_write(kiq_ring, 0);	/* gws mask lo */
+	amdgpu_ring_write(kiq_ring, 0);	/* gws mask hi */
+	amdgpu_ring_write(kiq_ring, 0);	/* oac mask */
+	amdgpu_ring_write(kiq_ring, 0);	/* gds heap base:0, gds heap size:0 */
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
+		uint64_t mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj);
+		uint64_t wptr_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+
+		/* map queues */
+		amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
+		/* Q_sel:0, vmid:0, vidmem: 1, engine:0, num_Q:1*/
+		amdgpu_ring_write(kiq_ring,
+				  PACKET3_MAP_QUEUES_NUM_QUEUES(1));
+		amdgpu_ring_write(kiq_ring,
+				  PACKET3_MAP_QUEUES_DOORBELL_OFFSET(ring->doorbell_index) |
+				  PACKET3_MAP_QUEUES_QUEUE(ring->queue) |
+				  PACKET3_MAP_QUEUES_PIPE(ring->pipe) |
+				  PACKET3_MAP_QUEUES_ME(ring->me == 1 ? 0 : 1)); /* doorbell */
+		amdgpu_ring_write(kiq_ring, lower_32_bits(mqd_addr));
+		amdgpu_ring_write(kiq_ring, upper_32_bits(mqd_addr));
+		amdgpu_ring_write(kiq_ring, lower_32_bits(wptr_addr));
+		amdgpu_ring_write(kiq_ring, upper_32_bits(wptr_addr));
+	}
+	/* write to scratch for completion */
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+	amdgpu_ring_write(kiq_ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(kiq_ring, 0xDEADBEEF);
+	amdgpu_ring_commit(kiq_ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i >= adev->usec_timeout) {
+		DRM_ERROR("KCQ enable failed (scratch(0x%04X)=0x%08X)\n",
+			  scratch, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_gfx_scratch_free(adev, scratch);
+
+	return r;
+}
+
+static int gfx_v8_0_deactivate_hqd(struct amdgpu_device *adev, u32 req)
+{
+	int i, r = 0;
+
+	if (RREG32(mmCP_HQD_ACTIVE) & CP_HQD_ACTIVE__ACTIVE_MASK) {
+		WREG32_FIELD(CP_HQD_DEQUEUE_REQUEST, DEQUEUE_REQ, req);
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!(RREG32(mmCP_HQD_ACTIVE) & CP_HQD_ACTIVE__ACTIVE_MASK))
+				break;
+			udelay(1);
+		}
+		if (i == adev->usec_timeout)
+			r = -ETIMEDOUT;
+	}
+	WREG32(mmCP_HQD_DEQUEUE_REQUEST, 0);
+	WREG32(mmCP_HQD_PQ_RPTR, 0);
+	WREG32(mmCP_HQD_PQ_WPTR, 0);
+
+	return r;
+}
+
+static int gfx_v8_0_mqd_init(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct vi_mqd *mqd = ring->mqd_ptr;
+	uint64_t hqd_gpu_addr, wb_gpu_addr, eop_base_addr;
+	uint32_t tmp;
+
+	mqd->header = 0xC0310800;
+	mqd->compute_pipelinestat_enable = 0x00000001;
+	mqd->compute_static_thread_mgmt_se0 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se1 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se2 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se3 = 0xffffffff;
+	mqd->compute_misc_reserved = 0x00000003;
+	mqd->dynamic_cu_mask_addr_lo = lower_32_bits(ring->mqd_gpu_addr
+						     + offsetof(struct vi_mqd_allocation, dynamic_cu_mask));
+	mqd->dynamic_cu_mask_addr_hi = upper_32_bits(ring->mqd_gpu_addr
+						     + offsetof(struct vi_mqd_allocation, dynamic_cu_mask));
+	eop_base_addr = ring->eop_gpu_addr >> 8;
+	mqd->cp_hqd_eop_base_addr_lo = eop_base_addr;
+	mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr);
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	tmp = RREG32(mmCP_HQD_EOP_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE,
+			(order_base_2(GFX8_MEC_HPD_SIZE / 4) - 1));
+
+	mqd->cp_hqd_eop_control = tmp;
+
+	/* enable doorbell? */
+	tmp = REG_SET_FIELD(RREG32(mmCP_HQD_PQ_DOORBELL_CONTROL),
+			    CP_HQD_PQ_DOORBELL_CONTROL,
+			    DOORBELL_EN,
+			    ring->use_doorbell ? 1 : 0);
+
+	mqd->cp_hqd_pq_doorbell_control = tmp;
+
+	/* set the pointer to the MQD */
+	mqd->cp_mqd_base_addr_lo = ring->mqd_gpu_addr & 0xfffffffc;
+	mqd->cp_mqd_base_addr_hi = upper_32_bits(ring->mqd_gpu_addr);
+
+	/* set MQD vmid to 0 */
+	tmp = RREG32(mmCP_MQD_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0);
+	mqd->cp_mqd_control = tmp;
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	hqd_gpu_addr = ring->gpu_addr >> 8;
+	mqd->cp_hqd_pq_base_lo = hqd_gpu_addr;
+	mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	tmp = RREG32(mmCP_HQD_PQ_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
+			    (order_base_2(ring->ring_size / 4) - 1));
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
+			((order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1) << 8));
+#ifdef __BIG_ENDIAN
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, ENDIAN_SWAP, 1);
+#endif
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 0);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, ROQ_PQ_IB_FLIP, 0);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+	mqd->cp_hqd_pq_control = tmp;
+
+	/* set the wb address whether it's enabled or not */
+	wb_gpu_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_rptr_report_addr_hi =
+		upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	wb_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	tmp = 0;
+	/* enable the doorbell if requested */
+	if (ring->use_doorbell) {
+		tmp = RREG32(mmCP_HQD_PQ_DOORBELL_CONTROL);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				DOORBELL_OFFSET, ring->doorbell_index);
+
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+					 DOORBELL_EN, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+					 DOORBELL_SOURCE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+					 DOORBELL_HIT, 0);
+	}
+
+	mqd->cp_hqd_pq_doorbell_control = tmp;
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	ring->wptr = 0;
+	mqd->cp_hqd_pq_wptr = ring->wptr;
+	mqd->cp_hqd_pq_rptr = RREG32(mmCP_HQD_PQ_RPTR);
+
+	/* set the vmid for the queue */
+	mqd->cp_hqd_vmid = 0;
+
+	tmp = RREG32(mmCP_HQD_PERSISTENT_STATE);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE, PRELOAD_SIZE, 0x53);
+	mqd->cp_hqd_persistent_state = tmp;
+
+	/* set MTYPE */
+	tmp = RREG32(mmCP_HQD_IB_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_IB_CONTROL, MIN_IB_AVAIL_SIZE, 3);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_IB_CONTROL, MTYPE, 3);
+	mqd->cp_hqd_ib_control = tmp;
+
+	tmp = RREG32(mmCP_HQD_IQ_TIMER);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_IQ_TIMER, MTYPE, 3);
+	mqd->cp_hqd_iq_timer = tmp;
+
+	tmp = RREG32(mmCP_HQD_CTX_SAVE_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_CTX_SAVE_CONTROL, MTYPE, 3);
+	mqd->cp_hqd_ctx_save_control = tmp;
+
+	/* defaults */
+	mqd->cp_hqd_eop_rptr = RREG32(mmCP_HQD_EOP_RPTR);
+	mqd->cp_hqd_eop_wptr = RREG32(mmCP_HQD_EOP_WPTR);
+	mqd->cp_hqd_pipe_priority = RREG32(mmCP_HQD_PIPE_PRIORITY);
+	mqd->cp_hqd_queue_priority = RREG32(mmCP_HQD_QUEUE_PRIORITY);
+	mqd->cp_hqd_quantum = RREG32(mmCP_HQD_QUANTUM);
+	mqd->cp_hqd_ctx_save_base_addr_lo = RREG32(mmCP_HQD_CTX_SAVE_BASE_ADDR_LO);
+	mqd->cp_hqd_ctx_save_base_addr_hi = RREG32(mmCP_HQD_CTX_SAVE_BASE_ADDR_HI);
+	mqd->cp_hqd_cntl_stack_offset = RREG32(mmCP_HQD_CNTL_STACK_OFFSET);
+	mqd->cp_hqd_cntl_stack_size = RREG32(mmCP_HQD_CNTL_STACK_SIZE);
+	mqd->cp_hqd_wg_state_offset = RREG32(mmCP_HQD_WG_STATE_OFFSET);
+	mqd->cp_hqd_ctx_save_size = RREG32(mmCP_HQD_CTX_SAVE_SIZE);
+	mqd->cp_hqd_eop_done_events = RREG32(mmCP_HQD_EOP_EVENTS);
+	mqd->cp_hqd_error = RREG32(mmCP_HQD_ERROR);
+	mqd->cp_hqd_eop_wptr_mem = RREG32(mmCP_HQD_EOP_WPTR_MEM);
+	mqd->cp_hqd_eop_dones = RREG32(mmCP_HQD_EOP_DONES);
+
+	/* activate the queue */
+	mqd->cp_hqd_active = 1;
+
+	return 0;
+}
+
+int gfx_v8_0_mqd_commit(struct amdgpu_device *adev,
+			struct vi_mqd *mqd)
+{
+	uint32_t mqd_reg;
+	uint32_t *mqd_data;
+
+	/* HQD registers extend from mmCP_MQD_BASE_ADDR to mmCP_HQD_ERROR */
+	mqd_data = &mqd->cp_mqd_base_addr_lo;
+
+	/* disable wptr polling */
+	WREG32_FIELD(CP_PQ_WPTR_POLL_CNTL, EN, 0);
+
+	/* program all HQD registers */
+	for (mqd_reg = mmCP_HQD_VMID; mqd_reg <= mmCP_HQD_EOP_CONTROL; mqd_reg++)
+		WREG32(mqd_reg, mqd_data[mqd_reg - mmCP_MQD_BASE_ADDR]);
+
+	/* Tonga errata: EOP RPTR/WPTR should be left unmodified.
+	 * This is safe since EOP RPTR==WPTR for any inactive HQD
+	 * on ASICs that do not support context-save.
+	 * EOP writes/reads can start anywhere in the ring.
+	 */
+	if (adev->asic_type != CHIP_TONGA) {
+		WREG32(mmCP_HQD_EOP_RPTR, mqd->cp_hqd_eop_rptr);
+		WREG32(mmCP_HQD_EOP_WPTR, mqd->cp_hqd_eop_wptr);
+		WREG32(mmCP_HQD_EOP_WPTR_MEM, mqd->cp_hqd_eop_wptr_mem);
+	}
+
+	for (mqd_reg = mmCP_HQD_EOP_EVENTS; mqd_reg <= mmCP_HQD_ERROR; mqd_reg++)
+		WREG32(mqd_reg, mqd_data[mqd_reg - mmCP_MQD_BASE_ADDR]);
+
+	/* activate the HQD */
+	for (mqd_reg = mmCP_MQD_BASE_ADDR; mqd_reg <= mmCP_HQD_ACTIVE; mqd_reg++)
+		WREG32(mqd_reg, mqd_data[mqd_reg - mmCP_MQD_BASE_ADDR]);
+
+	return 0;
+}
+
+static int gfx_v8_0_kiq_init_queue(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct vi_mqd *mqd = ring->mqd_ptr;
+	int mqd_idx = AMDGPU_MAX_COMPUTE_RINGS;
+
+	gfx_v8_0_kiq_setting(ring);
+
+	if (adev->in_gpu_reset) { /* for GPU_RESET case */
+		/* reset MQD to a clean status */
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(struct vi_mqd_allocation));
+
+		/* reset ring buffer */
+		ring->wptr = 0;
+		amdgpu_ring_clear_ring(ring);
+		mutex_lock(&adev->srbm_mutex);
+		vi_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		gfx_v8_0_mqd_commit(adev, mqd);
+		vi_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	} else {
+		memset((void *)mqd, 0, sizeof(struct vi_mqd_allocation));
+		((struct vi_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
+		((struct vi_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
+		mutex_lock(&adev->srbm_mutex);
+		vi_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		gfx_v8_0_mqd_init(ring);
+		gfx_v8_0_mqd_commit(adev, mqd);
+		vi_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(struct vi_mqd_allocation));
+	}
+
+	return 0;
+}
+
+static int gfx_v8_0_kcq_init_queue(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct vi_mqd *mqd = ring->mqd_ptr;
+	int mqd_idx = ring - &adev->gfx.compute_ring[0];
+
+	if (!adev->in_gpu_reset && !adev->gfx.in_suspend) {
+		memset((void *)mqd, 0, sizeof(struct vi_mqd_allocation));
+		((struct vi_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
+		((struct vi_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
+		mutex_lock(&adev->srbm_mutex);
+		vi_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		gfx_v8_0_mqd_init(ring);
+		vi_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(struct vi_mqd_allocation));
+	} else if (adev->in_gpu_reset) { /* for GPU_RESET case */
+		/* reset MQD to a clean status */
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(struct vi_mqd_allocation));
+		/* reset ring buffer */
+		ring->wptr = 0;
+		amdgpu_ring_clear_ring(ring);
+	} else {
+		amdgpu_ring_clear_ring(ring);
+	}
+	return 0;
+}
+
+static void gfx_v8_0_set_mec_doorbell_range(struct amdgpu_device *adev)
+{
+	if (adev->asic_type > CHIP_TONGA) {
+		WREG32(mmCP_MEC_DOORBELL_RANGE_LOWER, AMDGPU_DOORBELL_KIQ << 2);
+		WREG32(mmCP_MEC_DOORBELL_RANGE_UPPER, AMDGPU_DOORBELL_MEC_RING7 << 2);
+	}
+	/* enable doorbells */
+	WREG32_FIELD(CP_PQ_STATUS, DOORBELL_ENABLE, 1);
+}
+
+static int gfx_v8_0_kiq_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = NULL;
+	int r = 0, i;
+
+	gfx_v8_0_cp_compute_enable(adev, true);
+
+	ring = &adev->gfx.kiq.ring;
+
+	r = amdgpu_bo_reserve(ring->mqd_obj, false);
+	if (unlikely(r != 0))
+		goto done;
+
+	r = amdgpu_bo_kmap(ring->mqd_obj, &ring->mqd_ptr);
+	if (!r) {
+		r = gfx_v8_0_kiq_init_queue(ring);
+		amdgpu_bo_kunmap(ring->mqd_obj);
+		ring->mqd_ptr = NULL;
+	}
+	amdgpu_bo_unreserve(ring->mqd_obj);
+	if (r)
+		goto done;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+
+		r = amdgpu_bo_reserve(ring->mqd_obj, false);
+		if (unlikely(r != 0))
+			goto done;
+		r = amdgpu_bo_kmap(ring->mqd_obj, &ring->mqd_ptr);
+		if (!r) {
+			r = gfx_v8_0_kcq_init_queue(ring);
+			amdgpu_bo_kunmap(ring->mqd_obj);
+			ring->mqd_ptr = NULL;
+		}
+		amdgpu_bo_unreserve(ring->mqd_obj);
+		if (r)
+			goto done;
+	}
+
+	gfx_v8_0_set_mec_doorbell_range(adev);
+
+	r = gfx_v8_0_kiq_kcq_enable(adev);
+	if (r)
+		goto done;
+
+	/* Test KIQ */
+	ring = &adev->gfx.kiq.ring;
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		goto done;
+	}
+
+	/* Test KCQs */
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+		ring->ready = true;
+		r = amdgpu_ring_test_ring(ring);
+		if (r)
+			ring->ready = false;
+	}
+
+done:
+	return r;
+}
+
+static int gfx_v8_0_cp_resume(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!(adev->flags & AMD_IS_APU))
+		gfx_v8_0_enable_gui_idle_interrupt(adev, false);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+			/* legacy firmware loading */
+		r = gfx_v8_0_cp_gfx_load_microcode(adev);
+		if (r)
+			return r;
+
+		r = gfx_v8_0_cp_compute_load_microcode(adev);
+		if (r)
+			return r;
+	}
+
+	r = gfx_v8_0_cp_gfx_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v8_0_kiq_resume(adev);
+	if (r)
+		return r;
+
+	gfx_v8_0_enable_gui_idle_interrupt(adev, true);
+
+	return 0;
+}
+
+static void gfx_v8_0_cp_enable(struct amdgpu_device *adev, bool enable)
+{
+	gfx_v8_0_cp_gfx_enable(adev, enable);
+	gfx_v8_0_cp_compute_enable(adev, enable);
+}
+
+static int gfx_v8_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gfx_v8_0_init_golden_registers(adev);
+	gfx_v8_0_gpu_init(adev);
+
+	r = gfx_v8_0_rlc_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v8_0_cp_resume(adev);
+
+	return r;
+}
+
+static int gfx_v8_0_kcq_disable(struct amdgpu_ring *kiq_ring,struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	uint32_t scratch, tmp = 0;
+	int r, i;
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("Failed to get scratch reg (%d).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+
+	r = amdgpu_ring_alloc(kiq_ring, 10);
+	if (r) {
+		DRM_ERROR("Failed to lock KIQ (%d).\n", r);
+		amdgpu_gfx_scratch_free(adev, scratch);
+		return r;
+	}
+
+	/* unmap queues */
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_UNMAP_QUEUES, 4));
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+						PACKET3_UNMAP_QUEUES_ACTION(1) | /* RESET_QUEUES */
+						PACKET3_UNMAP_QUEUES_QUEUE_SEL(0) |
+						PACKET3_UNMAP_QUEUES_ENGINE_SEL(0) |
+						PACKET3_UNMAP_QUEUES_NUM_QUEUES(1));
+	amdgpu_ring_write(kiq_ring, PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(ring->doorbell_index));
+	amdgpu_ring_write(kiq_ring, 0);
+	amdgpu_ring_write(kiq_ring, 0);
+	amdgpu_ring_write(kiq_ring, 0);
+	/* write to scratch for completion */
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+	amdgpu_ring_write(kiq_ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(kiq_ring, 0xDEADBEEF);
+	amdgpu_ring_commit(kiq_ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i >= adev->usec_timeout) {
+		DRM_ERROR("KCQ disabled failed (scratch(0x%04X)=0x%08X)\n", scratch, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_gfx_scratch_free(adev, scratch);
+	return r;
+}
+
+static int gfx_v8_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
+
+	amdgpu_irq_put(adev, &adev->gfx.cp_ecc_error_irq, 0);
+
+	amdgpu_irq_put(adev, &adev->gfx.sq_irq, 0);
+
+	/* disable KCQ to avoid CPC touch memory not valid anymore */
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		gfx_v8_0_kcq_disable(&adev->gfx.kiq.ring, &adev->gfx.compute_ring[i]);
+
+	if (amdgpu_sriov_vf(adev)) {
+		pr_debug("For SRIOV client, shouldn't do anything.\n");
+		return 0;
+	}
+	gfx_v8_0_cp_enable(adev, false);
+	gfx_v8_0_rlc_stop(adev);
+
+	amdgpu_device_ip_set_powergating_state(adev,
+					       AMD_IP_BLOCK_TYPE_GFX,
+					       AMD_PG_STATE_UNGATE);
+
+	return 0;
+}
+
+static int gfx_v8_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	adev->gfx.in_suspend = true;
+	return gfx_v8_0_hw_fini(adev);
+}
+
+static int gfx_v8_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = gfx_v8_0_hw_init(adev);
+	adev->gfx.in_suspend = false;
+	return r;
+}
+
+static bool gfx_v8_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (REG_GET_FIELD(RREG32(mmGRBM_STATUS), GRBM_STATUS, GUI_ACTIVE))
+		return false;
+	else
+		return true;
+}
+
+static int gfx_v8_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (gfx_v8_0_is_idle(handle))
+			return 0;
+
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static bool gfx_v8_0_check_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+	u32 tmp;
+
+	/* GRBM_STATUS */
+	tmp = RREG32(mmGRBM_STATUS);
+	if (tmp & (GRBM_STATUS__PA_BUSY_MASK | GRBM_STATUS__SC_BUSY_MASK |
+		   GRBM_STATUS__BCI_BUSY_MASK | GRBM_STATUS__SX_BUSY_MASK |
+		   GRBM_STATUS__TA_BUSY_MASK | GRBM_STATUS__VGT_BUSY_MASK |
+		   GRBM_STATUS__DB_BUSY_MASK | GRBM_STATUS__CB_BUSY_MASK |
+		   GRBM_STATUS__GDS_BUSY_MASK | GRBM_STATUS__SPI_BUSY_MASK |
+		   GRBM_STATUS__IA_BUSY_MASK | GRBM_STATUS__IA_BUSY_NO_DMA_MASK |
+		   GRBM_STATUS__CP_BUSY_MASK | GRBM_STATUS__CP_COHERENCY_BUSY_MASK)) {
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_CP, 1);
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_GFX, 1);
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+						SRBM_SOFT_RESET, SOFT_RESET_GRBM, 1);
+	}
+
+	/* GRBM_STATUS2 */
+	tmp = RREG32(mmGRBM_STATUS2);
+	if (REG_GET_FIELD(tmp, GRBM_STATUS2, RLC_BUSY))
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+
+	if (REG_GET_FIELD(tmp, GRBM_STATUS2, CPF_BUSY) ||
+	    REG_GET_FIELD(tmp, GRBM_STATUS2, CPC_BUSY) ||
+	    REG_GET_FIELD(tmp, GRBM_STATUS2, CPG_BUSY)) {
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+						SOFT_RESET_CPF, 1);
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+						SOFT_RESET_CPC, 1);
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET,
+						SOFT_RESET_CPG, 1);
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET,
+						SOFT_RESET_GRBM, 1);
+	}
+
+	/* SRBM_STATUS */
+	tmp = RREG32(mmSRBM_STATUS);
+	if (REG_GET_FIELD(tmp, SRBM_STATUS, GRBM_RQ_PENDING))
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+						SRBM_SOFT_RESET, SOFT_RESET_GRBM, 1);
+	if (REG_GET_FIELD(tmp, SRBM_STATUS, SEM_BUSY))
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+						SRBM_SOFT_RESET, SOFT_RESET_SEM, 1);
+
+	if (grbm_soft_reset || srbm_soft_reset) {
+		adev->gfx.grbm_soft_reset = grbm_soft_reset;
+		adev->gfx.srbm_soft_reset = srbm_soft_reset;
+		return true;
+	} else {
+		adev->gfx.grbm_soft_reset = 0;
+		adev->gfx.srbm_soft_reset = 0;
+		return false;
+	}
+}
+
+static int gfx_v8_0_pre_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+
+	if ((!adev->gfx.grbm_soft_reset) &&
+	    (!adev->gfx.srbm_soft_reset))
+		return 0;
+
+	grbm_soft_reset = adev->gfx.grbm_soft_reset;
+	srbm_soft_reset = adev->gfx.srbm_soft_reset;
+
+	/* stop the rlc */
+	gfx_v8_0_rlc_stop(adev);
+
+	if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) ||
+	    REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_GFX))
+		/* Disable GFX parsing/prefetching */
+		gfx_v8_0_cp_gfx_enable(adev, false);
+
+	if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) ||
+	    REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPF) ||
+	    REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPC) ||
+	    REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPG)) {
+		int i;
+
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
+
+			mutex_lock(&adev->srbm_mutex);
+			vi_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+			gfx_v8_0_deactivate_hqd(adev, 2);
+			vi_srbm_select(adev, 0, 0, 0, 0);
+			mutex_unlock(&adev->srbm_mutex);
+		}
+		/* Disable MEC parsing/prefetching */
+		gfx_v8_0_cp_compute_enable(adev, false);
+	}
+
+       return 0;
+}
+
+static int gfx_v8_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+	u32 tmp;
+
+	if ((!adev->gfx.grbm_soft_reset) &&
+	    (!adev->gfx.srbm_soft_reset))
+		return 0;
+
+	grbm_soft_reset = adev->gfx.grbm_soft_reset;
+	srbm_soft_reset = adev->gfx.srbm_soft_reset;
+
+	if (grbm_soft_reset || srbm_soft_reset) {
+		tmp = RREG32(mmGMCON_DEBUG);
+		tmp = REG_SET_FIELD(tmp, GMCON_DEBUG, GFX_STALL, 1);
+		tmp = REG_SET_FIELD(tmp, GMCON_DEBUG, GFX_CLEAR, 1);
+		WREG32(mmGMCON_DEBUG, tmp);
+		udelay(50);
+	}
+
+	if (grbm_soft_reset) {
+		tmp = RREG32(mmGRBM_SOFT_RESET);
+		tmp |= grbm_soft_reset;
+		dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmGRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~grbm_soft_reset;
+		WREG32(mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmGRBM_SOFT_RESET);
+	}
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+	}
+
+	if (grbm_soft_reset || srbm_soft_reset) {
+		tmp = RREG32(mmGMCON_DEBUG);
+		tmp = REG_SET_FIELD(tmp, GMCON_DEBUG, GFX_STALL, 0);
+		tmp = REG_SET_FIELD(tmp, GMCON_DEBUG, GFX_CLEAR, 0);
+		WREG32(mmGMCON_DEBUG, tmp);
+	}
+
+	/* Wait a little for things to settle down */
+	udelay(50);
+
+	return 0;
+}
+
+static int gfx_v8_0_post_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+
+	if ((!adev->gfx.grbm_soft_reset) &&
+	    (!adev->gfx.srbm_soft_reset))
+		return 0;
+
+	grbm_soft_reset = adev->gfx.grbm_soft_reset;
+	srbm_soft_reset = adev->gfx.srbm_soft_reset;
+
+	if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) ||
+	    REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_GFX))
+		gfx_v8_0_cp_gfx_resume(adev);
+
+	if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) ||
+	    REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPF) ||
+	    REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPC) ||
+	    REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPG)) {
+		int i;
+
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
+
+			mutex_lock(&adev->srbm_mutex);
+			vi_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+			gfx_v8_0_deactivate_hqd(adev, 2);
+			vi_srbm_select(adev, 0, 0, 0, 0);
+			mutex_unlock(&adev->srbm_mutex);
+		}
+		gfx_v8_0_kiq_resume(adev);
+	}
+	gfx_v8_0_rlc_start(adev);
+
+	return 0;
+}
+
+/**
+ * gfx_v8_0_get_gpu_clock_counter - return GPU clock counter snapshot
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Fetches a GPU clock counter snapshot.
+ * Returns the 64 bit clock counter snapshot.
+ */
+static uint64_t gfx_v8_0_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
+	uint64_t clock;
+
+	mutex_lock(&adev->gfx.gpu_clock_mutex);
+	WREG32(mmRLC_CAPTURE_GPU_CLOCK_COUNT, 1);
+	clock = (uint64_t)RREG32(mmRLC_GPU_CLOCK_COUNT_LSB) |
+		((uint64_t)RREG32(mmRLC_GPU_CLOCK_COUNT_MSB) << 32ULL);
+	mutex_unlock(&adev->gfx.gpu_clock_mutex);
+	return clock;
+}
+
+static void gfx_v8_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
+					  uint32_t vmid,
+					  uint32_t gds_base, uint32_t gds_size,
+					  uint32_t gws_base, uint32_t gws_size,
+					  uint32_t oa_base, uint32_t oa_size)
+{
+	gds_base = gds_base >> AMDGPU_GDS_SHIFT;
+	gds_size = gds_size >> AMDGPU_GDS_SHIFT;
+
+	gws_base = gws_base >> AMDGPU_GWS_SHIFT;
+	gws_size = gws_size >> AMDGPU_GWS_SHIFT;
+
+	oa_base = oa_base >> AMDGPU_OA_SHIFT;
+	oa_size = oa_size >> AMDGPU_OA_SHIFT;
+
+	/* GDS Base */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].mem_base);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, gds_base);
+
+	/* GDS Size */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].mem_size);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, gds_size);
+
+	/* GWS */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].gws);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, gws_size << GDS_GWS_VMID0__SIZE__SHIFT | gws_base);
+
+	/* OA */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].oa);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, (1 << (oa_size + oa_base)) - (1 << oa_base));
+}
+
+static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t address)
+{
+	WREG32(mmSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(address << SQ_IND_INDEX__INDEX__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK));
+	return RREG32(mmSQ_IND_DATA);
+}
+
+static void wave_read_regs(struct amdgpu_device *adev, uint32_t simd,
+			   uint32_t wave, uint32_t thread,
+			   uint32_t regno, uint32_t num, uint32_t *out)
+{
+	WREG32(mmSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(regno << SQ_IND_INDEX__INDEX__SHIFT) |
+		(thread << SQ_IND_INDEX__THREAD_ID__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK) |
+		(SQ_IND_INDEX__AUTO_INCR_MASK));
+	while (num--)
+		*(out++) = RREG32(mmSQ_IND_DATA);
+}
+
+static void gfx_v8_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
+{
+	/* type 0 wave data */
+	dst[(*no_fields)++] = 0;
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_STATUS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_HW_ID);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW1);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_GPR_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_LDS_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TRAPSTS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_STS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TBA_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TBA_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TMA_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TMA_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_DBG0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_M0);
+}
+
+static void gfx_v8_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd,
+				     uint32_t wave, uint32_t start,
+				     uint32_t size, uint32_t *dst)
+{
+	wave_read_regs(
+		adev, simd, wave, 0,
+		start + SQIND_WAVE_SGPRS_OFFSET, size, dst);
+}
+
+
+static const struct amdgpu_gfx_funcs gfx_v8_0_gfx_funcs = {
+	.get_gpu_clock_counter = &gfx_v8_0_get_gpu_clock_counter,
+	.select_se_sh = &gfx_v8_0_select_se_sh,
+	.read_wave_data = &gfx_v8_0_read_wave_data,
+	.read_wave_sgprs = &gfx_v8_0_read_wave_sgprs,
+	.select_me_pipe_q = &gfx_v8_0_select_me_pipe_q
+};
+
+static int gfx_v8_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->gfx.num_gfx_rings = GFX8_NUM_GFX_RINGS;
+	adev->gfx.num_compute_rings = AMDGPU_MAX_COMPUTE_RINGS;
+	adev->gfx.funcs = &gfx_v8_0_gfx_funcs;
+	gfx_v8_0_set_ring_funcs(adev);
+	gfx_v8_0_set_irq_funcs(adev);
+	gfx_v8_0_set_gds_init(adev);
+	gfx_v8_0_set_rlc_funcs(adev);
+
+	return 0;
+}
+
+static int gfx_v8_0_late_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_reg_irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_inst_irq, 0);
+	if (r)
+		return r;
+
+	/* requires IBs so do in late init after IB pool is initialized */
+	r = gfx_v8_0_do_edc_gpr_workarounds(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.cp_ecc_error_irq, 0);
+	if (r) {
+		DRM_ERROR("amdgpu_irq_get() failed to get IRQ for EDC, r: %d.\n", r);
+		return r;
+	}
+
+	r = amdgpu_irq_get(adev, &adev->gfx.sq_irq, 0);
+	if (r) {
+		DRM_ERROR(
+			"amdgpu_irq_get() failed to get IRQ for SQ, r: %d.\n",
+			r);
+		return r;
+	}
+
+	return 0;
+}
+
+static void gfx_v8_0_enable_gfx_static_mg_power_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	if (((adev->asic_type == CHIP_POLARIS11) ||
+	    (adev->asic_type == CHIP_POLARIS12) ||
+	    (adev->asic_type == CHIP_VEGAM)) &&
+	    adev->powerplay.pp_funcs->set_powergating_by_smu)
+		/* Send msg to SMU via Powerplay */
+		amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_GFX, enable);
+
+	WREG32_FIELD(RLC_PG_CNTL, STATIC_PER_CU_PG_ENABLE, enable ? 1 : 0);
+}
+
+static void gfx_v8_0_enable_gfx_dynamic_mg_power_gating(struct amdgpu_device *adev,
+							bool enable)
+{
+	WREG32_FIELD(RLC_PG_CNTL, DYN_PER_CU_PG_ENABLE, enable ? 1 : 0);
+}
+
+static void polaris11_enable_gfx_quick_mg_power_gating(struct amdgpu_device *adev,
+		bool enable)
+{
+	WREG32_FIELD(RLC_PG_CNTL, QUICK_PG_ENABLE, enable ? 1 : 0);
+}
+
+static void cz_enable_gfx_cg_power_gating(struct amdgpu_device *adev,
+					  bool enable)
+{
+	WREG32_FIELD(RLC_PG_CNTL, GFX_POWER_GATING_ENABLE, enable ? 1 : 0);
+}
+
+static void cz_enable_gfx_pipeline_power_gating(struct amdgpu_device *adev,
+						bool enable)
+{
+	WREG32_FIELD(RLC_PG_CNTL, GFX_PIPELINE_PG_ENABLE, enable ? 1 : 0);
+
+	/* Read any GFX register to wake up GFX. */
+	if (!enable)
+		RREG32(mmDB_RENDER_CONTROL);
+}
+
+static void cz_update_gfx_cg_power_gating(struct amdgpu_device *adev,
+					  bool enable)
+{
+	if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_PG) && enable) {
+		cz_enable_gfx_cg_power_gating(adev, true);
+		if (adev->pg_flags & AMD_PG_SUPPORT_GFX_PIPELINE)
+			cz_enable_gfx_pipeline_power_gating(adev, true);
+	} else {
+		cz_enable_gfx_cg_power_gating(adev, false);
+		cz_enable_gfx_pipeline_power_gating(adev, false);
+	}
+}
+
+static int gfx_v8_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	bool enable = (state == AMD_PG_STATE_GATE);
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_SMG |
+				AMD_PG_SUPPORT_RLC_SMU_HS |
+				AMD_PG_SUPPORT_CP |
+				AMD_PG_SUPPORT_GFX_DMG))
+		adev->gfx.rlc.funcs->enter_safe_mode(adev);
+	switch (adev->asic_type) {
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+
+		if (adev->pg_flags & AMD_PG_SUPPORT_RLC_SMU_HS) {
+			cz_enable_sck_slow_down_on_power_up(adev, true);
+			cz_enable_sck_slow_down_on_power_down(adev, true);
+		} else {
+			cz_enable_sck_slow_down_on_power_up(adev, false);
+			cz_enable_sck_slow_down_on_power_down(adev, false);
+		}
+		if (adev->pg_flags & AMD_PG_SUPPORT_CP)
+			cz_enable_cp_power_gating(adev, true);
+		else
+			cz_enable_cp_power_gating(adev, false);
+
+		cz_update_gfx_cg_power_gating(adev, enable);
+
+		if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_SMG) && enable)
+			gfx_v8_0_enable_gfx_static_mg_power_gating(adev, true);
+		else
+			gfx_v8_0_enable_gfx_static_mg_power_gating(adev, false);
+
+		if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_DMG) && enable)
+			gfx_v8_0_enable_gfx_dynamic_mg_power_gating(adev, true);
+		else
+			gfx_v8_0_enable_gfx_dynamic_mg_power_gating(adev, false);
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_SMG) && enable)
+			gfx_v8_0_enable_gfx_static_mg_power_gating(adev, true);
+		else
+			gfx_v8_0_enable_gfx_static_mg_power_gating(adev, false);
+
+		if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_DMG) && enable)
+			gfx_v8_0_enable_gfx_dynamic_mg_power_gating(adev, true);
+		else
+			gfx_v8_0_enable_gfx_dynamic_mg_power_gating(adev, false);
+
+		if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_QUICK_MG) && enable)
+			polaris11_enable_gfx_quick_mg_power_gating(adev, true);
+		else
+			polaris11_enable_gfx_quick_mg_power_gating(adev, false);
+		break;
+	default:
+		break;
+	}
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_SMG |
+				AMD_PG_SUPPORT_RLC_SMU_HS |
+				AMD_PG_SUPPORT_CP |
+				AMD_PG_SUPPORT_GFX_DMG))
+		adev->gfx.rlc.funcs->exit_safe_mode(adev);
+	return 0;
+}
+
+static void gfx_v8_0_get_clockgating_state(void *handle, u32 *flags)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_GFX_MGCG */
+	data = RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__CPF_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_MGCG;
+
+	/* AMD_CG_SUPPORT_GFX_CGLG */
+	data = RREG32(mmRLC_CGCG_CGLS_CTRL);
+	if (data & RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CGCG;
+
+	/* AMD_CG_SUPPORT_GFX_CGLS */
+	if (data & RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CGLS;
+
+	/* AMD_CG_SUPPORT_GFX_CGTS */
+	data = RREG32(mmCGTS_SM_CTRL_REG);
+	if (!(data & CGTS_SM_CTRL_REG__OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_CGTS;
+
+	/* AMD_CG_SUPPORT_GFX_CGTS_LS */
+	if (!(data & CGTS_SM_CTRL_REG__LS_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_CGTS_LS;
+
+	/* AMD_CG_SUPPORT_GFX_RLC_LS */
+	data = RREG32(mmRLC_MEM_SLP_CNTL);
+	if (data & RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_RLC_LS | AMD_CG_SUPPORT_GFX_MGLS;
+
+	/* AMD_CG_SUPPORT_GFX_CP_LS */
+	data = RREG32(mmCP_MEM_SLP_CNTL);
+	if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CP_LS | AMD_CG_SUPPORT_GFX_MGLS;
+}
+
+static void gfx_v8_0_send_serdes_cmd(struct amdgpu_device *adev,
+				     uint32_t reg_addr, uint32_t cmd)
+{
+	uint32_t data;
+
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+
+	WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
+	WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
+
+	data = RREG32(mmRLC_SERDES_WR_CTRL);
+	if (adev->asic_type == CHIP_STONEY)
+		data &= ~(RLC_SERDES_WR_CTRL__WRITE_COMMAND_MASK |
+			  RLC_SERDES_WR_CTRL__READ_COMMAND_MASK |
+			  RLC_SERDES_WR_CTRL__P1_SELECT_MASK |
+			  RLC_SERDES_WR_CTRL__P2_SELECT_MASK |
+			  RLC_SERDES_WR_CTRL__RDDATA_RESET_MASK |
+			  RLC_SERDES_WR_CTRL__POWER_DOWN_MASK |
+			  RLC_SERDES_WR_CTRL__POWER_UP_MASK |
+			  RLC_SERDES_WR_CTRL__SHORT_FORMAT_MASK |
+			  RLC_SERDES_WR_CTRL__SRBM_OVERRIDE_MASK);
+	else
+		data &= ~(RLC_SERDES_WR_CTRL__WRITE_COMMAND_MASK |
+			  RLC_SERDES_WR_CTRL__READ_COMMAND_MASK |
+			  RLC_SERDES_WR_CTRL__P1_SELECT_MASK |
+			  RLC_SERDES_WR_CTRL__P2_SELECT_MASK |
+			  RLC_SERDES_WR_CTRL__RDDATA_RESET_MASK |
+			  RLC_SERDES_WR_CTRL__POWER_DOWN_MASK |
+			  RLC_SERDES_WR_CTRL__POWER_UP_MASK |
+			  RLC_SERDES_WR_CTRL__SHORT_FORMAT_MASK |
+			  RLC_SERDES_WR_CTRL__BPM_DATA_MASK |
+			  RLC_SERDES_WR_CTRL__REG_ADDR_MASK |
+			  RLC_SERDES_WR_CTRL__SRBM_OVERRIDE_MASK);
+	data |= (RLC_SERDES_WR_CTRL__RSVD_BPM_ADDR_MASK |
+		 (cmd << RLC_SERDES_WR_CTRL__BPM_DATA__SHIFT) |
+		 (reg_addr << RLC_SERDES_WR_CTRL__REG_ADDR__SHIFT) |
+		 (0xff << RLC_SERDES_WR_CTRL__BPM_ADDR__SHIFT));
+
+	WREG32(mmRLC_SERDES_WR_CTRL, data);
+}
+
+#define MSG_ENTER_RLC_SAFE_MODE     1
+#define MSG_EXIT_RLC_SAFE_MODE      0
+#define RLC_GPR_REG2__REQ_MASK 0x00000001
+#define RLC_GPR_REG2__REQ__SHIFT 0
+#define RLC_GPR_REG2__MESSAGE__SHIFT 0x00000001
+#define RLC_GPR_REG2__MESSAGE_MASK 0x0000001e
+
+static void iceland_enter_rlc_safe_mode(struct amdgpu_device *adev)
+{
+	u32 data;
+	unsigned i;
+
+	data = RREG32(mmRLC_CNTL);
+	if (!(data & RLC_CNTL__RLC_ENABLE_F32_MASK))
+		return;
+
+	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG)) {
+		data |= RLC_SAFE_MODE__CMD_MASK;
+		data &= ~RLC_SAFE_MODE__MESSAGE_MASK;
+		data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT);
+		WREG32(mmRLC_SAFE_MODE, data);
+
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if ((RREG32(mmRLC_GPM_STAT) &
+			     (RLC_GPM_STAT__GFX_CLOCK_STATUS_MASK |
+			      RLC_GPM_STAT__GFX_POWER_STATUS_MASK)) ==
+			    (RLC_GPM_STAT__GFX_CLOCK_STATUS_MASK |
+			     RLC_GPM_STAT__GFX_POWER_STATUS_MASK))
+				break;
+			udelay(1);
+		}
+
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!REG_GET_FIELD(RREG32(mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD))
+				break;
+			udelay(1);
+		}
+		adev->gfx.rlc.in_safe_mode = true;
+	}
+}
+
+static void iceland_exit_rlc_safe_mode(struct amdgpu_device *adev)
+{
+	u32 data = 0;
+	unsigned i;
+
+	data = RREG32(mmRLC_CNTL);
+	if (!(data & RLC_CNTL__RLC_ENABLE_F32_MASK))
+		return;
+
+	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG)) {
+		if (adev->gfx.rlc.in_safe_mode) {
+			data |= RLC_SAFE_MODE__CMD_MASK;
+			data &= ~RLC_SAFE_MODE__MESSAGE_MASK;
+			WREG32(mmRLC_SAFE_MODE, data);
+			adev->gfx.rlc.in_safe_mode = false;
+		}
+	}
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!REG_GET_FIELD(RREG32(mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD))
+			break;
+		udelay(1);
+	}
+}
+
+static const struct amdgpu_rlc_funcs iceland_rlc_funcs = {
+	.enter_safe_mode = iceland_enter_rlc_safe_mode,
+	.exit_safe_mode = iceland_exit_rlc_safe_mode
+};
+
+static void gfx_v8_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t temp, data;
+
+	adev->gfx.rlc.funcs->enter_safe_mode(adev);
+
+	/* It is disabled by HW by default */
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG)) {
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGLS) {
+			if (adev->cg_flags & AMD_CG_SUPPORT_GFX_RLC_LS)
+				/* 1 - RLC memory Light sleep */
+				WREG32_FIELD(RLC_MEM_SLP_CNTL, RLC_MEM_LS_EN, 1);
+
+			if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CP_LS)
+				WREG32_FIELD(CP_MEM_SLP_CNTL, CP_MEM_LS_EN, 1);
+		}
+
+		/* 3 - RLC_CGTT_MGCG_OVERRIDE */
+		temp = data = RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
+		if (adev->flags & AMD_IS_APU)
+			data &= ~(RLC_CGTT_MGCG_OVERRIDE__CPF_MASK |
+				  RLC_CGTT_MGCG_OVERRIDE__RLC_MASK |
+				  RLC_CGTT_MGCG_OVERRIDE__MGCG_MASK);
+		else
+			data &= ~(RLC_CGTT_MGCG_OVERRIDE__CPF_MASK |
+				  RLC_CGTT_MGCG_OVERRIDE__RLC_MASK |
+				  RLC_CGTT_MGCG_OVERRIDE__MGCG_MASK |
+				  RLC_CGTT_MGCG_OVERRIDE__GRBM_MASK);
+
+		if (temp != data)
+			WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* 4 - wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */
+		gfx_v8_0_wait_for_rlc_serdes(adev);
+
+		/* 5 - clear mgcg override */
+		gfx_v8_0_send_serdes_cmd(adev, BPM_REG_MGCG_OVERRIDE, CLE_BPM_SERDES_CMD);
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGTS) {
+			/* 6 - Enable CGTS(Tree Shade) MGCG /MGLS */
+			temp = data = RREG32(mmCGTS_SM_CTRL_REG);
+			data &= ~(CGTS_SM_CTRL_REG__SM_MODE_MASK);
+			data |= (0x2 << CGTS_SM_CTRL_REG__SM_MODE__SHIFT);
+			data |= CGTS_SM_CTRL_REG__SM_MODE_ENABLE_MASK;
+			data &= ~CGTS_SM_CTRL_REG__OVERRIDE_MASK;
+			if ((adev->cg_flags & AMD_CG_SUPPORT_GFX_MGLS) &&
+			    (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGTS_LS))
+				data &= ~CGTS_SM_CTRL_REG__LS_OVERRIDE_MASK;
+			data |= CGTS_SM_CTRL_REG__ON_MONITOR_ADD_EN_MASK;
+			data |= (0x96 << CGTS_SM_CTRL_REG__ON_MONITOR_ADD__SHIFT);
+			if (temp != data)
+				WREG32(mmCGTS_SM_CTRL_REG, data);
+		}
+		udelay(50);
+
+		/* 7 - wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */
+		gfx_v8_0_wait_for_rlc_serdes(adev);
+	} else {
+		/* 1 - MGCG_OVERRIDE[0] for CP and MGCG_OVERRIDE[1] for RLC */
+		temp = data = RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
+		data |= (RLC_CGTT_MGCG_OVERRIDE__CPF_MASK |
+				RLC_CGTT_MGCG_OVERRIDE__RLC_MASK |
+				RLC_CGTT_MGCG_OVERRIDE__MGCG_MASK |
+				RLC_CGTT_MGCG_OVERRIDE__GRBM_MASK);
+		if (temp != data)
+			WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* 2 - disable MGLS in RLC */
+		data = RREG32(mmRLC_MEM_SLP_CNTL);
+		if (data & RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK) {
+			data &= ~RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK;
+			WREG32(mmRLC_MEM_SLP_CNTL, data);
+		}
+
+		/* 3 - disable MGLS in CP */
+		data = RREG32(mmCP_MEM_SLP_CNTL);
+		if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK) {
+			data &= ~CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+			WREG32(mmCP_MEM_SLP_CNTL, data);
+		}
+
+		/* 4 - Disable CGTS(Tree Shade) MGCG and MGLS */
+		temp = data = RREG32(mmCGTS_SM_CTRL_REG);
+		data |= (CGTS_SM_CTRL_REG__OVERRIDE_MASK |
+				CGTS_SM_CTRL_REG__LS_OVERRIDE_MASK);
+		if (temp != data)
+			WREG32(mmCGTS_SM_CTRL_REG, data);
+
+		/* 5 - wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */
+		gfx_v8_0_wait_for_rlc_serdes(adev);
+
+		/* 6 - set mgcg override */
+		gfx_v8_0_send_serdes_cmd(adev, BPM_REG_MGCG_OVERRIDE, SET_BPM_SERDES_CMD);
+
+		udelay(50);
+
+		/* 7- wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */
+		gfx_v8_0_wait_for_rlc_serdes(adev);
+	}
+
+	adev->gfx.rlc.funcs->exit_safe_mode(adev);
+}
+
+static void gfx_v8_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t temp, temp1, data, data1;
+
+	temp = data = RREG32(mmRLC_CGCG_CGLS_CTRL);
+
+	adev->gfx.rlc.funcs->enter_safe_mode(adev);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)) {
+		temp1 = data1 =	RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
+		data1 &= ~RLC_CGTT_MGCG_OVERRIDE__CGCG_MASK;
+		if (temp1 != data1)
+			WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data1);
+
+		/* : wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */
+		gfx_v8_0_wait_for_rlc_serdes(adev);
+
+		/* 2 - clear cgcg override */
+		gfx_v8_0_send_serdes_cmd(adev, BPM_REG_CGCG_OVERRIDE, CLE_BPM_SERDES_CMD);
+
+		/* wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */
+		gfx_v8_0_wait_for_rlc_serdes(adev);
+
+		/* 3 - write cmd to set CGLS */
+		gfx_v8_0_send_serdes_cmd(adev, BPM_REG_CGLS_EN, SET_BPM_SERDES_CMD);
+
+		/* 4 - enable cgcg */
+		data |= RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK;
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS) {
+			/* enable cgls*/
+			data |= RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+
+			temp1 = data1 =	RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
+			data1 &= ~RLC_CGTT_MGCG_OVERRIDE__CGLS_MASK;
+
+			if (temp1 != data1)
+				WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data1);
+		} else {
+			data &= ~RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+		}
+
+		if (temp != data)
+			WREG32(mmRLC_CGCG_CGLS_CTRL, data);
+
+		/* 5 enable cntx_empty_int_enable/cntx_busy_int_enable/
+		 * Cmp_busy/GFX_Idle interrupts
+		 */
+		gfx_v8_0_enable_gui_idle_interrupt(adev, true);
+	} else {
+		/* disable cntx_empty_int_enable & GFX Idle interrupt */
+		gfx_v8_0_enable_gui_idle_interrupt(adev, false);
+
+		/* TEST CGCG */
+		temp1 = data1 =	RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
+		data1 |= (RLC_CGTT_MGCG_OVERRIDE__CGCG_MASK |
+				RLC_CGTT_MGCG_OVERRIDE__CGLS_MASK);
+		if (temp1 != data1)
+			WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data1);
+
+		/* read gfx register to wake up cgcg */
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+
+		/* wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */
+		gfx_v8_0_wait_for_rlc_serdes(adev);
+
+		/* write cmd to Set CGCG Overrride */
+		gfx_v8_0_send_serdes_cmd(adev, BPM_REG_CGCG_OVERRIDE, SET_BPM_SERDES_CMD);
+
+		/* wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */
+		gfx_v8_0_wait_for_rlc_serdes(adev);
+
+		/* write cmd to Clear CGLS */
+		gfx_v8_0_send_serdes_cmd(adev, BPM_REG_CGLS_EN, CLE_BPM_SERDES_CMD);
+
+		/* disable cgcg, cgls should be disabled too. */
+		data &= ~(RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK |
+			  RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK);
+		if (temp != data)
+			WREG32(mmRLC_CGCG_CGLS_CTRL, data);
+		/* enable interrupts again for PG */
+		gfx_v8_0_enable_gui_idle_interrupt(adev, true);
+	}
+
+	gfx_v8_0_wait_for_rlc_serdes(adev);
+
+	adev->gfx.rlc.funcs->exit_safe_mode(adev);
+}
+static int gfx_v8_0_update_gfx_clock_gating(struct amdgpu_device *adev,
+					    bool enable)
+{
+	if (enable) {
+		/* CGCG/CGLS should be enabled after MGCG/MGLS/TS(CG/LS)
+		 * ===  MGCG + MGLS + TS(CG/LS) ===
+		 */
+		gfx_v8_0_update_medium_grain_clock_gating(adev, enable);
+		gfx_v8_0_update_coarse_grain_clock_gating(adev, enable);
+	} else {
+		/* CGCG/CGLS should be disabled before MGCG/MGLS/TS(CG/LS)
+		 * ===  CGCG + CGLS ===
+		 */
+		gfx_v8_0_update_coarse_grain_clock_gating(adev, enable);
+		gfx_v8_0_update_medium_grain_clock_gating(adev, enable);
+	}
+	return 0;
+}
+
+static int gfx_v8_0_tonga_update_gfx_clock_gating(struct amdgpu_device *adev,
+					  enum amd_clockgating_state state)
+{
+	uint32_t msg_id, pp_state = 0;
+	uint32_t pp_support_state = 0;
+
+	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_CGLS)) {
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS) {
+			pp_support_state = PP_STATE_SUPPORT_LS;
+			pp_state = PP_STATE_LS;
+		}
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG) {
+			pp_support_state |= PP_STATE_SUPPORT_CG;
+			pp_state |= PP_STATE_CG;
+		}
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+
+		msg_id = PP_CG_MSG_ID(PP_GROUP_GFX,
+				PP_BLOCK_GFX_CG,
+				pp_support_state,
+				pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+
+	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_MGCG | AMD_CG_SUPPORT_GFX_MGLS)) {
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGLS) {
+			pp_support_state = PP_STATE_SUPPORT_LS;
+			pp_state = PP_STATE_LS;
+		}
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG) {
+			pp_support_state |= PP_STATE_SUPPORT_CG;
+			pp_state |= PP_STATE_CG;
+		}
+
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+
+		msg_id = PP_CG_MSG_ID(PP_GROUP_GFX,
+				PP_BLOCK_GFX_MG,
+				pp_support_state,
+				pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+
+	return 0;
+}
+
+static int gfx_v8_0_polaris_update_gfx_clock_gating(struct amdgpu_device *adev,
+					  enum amd_clockgating_state state)
+{
+
+	uint32_t msg_id, pp_state = 0;
+	uint32_t pp_support_state = 0;
+
+	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_CGLS)) {
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS) {
+			pp_support_state = PP_STATE_SUPPORT_LS;
+			pp_state = PP_STATE_LS;
+		}
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG) {
+			pp_support_state |= PP_STATE_SUPPORT_CG;
+			pp_state |= PP_STATE_CG;
+		}
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+
+		msg_id = PP_CG_MSG_ID(PP_GROUP_GFX,
+				PP_BLOCK_GFX_CG,
+				pp_support_state,
+				pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+
+	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_3D_CGCG | AMD_CG_SUPPORT_GFX_3D_CGLS)) {
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGLS) {
+			pp_support_state = PP_STATE_SUPPORT_LS;
+			pp_state = PP_STATE_LS;
+		}
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG) {
+			pp_support_state |= PP_STATE_SUPPORT_CG;
+			pp_state |= PP_STATE_CG;
+		}
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+
+		msg_id = PP_CG_MSG_ID(PP_GROUP_GFX,
+				PP_BLOCK_GFX_3D,
+				pp_support_state,
+				pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+
+	if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_MGCG | AMD_CG_SUPPORT_GFX_MGLS)) {
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGLS) {
+			pp_support_state = PP_STATE_SUPPORT_LS;
+			pp_state = PP_STATE_LS;
+		}
+
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG) {
+			pp_support_state |= PP_STATE_SUPPORT_CG;
+			pp_state |= PP_STATE_CG;
+		}
+
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+
+		msg_id = PP_CG_MSG_ID(PP_GROUP_GFX,
+				PP_BLOCK_GFX_MG,
+				pp_support_state,
+				pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_GFX_RLC_LS) {
+		pp_support_state = PP_STATE_SUPPORT_LS;
+
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+		else
+			pp_state = PP_STATE_LS;
+
+		msg_id = PP_CG_MSG_ID(PP_GROUP_GFX,
+				PP_BLOCK_GFX_RLC,
+				pp_support_state,
+				pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CP_LS) {
+		pp_support_state = PP_STATE_SUPPORT_LS;
+
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+		else
+			pp_state = PP_STATE_LS;
+		msg_id = PP_CG_MSG_ID(PP_GROUP_GFX,
+			PP_BLOCK_GFX_CP,
+			pp_support_state,
+			pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+
+	return 0;
+}
+
+static int gfx_v8_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		gfx_v8_0_update_gfx_clock_gating(adev,
+						 state == AMD_CG_STATE_GATE);
+		break;
+	case CHIP_TONGA:
+		gfx_v8_0_tonga_update_gfx_clock_gating(adev, state);
+		break;
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		gfx_v8_0_polaris_update_gfx_clock_gating(adev, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static u64 gfx_v8_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	return ring->adev->wb.wb[ring->rptr_offs];
+}
+
+static u64 gfx_v8_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		/* XXX check if swapping is necessary on BE */
+		return ring->adev->wb.wb[ring->wptr_offs];
+	else
+		return RREG32(mmCP_RB0_WPTR);
+}
+
+static void gfx_v8_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+	} else {
+		WREG32(mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
+		(void)RREG32(mmCP_RB0_WPTR);
+	}
+}
+
+static void gfx_v8_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	u32 ref_and_mask, reg_mem_engine;
+
+	if ((ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) ||
+	    (ring->funcs->type == AMDGPU_RING_TYPE_KIQ)) {
+		switch (ring->me) {
+		case 1:
+			ref_and_mask = GPU_HDP_FLUSH_DONE__CP2_MASK << ring->pipe;
+			break;
+		case 2:
+			ref_and_mask = GPU_HDP_FLUSH_DONE__CP6_MASK << ring->pipe;
+			break;
+		default:
+			return;
+		}
+		reg_mem_engine = 0;
+	} else {
+		ref_and_mask = GPU_HDP_FLUSH_DONE__CP0_MASK;
+		reg_mem_engine = WAIT_REG_MEM_ENGINE(1); /* pfp */
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_OPERATION(1) | /* write, wait, write */
+				 WAIT_REG_MEM_FUNCTION(3) |  /* == */
+				 reg_mem_engine));
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_REQ);
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_DONE);
+	amdgpu_ring_write(ring, ref_and_mask);
+	amdgpu_ring_write(ring, ref_and_mask);
+	amdgpu_ring_write(ring, 0x20); /* poll interval */
+}
+
+static void gfx_v8_0_ring_emit_vgt_flush(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE, 0));
+	amdgpu_ring_write(ring, EVENT_TYPE(VS_PARTIAL_FLUSH) |
+		EVENT_INDEX(4));
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE, 0));
+	amdgpu_ring_write(ring, EVENT_TYPE(VGT_FLUSH) |
+		EVENT_INDEX(0));
+}
+
+static void gfx_v8_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
+				      struct amdgpu_ib *ib,
+				      unsigned vmid, bool ctx_switch)
+{
+	u32 header, control = 0;
+
+	if (ib->flags & AMDGPU_IB_FLAG_CE)
+		header = PACKET3(PACKET3_INDIRECT_BUFFER_CONST, 2);
+	else
+		header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
+
+	control |= ib->length_dw | (vmid << 24);
+
+	if (amdgpu_sriov_vf(ring->adev) && (ib->flags & AMDGPU_IB_FLAG_PREEMPT)) {
+		control |= INDIRECT_BUFFER_PRE_ENB(1);
+
+		if (!(ib->flags & AMDGPU_IB_FLAG_CE))
+			gfx_v8_0_ring_emit_de_meta(ring);
+	}
+
+	amdgpu_ring_write(ring, header);
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+			  (2 << 0) |
+#endif
+			  (ib->gpu_addr & 0xFFFFFFFC));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFFFF);
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v8_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
+					  struct amdgpu_ib *ib,
+					  unsigned vmid, bool ctx_switch)
+{
+	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+				(2 << 0) |
+#endif
+				(ib->gpu_addr & 0xFFFFFFFC));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFFFF);
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v8_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr,
+					 u64 seq, unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+
+	/* EVENT_WRITE_EOP - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE_EOP, 4));
+	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
+				 EOP_TC_ACTION_EN |
+				 EOP_TC_WB_ACTION_EN |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) |
+			  DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+
+}
+
+static void gfx_v8_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_MEM_SPACE(1) | /* memory */
+				 WAIT_REG_MEM_FUNCTION(3) | /* equal */
+				 WAIT_REG_MEM_ENGINE(usepfp))); /* pfp or me */
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, 0xffffffff);
+	amdgpu_ring_write(ring, 4); /* poll interval */
+}
+
+static void gfx_v8_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned vmid, uint64_t pd_addr)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for the invalidate to complete */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_OPERATION(0) | /* wait */
+				 WAIT_REG_MEM_FUNCTION(0) |  /* always */
+				 WAIT_REG_MEM_ENGINE(0))); /* me */
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0); /* ref */
+	amdgpu_ring_write(ring, 0); /* mask */
+	amdgpu_ring_write(ring, 0x20); /* poll interval */
+
+	/* compute doesn't have PFP */
+	if (usepfp) {
+		/* sync PFP to ME, otherwise we might get invalid PFP reads */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
+		amdgpu_ring_write(ring, 0x0);
+	}
+}
+
+static u64 gfx_v8_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
+{
+	return ring->adev->wb.wb[ring->wptr_offs];
+}
+
+static void gfx_v8_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* XXX check if swapping is necessary on BE */
+	adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
+	WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+}
+
+static void gfx_v8_0_ring_set_pipe_percent(struct amdgpu_ring *ring,
+					   bool acquire)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int pipe_num, tmp, reg;
+	int pipe_percent = acquire ? SPI_WCL_PIPE_PERCENT_GFX__VALUE_MASK : 0x1;
+
+	pipe_num = ring->me * adev->gfx.mec.num_pipe_per_mec + ring->pipe;
+
+	/* first me only has 2 entries, GFX and HP3D */
+	if (ring->me > 0)
+		pipe_num -= 2;
+
+	reg = mmSPI_WCL_PIPE_PERCENT_GFX + pipe_num;
+	tmp = RREG32(reg);
+	tmp = REG_SET_FIELD(tmp, SPI_WCL_PIPE_PERCENT_GFX, VALUE, pipe_percent);
+	WREG32(reg, tmp);
+}
+
+static void gfx_v8_0_pipe_reserve_resources(struct amdgpu_device *adev,
+					    struct amdgpu_ring *ring,
+					    bool acquire)
+{
+	int i, pipe;
+	bool reserve;
+	struct amdgpu_ring *iring;
+
+	mutex_lock(&adev->gfx.pipe_reserve_mutex);
+	pipe = amdgpu_gfx_queue_to_bit(adev, ring->me, ring->pipe, 0);
+	if (acquire)
+		set_bit(pipe, adev->gfx.pipe_reserve_bitmap);
+	else
+		clear_bit(pipe, adev->gfx.pipe_reserve_bitmap);
+
+	if (!bitmap_weight(adev->gfx.pipe_reserve_bitmap, AMDGPU_MAX_COMPUTE_QUEUES)) {
+		/* Clear all reservations - everyone reacquires all resources */
+		for (i = 0; i < adev->gfx.num_gfx_rings; ++i)
+			gfx_v8_0_ring_set_pipe_percent(&adev->gfx.gfx_ring[i],
+						       true);
+
+		for (i = 0; i < adev->gfx.num_compute_rings; ++i)
+			gfx_v8_0_ring_set_pipe_percent(&adev->gfx.compute_ring[i],
+						       true);
+	} else {
+		/* Lower all pipes without a current reservation */
+		for (i = 0; i < adev->gfx.num_gfx_rings; ++i) {
+			iring = &adev->gfx.gfx_ring[i];
+			pipe = amdgpu_gfx_queue_to_bit(adev,
+						       iring->me,
+						       iring->pipe,
+						       0);
+			reserve = test_bit(pipe, adev->gfx.pipe_reserve_bitmap);
+			gfx_v8_0_ring_set_pipe_percent(iring, reserve);
+		}
+
+		for (i = 0; i < adev->gfx.num_compute_rings; ++i) {
+			iring = &adev->gfx.compute_ring[i];
+			pipe = amdgpu_gfx_queue_to_bit(adev,
+						       iring->me,
+						       iring->pipe,
+						       0);
+			reserve = test_bit(pipe, adev->gfx.pipe_reserve_bitmap);
+			gfx_v8_0_ring_set_pipe_percent(iring, reserve);
+		}
+	}
+
+	mutex_unlock(&adev->gfx.pipe_reserve_mutex);
+}
+
+static void gfx_v8_0_hqd_set_priority(struct amdgpu_device *adev,
+				      struct amdgpu_ring *ring,
+				      bool acquire)
+{
+	uint32_t pipe_priority = acquire ? 0x2 : 0x0;
+	uint32_t queue_priority = acquire ? 0xf : 0x0;
+
+	mutex_lock(&adev->srbm_mutex);
+	vi_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+
+	WREG32(mmCP_HQD_PIPE_PRIORITY, pipe_priority);
+	WREG32(mmCP_HQD_QUEUE_PRIORITY, queue_priority);
+
+	vi_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+static void gfx_v8_0_ring_set_priority_compute(struct amdgpu_ring *ring,
+					       enum drm_sched_priority priority)
+{
+	struct amdgpu_device *adev = ring->adev;
+	bool acquire = priority == DRM_SCHED_PRIORITY_HIGH_HW;
+
+	if (ring->funcs->type != AMDGPU_RING_TYPE_COMPUTE)
+		return;
+
+	gfx_v8_0_hqd_set_priority(adev, ring, acquire);
+	gfx_v8_0_pipe_reserve_resources(adev, ring, acquire);
+}
+
+static void gfx_v8_0_ring_emit_fence_compute(struct amdgpu_ring *ring,
+					     u64 addr, u64 seq,
+					     unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+
+	/* RELEASE_MEM - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RELEASE_MEM, 5));
+	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
+				 EOP_TC_ACTION_EN |
+				 EOP_TC_WB_ACTION_EN |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+}
+
+static void gfx_v8_0_ring_emit_fence_kiq(struct amdgpu_ring *ring, u64 addr,
+					 u64 seq, unsigned int flags)
+{
+	/* we only allocate 32bit for each seq wb address */
+	BUG_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	/* write fence seq to the "addr" */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				 WRITE_DATA_DST_SEL(5) | WR_CONFIRM));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	if (flags & AMDGPU_FENCE_FLAG_INT) {
+		/* set register to trigger INT */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+		amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+					 WRITE_DATA_DST_SEL(0) | WR_CONFIRM));
+		amdgpu_ring_write(ring, mmCPC_INT_STATUS);
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, 0x20000000); /* src_id is 178 */
+	}
+}
+
+static void gfx_v8_ring_emit_sb(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+	amdgpu_ring_write(ring, 0);
+}
+
+static void gfx_v8_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags)
+{
+	uint32_t dw2 = 0;
+
+	if (amdgpu_sriov_vf(ring->adev))
+		gfx_v8_0_ring_emit_ce_meta(ring);
+
+	dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */
+	if (flags & AMDGPU_HAVE_CTX_SWITCH) {
+		gfx_v8_0_ring_emit_vgt_flush(ring);
+		/* set load_global_config & load_global_uconfig */
+		dw2 |= 0x8001;
+		/* set load_cs_sh_regs */
+		dw2 |= 0x01000000;
+		/* set load_per_context_state & load_gfx_sh_regs for GFX */
+		dw2 |= 0x10002;
+
+		/* set load_ce_ram if preamble presented */
+		if (AMDGPU_PREAMBLE_IB_PRESENT & flags)
+			dw2 |= 0x10000000;
+	} else {
+		/* still load_ce_ram if this is the first time preamble presented
+		 * although there is no context switch happens.
+		 */
+		if (AMDGPU_PREAMBLE_IB_PRESENT_FIRST & flags)
+			dw2 |= 0x10000000;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, dw2);
+	amdgpu_ring_write(ring, 0);
+}
+
+static unsigned gfx_v8_0_ring_emit_init_cond_exec(struct amdgpu_ring *ring)
+{
+	unsigned ret;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_COND_EXEC, 3));
+	amdgpu_ring_write(ring, lower_32_bits(ring->cond_exe_gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ring->cond_exe_gpu_addr));
+	amdgpu_ring_write(ring, 0); /* discard following DWs if *cond_exec_gpu_addr==0 */
+	ret = ring->wptr & ring->buf_mask;
+	amdgpu_ring_write(ring, 0x55aa55aa); /* patch dummy value later */
+	return ret;
+}
+
+static void gfx_v8_0_ring_emit_patch_cond_exec(struct amdgpu_ring *ring, unsigned offset)
+{
+	unsigned cur;
+
+	BUG_ON(offset > ring->buf_mask);
+	BUG_ON(ring->ring[offset] != 0x55aa55aa);
+
+	cur = (ring->wptr & ring->buf_mask) - 1;
+	if (likely(cur > offset))
+		ring->ring[offset] = cur - offset;
+	else
+		ring->ring[offset] = (ring->ring_size >> 2) - offset + cur;
+}
+
+static void gfx_v8_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_COPY_DATA, 4));
+	amdgpu_ring_write(ring, 0 |	/* src: register*/
+				(5 << 8) |	/* dst: memory */
+				(1 << 20));	/* write confirm */
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, lower_32_bits(adev->wb.gpu_addr +
+				adev->virt.reg_val_offs * 4));
+	amdgpu_ring_write(ring, upper_32_bits(adev->wb.gpu_addr +
+				adev->virt.reg_val_offs * 4));
+}
+
+static void gfx_v8_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
+				  uint32_t val)
+{
+	uint32_t cmd;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_GFX:
+		cmd = WRITE_DATA_ENGINE_SEL(1) | WR_CONFIRM;
+		break;
+	case AMDGPU_RING_TYPE_KIQ:
+		cmd = 1 << 16; /* no inc addr */
+		break;
+	default:
+		cmd = WR_CONFIRM;
+		break;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, cmd);
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
+static void gfx_v8_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
+						 enum amdgpu_interrupt_state state)
+{
+	WREG32_FIELD(CP_INT_CNTL_RING0, TIME_STAMP_INT_ENABLE,
+		     state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1);
+}
+
+static void gfx_v8_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev,
+						     int me, int pipe,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 mec_int_cntl, mec_int_cntl_reg;
+
+	/*
+	 * amdgpu controls only the first MEC. That's why this function only
+	 * handles the setting of interrupts for this specific MEC. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+
+	if (me == 1) {
+		switch (pipe) {
+		case 0:
+			mec_int_cntl_reg = mmCP_ME1_PIPE0_INT_CNTL;
+			break;
+		case 1:
+			mec_int_cntl_reg = mmCP_ME1_PIPE1_INT_CNTL;
+			break;
+		case 2:
+			mec_int_cntl_reg = mmCP_ME1_PIPE2_INT_CNTL;
+			break;
+		case 3:
+			mec_int_cntl_reg = mmCP_ME1_PIPE3_INT_CNTL;
+			break;
+		default:
+			DRM_DEBUG("invalid pipe %d\n", pipe);
+			return;
+		}
+	} else {
+		DRM_DEBUG("invalid me %d\n", me);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		mec_int_cntl = RREG32(mec_int_cntl_reg);
+		mec_int_cntl &= ~CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK;
+		WREG32(mec_int_cntl_reg, mec_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		mec_int_cntl = RREG32(mec_int_cntl_reg);
+		mec_int_cntl |= CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK;
+		WREG32(mec_int_cntl_reg, mec_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static int gfx_v8_0_set_priv_reg_fault_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *source,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	WREG32_FIELD(CP_INT_CNTL_RING0, PRIV_REG_INT_ENABLE,
+		     state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1);
+
+	return 0;
+}
+
+static int gfx_v8_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      unsigned type,
+					      enum amdgpu_interrupt_state state)
+{
+	WREG32_FIELD(CP_INT_CNTL_RING0, PRIV_INSTR_INT_ENABLE,
+		     state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1);
+
+	return 0;
+}
+
+static int gfx_v8_0_set_eop_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CP_IRQ_GFX_EOP:
+		gfx_v8_0_set_gfx_eop_interrupt_state(adev, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 1, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 1, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 1, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE3_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 1, 3, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE0_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 2, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE1_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 2, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE2_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 2, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE3_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 2, 3, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v8_0_set_cp_ecc_int_state(struct amdgpu_device *adev,
+					 struct amdgpu_irq_src *source,
+					 unsigned int type,
+					 enum amdgpu_interrupt_state state)
+{
+	int enable_flag;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		enable_flag = 0;
+		break;
+
+	case AMDGPU_IRQ_STATE_ENABLE:
+		enable_flag = 1;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	WREG32_FIELD(CP_INT_CNTL, CP_ECC_ERROR_INT_ENABLE, enable_flag);
+	WREG32_FIELD(CP_INT_CNTL_RING0, CP_ECC_ERROR_INT_ENABLE, enable_flag);
+	WREG32_FIELD(CP_INT_CNTL_RING1, CP_ECC_ERROR_INT_ENABLE, enable_flag);
+	WREG32_FIELD(CP_INT_CNTL_RING2, CP_ECC_ERROR_INT_ENABLE, enable_flag);
+	WREG32_FIELD(CPC_INT_CNTL, CP_ECC_ERROR_INT_ENABLE, enable_flag);
+	WREG32_FIELD(CP_ME1_PIPE0_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME1_PIPE1_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME1_PIPE2_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME1_PIPE3_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME2_PIPE0_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME2_PIPE1_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME2_PIPE2_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+	WREG32_FIELD(CP_ME2_PIPE3_INT_CNTL, CP_ECC_ERROR_INT_ENABLE,
+		     enable_flag);
+
+	return 0;
+}
+
+static int gfx_v8_0_set_sq_int_state(struct amdgpu_device *adev,
+				     struct amdgpu_irq_src *source,
+				     unsigned int type,
+				     enum amdgpu_interrupt_state state)
+{
+	int enable_flag;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		enable_flag = 1;
+		break;
+
+	case AMDGPU_IRQ_STATE_ENABLE:
+		enable_flag = 0;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	WREG32_FIELD(SQ_INTERRUPT_MSG_CTRL, STALL,
+		     enable_flag);
+
+	return 0;
+}
+
+static int gfx_v8_0_eop_irq(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *source,
+			    struct amdgpu_iv_entry *entry)
+{
+	int i;
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+
+	DRM_DEBUG("IH: CP EOP\n");
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+
+	switch (me_id) {
+	case 0:
+		amdgpu_fence_process(&adev->gfx.gfx_ring[0]);
+		break;
+	case 1:
+	case 2:
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring[i];
+			/* Per-queue interrupt is supported for MEC starting from VI.
+			  * The interrupt can only be enabled/disabled per pipe instead of per queue.
+			  */
+			if ((ring->me == me_id) && (ring->pipe == pipe_id) && (ring->queue == queue_id))
+				amdgpu_fence_process(ring);
+		}
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v8_0_priv_reg_irq(struct amdgpu_device *adev,
+				 struct amdgpu_irq_src *source,
+				 struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal register access in command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int gfx_v8_0_priv_inst_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int gfx_v8_0_cp_ecc_error_irq(struct amdgpu_device *adev,
+				     struct amdgpu_irq_src *source,
+				     struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("CP EDC/ECC error detected.");
+	return 0;
+}
+
+static void gfx_v8_0_parse_sq_irq(struct amdgpu_device *adev, unsigned ih_data)
+{
+	u32 enc, se_id, sh_id, cu_id;
+	char type[20];
+	int sq_edc_source = -1;
+
+	enc = REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_CMN, ENCODING);
+	se_id = REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_CMN, SE_ID);
+
+	switch (enc) {
+		case 0:
+			DRM_INFO("SQ general purpose intr detected:"
+					"se_id %d, immed_overflow %d, host_reg_overflow %d,"
+					"host_cmd_overflow %d, cmd_timestamp %d,"
+					"reg_timestamp %d, thread_trace_buff_full %d,"
+					"wlt %d, thread_trace %d.\n",
+					se_id,
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, IMMED_OVERFLOW),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, HOST_REG_OVERFLOW),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, HOST_CMD_OVERFLOW),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, CMD_TIMESTAMP),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, REG_TIMESTAMP),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, THREAD_TRACE_BUF_FULL),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, WLT),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_AUTO, THREAD_TRACE)
+					);
+			break;
+		case 1:
+		case 2:
+
+			cu_id = REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_WAVE, CU_ID);
+			sh_id = REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_WAVE, SH_ID);
+
+			/*
+			 * This function can be called either directly from ISR
+			 * or from BH in which case we can access SQ_EDC_INFO
+			 * instance
+			 */
+			if (in_task()) {
+				mutex_lock(&adev->grbm_idx_mutex);
+				gfx_v8_0_select_se_sh(adev, se_id, sh_id, cu_id);
+
+				sq_edc_source = REG_GET_FIELD(RREG32(mmSQ_EDC_INFO), SQ_EDC_INFO, SOURCE);
+
+				gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+				mutex_unlock(&adev->grbm_idx_mutex);
+			}
+
+			if (enc == 1)
+				snprintf(type, sizeof(type), "instruction intr");
+			else
+				snprintf(type, sizeof(type), "EDC/ECC error");
+
+			DRM_INFO(
+				"SQ %s detected: "
+					"se_id %d, sh_id %d, cu_id %d, simd_id %d, wave_id %d, vm_id %d "
+					"trap %s, sq_ed_info.source %s.\n",
+					type, se_id, sh_id, cu_id,
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_WAVE, SIMD_ID),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_WAVE, WAVE_ID),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_WAVE, VM_ID),
+					REG_GET_FIELD(ih_data, SQ_INTERRUPT_WORD_WAVE, PRIV) ? "true" : "false",
+					(sq_edc_source != -1) ? sq_edc_source_names[sq_edc_source] : "unavailable"
+				);
+			break;
+		default:
+			DRM_ERROR("SQ invalid encoding type\n.");
+	}
+}
+
+static void gfx_v8_0_sq_irq_work_func(struct work_struct *work)
+{
+
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device, gfx.sq_work.work);
+	struct sq_work *sq_work = container_of(work, struct sq_work, work);
+
+	gfx_v8_0_parse_sq_irq(adev, sq_work->ih_data);
+}
+
+static int gfx_v8_0_sq_irq(struct amdgpu_device *adev,
+			   struct amdgpu_irq_src *source,
+			   struct amdgpu_iv_entry *entry)
+{
+	unsigned ih_data = entry->src_data[0];
+
+	/*
+	 * Try to submit work so SQ_EDC_INFO can be accessed from
+	 * BH. If previous work submission hasn't finished yet
+	 * just print whatever info is possible directly from the ISR.
+	 */
+	if (work_pending(&adev->gfx.sq_work.work)) {
+		gfx_v8_0_parse_sq_irq(adev, ih_data);
+	} else {
+		adev->gfx.sq_work.ih_data = ih_data;
+		schedule_work(&adev->gfx.sq_work.work);
+	}
+
+	return 0;
+}
+
+static int gfx_v8_0_kiq_set_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned int type,
+					    enum amdgpu_interrupt_state state)
+{
+	struct amdgpu_ring *ring = &(adev->gfx.kiq.ring);
+
+	switch (type) {
+	case AMDGPU_CP_KIQ_IRQ_DRIVER0:
+		WREG32_FIELD(CPC_INT_CNTL, GENERIC2_INT_ENABLE,
+			     state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1);
+		if (ring->me == 1)
+			WREG32_FIELD_OFFSET(CP_ME1_PIPE0_INT_CNTL,
+				     ring->pipe,
+				     GENERIC2_INT_ENABLE,
+				     state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1);
+		else
+			WREG32_FIELD_OFFSET(CP_ME2_PIPE0_INT_CNTL,
+				     ring->pipe,
+				     GENERIC2_INT_ENABLE,
+				     state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1);
+		break;
+	default:
+		BUG(); /* kiq only support GENERIC2_INT now */
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v8_0_kiq_irq(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *source,
+			    struct amdgpu_iv_entry *entry)
+{
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring = &(adev->gfx.kiq.ring);
+
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+	DRM_DEBUG("IH: CPC GENERIC2_INT, me:%d, pipe:%d, queue:%d\n",
+		   me_id, pipe_id, queue_id);
+
+	amdgpu_fence_process(ring);
+	return 0;
+}
+
+static const struct amd_ip_funcs gfx_v8_0_ip_funcs = {
+	.name = "gfx_v8_0",
+	.early_init = gfx_v8_0_early_init,
+	.late_init = gfx_v8_0_late_init,
+	.sw_init = gfx_v8_0_sw_init,
+	.sw_fini = gfx_v8_0_sw_fini,
+	.hw_init = gfx_v8_0_hw_init,
+	.hw_fini = gfx_v8_0_hw_fini,
+	.suspend = gfx_v8_0_suspend,
+	.resume = gfx_v8_0_resume,
+	.is_idle = gfx_v8_0_is_idle,
+	.wait_for_idle = gfx_v8_0_wait_for_idle,
+	.check_soft_reset = gfx_v8_0_check_soft_reset,
+	.pre_soft_reset = gfx_v8_0_pre_soft_reset,
+	.soft_reset = gfx_v8_0_soft_reset,
+	.post_soft_reset = gfx_v8_0_post_soft_reset,
+	.set_clockgating_state = gfx_v8_0_set_clockgating_state,
+	.set_powergating_state = gfx_v8_0_set_powergating_state,
+	.get_clockgating_state = gfx_v8_0_get_clockgating_state,
+};
+
+static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = {
+	.type = AMDGPU_RING_TYPE_GFX,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = false,
+	.get_rptr = gfx_v8_0_ring_get_rptr,
+	.get_wptr = gfx_v8_0_ring_get_wptr_gfx,
+	.set_wptr = gfx_v8_0_ring_set_wptr_gfx,
+	.emit_frame_size = /* maximum 215dw if count 16 IBs in */
+		5 +  /* COND_EXEC */
+		7 +  /* PIPELINE_SYNC */
+		VI_FLUSH_GPU_TLB_NUM_WREG * 5 + 9 + /* VM_FLUSH */
+		8 +  /* FENCE for VM_FLUSH */
+		20 + /* GDS switch */
+		4 + /* double SWITCH_BUFFER,
+		       the first COND_EXEC jump to the place just
+			   prior to this double SWITCH_BUFFER  */
+		5 + /* COND_EXEC */
+		7 +	 /*	HDP_flush */
+		4 +	 /*	VGT_flush */
+		14 + /*	CE_META */
+		31 + /*	DE_META */
+		3 + /* CNTX_CTRL */
+		5 + /* HDP_INVL */
+		8 + 8 + /* FENCE x2 */
+		2, /* SWITCH_BUFFER */
+	.emit_ib_size =	4, /* gfx_v8_0_ring_emit_ib_gfx */
+	.emit_ib = gfx_v8_0_ring_emit_ib_gfx,
+	.emit_fence = gfx_v8_0_ring_emit_fence_gfx,
+	.emit_pipeline_sync = gfx_v8_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v8_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v8_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v8_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v8_0_ring_test_ring,
+	.test_ib = gfx_v8_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_switch_buffer = gfx_v8_ring_emit_sb,
+	.emit_cntxcntl = gfx_v8_ring_emit_cntxcntl,
+	.init_cond_exec = gfx_v8_0_ring_emit_init_cond_exec,
+	.patch_cond_exec = gfx_v8_0_ring_emit_patch_cond_exec,
+	.emit_wreg = gfx_v8_0_ring_emit_wreg,
+};
+
+static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = {
+	.type = AMDGPU_RING_TYPE_COMPUTE,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = false,
+	.get_rptr = gfx_v8_0_ring_get_rptr,
+	.get_wptr = gfx_v8_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v8_0_ring_set_wptr_compute,
+	.emit_frame_size =
+		20 + /* gfx_v8_0_ring_emit_gds_switch */
+		7 + /* gfx_v8_0_ring_emit_hdp_flush */
+		5 + /* hdp_invalidate */
+		7 + /* gfx_v8_0_ring_emit_pipeline_sync */
+		VI_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + /* gfx_v8_0_ring_emit_vm_flush */
+		7 + 7 + 7, /* gfx_v8_0_ring_emit_fence_compute x3 for user fence, vm fence */
+	.emit_ib_size =	4, /* gfx_v8_0_ring_emit_ib_compute */
+	.emit_ib = gfx_v8_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v8_0_ring_emit_fence_compute,
+	.emit_pipeline_sync = gfx_v8_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v8_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v8_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v8_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v8_0_ring_test_ring,
+	.test_ib = gfx_v8_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.set_priority = gfx_v8_0_ring_set_priority_compute,
+	.emit_wreg = gfx_v8_0_ring_emit_wreg,
+};
+
+static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_kiq = {
+	.type = AMDGPU_RING_TYPE_KIQ,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = false,
+	.get_rptr = gfx_v8_0_ring_get_rptr,
+	.get_wptr = gfx_v8_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v8_0_ring_set_wptr_compute,
+	.emit_frame_size =
+		20 + /* gfx_v8_0_ring_emit_gds_switch */
+		7 + /* gfx_v8_0_ring_emit_hdp_flush */
+		5 + /* hdp_invalidate */
+		7 + /* gfx_v8_0_ring_emit_pipeline_sync */
+		17 + /* gfx_v8_0_ring_emit_vm_flush */
+		7 + 7 + 7, /* gfx_v8_0_ring_emit_fence_kiq x3 for user fence, vm fence */
+	.emit_ib_size =	4, /* gfx_v8_0_ring_emit_ib_compute */
+	.emit_ib = gfx_v8_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v8_0_ring_emit_fence_kiq,
+	.test_ring = gfx_v8_0_ring_test_ring,
+	.test_ib = gfx_v8_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_rreg = gfx_v8_0_ring_emit_rreg,
+	.emit_wreg = gfx_v8_0_ring_emit_wreg,
+};
+
+static void gfx_v8_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	adev->gfx.kiq.ring.funcs = &gfx_v8_0_ring_funcs_kiq;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		adev->gfx.gfx_ring[i].funcs = &gfx_v8_0_ring_funcs_gfx;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		adev->gfx.compute_ring[i].funcs = &gfx_v8_0_ring_funcs_compute;
+}
+
+static const struct amdgpu_irq_src_funcs gfx_v8_0_eop_irq_funcs = {
+	.set = gfx_v8_0_set_eop_interrupt_state,
+	.process = gfx_v8_0_eop_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v8_0_priv_reg_irq_funcs = {
+	.set = gfx_v8_0_set_priv_reg_fault_state,
+	.process = gfx_v8_0_priv_reg_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v8_0_priv_inst_irq_funcs = {
+	.set = gfx_v8_0_set_priv_inst_fault_state,
+	.process = gfx_v8_0_priv_inst_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v8_0_kiq_irq_funcs = {
+	.set = gfx_v8_0_kiq_set_interrupt_state,
+	.process = gfx_v8_0_kiq_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v8_0_cp_ecc_error_irq_funcs = {
+	.set = gfx_v8_0_set_cp_ecc_int_state,
+	.process = gfx_v8_0_cp_ecc_error_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v8_0_sq_irq_funcs = {
+	.set = gfx_v8_0_set_sq_int_state,
+	.process = gfx_v8_0_sq_irq,
+};
+
+static void gfx_v8_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.eop_irq.num_types = AMDGPU_CP_IRQ_LAST;
+	adev->gfx.eop_irq.funcs = &gfx_v8_0_eop_irq_funcs;
+
+	adev->gfx.priv_reg_irq.num_types = 1;
+	adev->gfx.priv_reg_irq.funcs = &gfx_v8_0_priv_reg_irq_funcs;
+
+	adev->gfx.priv_inst_irq.num_types = 1;
+	adev->gfx.priv_inst_irq.funcs = &gfx_v8_0_priv_inst_irq_funcs;
+
+	adev->gfx.kiq.irq.num_types = AMDGPU_CP_KIQ_IRQ_LAST;
+	adev->gfx.kiq.irq.funcs = &gfx_v8_0_kiq_irq_funcs;
+
+	adev->gfx.cp_ecc_error_irq.num_types = 1;
+	adev->gfx.cp_ecc_error_irq.funcs = &gfx_v8_0_cp_ecc_error_irq_funcs;
+
+	adev->gfx.sq_irq.num_types = 1;
+	adev->gfx.sq_irq.funcs = &gfx_v8_0_sq_irq_funcs;
+}
+
+static void gfx_v8_0_set_rlc_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.rlc.funcs = &iceland_rlc_funcs;
+}
+
+static void gfx_v8_0_set_gds_init(struct amdgpu_device *adev)
+{
+	/* init asci gds info */
+	adev->gds.mem.total_size = RREG32(mmGDS_VMID0_SIZE);
+	adev->gds.gws.total_size = 64;
+	adev->gds.oa.total_size = 16;
+
+	if (adev->gds.mem.total_size == 64 * 1024) {
+		adev->gds.mem.gfx_partition_size = 4096;
+		adev->gds.mem.cs_partition_size = 4096;
+
+		adev->gds.gws.gfx_partition_size = 4;
+		adev->gds.gws.cs_partition_size = 4;
+
+		adev->gds.oa.gfx_partition_size = 4;
+		adev->gds.oa.cs_partition_size = 1;
+	} else {
+		adev->gds.mem.gfx_partition_size = 1024;
+		adev->gds.mem.cs_partition_size = 1024;
+
+		adev->gds.gws.gfx_partition_size = 16;
+		adev->gds.gws.cs_partition_size = 16;
+
+		adev->gds.oa.gfx_partition_size = 4;
+		adev->gds.oa.cs_partition_size = 4;
+	}
+}
+
+static void gfx_v8_0_set_user_cu_inactive_bitmap(struct amdgpu_device *adev,
+						 u32 bitmap)
+{
+	u32 data;
+
+	if (!bitmap)
+		return;
+
+	data = bitmap << GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT;
+	data &= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK;
+
+	WREG32(mmGC_USER_SHADER_ARRAY_CONFIG, data);
+}
+
+static u32 gfx_v8_0_get_cu_active_bitmap(struct amdgpu_device *adev)
+{
+	u32 data, mask;
+
+	data =  RREG32(mmCC_GC_SHADER_ARRAY_CONFIG) |
+		RREG32(mmGC_USER_SHADER_ARRAY_CONFIG);
+
+	mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_cu_per_sh);
+
+	return ~REG_GET_FIELD(data, CC_GC_SHADER_ARRAY_CONFIG, INACTIVE_CUS) & mask;
+}
+
+static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev)
+{
+	int i, j, k, counter, active_cu_number = 0;
+	u32 mask, bitmap, ao_bitmap, ao_cu_mask = 0;
+	struct amdgpu_cu_info *cu_info = &adev->gfx.cu_info;
+	unsigned disable_masks[4 * 2];
+	u32 ao_cu_num;
+
+	memset(cu_info, 0, sizeof(*cu_info));
+
+	if (adev->flags & AMD_IS_APU)
+		ao_cu_num = 2;
+	else
+		ao_cu_num = adev->gfx.config.max_cu_per_sh;
+
+	amdgpu_gfx_parse_disable_cu(disable_masks, 4, 2);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			mask = 1;
+			ao_bitmap = 0;
+			counter = 0;
+			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+			if (i < 4 && j < 2)
+				gfx_v8_0_set_user_cu_inactive_bitmap(
+					adev, disable_masks[i * 2 + j]);
+			bitmap = gfx_v8_0_get_cu_active_bitmap(adev);
+			cu_info->bitmap[i][j] = bitmap;
+
+			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) {
+				if (bitmap & mask) {
+					if (counter < ao_cu_num)
+						ao_bitmap |= mask;
+					counter ++;
+				}
+				mask <<= 1;
+			}
+			active_cu_number += counter;
+			if (i < 2 && j < 2)
+				ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8));
+			cu_info->ao_cu_bitmap[i][j] = ao_bitmap;
+		}
+	}
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	cu_info->number = active_cu_number;
+	cu_info->ao_cu_mask = ao_cu_mask;
+	cu_info->simd_per_cu = NUM_SIMD_PER_CU;
+	cu_info->max_waves_per_simd = 10;
+	cu_info->max_scratch_slots_per_cu = 32;
+	cu_info->wave_front_size = 64;
+	cu_info->lds_size = 64;
+}
+
+const struct amdgpu_ip_block_version gfx_v8_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GFX,
+	.major = 8,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gfx_v8_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version gfx_v8_1_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GFX,
+	.major = 8,
+	.minor = 1,
+	.rev = 0,
+	.funcs = &gfx_v8_0_ip_funcs,
+};
+
+static void gfx_v8_0_ring_emit_ce_meta(struct amdgpu_ring *ring)
+{
+	uint64_t ce_payload_addr;
+	int cnt_ce;
+	union {
+		struct vi_ce_ib_state regular;
+		struct vi_ce_ib_state_chained_ib chained;
+	} ce_payload = {};
+
+	if (ring->adev->virt.chained_ib_support) {
+		ce_payload_addr = amdgpu_csa_vaddr(ring->adev) +
+			offsetof(struct vi_gfx_meta_data_chained_ib, ce_payload);
+		cnt_ce = (sizeof(ce_payload.chained) >> 2) + 4 - 2;
+	} else {
+		ce_payload_addr = amdgpu_csa_vaddr(ring->adev) +
+			offsetof(struct vi_gfx_meta_data, ce_payload);
+		cnt_ce = (sizeof(ce_payload.regular) >> 2) + 4 - 2;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, cnt_ce));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(2) |
+				WRITE_DATA_DST_SEL(8) |
+				WR_CONFIRM) |
+				WRITE_DATA_CACHE_POLICY(0));
+	amdgpu_ring_write(ring, lower_32_bits(ce_payload_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ce_payload_addr));
+	amdgpu_ring_write_multiple(ring, (void *)&ce_payload, cnt_ce - 2);
+}
+
+static void gfx_v8_0_ring_emit_de_meta(struct amdgpu_ring *ring)
+{
+	uint64_t de_payload_addr, gds_addr, csa_addr;
+	int cnt_de;
+	union {
+		struct vi_de_ib_state regular;
+		struct vi_de_ib_state_chained_ib chained;
+	} de_payload = {};
+
+	csa_addr = amdgpu_csa_vaddr(ring->adev);
+	gds_addr = csa_addr + 4096;
+	if (ring->adev->virt.chained_ib_support) {
+		de_payload.chained.gds_backup_addrlo = lower_32_bits(gds_addr);
+		de_payload.chained.gds_backup_addrhi = upper_32_bits(gds_addr);
+		de_payload_addr = csa_addr + offsetof(struct vi_gfx_meta_data_chained_ib, de_payload);
+		cnt_de = (sizeof(de_payload.chained) >> 2) + 4 - 2;
+	} else {
+		de_payload.regular.gds_backup_addrlo = lower_32_bits(gds_addr);
+		de_payload.regular.gds_backup_addrhi = upper_32_bits(gds_addr);
+		de_payload_addr = csa_addr + offsetof(struct vi_gfx_meta_data, de_payload);
+		cnt_de = (sizeof(de_payload.regular) >> 2) + 4 - 2;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, cnt_de));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(1) |
+				WRITE_DATA_DST_SEL(8) |
+				WR_CONFIRM) |
+				WRITE_DATA_CACHE_POLICY(0));
+	amdgpu_ring_write(ring, lower_32_bits(de_payload_addr));
+	amdgpu_ring_write(ring, upper_32_bits(de_payload_addr));
+	amdgpu_ring_write_multiple(ring, (void *)&de_payload, cnt_de - 2);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/gfx_v8_0.h b/sys/dev/pci/drm/amd/amdgpu/gfx_v8_0.h
new file mode 100644
index 00000000000..ec3f11fa986
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gfx_v8_0.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V8_0_H__
+#define __GFX_V8_0_H__
+
+extern const struct amdgpu_ip_block_version gfx_v8_0_ip_block;
+extern const struct amdgpu_ip_block_version gfx_v8_1_ip_block;
+
+struct amdgpu_device;
+struct vi_mqd;
+
+int gfx_v8_0_mqd_commit(struct amdgpu_device *adev, struct vi_mqd *mqd);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/gfx_v9_0.c b/sys/dev/pci/drm/amd/amdgpu/gfx_v9_0.c
new file mode 100644
index 00000000000..f67571e43a9
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gfx_v9_0.c
@@ -0,0 +1,4928 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "amdgpu_atomfirmware.h"
+
+#include "gc/gc_9_0_offset.h"
+#include "gc/gc_9_0_sh_mask.h"
+#include "vega10_enum.h"
+#include "hdp/hdp_4_0_offset.h"
+
+#include "soc15_common.h"
+#include "clearstate_gfx9.h"
+#include "v9_structs.h"
+
+#include "ivsrcid/gfx/irqsrcs_gfx_9_0.h"
+
+#define GFX9_NUM_GFX_RINGS     1
+#define GFX9_MEC_HPD_SIZE 2048
+#define RLCG_UCODE_LOADING_START_ADDRESS 0x00002000L
+#define RLC_SAVE_RESTORE_ADDR_STARTING_OFFSET 0x00000000L
+
+#define mmPWR_MISC_CNTL_STATUS					0x0183
+#define mmPWR_MISC_CNTL_STATUS_BASE_IDX				0
+#define PWR_MISC_CNTL_STATUS__PWR_GFX_RLC_CGPG_EN__SHIFT	0x0
+#define PWR_MISC_CNTL_STATUS__PWR_GFXOFF_STATUS__SHIFT		0x1
+#define PWR_MISC_CNTL_STATUS__PWR_GFX_RLC_CGPG_EN_MASK		0x00000001L
+#define PWR_MISC_CNTL_STATUS__PWR_GFXOFF_STATUS_MASK		0x00000006L
+
+MODULE_FIRMWARE("amdgpu/vega10_ce.bin");
+MODULE_FIRMWARE("amdgpu/vega10_pfp.bin");
+MODULE_FIRMWARE("amdgpu/vega10_me.bin");
+MODULE_FIRMWARE("amdgpu/vega10_mec.bin");
+MODULE_FIRMWARE("amdgpu/vega10_mec2.bin");
+MODULE_FIRMWARE("amdgpu/vega10_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/vega12_ce.bin");
+MODULE_FIRMWARE("amdgpu/vega12_pfp.bin");
+MODULE_FIRMWARE("amdgpu/vega12_me.bin");
+MODULE_FIRMWARE("amdgpu/vega12_mec.bin");
+MODULE_FIRMWARE("amdgpu/vega12_mec2.bin");
+MODULE_FIRMWARE("amdgpu/vega12_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/vega20_ce.bin");
+MODULE_FIRMWARE("amdgpu/vega20_pfp.bin");
+MODULE_FIRMWARE("amdgpu/vega20_me.bin");
+MODULE_FIRMWARE("amdgpu/vega20_mec.bin");
+MODULE_FIRMWARE("amdgpu/vega20_mec2.bin");
+MODULE_FIRMWARE("amdgpu/vega20_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/raven_ce.bin");
+MODULE_FIRMWARE("amdgpu/raven_pfp.bin");
+MODULE_FIRMWARE("amdgpu/raven_me.bin");
+MODULE_FIRMWARE("amdgpu/raven_mec.bin");
+MODULE_FIRMWARE("amdgpu/raven_mec2.bin");
+MODULE_FIRMWARE("amdgpu/raven_rlc.bin");
+
+static const struct soc15_reg_golden golden_settings_gc_9_0[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0xf00fffff, 0x00000420),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_GPU_ID, 0x0000000f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_3, 0x00000003, 0x82400024),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE, 0x3fffffff, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSH_MEM_CONFIG, 0x00001000, 0x00001000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_CU_0, 0x0007ffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_CU_1, 0x0007ffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_EN_CU_0, 0x01ffffff, 0x0000ff87),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_EN_CU_1, 0x01ffffff, 0x0000ff8f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQC_CONFIG, 0x03000000, 0x020a2000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfffffeef, 0x010b0000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_HI, 0xffffffff, 0x4a2c0e68),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_LO, 0xffffffff, 0xb5d3f197),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_CACHE_INVALIDATION, 0x3fff3af3, 0x19200000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_0_vg10[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL, 0x0000f000, 0x00012107),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_3, 0x30000000, 0x10000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPC_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPG_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0xffff77ff, 0x2a114042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG_READ, 0xffff77ff, 0x2a114042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmIA_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0x00008000, 0x00048000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_0, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_1, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_2, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_PREWALKER_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRMI_UTCL1_CNTL2, 0x00030000, 0x00020000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0x0000000f, 0x01000107),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTD_CNTL, 0x00001800, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmWD_UTCL1_CNTL, 0x08000000, 0x08000080)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_0_vg20[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_DCC_CONFIG, 0x0f000080, 0x04000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_2, 0x0f000000, 0x0a000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_3, 0x30000000, 0x10000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0xf3e777ff, 0x22014042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG_READ, 0xf3e777ff, 0x22014042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0x00003e00, 0x00000400),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0xff840000, 0x04040000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRMI_UTCL1_CNTL2, 0x00030000, 0x00030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0xffff010f, 0x01000107),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0x000b0000, 0x000b0000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTD_CNTL, 0x01000000, 0x01000000)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_1[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL, 0xfffdf3cf, 0x00014104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPC_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPG_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0xf00fffff, 0x00000420),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_GPU_ID, 0x0000000f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmIA_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_3, 0x00000003, 0x82400024),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE, 0x3fffffff, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_0, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_1, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_GPM_UTCL1_CNTL_2, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_PREWALKER_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRLC_SPM_UTCL1_CNTL, 0x08000000, 0x08000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfffffeef, 0x010b0000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00003120),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_CACHE_INVALIDATION, 0x3fff3af3, 0x19200000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000000ff),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmWD_UTCL1_CNTL, 0x08000000, 0x08000080)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_1_rv1[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_3, 0x30000000, 0x10000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0xffff77ff, 0x24000042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG_READ, 0xffff77ff, 0x24000042),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0xffffffff, 0x04048000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_MODE_CNTL_1, 0x06000000, 0x06000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmRMI_UTCL1_CNTL2, 0x00030000, 0x00020000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTD_CNTL, 0x01bd9f33, 0x00000800)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_x_common[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_CAM_INDEX, 0xffffffff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGRBM_CAM_DATA, 0xffffffff, 0x2544c382)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_2_1[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG2, 0xf00fffff, 0x00000420),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_GPU_ID, 0x0000000f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_3, 0x00000003, 0x82400024),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE, 0x3fffffff, 0x00000001),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSH_MEM_CONFIG, 0x00001000, 0x00001000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_CU_0, 0x0007ffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_CU_1, 0x0007ffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_EN_CU_0, 0x01ffffff, 0x0000ff87),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_RESOURCE_RESERVE_EN_CU_1, 0x01ffffff, 0x0000ff8f),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQC_CONFIG, 0x03000000, 0x020a2000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfffffeef, 0x010b0000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_HI, 0xffffffff, 0x4a2c0e68),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_LO, 0xffffffff, 0xb5d3f197),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_CACHE_INVALIDATION, 0x3fff3af3, 0x19200000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff)
+};
+
+static const struct soc15_reg_golden golden_settings_gc_9_2_1_vg12[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_DCC_CONFIG, 0x00000080, 0x04000080),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL, 0xfffdf3cf, 0x00014104),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_2, 0x0f000000, 0x0a000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0xffff77ff, 0x24104041),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG_READ, 0xffff77ff, 0x24104041),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_1, 0xffffffff, 0x04040000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0xffff03ff, 0x01000107),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_LO, 0xffffffff, 0x76325410),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTD_CNTL, 0x01bd9f33, 0x01000000)
+};
+
+static const u32 GFX_RLC_SRM_INDEX_CNTL_ADDR_OFFSETS[] =
+{
+	mmRLC_SRM_INDEX_CNTL_ADDR_0 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_1 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_2 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_3 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_4 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_5 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_6 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+	mmRLC_SRM_INDEX_CNTL_ADDR_7 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
+};
+
+static const u32 GFX_RLC_SRM_INDEX_CNTL_DATA_OFFSETS[] =
+{
+	mmRLC_SRM_INDEX_CNTL_DATA_0 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_1 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_2 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_3 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_4 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_5 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_6 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+	mmRLC_SRM_INDEX_CNTL_DATA_7 - mmRLC_SRM_INDEX_CNTL_DATA_0,
+};
+
+#define VEGA10_GB_ADDR_CONFIG_GOLDEN 0x2a114042
+#define VEGA12_GB_ADDR_CONFIG_GOLDEN 0x24104041
+#define RAVEN_GB_ADDR_CONFIG_GOLDEN 0x24000042
+
+static void gfx_v9_0_set_ring_funcs(struct amdgpu_device *adev);
+static void gfx_v9_0_set_irq_funcs(struct amdgpu_device *adev);
+static void gfx_v9_0_set_gds_init(struct amdgpu_device *adev);
+static void gfx_v9_0_set_rlc_funcs(struct amdgpu_device *adev);
+static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev,
+                                 struct amdgpu_cu_info *cu_info);
+static uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev);
+static void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 instance);
+static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring);
+
+static void gfx_v9_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+		soc15_program_register_sequence(adev,
+						 golden_settings_gc_9_0,
+						 ARRAY_SIZE(golden_settings_gc_9_0));
+		soc15_program_register_sequence(adev,
+						 golden_settings_gc_9_0_vg10,
+						 ARRAY_SIZE(golden_settings_gc_9_0_vg10));
+		break;
+	case CHIP_VEGA12:
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_9_2_1,
+						ARRAY_SIZE(golden_settings_gc_9_2_1));
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_9_2_1_vg12,
+						ARRAY_SIZE(golden_settings_gc_9_2_1_vg12));
+		break;
+	case CHIP_VEGA20:
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_9_0,
+						ARRAY_SIZE(golden_settings_gc_9_0));
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_9_0_vg20,
+						ARRAY_SIZE(golden_settings_gc_9_0_vg20));
+		break;
+	case CHIP_RAVEN:
+		soc15_program_register_sequence(adev,
+						 golden_settings_gc_9_1,
+						 ARRAY_SIZE(golden_settings_gc_9_1));
+		soc15_program_register_sequence(adev,
+						 golden_settings_gc_9_1_rv1,
+						 ARRAY_SIZE(golden_settings_gc_9_1_rv1));
+		break;
+	default:
+		break;
+	}
+
+	soc15_program_register_sequence(adev, golden_settings_gc_9_x_common,
+					(const u32)ARRAY_SIZE(golden_settings_gc_9_x_common));
+}
+
+static void gfx_v9_0_scratch_init(struct amdgpu_device *adev)
+{
+	adev->gfx.scratch.num_reg = 8;
+	adev->gfx.scratch.reg_base = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG0);
+	adev->gfx.scratch.free_mask = (1u << adev->gfx.scratch.num_reg) - 1;
+}
+
+static void gfx_v9_0_write_data_to_reg(struct amdgpu_ring *ring, int eng_sel,
+				       bool wc, uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, WRITE_DATA_ENGINE_SEL(eng_sel) |
+				WRITE_DATA_DST_SEL(0) |
+				(wc ? WR_CONFIRM : 0));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
+static void gfx_v9_0_wait_reg_mem(struct amdgpu_ring *ring, int eng_sel,
+				  int mem_space, int opt, uint32_t addr0,
+				  uint32_t addr1, uint32_t ref, uint32_t mask,
+				  uint32_t inv)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring,
+				 /* memory (1) or register (0) */
+				 (WAIT_REG_MEM_MEM_SPACE(mem_space) |
+				 WAIT_REG_MEM_OPERATION(opt) | /* wait */
+				 WAIT_REG_MEM_FUNCTION(3) |  /* equal */
+				 WAIT_REG_MEM_ENGINE(eng_sel)));
+
+	if (mem_space)
+		BUG_ON(addr0 & 0x3); /* Dword align */
+	amdgpu_ring_write(ring, addr0);
+	amdgpu_ring_write(ring, addr1);
+	amdgpu_ring_write(ring, ref);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, inv); /* poll interval */
+}
+
+static int gfx_v9_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t scratch;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		amdgpu_gfx_scratch_free(adev, scratch);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+	amdgpu_ring_write(ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
+			  ring->idx, scratch, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_gfx_scratch_free(adev, scratch);
+	return r;
+}
+
+static int gfx_v9_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+
+	unsigned index;
+	uint64_t gpu_addr;
+	uint32_t tmp;
+	long r;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, 16, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err1;
+	}
+	ib.ptr[0] = PACKET3(PACKET3_WRITE_DATA, 3);
+	ib.ptr[1] = WRITE_DATA_DST_SEL(5) | WR_CONFIRM;
+	ib.ptr[2] = lower_32_bits(gpu_addr);
+	ib.ptr[3] = upper_32_bits(gpu_addr);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.length_dw = 5;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err2;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+			DRM_ERROR("amdgpu: IB test timed out.\n");
+			r = -ETIMEDOUT;
+			goto err2;
+	} else if (r < 0) {
+			DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+			goto err2;
+	}
+
+	tmp = adev->wb.wb[index];
+	if (tmp == 0xDEADBEEF) {
+			DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+			r = 0;
+	} else {
+			DRM_ERROR("ib test on ring %d failed\n", ring->idx);
+			r = -EINVAL;
+	}
+
+err2:
+	amdgpu_ib_free(adev, &ib, NULL);
+	dma_fence_put(f);
+err1:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+
+static void gfx_v9_0_free_microcode(struct amdgpu_device *adev)
+{
+	release_firmware(adev->gfx.pfp_fw);
+	adev->gfx.pfp_fw = NULL;
+	release_firmware(adev->gfx.me_fw);
+	adev->gfx.me_fw = NULL;
+	release_firmware(adev->gfx.ce_fw);
+	adev->gfx.ce_fw = NULL;
+	release_firmware(adev->gfx.rlc_fw);
+	adev->gfx.rlc_fw = NULL;
+	release_firmware(adev->gfx.mec_fw);
+	adev->gfx.mec_fw = NULL;
+	release_firmware(adev->gfx.mec2_fw);
+	adev->gfx.mec2_fw = NULL;
+
+	kfree(adev->gfx.rlc.register_list_format);
+}
+
+static void gfx_v9_0_init_rlc_ext_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_1 *rlc_hdr;
+
+	rlc_hdr = (const struct rlc_firmware_header_v2_1 *)adev->gfx.rlc_fw->data;
+	adev->gfx.rlc_srlc_fw_version = le32_to_cpu(rlc_hdr->save_restore_list_cntl_ucode_ver);
+	adev->gfx.rlc_srlc_feature_version = le32_to_cpu(rlc_hdr->save_restore_list_cntl_feature_ver);
+	adev->gfx.rlc.save_restore_list_cntl_size_bytes = le32_to_cpu(rlc_hdr->save_restore_list_cntl_size_bytes);
+	adev->gfx.rlc.save_restore_list_cntl = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->save_restore_list_cntl_offset_bytes);
+	adev->gfx.rlc_srlg_fw_version = le32_to_cpu(rlc_hdr->save_restore_list_gpm_ucode_ver);
+	adev->gfx.rlc_srlg_feature_version = le32_to_cpu(rlc_hdr->save_restore_list_gpm_feature_ver);
+	adev->gfx.rlc.save_restore_list_gpm_size_bytes = le32_to_cpu(rlc_hdr->save_restore_list_gpm_size_bytes);
+	adev->gfx.rlc.save_restore_list_gpm = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->save_restore_list_gpm_offset_bytes);
+	adev->gfx.rlc_srls_fw_version = le32_to_cpu(rlc_hdr->save_restore_list_srm_ucode_ver);
+	adev->gfx.rlc_srls_feature_version = le32_to_cpu(rlc_hdr->save_restore_list_srm_feature_ver);
+	adev->gfx.rlc.save_restore_list_srm_size_bytes = le32_to_cpu(rlc_hdr->save_restore_list_srm_size_bytes);
+	adev->gfx.rlc.save_restore_list_srm = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->save_restore_list_srm_offset_bytes);
+	adev->gfx.rlc.reg_list_format_direct_reg_list_length =
+			le32_to_cpu(rlc_hdr->reg_list_format_direct_reg_list_length);
+}
+
+static int gfx_v9_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+	struct amdgpu_firmware_info *info = NULL;
+	const struct common_firmware_header *header = NULL;
+	const struct gfx_firmware_header_v1_0 *cp_hdr;
+	const struct rlc_firmware_header_v2_0 *rlc_hdr;
+	unsigned int *tmp = NULL;
+	unsigned int i = 0;
+	uint16_t version_major;
+	uint16_t version_minor;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+		chip_name = "vega10";
+		break;
+	case CHIP_VEGA12:
+		chip_name = "vega12";
+		break;
+	case CHIP_VEGA20:
+		chip_name = "vega20";
+		break;
+	case CHIP_RAVEN:
+		chip_name = "raven";
+		break;
+	default:
+		BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_pfp.bin", chip_name);
+	err = request_firmware(&adev->gfx.pfp_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.pfp_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+	adev->gfx.pfp_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.pfp_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_me.bin", chip_name);
+	err = request_firmware(&adev->gfx.me_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.me_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+	adev->gfx.me_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.me_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ce.bin", chip_name);
+	err = request_firmware(&adev->gfx.ce_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.ce_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+	adev->gfx.ce_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.ce_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc.bin", chip_name);
+	err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.rlc_fw);
+	rlc_hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+
+	version_major = le16_to_cpu(rlc_hdr->header.header_version_major);
+	version_minor = le16_to_cpu(rlc_hdr->header.header_version_minor);
+	if (version_major == 2 && version_minor == 1)
+		adev->gfx.rlc.is_rlc_v2_1 = true;
+
+	adev->gfx.rlc_fw_version = le32_to_cpu(rlc_hdr->header.ucode_version);
+	adev->gfx.rlc_feature_version = le32_to_cpu(rlc_hdr->ucode_feature_version);
+	adev->gfx.rlc.save_and_restore_offset =
+			le32_to_cpu(rlc_hdr->save_and_restore_offset);
+	adev->gfx.rlc.clear_state_descriptor_offset =
+			le32_to_cpu(rlc_hdr->clear_state_descriptor_offset);
+	adev->gfx.rlc.avail_scratch_ram_locations =
+			le32_to_cpu(rlc_hdr->avail_scratch_ram_locations);
+	adev->gfx.rlc.reg_restore_list_size =
+			le32_to_cpu(rlc_hdr->reg_restore_list_size);
+	adev->gfx.rlc.reg_list_format_start =
+			le32_to_cpu(rlc_hdr->reg_list_format_start);
+	adev->gfx.rlc.reg_list_format_separate_start =
+			le32_to_cpu(rlc_hdr->reg_list_format_separate_start);
+	adev->gfx.rlc.starting_offsets_start =
+			le32_to_cpu(rlc_hdr->starting_offsets_start);
+	adev->gfx.rlc.reg_list_format_size_bytes =
+			le32_to_cpu(rlc_hdr->reg_list_format_size_bytes);
+	adev->gfx.rlc.reg_list_size_bytes =
+			le32_to_cpu(rlc_hdr->reg_list_size_bytes);
+	adev->gfx.rlc.register_list_format =
+			kmalloc(adev->gfx.rlc.reg_list_format_size_bytes +
+				adev->gfx.rlc.reg_list_size_bytes, GFP_KERNEL);
+	if (!adev->gfx.rlc.register_list_format) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	tmp = (unsigned int *)((uintptr_t)rlc_hdr +
+			le32_to_cpu(rlc_hdr->reg_list_format_array_offset_bytes));
+	for (i = 0 ; i < (rlc_hdr->reg_list_format_size_bytes >> 2); i++)
+		adev->gfx.rlc.register_list_format[i] =	le32_to_cpu(tmp[i]);
+
+	adev->gfx.rlc.register_restore = adev->gfx.rlc.register_list_format + i;
+
+	tmp = (unsigned int *)((uintptr_t)rlc_hdr +
+			le32_to_cpu(rlc_hdr->reg_list_array_offset_bytes));
+	for (i = 0 ; i < (rlc_hdr->reg_list_size_bytes >> 2); i++)
+		adev->gfx.rlc.register_restore[i] = le32_to_cpu(tmp[i]);
+
+	if (adev->gfx.rlc.is_rlc_v2_1)
+		gfx_v9_0_init_rlc_ext_microcode(adev);
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec.bin", chip_name);
+	err = request_firmware(&adev->gfx.mec_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.mec_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+	adev->gfx.mec_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.mec_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec2.bin", chip_name);
+	err = request_firmware(&adev->gfx.mec2_fw, fw_name, adev->dev);
+	if (!err) {
+		err = amdgpu_ucode_validate(adev->gfx.mec2_fw);
+		if (err)
+			goto out;
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.mec2_fw->data;
+		adev->gfx.mec2_fw_version =
+		le32_to_cpu(cp_hdr->header.ucode_version);
+		adev->gfx.mec2_feature_version =
+		le32_to_cpu(cp_hdr->ucode_feature_version);
+	} else {
+		err = 0;
+		adev->gfx.mec2_fw = NULL;
+	}
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_PFP];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_PFP;
+		info->fw = adev->gfx.pfp_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_ME];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_ME;
+		info->fw = adev->gfx.me_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_CE];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_CE;
+		info->fw = adev->gfx.ce_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_G];
+		info->ucode_id = AMDGPU_UCODE_ID_RLC_G;
+		info->fw = adev->gfx.rlc_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		if (adev->gfx.rlc.is_rlc_v2_1 &&
+		    adev->gfx.rlc.save_restore_list_cntl_size_bytes &&
+		    adev->gfx.rlc.save_restore_list_gpm_size_bytes &&
+		    adev->gfx.rlc.save_restore_list_srm_size_bytes) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL];
+			info->ucode_id = AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				roundup2(adev->gfx.rlc.save_restore_list_cntl_size_bytes, PAGE_SIZE);
+
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM];
+			info->ucode_id = AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				roundup2(adev->gfx.rlc.save_restore_list_gpm_size_bytes, PAGE_SIZE);
+
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM];
+			info->ucode_id = AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM;
+			info->fw = adev->gfx.rlc_fw;
+			adev->firmware.fw_size +=
+				roundup2(adev->gfx.rlc.save_restore_list_srm_size_bytes, PAGE_SIZE);
+		}
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC1];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_MEC1;
+		info->fw = adev->gfx.mec_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		cp_hdr = (const struct gfx_firmware_header_v1_0 *)info->fw->data;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(header->ucode_size_bytes) - le32_to_cpu(cp_hdr->jt_size) * 4, PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC1_JT];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_MEC1_JT;
+		info->fw = adev->gfx.mec_fw;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(cp_hdr->jt_size) * 4, PAGE_SIZE);
+
+		if (adev->gfx.mec2_fw) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC2];
+			info->ucode_id = AMDGPU_UCODE_ID_CP_MEC2;
+			info->fw = adev->gfx.mec2_fw;
+			header = (const struct common_firmware_header *)info->fw->data;
+			cp_hdr = (const struct gfx_firmware_header_v1_0 *)info->fw->data;
+			adev->firmware.fw_size +=
+				roundup2(le32_to_cpu(header->ucode_size_bytes) - le32_to_cpu(cp_hdr->jt_size) * 4, PAGE_SIZE);
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC2_JT];
+			info->ucode_id = AMDGPU_UCODE_ID_CP_MEC2_JT;
+			info->fw = adev->gfx.mec2_fw;
+			adev->firmware.fw_size +=
+				roundup2(le32_to_cpu(cp_hdr->jt_size) * 4, PAGE_SIZE);
+		}
+
+	}
+
+out:
+	if (err) {
+		dev_err(adev->dev,
+			"gfx9: Failed to load firmware \"%s\"\n",
+			fw_name);
+		release_firmware(adev->gfx.pfp_fw);
+		adev->gfx.pfp_fw = NULL;
+		release_firmware(adev->gfx.me_fw);
+		adev->gfx.me_fw = NULL;
+		release_firmware(adev->gfx.ce_fw);
+		adev->gfx.ce_fw = NULL;
+		release_firmware(adev->gfx.rlc_fw);
+		adev->gfx.rlc_fw = NULL;
+		release_firmware(adev->gfx.mec_fw);
+		adev->gfx.mec_fw = NULL;
+		release_firmware(adev->gfx.mec2_fw);
+		adev->gfx.mec2_fw = NULL;
+	}
+	return err;
+}
+
+static u32 gfx_v9_0_get_csb_size(struct amdgpu_device *adev)
+{
+	u32 count = 0;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	/* begin clear state */
+	count += 2;
+	/* context control state */
+	count += 3;
+
+	for (sect = gfx9_cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT)
+				count += 2 + ext->reg_count;
+			else
+				return 0;
+		}
+	}
+
+	/* end clear state */
+	count += 2;
+	/* clear state */
+	count += 2;
+
+	return count;
+}
+
+static void gfx_v9_0_get_csb_buffer(struct amdgpu_device *adev,
+				    volatile u32 *buffer)
+{
+	u32 count = 0, i;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	if (adev->gfx.rlc.cs_data == NULL)
+		return;
+	if (buffer == NULL)
+		return;
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	buffer[count++] = cpu_to_le32(0x80000000);
+	buffer[count++] = cpu_to_le32(0x80000000);
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				buffer[count++] =
+					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
+				buffer[count++] = cpu_to_le32(ext->reg_index -
+						PACKET3_SET_CONTEXT_REG_START);
+				for (i = 0; i < ext->reg_count; i++)
+					buffer[count++] = cpu_to_le32(ext->extent[i]);
+			} else {
+				return;
+			}
+		}
+	}
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
+	buffer[count++] = cpu_to_le32(0);
+}
+
+static void gfx_v9_0_init_lbpw(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	/* set mmRLC_LB_THR_CONFIG_1/2/3/4 */
+	WREG32_SOC15(GC, 0, mmRLC_LB_THR_CONFIG_1, 0x0000007F);
+	WREG32_SOC15(GC, 0, mmRLC_LB_THR_CONFIG_2, 0x0333A5A7);
+	WREG32_SOC15(GC, 0, mmRLC_LB_THR_CONFIG_3, 0x00000077);
+	WREG32_SOC15(GC, 0, mmRLC_LB_THR_CONFIG_4, (0x30 | 0x40 << 8 | 0x02FA << 16));
+
+	/* set mmRLC_LB_CNTR_INIT = 0x0000_0000 */
+	WREG32_SOC15(GC, 0, mmRLC_LB_CNTR_INIT, 0x00000000);
+
+	/* set mmRLC_LB_CNTR_MAX = 0x0000_0500 */
+	WREG32_SOC15(GC, 0, mmRLC_LB_CNTR_MAX, 0x00000500);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	/* set mmRLC_LB_INIT_CU_MASK thru broadcast mode to enable all SE/SH*/
+	gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	WREG32_SOC15(GC, 0, mmRLC_LB_INIT_CU_MASK, 0xffffffff);
+
+	/* set mmRLC_LB_PARAMS = 0x003F_1006 */
+	data = REG_SET_FIELD(0, RLC_LB_PARAMS, FIFO_SAMPLES, 0x0003);
+	data |= REG_SET_FIELD(data, RLC_LB_PARAMS, PG_IDLE_SAMPLES, 0x0010);
+	data |= REG_SET_FIELD(data, RLC_LB_PARAMS, PG_IDLE_SAMPLE_INTERVAL, 0x033F);
+	WREG32_SOC15(GC, 0, mmRLC_LB_PARAMS, data);
+
+	/* set mmRLC_GPM_GENERAL_7[31-16] = 0x00C0 */
+	data = RREG32_SOC15(GC, 0, mmRLC_GPM_GENERAL_7);
+	data &= 0x0000FFFF;
+	data |= 0x00C00000;
+	WREG32_SOC15(GC, 0, mmRLC_GPM_GENERAL_7, data);
+
+	/* set RLC_LB_ALWAYS_ACTIVE_CU_MASK = 0xFFF */
+	WREG32_SOC15(GC, 0, mmRLC_LB_ALWAYS_ACTIVE_CU_MASK, 0xFFF);
+
+	/* set RLC_LB_CNTL = 0x8000_0095, 31 bit is reserved,
+	 * but used for RLC_LB_CNTL configuration */
+	data = RLC_LB_CNTL__LB_CNT_SPIM_ACTIVE_MASK;
+	data |= REG_SET_FIELD(data, RLC_LB_CNTL, CU_MASK_USED_OFF_HYST, 0x09);
+	data |= REG_SET_FIELD(data, RLC_LB_CNTL, RESERVED, 0x80000);
+	WREG32_SOC15(GC, 0, mmRLC_LB_CNTL, data);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+static void gfx_v9_0_enable_lbpw(struct amdgpu_device *adev, bool enable)
+{
+	WREG32_FIELD15(GC, 0, RLC_LB_CNTL, LOAD_BALANCE_ENABLE, enable ? 1 : 0);
+}
+
+static void rv_init_cp_jump_table(struct amdgpu_device *adev)
+{
+	const __le32 *fw_data;
+	volatile u32 *dst_ptr;
+	int me, i, max_me = 5;
+	u32 bo_offset = 0;
+	u32 table_offset, table_size;
+
+	/* write the cp table buffer */
+	dst_ptr = adev->gfx.rlc.cp_table_ptr;
+	for (me = 0; me < max_me; me++) {
+		if (me == 0) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.ce_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 1) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.pfp_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 2) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.me_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 3) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else  if (me == 4) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec2_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		}
+
+		for (i = 0; i < table_size; i ++) {
+			dst_ptr[bo_offset + i] =
+				cpu_to_le32(le32_to_cpu(fw_data[table_offset + i]));
+		}
+
+		bo_offset += table_size;
+	}
+}
+
+static void gfx_v9_0_rlc_fini(struct amdgpu_device *adev)
+{
+	/* clear state block */
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj,
+			&adev->gfx.rlc.clear_state_gpu_addr,
+			(void **)&adev->gfx.rlc.cs_ptr);
+
+	/* jump table block */
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj,
+			&adev->gfx.rlc.cp_table_gpu_addr,
+			(void **)&adev->gfx.rlc.cp_table_ptr);
+}
+
+static int gfx_v9_0_rlc_init(struct amdgpu_device *adev)
+{
+	volatile u32 *dst_ptr;
+	u32 dws;
+	const struct cs_section_def *cs_data;
+	int r;
+
+	adev->gfx.rlc.cs_data = gfx9_cs_data;
+
+	cs_data = adev->gfx.rlc.cs_data;
+
+	if (cs_data) {
+		/* clear state block */
+		adev->gfx.rlc.clear_state_size = dws = gfx_v9_0_get_csb_size(adev);
+		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.clear_state_obj,
+					      &adev->gfx.rlc.clear_state_gpu_addr,
+					      (void **)&adev->gfx.rlc.cs_ptr);
+		if (r) {
+			dev_err(adev->dev, "(%d) failed to create rlc csb bo\n",
+				r);
+			gfx_v9_0_rlc_fini(adev);
+			return r;
+		}
+		/* set up the cs buffer */
+		dst_ptr = adev->gfx.rlc.cs_ptr;
+		gfx_v9_0_get_csb_buffer(adev, dst_ptr);
+		amdgpu_bo_kunmap(adev->gfx.rlc.clear_state_obj);
+		amdgpu_bo_unpin(adev->gfx.rlc.clear_state_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
+	}
+
+	if (adev->asic_type == CHIP_RAVEN) {
+		/* TODO: double check the cp_table_size for RV */
+		adev->gfx.rlc.cp_table_size = roundup2(96 * 5 * 4, 2048) + (64 * 1024); /* JT + GDS */
+		r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
+					      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.cp_table_obj,
+					      &adev->gfx.rlc.cp_table_gpu_addr,
+					      (void **)&adev->gfx.rlc.cp_table_ptr);
+		if (r) {
+			dev_err(adev->dev,
+				"(%d) failed to create cp table bo\n", r);
+			gfx_v9_0_rlc_fini(adev);
+			return r;
+		}
+
+		rv_init_cp_jump_table(adev);
+		amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
+
+		gfx_v9_0_init_lbpw(adev);
+	}
+
+	return 0;
+}
+
+static int gfx_v9_0_csb_vram_pin(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, false);
+	if (unlikely(r != 0))
+		return r;
+
+	r = amdgpu_bo_pin(adev->gfx.rlc.clear_state_obj,
+			AMDGPU_GEM_DOMAIN_VRAM);
+	if (!r)
+		adev->gfx.rlc.clear_state_gpu_addr =
+			amdgpu_bo_gpu_offset(adev->gfx.rlc.clear_state_obj);
+
+	amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
+
+	return r;
+}
+
+static void gfx_v9_0_csb_vram_unpin(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!adev->gfx.rlc.clear_state_obj)
+		return;
+
+	r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, true);
+	if (likely(r == 0)) {
+		amdgpu_bo_unpin(adev->gfx.rlc.clear_state_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
+	}
+}
+
+static void gfx_v9_0_mec_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.mec.mec_fw_obj, NULL, NULL);
+}
+
+static int gfx_v9_0_mec_init(struct amdgpu_device *adev)
+{
+	int r;
+	u32 *hpd;
+	const __le32 *fw_data;
+	unsigned fw_size;
+	u32 *fw;
+	size_t mec_hpd_size;
+
+	const struct gfx_firmware_header_v1_0 *mec_hdr;
+
+	bitmap_zero(adev->gfx.mec.queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
+
+	/* take ownership of the relevant compute queues */
+	amdgpu_gfx_compute_queue_acquire(adev);
+	mec_hpd_size = adev->gfx.num_compute_rings * GFX9_MEC_HPD_SIZE;
+
+	r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.mec.hpd_eop_obj,
+				      &adev->gfx.mec.hpd_eop_gpu_addr,
+				      (void **)&hpd);
+	if (r) {
+		dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
+		gfx_v9_0_mec_fini(adev);
+		return r;
+	}
+
+	memset(hpd, 0, adev->gfx.mec.hpd_eop_obj->tbo.mem.size);
+
+	amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
+	amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+
+	mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+
+	fw_data = (const __le32 *)
+		(adev->gfx.mec_fw->data +
+		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes) / 4;
+
+	r = amdgpu_bo_create_reserved(adev, mec_hdr->header.ucode_size_bytes,
+				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.mec.mec_fw_obj,
+				      &adev->gfx.mec.mec_fw_gpu_addr,
+				      (void **)&fw);
+	if (r) {
+		dev_warn(adev->dev, "(%d) create mec firmware bo failed\n", r);
+		gfx_v9_0_mec_fini(adev);
+		return r;
+	}
+
+	memcpy(fw, fw_data, fw_size);
+
+	amdgpu_bo_kunmap(adev->gfx.mec.mec_fw_obj);
+	amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_obj);
+
+	return 0;
+}
+
+static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t address)
+{
+	WREG32_SOC15(GC, 0, mmSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(address << SQ_IND_INDEX__INDEX__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK));
+	return RREG32_SOC15(GC, 0, mmSQ_IND_DATA);
+}
+
+static void wave_read_regs(struct amdgpu_device *adev, uint32_t simd,
+			   uint32_t wave, uint32_t thread,
+			   uint32_t regno, uint32_t num, uint32_t *out)
+{
+	WREG32_SOC15(GC, 0, mmSQ_IND_INDEX,
+		(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
+		(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
+		(regno << SQ_IND_INDEX__INDEX__SHIFT) |
+		(thread << SQ_IND_INDEX__THREAD_ID__SHIFT) |
+		(SQ_IND_INDEX__FORCE_READ_MASK) |
+		(SQ_IND_INDEX__AUTO_INCR_MASK));
+	while (num--)
+		*(out++) = RREG32_SOC15(GC, 0, mmSQ_IND_DATA);
+}
+
+static void gfx_v9_0_read_wave_data(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields)
+{
+	/* type 1 wave data */
+	dst[(*no_fields)++] = 1;
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_STATUS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_PC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_LO);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_EXEC_HI);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_HW_ID);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_INST_DW1);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_GPR_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_LDS_ALLOC);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_TRAPSTS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_STS);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_IB_DBG0);
+	dst[(*no_fields)++] = wave_read_ind(adev, simd, wave, ixSQ_WAVE_M0);
+}
+
+static void gfx_v9_0_read_wave_sgprs(struct amdgpu_device *adev, uint32_t simd,
+				     uint32_t wave, uint32_t start,
+				     uint32_t size, uint32_t *dst)
+{
+	wave_read_regs(
+		adev, simd, wave, 0,
+		start + SQIND_WAVE_SGPRS_OFFSET, size, dst);
+}
+
+static void gfx_v9_0_read_wave_vgprs(struct amdgpu_device *adev, uint32_t simd,
+				     uint32_t wave, uint32_t thread,
+				     uint32_t start, uint32_t size,
+				     uint32_t *dst)
+{
+	wave_read_regs(
+		adev, simd, wave, thread,
+		start + SQIND_WAVE_VGPRS_OFFSET, size, dst);
+}
+
+static void gfx_v9_0_select_me_pipe_q(struct amdgpu_device *adev,
+				  u32 me, u32 pipe, u32 q)
+{
+	soc15_grbm_select(adev, me, pipe, q, 0);
+}
+
+static const struct amdgpu_gfx_funcs gfx_v9_0_gfx_funcs = {
+	.get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter,
+	.select_se_sh = &gfx_v9_0_select_se_sh,
+	.read_wave_data = &gfx_v9_0_read_wave_data,
+	.read_wave_sgprs = &gfx_v9_0_read_wave_sgprs,
+	.read_wave_vgprs = &gfx_v9_0_read_wave_vgprs,
+	.select_me_pipe_q = &gfx_v9_0_select_me_pipe_q
+};
+
+static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev)
+{
+	u32 gb_addr_config;
+	int err;
+
+	adev->gfx.funcs = &gfx_v9_0_gfx_funcs;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		gb_addr_config = VEGA10_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_VEGA12:
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		gb_addr_config = VEGA12_GB_ADDR_CONFIG_GOLDEN;
+		DRM_INFO("fix gfx.config for vega12\n");
+		break;
+	case CHIP_VEGA20:
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		gb_addr_config = RREG32_SOC15(GC, 0, mmGB_ADDR_CONFIG);
+		gb_addr_config &= ~0xf3e777ff;
+		gb_addr_config |= 0x22014042;
+		/* check vbios table if gpu info is not available */
+		err = amdgpu_atomfirmware_get_gfx_info(adev);
+		if (err)
+			return err;
+		break;
+	case CHIP_RAVEN:
+		adev->gfx.config.max_hw_contexts = 8;
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x4C0;
+		gb_addr_config = RAVEN_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	default:
+		BUG();
+		break;
+	}
+
+	adev->gfx.config.gb_addr_config = gb_addr_config;
+
+	adev->gfx.config.gb_addr_config_fields.num_pipes = 1 <<
+			REG_GET_FIELD(
+					adev->gfx.config.gb_addr_config,
+					GB_ADDR_CONFIG,
+					NUM_PIPES);
+
+	adev->gfx.config.max_tile_pipes =
+		adev->gfx.config.gb_addr_config_fields.num_pipes;
+
+	adev->gfx.config.gb_addr_config_fields.num_banks = 1 <<
+			REG_GET_FIELD(
+					adev->gfx.config.gb_addr_config,
+					GB_ADDR_CONFIG,
+					NUM_BANKS);
+	adev->gfx.config.gb_addr_config_fields.max_compress_frags = 1 <<
+			REG_GET_FIELD(
+					adev->gfx.config.gb_addr_config,
+					GB_ADDR_CONFIG,
+					MAX_COMPRESSED_FRAGS);
+	adev->gfx.config.gb_addr_config_fields.num_rb_per_se = 1 <<
+			REG_GET_FIELD(
+					adev->gfx.config.gb_addr_config,
+					GB_ADDR_CONFIG,
+					NUM_RB_PER_SE);
+	adev->gfx.config.gb_addr_config_fields.num_se = 1 <<
+			REG_GET_FIELD(
+					adev->gfx.config.gb_addr_config,
+					GB_ADDR_CONFIG,
+					NUM_SHADER_ENGINES);
+	adev->gfx.config.gb_addr_config_fields.pipe_interleave_size = 1 << (8 +
+			REG_GET_FIELD(
+					adev->gfx.config.gb_addr_config,
+					GB_ADDR_CONFIG,
+					PIPE_INTERLEAVE_SIZE));
+
+	return 0;
+}
+
+static int gfx_v9_0_ngg_create_buf(struct amdgpu_device *adev,
+				   struct amdgpu_ngg_buf *ngg_buf,
+				   int size_se,
+				   int default_size_se)
+{
+	int r;
+
+	if (size_se < 0) {
+		dev_err(adev->dev, "Buffer size is invalid: %d\n", size_se);
+		return -EINVAL;
+	}
+	size_se = size_se ? size_se : default_size_se;
+
+	ngg_buf->size = size_se * adev->gfx.config.max_shader_engines;
+	r = amdgpu_bo_create_kernel(adev, ngg_buf->size,
+				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
+				    &ngg_buf->bo,
+				    &ngg_buf->gpu_addr,
+				    NULL);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to create NGG buffer\n", r);
+		return r;
+	}
+	ngg_buf->bo_size = amdgpu_bo_size(ngg_buf->bo);
+
+	return r;
+}
+
+static int gfx_v9_0_ngg_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < NGG_BUF_MAX; i++)
+		amdgpu_bo_free_kernel(&adev->gfx.ngg.buf[i].bo,
+				      &adev->gfx.ngg.buf[i].gpu_addr,
+				      NULL);
+
+	memset(&adev->gfx.ngg.buf[0], 0,
+			sizeof(struct amdgpu_ngg_buf) * NGG_BUF_MAX);
+
+	adev->gfx.ngg.init = false;
+
+	return 0;
+}
+
+static int gfx_v9_0_ngg_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!amdgpu_ngg || adev->gfx.ngg.init == true)
+		return 0;
+
+	/* GDS reserve memory: 64 bytes alignment */
+	adev->gfx.ngg.gds_reserve_size = roundup2(5 * 4, 0x40);
+	adev->gds.mem.total_size -= adev->gfx.ngg.gds_reserve_size;
+	adev->gds.mem.gfx_partition_size -= adev->gfx.ngg.gds_reserve_size;
+	adev->gfx.ngg.gds_reserve_addr = RREG32_SOC15(GC, 0, mmGDS_VMID0_BASE);
+	adev->gfx.ngg.gds_reserve_addr += RREG32_SOC15(GC, 0, mmGDS_VMID0_SIZE);
+
+	/* Primitive Buffer */
+	r = gfx_v9_0_ngg_create_buf(adev, &adev->gfx.ngg.buf[NGG_PRIM],
+				    amdgpu_prim_buf_per_se,
+				    64 * 1024);
+	if (r) {
+		dev_err(adev->dev, "Failed to create Primitive Buffer\n");
+		goto err;
+	}
+
+	/* Position Buffer */
+	r = gfx_v9_0_ngg_create_buf(adev, &adev->gfx.ngg.buf[NGG_POS],
+				    amdgpu_pos_buf_per_se,
+				    256 * 1024);
+	if (r) {
+		dev_err(adev->dev, "Failed to create Position Buffer\n");
+		goto err;
+	}
+
+	/* Control Sideband */
+	r = gfx_v9_0_ngg_create_buf(adev, &adev->gfx.ngg.buf[NGG_CNTL],
+				    amdgpu_cntl_sb_buf_per_se,
+				    256);
+	if (r) {
+		dev_err(adev->dev, "Failed to create Control Sideband Buffer\n");
+		goto err;
+	}
+
+	/* Parameter Cache, not created by default */
+	if (amdgpu_param_buf_per_se <= 0)
+		goto out;
+
+	r = gfx_v9_0_ngg_create_buf(adev, &adev->gfx.ngg.buf[NGG_PARAM],
+				    amdgpu_param_buf_per_se,
+				    512 * 1024);
+	if (r) {
+		dev_err(adev->dev, "Failed to create Parameter Cache\n");
+		goto err;
+	}
+
+out:
+	adev->gfx.ngg.init = true;
+	return 0;
+err:
+	gfx_v9_0_ngg_fini(adev);
+	return r;
+}
+
+static int gfx_v9_0_ngg_en(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->gfx.gfx_ring[0];
+	int r;
+	u32 data, base;
+
+	if (!amdgpu_ngg)
+		return 0;
+
+	/* Program buffer size */
+	data = REG_SET_FIELD(0, WD_BUF_RESOURCE_1, INDEX_BUF_SIZE,
+			     adev->gfx.ngg.buf[NGG_PRIM].size >> 8);
+	data = REG_SET_FIELD(data, WD_BUF_RESOURCE_1, POS_BUF_SIZE,
+			     adev->gfx.ngg.buf[NGG_POS].size >> 8);
+	WREG32_SOC15(GC, 0, mmWD_BUF_RESOURCE_1, data);
+
+	data = REG_SET_FIELD(0, WD_BUF_RESOURCE_2, CNTL_SB_BUF_SIZE,
+			     adev->gfx.ngg.buf[NGG_CNTL].size >> 8);
+	data = REG_SET_FIELD(data, WD_BUF_RESOURCE_2, PARAM_BUF_SIZE,
+			     adev->gfx.ngg.buf[NGG_PARAM].size >> 10);
+	WREG32_SOC15(GC, 0, mmWD_BUF_RESOURCE_2, data);
+
+	/* Program buffer base address */
+	base = lower_32_bits(adev->gfx.ngg.buf[NGG_PRIM].gpu_addr);
+	data = REG_SET_FIELD(0, WD_INDEX_BUF_BASE, BASE, base);
+	WREG32_SOC15(GC, 0, mmWD_INDEX_BUF_BASE, data);
+
+	base = upper_32_bits(adev->gfx.ngg.buf[NGG_PRIM].gpu_addr);
+	data = REG_SET_FIELD(0, WD_INDEX_BUF_BASE_HI, BASE_HI, base);
+	WREG32_SOC15(GC, 0, mmWD_INDEX_BUF_BASE_HI, data);
+
+	base = lower_32_bits(adev->gfx.ngg.buf[NGG_POS].gpu_addr);
+	data = REG_SET_FIELD(0, WD_POS_BUF_BASE, BASE, base);
+	WREG32_SOC15(GC, 0, mmWD_POS_BUF_BASE, data);
+
+	base = upper_32_bits(adev->gfx.ngg.buf[NGG_POS].gpu_addr);
+	data = REG_SET_FIELD(0, WD_POS_BUF_BASE_HI, BASE_HI, base);
+	WREG32_SOC15(GC, 0, mmWD_POS_BUF_BASE_HI, data);
+
+	base = lower_32_bits(adev->gfx.ngg.buf[NGG_CNTL].gpu_addr);
+	data = REG_SET_FIELD(0, WD_CNTL_SB_BUF_BASE, BASE, base);
+	WREG32_SOC15(GC, 0, mmWD_CNTL_SB_BUF_BASE, data);
+
+	base = upper_32_bits(adev->gfx.ngg.buf[NGG_CNTL].gpu_addr);
+	data = REG_SET_FIELD(0, WD_CNTL_SB_BUF_BASE_HI, BASE_HI, base);
+	WREG32_SOC15(GC, 0, mmWD_CNTL_SB_BUF_BASE_HI, data);
+
+	/* Clear GDS reserved memory */
+	r = amdgpu_ring_alloc(ring, 17);
+	if (r) {
+		DRM_ERROR("amdgpu: NGG failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+
+	gfx_v9_0_write_data_to_reg(ring, 0, false,
+				   SOC15_REG_OFFSET(GC, 0, mmGDS_VMID0_SIZE),
+			           (adev->gds.mem.total_size +
+				    adev->gfx.ngg.gds_reserve_size) >>
+				   AMDGPU_GDS_SHIFT);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_DMA_DATA, 5));
+	amdgpu_ring_write(ring, (PACKET3_DMA_DATA_CP_SYNC |
+				PACKET3_DMA_DATA_DST_SEL(1) |
+				PACKET3_DMA_DATA_SRC_SEL(2)));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, adev->gfx.ngg.gds_reserve_addr);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET3_DMA_DATA_CMD_RAW_WAIT |
+				adev->gfx.ngg.gds_reserve_size);
+
+	gfx_v9_0_write_data_to_reg(ring, 0, false,
+				   SOC15_REG_OFFSET(GC, 0, mmGDS_VMID0_SIZE), 0);
+
+	amdgpu_ring_commit(ring);
+
+	return 0;
+}
+
+static int gfx_v9_0_compute_ring_init(struct amdgpu_device *adev, int ring_id,
+				      int mec, int pipe, int queue)
+{
+	int r;
+	unsigned irq_type;
+	struct amdgpu_ring *ring = &adev->gfx.compute_ring[ring_id];
+
+	ring = &adev->gfx.compute_ring[ring_id];
+
+	/* mec0 is me1 */
+	ring->me = mec + 1;
+	ring->pipe = pipe;
+	ring->queue = queue;
+
+	ring->ring_obj = NULL;
+	ring->use_doorbell = true;
+	ring->doorbell_index = (AMDGPU_DOORBELL_MEC_RING0 + ring_id) << 1;
+	ring->eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr
+				+ (ring_id * GFX9_MEC_HPD_SIZE);
+	snprintf(ring->name, sizeof(ring->name), "comp_%d.%d.%d", ring->me, ring->pipe, ring->queue);
+
+	irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP
+		+ ((ring->me - 1) * adev->gfx.mec.num_pipe_per_mec)
+		+ ring->pipe;
+
+	/* type-2 packets are deprecated on MEC, use type-3 instead */
+	r = amdgpu_ring_init(adev, ring, 1024,
+			     &adev->gfx.eop_irq, irq_type);
+	if (r)
+		return r;
+
+
+	return 0;
+}
+
+static int gfx_v9_0_sw_init(void *handle)
+{
+	int i, j, k, r, ring_id;
+	struct amdgpu_ring *ring;
+	struct amdgpu_kiq *kiq;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+	case CHIP_RAVEN:
+		adev->gfx.mec.num_mec = 2;
+		break;
+	default:
+		adev->gfx.mec.num_mec = 1;
+		break;
+	}
+
+	adev->gfx.mec.num_pipe_per_mec = 4;
+	adev->gfx.mec.num_queue_per_pipe = 8;
+
+	/* KIQ event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_9_0__SRCID__CP_IB2_INTERRUPT_PKT, &adev->gfx.kiq.irq);
+	if (r)
+		return r;
+
+	/* EOP Event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_9_0__SRCID__CP_EOP_INTERRUPT, &adev->gfx.eop_irq);
+	if (r)
+		return r;
+
+	/* Privileged reg */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_9_0__SRCID__CP_PRIV_REG_FAULT,
+			      &adev->gfx.priv_reg_irq);
+	if (r)
+		return r;
+
+	/* Privileged inst */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_GRBM_CP, GFX_9_0__SRCID__CP_PRIV_INSTR_FAULT,
+			      &adev->gfx.priv_inst_irq);
+	if (r)
+		return r;
+
+	adev->gfx.gfx_current_status = AMDGPU_GFX_NORMAL_MODE;
+
+	gfx_v9_0_scratch_init(adev);
+
+	r = gfx_v9_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load gfx firmware!\n");
+		return r;
+	}
+
+	r = gfx_v9_0_rlc_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init rlc BOs!\n");
+		return r;
+	}
+
+	r = gfx_v9_0_mec_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init MEC BOs!\n");
+		return r;
+	}
+
+	/* set up the gfx ring */
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		ring = &adev->gfx.gfx_ring[i];
+		ring->ring_obj = NULL;
+		if (!i)
+			snprintf(ring->name, sizeof(ring->name), "gfx");
+		else
+			snprintf(ring->name, sizeof(ring->name), "gfx_%d", i);
+		ring->use_doorbell = true;
+		ring->doorbell_index = AMDGPU_DOORBELL64_GFX_RING0 << 1;
+		r = amdgpu_ring_init(adev, ring, 1024,
+				     &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP);
+		if (r)
+			return r;
+	}
+
+	/* set up the compute queues - allocate horizontally across pipes */
+	ring_id = 0;
+	for (i = 0; i < adev->gfx.mec.num_mec; ++i) {
+		for (j = 0; j < adev->gfx.mec.num_queue_per_pipe; j++) {
+			for (k = 0; k < adev->gfx.mec.num_pipe_per_mec; k++) {
+				if (!amdgpu_gfx_is_mec_queue_enabled(adev, i, k, j))
+					continue;
+
+				r = gfx_v9_0_compute_ring_init(adev,
+							       ring_id,
+							       i, k, j);
+				if (r)
+					return r;
+
+				ring_id++;
+			}
+		}
+	}
+
+	r = amdgpu_gfx_kiq_init(adev, GFX9_MEC_HPD_SIZE);
+	if (r) {
+		DRM_ERROR("Failed to init KIQ BOs!\n");
+		return r;
+	}
+
+	kiq = &adev->gfx.kiq;
+	r = amdgpu_gfx_kiq_init_ring(adev, &kiq->ring, &kiq->irq);
+	if (r)
+		return r;
+
+	/* create MQD for all compute queues as wel as KIQ for SRIOV case */
+	r = amdgpu_gfx_compute_mqd_sw_init(adev, sizeof(struct v9_mqd_allocation));
+	if (r)
+		return r;
+
+	/* reserve GDS, GWS and OA resource for gfx */
+	r = amdgpu_bo_create_kernel(adev, adev->gds.mem.gfx_partition_size,
+				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GDS,
+				    &adev->gds.gds_gfx_bo, NULL, NULL);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_create_kernel(adev, adev->gds.gws.gfx_partition_size,
+				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GWS,
+				    &adev->gds.gws_gfx_bo, NULL, NULL);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_create_kernel(adev, adev->gds.oa.gfx_partition_size,
+				    PAGE_SIZE, AMDGPU_GEM_DOMAIN_OA,
+				    &adev->gds.oa_gfx_bo, NULL, NULL);
+	if (r)
+		return r;
+
+	adev->gfx.ce_ram_size = 0x8000;
+
+	r = gfx_v9_0_gpu_early_init(adev);
+	if (r)
+		return r;
+
+	r = gfx_v9_0_ngg_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+
+static int gfx_v9_0_sw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_bo_free_kernel(&adev->gds.oa_gfx_bo, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gds.gws_gfx_bo, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gds.gds_gfx_bo, NULL, NULL);
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
+
+	amdgpu_gfx_compute_mqd_sw_fini(adev);
+	amdgpu_gfx_kiq_free_ring(&adev->gfx.kiq.ring, &adev->gfx.kiq.irq);
+	amdgpu_gfx_kiq_fini(adev);
+
+	gfx_v9_0_mec_fini(adev);
+	gfx_v9_0_ngg_fini(adev);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj,
+				&adev->gfx.rlc.clear_state_gpu_addr,
+				(void **)&adev->gfx.rlc.cs_ptr);
+	if (adev->asic_type == CHIP_RAVEN) {
+		amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj,
+				&adev->gfx.rlc.cp_table_gpu_addr,
+				(void **)&adev->gfx.rlc.cp_table_ptr);
+	}
+	gfx_v9_0_free_microcode(adev);
+
+	return 0;
+}
+
+
+static void gfx_v9_0_tiling_mode_table_init(struct amdgpu_device *adev)
+{
+	/* TODO */
+}
+
+static void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 instance)
+{
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX, instance);
+
+	if (se_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
+
+	if (sh_num == 0xffffffff)
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
+
+	WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, data);
+}
+
+static u32 gfx_v9_0_get_rb_active_bitmap(struct amdgpu_device *adev)
+{
+	u32 data, mask;
+
+	data = RREG32_SOC15(GC, 0, mmCC_RB_BACKEND_DISABLE);
+	data |= RREG32_SOC15(GC, 0, mmGC_USER_RB_BACKEND_DISABLE);
+
+	data &= CC_RB_BACKEND_DISABLE__BACKEND_DISABLE_MASK;
+	data >>= GC_USER_RB_BACKEND_DISABLE__BACKEND_DISABLE__SHIFT;
+
+	mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_backends_per_se /
+					 adev->gfx.config.max_sh_per_se);
+
+	return (~data) & mask;
+}
+
+static void gfx_v9_0_setup_rb(struct amdgpu_device *adev)
+{
+	int i, j;
+	u32 data;
+	u32 active_rbs = 0;
+	u32 rb_bitmap_width_per_sh = adev->gfx.config.max_backends_per_se /
+					adev->gfx.config.max_sh_per_se;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v9_0_select_se_sh(adev, i, j, 0xffffffff);
+			data = gfx_v9_0_get_rb_active_bitmap(adev);
+			active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
+					       rb_bitmap_width_per_sh);
+		}
+	}
+	gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	adev->gfx.config.backend_enable_mask = active_rbs;
+	adev->gfx.config.num_rbs = hweight32(active_rbs);
+}
+
+#define DEFAULT_SH_MEM_BASES	(0x6000)
+#define FIRST_COMPUTE_VMID	(8)
+#define LAST_COMPUTE_VMID	(16)
+static void gfx_v9_0_init_compute_vmid(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t sh_mem_config;
+	uint32_t sh_mem_bases;
+
+	/*
+	 * Configure apertures:
+	 * LDS:         0x60000000'00000000 - 0x60000001'00000000 (4GB)
+	 * Scratch:     0x60000001'00000000 - 0x60000002'00000000 (4GB)
+	 * GPUVM:       0x60010000'00000000 - 0x60020000'00000000 (1TB)
+	 */
+	sh_mem_bases = DEFAULT_SH_MEM_BASES | (DEFAULT_SH_MEM_BASES << 16);
+
+	sh_mem_config = SH_MEM_ADDRESS_MODE_64 |
+			SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
+			SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT;
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = FIRST_COMPUTE_VMID; i < LAST_COMPUTE_VMID; i++) {
+		soc15_grbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, sh_mem_config);
+		WREG32_SOC15(GC, 0, mmSH_MEM_BASES, sh_mem_bases);
+	}
+	soc15_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void gfx_v9_0_gpu_init(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	int i;
+
+	WREG32_FIELD15(GC, 0, GRBM_CNTL, READ_TIMEOUT, 0xff);
+
+	gfx_v9_0_tiling_mode_table_init(adev);
+
+	gfx_v9_0_setup_rb(adev);
+	gfx_v9_0_get_cu_info(adev, &adev->gfx.cu_info);
+	adev->gfx.config.db_debug2 = RREG32_SOC15(GC, 0, mmDB_DEBUG2);
+
+	/* XXX SH_MEM regs */
+	/* where to put LDS, scratch, GPUVM in FSA64 space */
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < adev->vm_manager.id_mgr[AMDGPU_GFXHUB].num_ids; i++) {
+		soc15_grbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		if (i == 0) {
+			tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE,
+					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
+			WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, tmp);
+			WREG32_SOC15(GC, 0, mmSH_MEM_BASES, 0);
+		} else {
+			tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE,
+					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
+			WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, tmp);
+			tmp = REG_SET_FIELD(0, SH_MEM_BASES, PRIVATE_BASE,
+				(adev->gmc.private_aperture_start >> 48));
+			tmp = REG_SET_FIELD(tmp, SH_MEM_BASES, SHARED_BASE,
+				(adev->gmc.shared_aperture_start >> 48));
+			WREG32_SOC15(GC, 0, mmSH_MEM_BASES, tmp);
+		}
+	}
+	soc15_grbm_select(adev, 0, 0, 0, 0);
+
+	mutex_unlock(&adev->srbm_mutex);
+
+	gfx_v9_0_init_compute_vmid(adev);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	/*
+	 * making sure that the following register writes will be broadcasted
+	 * to all the shaders
+	 */
+	gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+
+	WREG32_SOC15(GC, 0, mmPA_SC_FIFO_SIZE,
+		   (adev->gfx.config.sc_prim_fifo_size_frontend <<
+			PA_SC_FIFO_SIZE__SC_FRONTEND_PRIM_FIFO_SIZE__SHIFT) |
+		   (adev->gfx.config.sc_prim_fifo_size_backend <<
+			PA_SC_FIFO_SIZE__SC_BACKEND_PRIM_FIFO_SIZE__SHIFT) |
+		   (adev->gfx.config.sc_hiz_tile_fifo_size <<
+			PA_SC_FIFO_SIZE__SC_HIZ_TILE_FIFO_SIZE__SHIFT) |
+		   (adev->gfx.config.sc_earlyz_tile_fifo_size <<
+			PA_SC_FIFO_SIZE__SC_EARLYZ_TILE_FIFO_SIZE__SHIFT));
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+}
+
+static void gfx_v9_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
+{
+	u32 i, j, k;
+	u32 mask;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v9_0_select_se_sh(adev, i, j, 0xffffffff);
+			for (k = 0; k < adev->usec_timeout; k++) {
+				if (RREG32_SOC15(GC, 0, mmRLC_SERDES_CU_MASTER_BUSY) == 0)
+					break;
+				udelay(1);
+			}
+			if (k == adev->usec_timeout) {
+				gfx_v9_0_select_se_sh(adev, 0xffffffff,
+						      0xffffffff, 0xffffffff);
+				mutex_unlock(&adev->grbm_idx_mutex);
+				DRM_INFO("Timeout wait for RLC serdes %u,%u\n",
+					 i, j);
+				return;
+			}
+		}
+	}
+	gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__GC_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC0_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC1_MASTER_BUSY_MASK;
+	for (k = 0; k < adev->usec_timeout; k++) {
+		if ((RREG32_SOC15(GC, 0, mmRLC_SERDES_NONCU_MASTER_BUSY) & mask) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v9_0_enable_gui_idle_interrupt(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 tmp = RREG32_SOC15(GC, 0, mmCP_INT_CNTL_RING0);
+
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE, enable ? 1 : 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE, enable ? 1 : 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE, enable ? 1 : 0);
+	tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE, enable ? 1 : 0);
+
+	WREG32_SOC15(GC, 0, mmCP_INT_CNTL_RING0, tmp);
+}
+
+static void gfx_v9_0_init_csb(struct amdgpu_device *adev)
+{
+	/* csib */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_CSIB_ADDR_HI),
+			adev->gfx.rlc.clear_state_gpu_addr >> 32);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_CSIB_ADDR_LO),
+			adev->gfx.rlc.clear_state_gpu_addr & 0xfffffffc);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_CSIB_LENGTH),
+			adev->gfx.rlc.clear_state_size);
+}
+
+static void gfx_v9_1_parse_ind_reg_list(int *register_list_format,
+				int indirect_offset,
+				int list_size,
+				int *unique_indirect_regs,
+				int unique_indirect_reg_count,
+				int *indirect_start_offsets,
+				int *indirect_start_offsets_count,
+				int max_start_offsets_count)
+{
+	int idx;
+
+	for (; indirect_offset < list_size; indirect_offset++) {
+		WARN_ON(*indirect_start_offsets_count >= max_start_offsets_count);
+		indirect_start_offsets[*indirect_start_offsets_count] = indirect_offset;
+		*indirect_start_offsets_count = *indirect_start_offsets_count + 1;
+
+		while (register_list_format[indirect_offset] != 0xFFFFFFFF) {
+			indirect_offset += 2;
+
+			/* look for the matching indice */
+			for (idx = 0; idx < unique_indirect_reg_count; idx++) {
+				if (unique_indirect_regs[idx] ==
+					register_list_format[indirect_offset] ||
+					!unique_indirect_regs[idx])
+					break;
+			}
+
+			BUG_ON(idx >= unique_indirect_reg_count);
+
+			if (!unique_indirect_regs[idx])
+				unique_indirect_regs[idx] = register_list_format[indirect_offset];
+
+			indirect_offset++;
+		}
+	}
+}
+
+static int gfx_v9_1_init_rlc_save_restore_list(struct amdgpu_device *adev)
+{
+	int unique_indirect_regs[] = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0};
+	int unique_indirect_reg_count = 0;
+
+	int indirect_start_offsets[] = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0};
+	int indirect_start_offsets_count = 0;
+
+	int list_size = 0;
+	int i = 0, j = 0;
+	u32 tmp = 0;
+
+	u32 *register_list_format =
+		kmalloc(adev->gfx.rlc.reg_list_format_size_bytes, GFP_KERNEL);
+	if (!register_list_format)
+		return -ENOMEM;
+	memcpy(register_list_format, adev->gfx.rlc.register_list_format,
+		adev->gfx.rlc.reg_list_format_size_bytes);
+
+	/* setup unique_indirect_regs array and indirect_start_offsets array */
+	unique_indirect_reg_count = ARRAY_SIZE(unique_indirect_regs);
+	gfx_v9_1_parse_ind_reg_list(register_list_format,
+				    adev->gfx.rlc.reg_list_format_direct_reg_list_length,
+				    adev->gfx.rlc.reg_list_format_size_bytes >> 2,
+				    unique_indirect_regs,
+				    unique_indirect_reg_count,
+				    indirect_start_offsets,
+				    &indirect_start_offsets_count,
+				    ARRAY_SIZE(indirect_start_offsets));
+
+	/* enable auto inc in case it is disabled */
+	tmp = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_CNTL));
+	tmp |= RLC_SRM_CNTL__AUTO_INCR_ADDR_MASK;
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_CNTL), tmp);
+
+	/* write register_restore table to offset 0x0 using RLC_SRM_ARAM_ADDR/DATA */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_ARAM_ADDR),
+		RLC_SAVE_RESTORE_ADDR_STARTING_OFFSET);
+	for (i = 0; i < adev->gfx.rlc.reg_list_size_bytes >> 2; i++)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_ARAM_DATA),
+			adev->gfx.rlc.register_restore[i]);
+
+	/* load indirect register */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_SCRATCH_ADDR),
+		adev->gfx.rlc.reg_list_format_start);
+
+	/* direct register portion */
+	for (i = 0; i < adev->gfx.rlc.reg_list_format_direct_reg_list_length; i++)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_SCRATCH_DATA),
+			register_list_format[i]);
+
+	/* indirect register portion */
+	while (i < (adev->gfx.rlc.reg_list_format_size_bytes >> 2)) {
+		if (register_list_format[i] == 0xFFFFFFFF) {
+			WREG32_SOC15(GC, 0, mmRLC_GPM_SCRATCH_DATA, register_list_format[i++]);
+			continue;
+		}
+
+		WREG32_SOC15(GC, 0, mmRLC_GPM_SCRATCH_DATA, register_list_format[i++]);
+		WREG32_SOC15(GC, 0, mmRLC_GPM_SCRATCH_DATA, register_list_format[i++]);
+
+		for (j = 0; j < unique_indirect_reg_count; j++) {
+			if (register_list_format[i] == unique_indirect_regs[j]) {
+				WREG32_SOC15(GC, 0, mmRLC_GPM_SCRATCH_DATA, j);
+				break;
+			}
+		}
+
+		BUG_ON(j >= unique_indirect_reg_count);
+
+		i++;
+	}
+
+	/* set save/restore list size */
+	list_size = adev->gfx.rlc.reg_list_size_bytes >> 2;
+	list_size = list_size >> 1;
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_SCRATCH_ADDR),
+		adev->gfx.rlc.reg_restore_list_size);
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_SCRATCH_DATA), list_size);
+
+	/* write the starting offsets to RLC scratch ram */
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_SCRATCH_ADDR),
+		adev->gfx.rlc.starting_offsets_start);
+	for (i = 0; i < ARRAY_SIZE(indirect_start_offsets); i++)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_SCRATCH_DATA),
+		       indirect_start_offsets[i]);
+
+	/* load unique indirect regs*/
+	for (i = 0; i < ARRAY_SIZE(unique_indirect_regs); i++) {
+		if (unique_indirect_regs[i] != 0) {
+			WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_INDEX_CNTL_ADDR_0)
+			       + GFX_RLC_SRM_INDEX_CNTL_ADDR_OFFSETS[i],
+			       unique_indirect_regs[i] & 0x3FFFF);
+
+			WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_SRM_INDEX_CNTL_DATA_0)
+			       + GFX_RLC_SRM_INDEX_CNTL_DATA_OFFSETS[i],
+			       unique_indirect_regs[i] >> 20);
+		}
+	}
+
+	kfree(register_list_format);
+	return 0;
+}
+
+static void gfx_v9_0_enable_save_restore_machine(struct amdgpu_device *adev)
+{
+	WREG32_FIELD15(GC, 0, RLC_SRM_CNTL, SRM_ENABLE, 1);
+}
+
+static void pwr_10_0_gfxip_control_over_cgpg(struct amdgpu_device *adev,
+					     bool enable)
+{
+	uint32_t data = 0;
+	uint32_t default_data = 0;
+
+	default_data = data = RREG32(SOC15_REG_OFFSET(PWR, 0, mmPWR_MISC_CNTL_STATUS));
+	if (enable == true) {
+		/* enable GFXIP control over CGPG */
+		data |= PWR_MISC_CNTL_STATUS__PWR_GFX_RLC_CGPG_EN_MASK;
+		if(default_data != data)
+			WREG32(SOC15_REG_OFFSET(PWR, 0, mmPWR_MISC_CNTL_STATUS), data);
+
+		/* update status */
+		data &= ~PWR_MISC_CNTL_STATUS__PWR_GFXOFF_STATUS_MASK;
+		data |= (2 << PWR_MISC_CNTL_STATUS__PWR_GFXOFF_STATUS__SHIFT);
+		if(default_data != data)
+			WREG32(SOC15_REG_OFFSET(PWR, 0, mmPWR_MISC_CNTL_STATUS), data);
+	} else {
+		/* restore GFXIP control over GCPG */
+		data &= ~PWR_MISC_CNTL_STATUS__PWR_GFX_RLC_CGPG_EN_MASK;
+		if(default_data != data)
+			WREG32(SOC15_REG_OFFSET(PWR, 0, mmPWR_MISC_CNTL_STATUS), data);
+	}
+}
+
+static void gfx_v9_0_init_gfx_power_gating(struct amdgpu_device *adev)
+{
+	uint32_t data = 0;
+
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_PG |
+			      AMD_PG_SUPPORT_GFX_SMG |
+			      AMD_PG_SUPPORT_GFX_DMG)) {
+		/* init IDLE_POLL_COUNT = 60 */
+		data = RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_RB_WPTR_POLL_CNTL));
+		data &= ~CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT_MASK;
+		data |= (0x60 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT);
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_RB_WPTR_POLL_CNTL), data);
+
+		/* init RLC PG Delay */
+		data = 0;
+		data |= (0x10 << RLC_PG_DELAY__POWER_UP_DELAY__SHIFT);
+		data |= (0x10 << RLC_PG_DELAY__POWER_DOWN_DELAY__SHIFT);
+		data |= (0x10 << RLC_PG_DELAY__CMD_PROPAGATE_DELAY__SHIFT);
+		data |= (0x40 << RLC_PG_DELAY__MEM_SLEEP_DELAY__SHIFT);
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_DELAY), data);
+
+		data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_DELAY_2));
+		data &= ~RLC_PG_DELAY_2__SERDES_CMD_DELAY_MASK;
+		data |= (0x4 << RLC_PG_DELAY_2__SERDES_CMD_DELAY__SHIFT);
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_DELAY_2), data);
+
+		data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_DELAY_3));
+		data &= ~RLC_PG_DELAY_3__CGCG_ACTIVE_BEFORE_CGPG_MASK;
+		data |= (0xff << RLC_PG_DELAY_3__CGCG_ACTIVE_BEFORE_CGPG__SHIFT);
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_DELAY_3), data);
+
+		data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_AUTO_PG_CTRL));
+		data &= ~RLC_AUTO_PG_CTRL__GRBM_REG_SAVE_GFX_IDLE_THRESHOLD_MASK;
+
+		/* program GRBM_REG_SAVE_GFX_IDLE_THRESHOLD to 0x55f0 */
+		data |= (0x55f0 << RLC_AUTO_PG_CTRL__GRBM_REG_SAVE_GFX_IDLE_THRESHOLD__SHIFT);
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_AUTO_PG_CTRL), data);
+
+		pwr_10_0_gfxip_control_over_cgpg(adev, true);
+	}
+}
+
+static void gfx_v9_0_enable_sck_slow_down_on_power_up(struct amdgpu_device *adev,
+						bool enable)
+{
+	uint32_t data = 0;
+	uint32_t default_data = 0;
+
+	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     SMU_CLK_SLOWDOWN_ON_PU_ENABLE,
+			     enable ? 1 : 0);
+	if (default_data != data)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
+}
+
+static void gfx_v9_0_enable_sck_slow_down_on_power_down(struct amdgpu_device *adev,
+						bool enable)
+{
+	uint32_t data = 0;
+	uint32_t default_data = 0;
+
+	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     SMU_CLK_SLOWDOWN_ON_PD_ENABLE,
+			     enable ? 1 : 0);
+	if(default_data != data)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
+}
+
+static void gfx_v9_0_enable_cp_power_gating(struct amdgpu_device *adev,
+					bool enable)
+{
+	uint32_t data = 0;
+	uint32_t default_data = 0;
+
+	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     CP_PG_DISABLE,
+			     enable ? 0 : 1);
+	if(default_data != data)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
+}
+
+static void gfx_v9_0_enable_gfx_cg_power_gating(struct amdgpu_device *adev,
+						bool enable)
+{
+	uint32_t data, default_data;
+
+	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     GFX_POWER_GATING_ENABLE,
+			     enable ? 1 : 0);
+	if(default_data != data)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
+}
+
+static void gfx_v9_0_enable_gfx_pipeline_powergating(struct amdgpu_device *adev,
+						bool enable)
+{
+	uint32_t data, default_data;
+
+	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     GFX_PIPELINE_PG_ENABLE,
+			     enable ? 1 : 0);
+	if(default_data != data)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
+
+	if (!enable)
+		/* read any GFX register to wake up GFX */
+		data = RREG32(SOC15_REG_OFFSET(GC, 0, mmDB_RENDER_CONTROL));
+}
+
+static void gfx_v9_0_enable_gfx_static_mg_power_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t data, default_data;
+
+	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     STATIC_PER_CU_PG_ENABLE,
+			     enable ? 1 : 0);
+	if(default_data != data)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
+}
+
+static void gfx_v9_0_enable_gfx_dynamic_mg_power_gating(struct amdgpu_device *adev,
+						bool enable)
+{
+	uint32_t data, default_data;
+
+	default_data = data = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL));
+	data = REG_SET_FIELD(data, RLC_PG_CNTL,
+			     DYN_PER_CU_PG_ENABLE,
+			     enable ? 1 : 0);
+	if(default_data != data)
+		WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_PG_CNTL), data);
+}
+
+static void gfx_v9_0_init_pg(struct amdgpu_device *adev)
+{
+	gfx_v9_0_init_csb(adev);
+
+	/*
+	 * Rlc save restore list is workable since v2_1.
+	 * And it's needed by gfxoff feature.
+	 */
+	if (adev->gfx.rlc.is_rlc_v2_1) {
+		gfx_v9_1_init_rlc_save_restore_list(adev);
+		gfx_v9_0_enable_save_restore_machine(adev);
+	}
+
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_PG |
+			      AMD_PG_SUPPORT_GFX_SMG |
+			      AMD_PG_SUPPORT_GFX_DMG |
+			      AMD_PG_SUPPORT_CP |
+			      AMD_PG_SUPPORT_GDS |
+			      AMD_PG_SUPPORT_RLC_SMU_HS)) {
+		WREG32(mmRLC_JUMP_TABLE_RESTORE,
+		       adev->gfx.rlc.cp_table_gpu_addr >> 8);
+		gfx_v9_0_init_gfx_power_gating(adev);
+	}
+}
+
+void gfx_v9_0_rlc_stop(struct amdgpu_device *adev)
+{
+	WREG32_FIELD15(GC, 0, RLC_CNTL, RLC_ENABLE_F32, 0);
+	gfx_v9_0_enable_gui_idle_interrupt(adev, false);
+	gfx_v9_0_wait_for_rlc_serdes(adev);
+}
+
+static void gfx_v9_0_rlc_reset(struct amdgpu_device *adev)
+{
+	WREG32_FIELD15(GC, 0, GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+	udelay(50);
+	WREG32_FIELD15(GC, 0, GRBM_SOFT_RESET, SOFT_RESET_RLC, 0);
+	udelay(50);
+}
+
+static void gfx_v9_0_rlc_start(struct amdgpu_device *adev)
+{
+#ifdef AMDGPU_RLC_DEBUG_RETRY
+	u32 rlc_ucode_ver;
+#endif
+
+	WREG32_FIELD15(GC, 0, RLC_CNTL, RLC_ENABLE_F32, 1);
+	udelay(50);
+
+	/* carrizo do enable cp interrupt after cp inited */
+	if (!(adev->flags & AMD_IS_APU)) {
+		gfx_v9_0_enable_gui_idle_interrupt(adev, true);
+		udelay(50);
+	}
+
+#ifdef AMDGPU_RLC_DEBUG_RETRY
+	/* RLC_GPM_GENERAL_6 : RLC Ucode version */
+	rlc_ucode_ver = RREG32_SOC15(GC, 0, mmRLC_GPM_GENERAL_6);
+	if(rlc_ucode_ver == 0x108) {
+		DRM_INFO("Using rlc debug ucode. mmRLC_GPM_GENERAL_6 ==0x08%x / fw_ver == %i \n",
+				rlc_ucode_ver, adev->gfx.rlc_fw_version);
+		/* RLC_GPM_TIMER_INT_3 : Timer interval in RefCLK cycles,
+		 * default is 0x9C4 to create a 100us interval */
+		WREG32_SOC15(GC, 0, mmRLC_GPM_TIMER_INT_3, 0x9C4);
+		/* RLC_GPM_GENERAL_12 : Minimum gap between wptr and rptr
+		 * to disable the page fault retry interrupts, default is
+		 * 0x100 (256) */
+		WREG32_SOC15(GC, 0, mmRLC_GPM_GENERAL_12, 0x100);
+	}
+#endif
+}
+
+static int gfx_v9_0_rlc_load_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_0 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.rlc_fw)
+		return -EINVAL;
+
+	hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+	amdgpu_ucode_print_rlc_hdr(&hdr->header);
+
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			   le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+
+	WREG32_SOC15(GC, 0, mmRLC_GPM_UCODE_ADDR,
+			RLCG_UCODE_LOADING_START_ADDRESS);
+	for (i = 0; i < fw_size; i++)
+		WREG32_SOC15(GC, 0, mmRLC_GPM_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32_SOC15(GC, 0, mmRLC_GPM_UCODE_ADDR, adev->gfx.rlc_fw_version);
+
+	return 0;
+}
+
+static int gfx_v9_0_rlc_resume(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (amdgpu_sriov_vf(adev)) {
+		gfx_v9_0_init_csb(adev);
+		return 0;
+	}
+
+	gfx_v9_0_rlc_stop(adev);
+
+	/* disable CG */
+	WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL, 0);
+
+	gfx_v9_0_rlc_reset(adev);
+
+	gfx_v9_0_init_pg(adev);
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+		/* legacy rlc firmware loading */
+		r = gfx_v9_0_rlc_load_microcode(adev);
+		if (r)
+			return r;
+	}
+
+	if (adev->asic_type == CHIP_RAVEN) {
+		if (amdgpu_lbpw != 0)
+			gfx_v9_0_enable_lbpw(adev, true);
+		else
+			gfx_v9_0_enable_lbpw(adev, false);
+	}
+
+	gfx_v9_0_rlc_start(adev);
+
+	return 0;
+}
+
+static void gfx_v9_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+	u32 tmp = RREG32_SOC15(GC, 0, mmCP_ME_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_HALT, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, enable ? 0 : 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, enable ? 0 : 1);
+	if (!enable) {
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+			adev->gfx.gfx_ring[i].ready = false;
+	}
+	WREG32_SOC15(GC, 0, mmCP_ME_CNTL, tmp);
+	udelay(50);
+}
+
+static int gfx_v9_0_cp_gfx_load_microcode(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v1_0 *pfp_hdr;
+	const struct gfx_firmware_header_v1_0 *ce_hdr;
+	const struct gfx_firmware_header_v1_0 *me_hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.me_fw || !adev->gfx.pfp_fw || !adev->gfx.ce_fw)
+		return -EINVAL;
+
+	pfp_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.pfp_fw->data;
+	ce_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.ce_fw->data;
+	me_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.me_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&pfp_hdr->header);
+	amdgpu_ucode_print_gfx_hdr(&ce_hdr->header);
+	amdgpu_ucode_print_gfx_hdr(&me_hdr->header);
+
+	gfx_v9_0_cp_gfx_enable(adev, false);
+
+	/* PFP */
+	fw_data = (const __le32 *)
+		(adev->gfx.pfp_fw->data +
+		 le32_to_cpu(pfp_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(pfp_hdr->header.ucode_size_bytes) / 4;
+	WREG32_SOC15(GC, 0, mmCP_PFP_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32_SOC15(GC, 0, mmCP_PFP_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32_SOC15(GC, 0, mmCP_PFP_UCODE_ADDR, adev->gfx.pfp_fw_version);
+
+	/* CE */
+	fw_data = (const __le32 *)
+		(adev->gfx.ce_fw->data +
+		 le32_to_cpu(ce_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(ce_hdr->header.ucode_size_bytes) / 4;
+	WREG32_SOC15(GC, 0, mmCP_CE_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32_SOC15(GC, 0, mmCP_CE_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32_SOC15(GC, 0, mmCP_CE_UCODE_ADDR, adev->gfx.ce_fw_version);
+
+	/* ME */
+	fw_data = (const __le32 *)
+		(adev->gfx.me_fw->data +
+		 le32_to_cpu(me_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(me_hdr->header.ucode_size_bytes) / 4;
+	WREG32_SOC15(GC, 0, mmCP_ME_RAM_WADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32_SOC15(GC, 0, mmCP_ME_RAM_DATA, le32_to_cpup(fw_data++));
+	WREG32_SOC15(GC, 0, mmCP_ME_RAM_WADDR, adev->gfx.me_fw_version);
+
+	return 0;
+}
+
+static int gfx_v9_0_cp_gfx_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->gfx.gfx_ring[0];
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+	int r, i, tmp;
+
+	/* init the CP */
+	WREG32_SOC15(GC, 0, mmCP_MAX_CONTEXT, adev->gfx.config.max_hw_contexts - 1);
+	WREG32_SOC15(GC, 0, mmCP_DEVICE_ID, 1);
+
+	gfx_v9_0_cp_gfx_enable(adev, true);
+
+	r = amdgpu_ring_alloc(ring, gfx_v9_0_get_csb_size(adev) + 4 + 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, 0x80000000);
+	amdgpu_ring_write(ring, 0x80000000);
+
+	for (sect = gfx9_cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				amdgpu_ring_write(ring,
+				       PACKET3(PACKET3_SET_CONTEXT_REG,
+					       ext->reg_count));
+				amdgpu_ring_write(ring,
+				       ext->reg_index - PACKET3_SET_CONTEXT_REG_START);
+				for (i = 0; i < ext->reg_count; i++)
+					amdgpu_ring_write(ring, ext->extent[i]);
+			}
+		}
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CLEAR_STATE, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_BASE, 2));
+	amdgpu_ring_write(ring, PACKET3_BASE_INDEX(CE_PARTITION_BASE));
+	amdgpu_ring_write(ring, 0x8000);
+	amdgpu_ring_write(ring, 0x8000);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG,1));
+	tmp = (PACKET3_SET_UCONFIG_REG_INDEX_TYPE |
+		(SOC15_REG_OFFSET(GC, 0, mmVGT_INDEX_TYPE) - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_commit(ring);
+
+	return 0;
+}
+
+static int gfx_v9_0_cp_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 tmp;
+	u32 rb_bufsz;
+	u64 rb_addr, rptr_addr, wptr_gpu_addr;
+
+	/* Set the write pointer delay */
+	WREG32_SOC15(GC, 0, mmCP_RB_WPTR_DELAY, 0);
+
+	/* set the RB to use vmid 0 */
+	WREG32_SOC15(GC, 0, mmCP_RB_VMID, 0);
+
+	/* Set ring buffer size */
+	ring = &adev->gfx.gfx_ring[0];
+	rb_bufsz = order_base_2(ring->ring_size / 8);
+	tmp = REG_SET_FIELD(0, CP_RB0_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, RB_BLKSZ, rb_bufsz - 2);
+#ifdef __BIG_ENDIAN
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, BUF_SWAP, 1);
+#endif
+	WREG32_SOC15(GC, 0, mmCP_RB0_CNTL, tmp);
+
+	/* Initialize the ring buffer's write pointers */
+	ring->wptr = 0;
+	WREG32_SOC15(GC, 0, mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
+	WREG32_SOC15(GC, 0, mmCP_RB0_WPTR_HI, upper_32_bits(ring->wptr));
+
+	/* set the wb address wether it's enabled or not */
+	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	WREG32_SOC15(GC, 0, mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
+	WREG32_SOC15(GC, 0, mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & CP_RB_RPTR_ADDR_HI__RB_RPTR_ADDR_HI_MASK);
+
+	wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	WREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_ADDR_LO, lower_32_bits(wptr_gpu_addr));
+	WREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_ADDR_HI, upper_32_bits(wptr_gpu_addr));
+
+	mdelay(1);
+	WREG32_SOC15(GC, 0, mmCP_RB0_CNTL, tmp);
+
+	rb_addr = ring->gpu_addr >> 8;
+	WREG32_SOC15(GC, 0, mmCP_RB0_BASE, rb_addr);
+	WREG32_SOC15(GC, 0, mmCP_RB0_BASE_HI, upper_32_bits(rb_addr));
+
+	tmp = RREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_CONTROL);
+	if (ring->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, ring->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL, DOORBELL_EN, 0);
+	}
+	WREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_CONTROL, tmp);
+
+	tmp = REG_SET_FIELD(0, CP_RB_DOORBELL_RANGE_LOWER,
+			DOORBELL_RANGE_LOWER, ring->doorbell_index);
+	WREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_RANGE_LOWER, tmp);
+
+	WREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_RANGE_UPPER,
+		       CP_RB_DOORBELL_RANGE_UPPER__DOORBELL_RANGE_UPPER_MASK);
+
+
+	/* start the ring */
+	gfx_v9_0_cp_gfx_start(adev);
+	ring->ready = true;
+
+	return 0;
+}
+
+static void gfx_v9_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+
+	if (enable) {
+		WREG32_SOC15(GC, 0, mmCP_MEC_CNTL, 0);
+	} else {
+		WREG32_SOC15(GC, 0, mmCP_MEC_CNTL,
+			(CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK));
+		for (i = 0; i < adev->gfx.num_compute_rings; i++)
+			adev->gfx.compute_ring[i].ready = false;
+		adev->gfx.kiq.ring.ready = false;
+	}
+	udelay(50);
+}
+
+static int gfx_v9_0_cp_compute_load_microcode(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v1_0 *mec_hdr;
+	const __le32 *fw_data;
+	unsigned i;
+	u32 tmp;
+
+	if (!adev->gfx.mec_fw)
+		return -EINVAL;
+
+	gfx_v9_0_cp_compute_enable(adev, false);
+
+	mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+	amdgpu_ucode_print_gfx_hdr(&mec_hdr->header);
+
+	fw_data = (const __le32 *)
+		(adev->gfx.mec_fw->data +
+		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
+	tmp = 0;
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, CACHE_POLICY, 0);
+	WREG32_SOC15(GC, 0, mmCP_CPC_IC_BASE_CNTL, tmp);
+
+	WREG32_SOC15(GC, 0, mmCP_CPC_IC_BASE_LO,
+		adev->gfx.mec.mec_fw_gpu_addr & 0xFFFFF000);
+	WREG32_SOC15(GC, 0, mmCP_CPC_IC_BASE_HI,
+		upper_32_bits(adev->gfx.mec.mec_fw_gpu_addr));
+
+	/* MEC1 */
+	WREG32_SOC15(GC, 0, mmCP_MEC_ME1_UCODE_ADDR,
+			 mec_hdr->jt_offset);
+	for (i = 0; i < mec_hdr->jt_size; i++)
+		WREG32_SOC15(GC, 0, mmCP_MEC_ME1_UCODE_DATA,
+			le32_to_cpup(fw_data + mec_hdr->jt_offset + i));
+
+	WREG32_SOC15(GC, 0, mmCP_MEC_ME1_UCODE_ADDR,
+			adev->gfx.mec_fw_version);
+	/* Todo : Loading MEC2 firmware is only necessary if MEC2 should run different microcode than MEC1. */
+
+	return 0;
+}
+
+/* KIQ functions */
+static void gfx_v9_0_kiq_setting(struct amdgpu_ring *ring)
+{
+	uint32_t tmp;
+	struct amdgpu_device *adev = ring->adev;
+
+	/* tell RLC which is KIQ queue */
+	tmp = RREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS);
+	tmp &= 0xffffff00;
+	tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue);
+	WREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS, tmp);
+	tmp |= 0x80;
+	WREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS, tmp);
+}
+
+static int gfx_v9_0_kiq_kcq_enable(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq.ring;
+	uint32_t scratch, tmp = 0;
+	uint64_t queue_mask = 0;
+	int r, i;
+
+	for (i = 0; i < AMDGPU_MAX_COMPUTE_QUEUES; ++i) {
+		if (!test_bit(i, adev->gfx.mec.queue_bitmap))
+			continue;
+
+		/* This situation may be hit in the future if a new HW
+		 * generation exposes more than 64 queues. If so, the
+		 * definition of queue_mask needs updating */
+		if (WARN_ON(i >= (sizeof(queue_mask)*8))) {
+			DRM_ERROR("Invalid KCQ enabled: %d\n", i);
+			break;
+		}
+
+		queue_mask |= (1ull << i);
+	}
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("Failed to get scratch reg (%d).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+
+	r = amdgpu_ring_alloc(kiq_ring, (7 * adev->gfx.num_compute_rings) + 11);
+	if (r) {
+		DRM_ERROR("Failed to lock KIQ (%d).\n", r);
+		amdgpu_gfx_scratch_free(adev, scratch);
+		return r;
+	}
+
+	/* set resources */
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_RESOURCES, 6));
+	amdgpu_ring_write(kiq_ring, PACKET3_SET_RESOURCES_VMID_MASK(0) |
+			  PACKET3_SET_RESOURCES_QUEUE_TYPE(0));	/* vmid_mask:0 queue_type:0 (KIQ) */
+	amdgpu_ring_write(kiq_ring, lower_32_bits(queue_mask));	/* queue mask lo */
+	amdgpu_ring_write(kiq_ring, upper_32_bits(queue_mask));	/* queue mask hi */
+	amdgpu_ring_write(kiq_ring, 0);	/* gws mask lo */
+	amdgpu_ring_write(kiq_ring, 0);	/* gws mask hi */
+	amdgpu_ring_write(kiq_ring, 0);	/* oac mask */
+	amdgpu_ring_write(kiq_ring, 0);	/* gds heap base:0, gds heap size:0 */
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
+		uint64_t mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj);
+		uint64_t wptr_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+
+		amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_MAP_QUEUES, 5));
+		/* Q_sel:0, vmid:0, vidmem: 1, engine:0, num_Q:1*/
+		amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+				  PACKET3_MAP_QUEUES_QUEUE_SEL(0) | /* Queue_Sel */
+				  PACKET3_MAP_QUEUES_VMID(0) | /* VMID */
+				  PACKET3_MAP_QUEUES_QUEUE(ring->queue) |
+				  PACKET3_MAP_QUEUES_PIPE(ring->pipe) |
+				  PACKET3_MAP_QUEUES_ME((ring->me == 1 ? 0 : 1)) |
+				  PACKET3_MAP_QUEUES_QUEUE_TYPE(0) | /*queue_type: normal compute queue */
+				  PACKET3_MAP_QUEUES_ALLOC_FORMAT(0) | /* alloc format: all_on_one_pipe */
+				  PACKET3_MAP_QUEUES_ENGINE_SEL(0) | /* engine_sel: compute */
+				  PACKET3_MAP_QUEUES_NUM_QUEUES(1)); /* num_queues: must be 1 */
+		amdgpu_ring_write(kiq_ring, PACKET3_MAP_QUEUES_DOORBELL_OFFSET(ring->doorbell_index));
+		amdgpu_ring_write(kiq_ring, lower_32_bits(mqd_addr));
+		amdgpu_ring_write(kiq_ring, upper_32_bits(mqd_addr));
+		amdgpu_ring_write(kiq_ring, lower_32_bits(wptr_addr));
+		amdgpu_ring_write(kiq_ring, upper_32_bits(wptr_addr));
+	}
+	/* write to scratch for completion */
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+	amdgpu_ring_write(kiq_ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(kiq_ring, 0xDEADBEEF);
+	amdgpu_ring_commit(kiq_ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i >= adev->usec_timeout) {
+		DRM_ERROR("KCQ enable failed (scratch(0x%04X)=0x%08X)\n",
+			  scratch, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_gfx_scratch_free(adev, scratch);
+
+	return r;
+}
+
+static int gfx_v9_0_mqd_init(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v9_mqd *mqd = ring->mqd_ptr;
+	uint64_t hqd_gpu_addr, wb_gpu_addr, eop_base_addr;
+	uint32_t tmp;
+
+	mqd->header = 0xC0310800;
+	mqd->compute_pipelinestat_enable = 0x00000001;
+	mqd->compute_static_thread_mgmt_se0 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se1 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se2 = 0xffffffff;
+	mqd->compute_static_thread_mgmt_se3 = 0xffffffff;
+	mqd->compute_misc_reserved = 0x00000003;
+
+	mqd->dynamic_cu_mask_addr_lo =
+		lower_32_bits(ring->mqd_gpu_addr
+			      + offsetof(struct v9_mqd_allocation, dynamic_cu_mask));
+	mqd->dynamic_cu_mask_addr_hi =
+		upper_32_bits(ring->mqd_gpu_addr
+			      + offsetof(struct v9_mqd_allocation, dynamic_cu_mask));
+
+	eop_base_addr = ring->eop_gpu_addr >> 8;
+	mqd->cp_hqd_eop_base_addr_lo = eop_base_addr;
+	mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr);
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	tmp = RREG32_SOC15(GC, 0, mmCP_HQD_EOP_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE,
+			(order_base_2(GFX9_MEC_HPD_SIZE / 4) - 1));
+
+	mqd->cp_hqd_eop_control = tmp;
+
+	/* enable doorbell? */
+	tmp = RREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL);
+
+	if (ring->use_doorbell) {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_OFFSET, ring->doorbell_index);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_EN, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_SOURCE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				    DOORBELL_HIT, 0);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+					 DOORBELL_EN, 0);
+	}
+
+	mqd->cp_hqd_pq_doorbell_control = tmp;
+
+	/* disable the queue if it's active */
+	ring->wptr = 0;
+	mqd->cp_hqd_dequeue_request = 0;
+	mqd->cp_hqd_pq_rptr = 0;
+	mqd->cp_hqd_pq_wptr_lo = 0;
+	mqd->cp_hqd_pq_wptr_hi = 0;
+
+	/* set the pointer to the MQD */
+	mqd->cp_mqd_base_addr_lo = ring->mqd_gpu_addr & 0xfffffffc;
+	mqd->cp_mqd_base_addr_hi = upper_32_bits(ring->mqd_gpu_addr);
+
+	/* set MQD vmid to 0 */
+	tmp = RREG32_SOC15(GC, 0, mmCP_MQD_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0);
+	mqd->cp_mqd_control = tmp;
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	hqd_gpu_addr = ring->gpu_addr >> 8;
+	mqd->cp_hqd_pq_base_lo = hqd_gpu_addr;
+	mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	tmp = RREG32_SOC15(GC, 0, mmCP_HQD_PQ_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
+			    (order_base_2(ring->ring_size / 4) - 1));
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
+			((order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1) << 8));
+#ifdef __BIG_ENDIAN
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, ENDIAN_SWAP, 1);
+#endif
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 0);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, ROQ_PQ_IB_FLIP, 0);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+	mqd->cp_hqd_pq_control = tmp;
+
+	/* set the wb address whether it's enabled or not */
+	wb_gpu_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_rptr_report_addr_hi =
+		upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	wb_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+	mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc;
+	mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+
+	tmp = 0;
+	/* enable the doorbell if requested */
+	if (ring->use_doorbell) {
+		tmp = RREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+				DOORBELL_OFFSET, ring->doorbell_index);
+
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+					 DOORBELL_EN, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+					 DOORBELL_SOURCE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+					 DOORBELL_HIT, 0);
+	}
+
+	mqd->cp_hqd_pq_doorbell_control = tmp;
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	ring->wptr = 0;
+	mqd->cp_hqd_pq_rptr = RREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR);
+
+	/* set the vmid for the queue */
+	mqd->cp_hqd_vmid = 0;
+
+	tmp = RREG32_SOC15(GC, 0, mmCP_HQD_PERSISTENT_STATE);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE, PRELOAD_SIZE, 0x53);
+	mqd->cp_hqd_persistent_state = tmp;
+
+	/* set MIN_IB_AVAIL_SIZE */
+	tmp = RREG32_SOC15(GC, 0, mmCP_HQD_IB_CONTROL);
+	tmp = REG_SET_FIELD(tmp, CP_HQD_IB_CONTROL, MIN_IB_AVAIL_SIZE, 3);
+	mqd->cp_hqd_ib_control = tmp;
+
+	/* activate the queue */
+	mqd->cp_hqd_active = 1;
+
+	return 0;
+}
+
+static int gfx_v9_0_kiq_init_register(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v9_mqd *mqd = ring->mqd_ptr;
+	int j;
+
+	/* disable wptr polling */
+	WREG32_FIELD15(GC, 0, CP_PQ_WPTR_POLL_CNTL, EN, 0);
+
+	WREG32_SOC15(GC, 0, mmCP_HQD_EOP_BASE_ADDR,
+	       mqd->cp_hqd_eop_base_addr_lo);
+	WREG32_SOC15(GC, 0, mmCP_HQD_EOP_BASE_ADDR_HI,
+	       mqd->cp_hqd_eop_base_addr_hi);
+
+	/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+	WREG32_SOC15(GC, 0, mmCP_HQD_EOP_CONTROL,
+	       mqd->cp_hqd_eop_control);
+
+	/* enable doorbell? */
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL,
+	       mqd->cp_hqd_pq_doorbell_control);
+
+	/* disable the queue if it's active */
+	if (RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1) {
+		WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, 1);
+		for (j = 0; j < adev->usec_timeout; j++) {
+			if (!(RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+		WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST,
+		       mqd->cp_hqd_dequeue_request);
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR,
+		       mqd->cp_hqd_pq_rptr);
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_LO,
+		       mqd->cp_hqd_pq_wptr_lo);
+		WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_HI,
+		       mqd->cp_hqd_pq_wptr_hi);
+	}
+
+	/* set the pointer to the MQD */
+	WREG32_SOC15(GC, 0, mmCP_MQD_BASE_ADDR,
+	       mqd->cp_mqd_base_addr_lo);
+	WREG32_SOC15(GC, 0, mmCP_MQD_BASE_ADDR_HI,
+	       mqd->cp_mqd_base_addr_hi);
+
+	/* set MQD vmid to 0 */
+	WREG32_SOC15(GC, 0, mmCP_MQD_CONTROL,
+	       mqd->cp_mqd_control);
+
+	/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_BASE,
+	       mqd->cp_hqd_pq_base_lo);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_BASE_HI,
+	       mqd->cp_hqd_pq_base_hi);
+
+	/* set up the HQD, this is similar to CP_RB0_CNTL */
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_CONTROL,
+	       mqd->cp_hqd_pq_control);
+
+	/* set the wb address whether it's enabled or not */
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR_REPORT_ADDR,
+				mqd->cp_hqd_pq_rptr_report_addr_lo);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
+				mqd->cp_hqd_pq_rptr_report_addr_hi);
+
+	/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR,
+	       mqd->cp_hqd_pq_wptr_poll_addr_lo);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+	       mqd->cp_hqd_pq_wptr_poll_addr_hi);
+
+	/* enable the doorbell if requested */
+	if (ring->use_doorbell) {
+		WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_LOWER,
+					(AMDGPU_DOORBELL64_KIQ *2) << 2);
+		WREG32_SOC15(GC, 0, mmCP_MEC_DOORBELL_RANGE_UPPER,
+					(AMDGPU_DOORBELL64_USERQUEUE_END * 2) << 2);
+	}
+
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL,
+	       mqd->cp_hqd_pq_doorbell_control);
+
+	/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_LO,
+	       mqd->cp_hqd_pq_wptr_lo);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_HI,
+	       mqd->cp_hqd_pq_wptr_hi);
+
+	/* set the vmid for the queue */
+	WREG32_SOC15(GC, 0, mmCP_HQD_VMID, mqd->cp_hqd_vmid);
+
+	WREG32_SOC15(GC, 0, mmCP_HQD_PERSISTENT_STATE,
+	       mqd->cp_hqd_persistent_state);
+
+	/* activate the queue */
+	WREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE,
+	       mqd->cp_hqd_active);
+
+	if (ring->use_doorbell)
+		WREG32_FIELD15(GC, 0, CP_PQ_STATUS, DOORBELL_ENABLE, 1);
+
+	return 0;
+}
+
+static int gfx_v9_0_kiq_fini_register(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int j;
+
+	/* disable the queue if it's active */
+	if (RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1) {
+
+		WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, 1);
+
+		for (j = 0; j < adev->usec_timeout; j++) {
+			if (!(RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE) & 1))
+				break;
+			udelay(1);
+		}
+
+		if (j == AMDGPU_MAX_USEC_TIMEOUT) {
+			DRM_DEBUG("KIQ dequeue request failed.\n");
+
+			/* Manual disable if dequeue request times out */
+			WREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE, 0);
+		}
+
+		WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST,
+		      0);
+	}
+
+	WREG32_SOC15(GC, 0, mmCP_HQD_IQ_TIMER, 0);
+	WREG32_SOC15(GC, 0, mmCP_HQD_IB_CONTROL, 0);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PERSISTENT_STATE, 0);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL, 0x40000000);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL, 0);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_RPTR, 0);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_HI, 0);
+	WREG32_SOC15(GC, 0, mmCP_HQD_PQ_WPTR_LO, 0);
+
+	return 0;
+}
+
+static int gfx_v9_0_kiq_init_queue(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v9_mqd *mqd = ring->mqd_ptr;
+	int mqd_idx = AMDGPU_MAX_COMPUTE_RINGS;
+
+	gfx_v9_0_kiq_setting(ring);
+
+	if (adev->in_gpu_reset) { /* for GPU_RESET case */
+		/* reset MQD to a clean status */
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(struct v9_mqd_allocation));
+
+		/* reset ring buffer */
+		ring->wptr = 0;
+		amdgpu_ring_clear_ring(ring);
+
+		mutex_lock(&adev->srbm_mutex);
+		soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		gfx_v9_0_kiq_init_register(ring);
+		soc15_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	} else {
+		memset((void *)mqd, 0, sizeof(struct v9_mqd_allocation));
+		((struct v9_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
+		((struct v9_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
+		mutex_lock(&adev->srbm_mutex);
+		soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		gfx_v9_0_mqd_init(ring);
+		gfx_v9_0_kiq_init_register(ring);
+		soc15_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(struct v9_mqd_allocation));
+	}
+
+	return 0;
+}
+
+static int gfx_v9_0_kcq_init_queue(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct v9_mqd *mqd = ring->mqd_ptr;
+	int mqd_idx = ring - &adev->gfx.compute_ring[0];
+
+	if (!adev->in_gpu_reset && !adev->gfx.in_suspend) {
+		memset((void *)mqd, 0, sizeof(struct v9_mqd_allocation));
+		((struct v9_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
+		((struct v9_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
+		mutex_lock(&adev->srbm_mutex);
+		soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+		gfx_v9_0_mqd_init(ring);
+		soc15_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy(adev->gfx.mec.mqd_backup[mqd_idx], mqd, sizeof(struct v9_mqd_allocation));
+	} else if (adev->in_gpu_reset) { /* for GPU_RESET case */
+		/* reset MQD to a clean status */
+		if (adev->gfx.mec.mqd_backup[mqd_idx])
+			memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(struct v9_mqd_allocation));
+
+		/* reset ring buffer */
+		ring->wptr = 0;
+		amdgpu_ring_clear_ring(ring);
+	} else {
+		amdgpu_ring_clear_ring(ring);
+	}
+
+	return 0;
+}
+
+static int gfx_v9_0_kiq_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = NULL;
+	int r = 0, i;
+
+	gfx_v9_0_cp_compute_enable(adev, true);
+
+	ring = &adev->gfx.kiq.ring;
+
+	r = amdgpu_bo_reserve(ring->mqd_obj, false);
+	if (unlikely(r != 0))
+		goto done;
+
+	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
+	if (!r) {
+		r = gfx_v9_0_kiq_init_queue(ring);
+		amdgpu_bo_kunmap(ring->mqd_obj);
+		ring->mqd_ptr = NULL;
+	}
+	amdgpu_bo_unreserve(ring->mqd_obj);
+	if (r)
+		goto done;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+
+		r = amdgpu_bo_reserve(ring->mqd_obj, false);
+		if (unlikely(r != 0))
+			goto done;
+		r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&ring->mqd_ptr);
+		if (!r) {
+			r = gfx_v9_0_kcq_init_queue(ring);
+			amdgpu_bo_kunmap(ring->mqd_obj);
+			ring->mqd_ptr = NULL;
+		}
+		amdgpu_bo_unreserve(ring->mqd_obj);
+		if (r)
+			goto done;
+	}
+
+	r = gfx_v9_0_kiq_kcq_enable(adev);
+done:
+	return r;
+}
+
+static int gfx_v9_0_cp_resume(struct amdgpu_device *adev)
+{
+	int r, i;
+	struct amdgpu_ring *ring;
+
+	if (!(adev->flags & AMD_IS_APU))
+		gfx_v9_0_enable_gui_idle_interrupt(adev, false);
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+		/* legacy firmware loading */
+		r = gfx_v9_0_cp_gfx_load_microcode(adev);
+		if (r)
+			return r;
+
+		r = gfx_v9_0_cp_compute_load_microcode(adev);
+		if (r)
+			return r;
+	}
+
+	r = gfx_v9_0_cp_gfx_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v9_0_kiq_resume(adev);
+	if (r)
+		return r;
+
+	ring = &adev->gfx.gfx_ring[0];
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		return r;
+	}
+
+	ring = &adev->gfx.kiq.ring;
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r)
+		ring->ready = false;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		ring = &adev->gfx.compute_ring[i];
+
+		ring->ready = true;
+		r = amdgpu_ring_test_ring(ring);
+		if (r)
+			ring->ready = false;
+	}
+
+	gfx_v9_0_enable_gui_idle_interrupt(adev, true);
+
+	return 0;
+}
+
+static void gfx_v9_0_cp_enable(struct amdgpu_device *adev, bool enable)
+{
+	gfx_v9_0_cp_gfx_enable(adev, enable);
+	gfx_v9_0_cp_compute_enable(adev, enable);
+}
+
+static int gfx_v9_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gfx_v9_0_init_golden_registers(adev);
+
+	gfx_v9_0_gpu_init(adev);
+
+	r = gfx_v9_0_csb_vram_pin(adev);
+	if (r)
+		return r;
+
+	r = gfx_v9_0_rlc_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v9_0_cp_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v9_0_ngg_en(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int gfx_v9_0_kcq_disable(struct amdgpu_ring *kiq_ring,struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = kiq_ring->adev;
+	uint32_t scratch, tmp = 0;
+	int r, i;
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("Failed to get scratch reg (%d).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+
+	r = amdgpu_ring_alloc(kiq_ring, 10);
+	if (r) {
+		DRM_ERROR("Failed to lock KIQ (%d).\n", r);
+		amdgpu_gfx_scratch_free(adev, scratch);
+		return r;
+	}
+
+	/* unmap queues */
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_UNMAP_QUEUES, 4));
+	amdgpu_ring_write(kiq_ring, /* Q_sel: 0, vmid: 0, engine: 0, num_Q: 1 */
+						PACKET3_UNMAP_QUEUES_ACTION(1) | /* RESET_QUEUES */
+						PACKET3_UNMAP_QUEUES_QUEUE_SEL(0) |
+						PACKET3_UNMAP_QUEUES_ENGINE_SEL(0) |
+						PACKET3_UNMAP_QUEUES_NUM_QUEUES(1));
+	amdgpu_ring_write(kiq_ring, PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(ring->doorbell_index));
+	amdgpu_ring_write(kiq_ring, 0);
+	amdgpu_ring_write(kiq_ring, 0);
+	amdgpu_ring_write(kiq_ring, 0);
+	/* write to scratch for completion */
+	amdgpu_ring_write(kiq_ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+	amdgpu_ring_write(kiq_ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(kiq_ring, 0xDEADBEEF);
+	amdgpu_ring_commit(kiq_ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i >= adev->usec_timeout) {
+		DRM_ERROR("KCQ disabled failed (scratch(0x%04X)=0x%08X)\n", scratch, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_gfx_scratch_free(adev, scratch);
+	return r;
+}
+
+static int gfx_v9_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_GFX,
+					       AMD_PG_STATE_UNGATE);
+
+	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
+	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
+
+	/* disable KCQ to avoid CPC touch memory not valid anymore */
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		gfx_v9_0_kcq_disable(&adev->gfx.kiq.ring, &adev->gfx.compute_ring[i]);
+
+	if (amdgpu_sriov_vf(adev)) {
+		gfx_v9_0_cp_gfx_enable(adev, false);
+		/* must disable polling for SRIOV when hw finished, otherwise
+		 * CPC engine may still keep fetching WB address which is already
+		 * invalid after sw finished and trigger DMAR reading error in
+		 * hypervisor side.
+		 */
+		WREG32_FIELD15(GC, 0, CP_PQ_WPTR_POLL_CNTL, EN, 0);
+		return 0;
+	}
+
+	/* Use deinitialize sequence from CAIL when unbinding device from driver,
+	 * otherwise KIQ is hanging when binding back
+	 */
+	if (!adev->in_gpu_reset && !adev->gfx.in_suspend) {
+		mutex_lock(&adev->srbm_mutex);
+		soc15_grbm_select(adev, adev->gfx.kiq.ring.me,
+				adev->gfx.kiq.ring.pipe,
+				adev->gfx.kiq.ring.queue, 0);
+		gfx_v9_0_kiq_fini_register(&adev->gfx.kiq.ring);
+		soc15_grbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	}
+
+	gfx_v9_0_cp_enable(adev, false);
+	gfx_v9_0_rlc_stop(adev);
+
+	gfx_v9_0_csb_vram_unpin(adev);
+
+	return 0;
+}
+
+static int gfx_v9_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->gfx.in_suspend = true;
+	return gfx_v9_0_hw_fini(adev);
+}
+
+static int gfx_v9_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r;
+
+	r = gfx_v9_0_hw_init(adev);
+	adev->gfx.in_suspend = false;
+	return r;
+}
+
+static bool gfx_v9_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (REG_GET_FIELD(RREG32_SOC15(GC, 0, mmGRBM_STATUS),
+				GRBM_STATUS, GUI_ACTIVE))
+		return false;
+	else
+		return true;
+}
+
+static int gfx_v9_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (gfx_v9_0_is_idle(handle))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int gfx_v9_0_soft_reset(void *handle)
+{
+	u32 grbm_soft_reset = 0;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* GRBM_STATUS */
+	tmp = RREG32_SOC15(GC, 0, mmGRBM_STATUS);
+	if (tmp & (GRBM_STATUS__PA_BUSY_MASK | GRBM_STATUS__SC_BUSY_MASK |
+		   GRBM_STATUS__BCI_BUSY_MASK | GRBM_STATUS__SX_BUSY_MASK |
+		   GRBM_STATUS__TA_BUSY_MASK | GRBM_STATUS__VGT_BUSY_MASK |
+		   GRBM_STATUS__DB_BUSY_MASK | GRBM_STATUS__CB_BUSY_MASK |
+		   GRBM_STATUS__GDS_BUSY_MASK | GRBM_STATUS__SPI_BUSY_MASK |
+		   GRBM_STATUS__IA_BUSY_MASK | GRBM_STATUS__IA_BUSY_NO_DMA_MASK)) {
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_CP, 1);
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_GFX, 1);
+	}
+
+	if (tmp & (GRBM_STATUS__CP_BUSY_MASK | GRBM_STATUS__CP_COHERENCY_BUSY_MASK)) {
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_CP, 1);
+	}
+
+	/* GRBM_STATUS2 */
+	tmp = RREG32_SOC15(GC, 0, mmGRBM_STATUS2);
+	if (REG_GET_FIELD(tmp, GRBM_STATUS2, RLC_BUSY))
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+
+
+	if (grbm_soft_reset) {
+		/* stop the rlc */
+		gfx_v9_0_rlc_stop(adev);
+
+		/* Disable GFX parsing/prefetching */
+		gfx_v9_0_cp_gfx_enable(adev, false);
+
+		/* Disable MEC parsing/prefetching */
+		gfx_v9_0_cp_compute_enable(adev, false);
+
+		if (grbm_soft_reset) {
+			tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+			tmp |= grbm_soft_reset;
+			dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+			WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+			tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+
+			udelay(50);
+
+			tmp &= ~grbm_soft_reset;
+			WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
+			tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
+		}
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+	return 0;
+}
+
+static uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
+	uint64_t clock;
+
+	mutex_lock(&adev->gfx.gpu_clock_mutex);
+	WREG32_SOC15(GC, 0, mmRLC_CAPTURE_GPU_CLOCK_COUNT, 1);
+	clock = (uint64_t)RREG32_SOC15(GC, 0, mmRLC_GPU_CLOCK_COUNT_LSB) |
+		((uint64_t)RREG32_SOC15(GC, 0, mmRLC_GPU_CLOCK_COUNT_MSB) << 32ULL);
+	mutex_unlock(&adev->gfx.gpu_clock_mutex);
+	return clock;
+}
+
+static void gfx_v9_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
+					  uint32_t vmid,
+					  uint32_t gds_base, uint32_t gds_size,
+					  uint32_t gws_base, uint32_t gws_size,
+					  uint32_t oa_base, uint32_t oa_size)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	gds_base = gds_base >> AMDGPU_GDS_SHIFT;
+	gds_size = gds_size >> AMDGPU_GDS_SHIFT;
+
+	gws_base = gws_base >> AMDGPU_GWS_SHIFT;
+	gws_size = gws_size >> AMDGPU_GWS_SHIFT;
+
+	oa_base = oa_base >> AMDGPU_OA_SHIFT;
+	oa_size = oa_size >> AMDGPU_OA_SHIFT;
+
+	/* GDS Base */
+	gfx_v9_0_write_data_to_reg(ring, 0, false,
+				   SOC15_REG_OFFSET(GC, 0, mmGDS_VMID0_BASE) + 2 * vmid,
+				   gds_base);
+
+	/* GDS Size */
+	gfx_v9_0_write_data_to_reg(ring, 0, false,
+				   SOC15_REG_OFFSET(GC, 0, mmGDS_VMID0_SIZE) + 2 * vmid,
+				   gds_size);
+
+	/* GWS */
+	gfx_v9_0_write_data_to_reg(ring, 0, false,
+				   SOC15_REG_OFFSET(GC, 0, mmGDS_GWS_VMID0) + vmid,
+				   gws_size << GDS_GWS_VMID0__SIZE__SHIFT | gws_base);
+
+	/* OA */
+	gfx_v9_0_write_data_to_reg(ring, 0, false,
+				   SOC15_REG_OFFSET(GC, 0, mmGDS_OA_VMID0) + vmid,
+				   (1 << (oa_size + oa_base)) - (1 << oa_base));
+}
+
+static int gfx_v9_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->gfx.num_gfx_rings = GFX9_NUM_GFX_RINGS;
+	adev->gfx.num_compute_rings = AMDGPU_MAX_COMPUTE_RINGS;
+	gfx_v9_0_set_ring_funcs(adev);
+	gfx_v9_0_set_irq_funcs(adev);
+	gfx_v9_0_set_gds_init(adev);
+	gfx_v9_0_set_rlc_funcs(adev);
+
+	return 0;
+}
+
+static int gfx_v9_0_late_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_reg_irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_get(adev, &adev->gfx.priv_inst_irq, 0);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static void gfx_v9_0_enter_rlc_safe_mode(struct amdgpu_device *adev)
+{
+	uint32_t rlc_setting, data;
+	unsigned i;
+
+	if (adev->gfx.rlc.in_safe_mode)
+		return;
+
+	/* if RLC is not enabled, do nothing */
+	rlc_setting = RREG32_SOC15(GC, 0, mmRLC_CNTL);
+	if (!(rlc_setting & RLC_CNTL__RLC_ENABLE_F32_MASK))
+		return;
+
+	if (adev->cg_flags &
+	    (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG |
+	     AMD_CG_SUPPORT_GFX_3D_CGCG)) {
+		data = RLC_SAFE_MODE__CMD_MASK;
+		data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT);
+		WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE, data);
+
+		/* wait for RLC_SAFE_MODE */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (!REG_GET_FIELD(RREG32_SOC15(GC, 0, mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD))
+				break;
+			udelay(1);
+		}
+		adev->gfx.rlc.in_safe_mode = true;
+	}
+}
+
+static void gfx_v9_0_exit_rlc_safe_mode(struct amdgpu_device *adev)
+{
+	uint32_t rlc_setting, data;
+
+	if (!adev->gfx.rlc.in_safe_mode)
+		return;
+
+	/* if RLC is not enabled, do nothing */
+	rlc_setting = RREG32_SOC15(GC, 0, mmRLC_CNTL);
+	if (!(rlc_setting & RLC_CNTL__RLC_ENABLE_F32_MASK))
+		return;
+
+	if (adev->cg_flags &
+	    (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG)) {
+		/*
+		 * Try to exit safe mode only if it is already in safe
+		 * mode.
+		 */
+		data = RLC_SAFE_MODE__CMD_MASK;
+		WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE, data);
+		adev->gfx.rlc.in_safe_mode = false;
+	}
+}
+
+static void gfx_v9_0_update_gfx_cg_power_gating(struct amdgpu_device *adev,
+						bool enable)
+{
+	gfx_v9_0_enter_rlc_safe_mode(adev);
+
+	if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_PG) && enable) {
+		gfx_v9_0_enable_gfx_cg_power_gating(adev, true);
+		if (adev->pg_flags & AMD_PG_SUPPORT_GFX_PIPELINE)
+			gfx_v9_0_enable_gfx_pipeline_powergating(adev, true);
+	} else {
+		gfx_v9_0_enable_gfx_cg_power_gating(adev, false);
+		gfx_v9_0_enable_gfx_pipeline_powergating(adev, false);
+	}
+
+	gfx_v9_0_exit_rlc_safe_mode(adev);
+}
+
+static void gfx_v9_0_update_gfx_mg_power_gating(struct amdgpu_device *adev,
+						bool enable)
+{
+	/* TODO: double check if we need to perform under safe mode */
+	/* gfx_v9_0_enter_rlc_safe_mode(adev); */
+
+	if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_SMG) && enable)
+		gfx_v9_0_enable_gfx_static_mg_power_gating(adev, true);
+	else
+		gfx_v9_0_enable_gfx_static_mg_power_gating(adev, false);
+
+	if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_DMG) && enable)
+		gfx_v9_0_enable_gfx_dynamic_mg_power_gating(adev, true);
+	else
+		gfx_v9_0_enable_gfx_dynamic_mg_power_gating(adev, false);
+
+	/* gfx_v9_0_exit_rlc_safe_mode(adev); */
+}
+
+static void gfx_v9_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t data, def;
+
+	/* It is disabled by HW by default */
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG)) {
+		/* 1 - RLC_CGTT_MGCG_OVERRIDE */
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+
+		if (adev->asic_type != CHIP_VEGA12)
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__CPF_CGTT_SCLK_OVERRIDE_MASK;
+
+		data &= ~(RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
+			  RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK |
+			  RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGLS_OVERRIDE_MASK);
+
+		/* only for Vega10 & Raven1 */
+		data |= RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK;
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* MGLS is a global flag to control all MGLS in GFX */
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGLS) {
+			/* 2 - RLC memory Light sleep */
+			if (adev->cg_flags & AMD_CG_SUPPORT_GFX_RLC_LS) {
+				def = data = RREG32_SOC15(GC, 0, mmRLC_MEM_SLP_CNTL);
+				data |= RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK;
+				if (def != data)
+					WREG32_SOC15(GC, 0, mmRLC_MEM_SLP_CNTL, data);
+			}
+			/* 3 - CP memory Light sleep */
+			if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CP_LS) {
+				def = data = RREG32_SOC15(GC, 0, mmCP_MEM_SLP_CNTL);
+				data |= CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+				if (def != data)
+					WREG32_SOC15(GC, 0, mmCP_MEM_SLP_CNTL, data);
+			}
+		}
+	} else {
+		/* 1 - MGCG_OVERRIDE */
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+
+		if (adev->asic_type != CHIP_VEGA12)
+			data |= RLC_CGTT_MGCG_OVERRIDE__CPF_CGTT_SCLK_OVERRIDE_MASK;
+
+		data |= (RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK |
+			 RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
+			 RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK |
+			 RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGLS_OVERRIDE_MASK);
+
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* 2 - disable MGLS in RLC */
+		data = RREG32_SOC15(GC, 0, mmRLC_MEM_SLP_CNTL);
+		if (data & RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK) {
+			data &= ~RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK;
+			WREG32_SOC15(GC, 0, mmRLC_MEM_SLP_CNTL, data);
+		}
+
+		/* 3 - disable MGLS in CP */
+		data = RREG32_SOC15(GC, 0, mmCP_MEM_SLP_CNTL);
+		if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK) {
+			data &= ~CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+			WREG32_SOC15(GC, 0, mmCP_MEM_SLP_CNTL, data);
+		}
+	}
+}
+
+static void gfx_v9_0_update_3d_clock_gating(struct amdgpu_device *adev,
+					   bool enable)
+{
+	uint32_t data, def;
+
+	adev->gfx.rlc.funcs->enter_safe_mode(adev);
+
+	/* Enable 3D CGCG/CGLS */
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG)) {
+		/* write cmd to clear cgcg/cgls ov */
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+		/* unset CGCG override */
+		data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_GFX3D_CG_OVERRIDE_MASK;
+		/* update CGCG and CGLS override bits */
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* enable 3Dcgcg FSM(0x0000363f) */
+		def = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D);
+
+		data = (0x36 << RLC_CGCG_CGLS_CTRL_3D__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
+			RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK;
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGLS)
+			data |= (0x000F << RLC_CGCG_CGLS_CTRL_3D__CGLS_REP_COMPANSAT_DELAY__SHIFT) |
+				RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK;
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D, data);
+
+		/* set IDLE_POLL_COUNT(0x00900100) */
+		def = RREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_CNTL);
+		data = (0x0100 << CP_RB_WPTR_POLL_CNTL__POLL_FREQUENCY__SHIFT) |
+			(0x0090 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT);
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_CNTL, data);
+	} else {
+		/* Disable CGCG/CGLS */
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D);
+		/* disable cgcg, cgls should be disabled */
+		data &= ~(RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK |
+			  RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK);
+		/* disable cgcg and cgls in FSM */
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D, data);
+	}
+
+	adev->gfx.rlc.funcs->exit_safe_mode(adev);
+}
+
+static void gfx_v9_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	adev->gfx.rlc.funcs->enter_safe_mode(adev);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)) {
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+		/* unset CGCG override */
+		data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_CGCG_OVERRIDE_MASK;
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS)
+			data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_CGLS_OVERRIDE_MASK;
+		else
+			data |= RLC_CGTT_MGCG_OVERRIDE__GFXIP_CGLS_OVERRIDE_MASK;
+		/* update CGCG and CGLS override bits */
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		/* enable cgcg FSM(0x0000363F) */
+		def = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL);
+
+		data = (0x36 << RLC_CGCG_CGLS_CTRL__CGCG_GFX_IDLE_THRESHOLD__SHIFT) |
+			RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK;
+		if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGLS)
+			data |= (0x000F << RLC_CGCG_CGLS_CTRL__CGLS_REP_COMPANSAT_DELAY__SHIFT) |
+				RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL, data);
+
+		/* set IDLE_POLL_COUNT(0x00900100) */
+		def = RREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_CNTL);
+		data = (0x0100 << CP_RB_WPTR_POLL_CNTL__POLL_FREQUENCY__SHIFT) |
+			(0x0090 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT);
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmCP_RB_WPTR_POLL_CNTL, data);
+	} else {
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL);
+		/* reset CGCG/CGLS bits */
+		data &= ~(RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK | RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK);
+		/* disable cgcg and cgls in FSM */
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL, data);
+	}
+
+	adev->gfx.rlc.funcs->exit_safe_mode(adev);
+}
+
+static int gfx_v9_0_update_gfx_clock_gating(struct amdgpu_device *adev,
+					    bool enable)
+{
+	if (enable) {
+		/* CGCG/CGLS should be enabled after MGCG/MGLS
+		 * ===  MGCG + MGLS ===
+		 */
+		gfx_v9_0_update_medium_grain_clock_gating(adev, enable);
+		/* ===  CGCG /CGLS for GFX 3D Only === */
+		gfx_v9_0_update_3d_clock_gating(adev, enable);
+		/* ===  CGCG + CGLS === */
+		gfx_v9_0_update_coarse_grain_clock_gating(adev, enable);
+	} else {
+		/* CGCG/CGLS should be disabled before MGCG/MGLS
+		 * ===  CGCG + CGLS ===
+		 */
+		gfx_v9_0_update_coarse_grain_clock_gating(adev, enable);
+		/* ===  CGCG /CGLS for GFX 3D Only === */
+		gfx_v9_0_update_3d_clock_gating(adev, enable);
+		/* ===  MGCG + MGLS === */
+		gfx_v9_0_update_medium_grain_clock_gating(adev, enable);
+	}
+	return 0;
+}
+
+static const struct amdgpu_rlc_funcs gfx_v9_0_rlc_funcs = {
+	.enter_safe_mode = gfx_v9_0_enter_rlc_safe_mode,
+	.exit_safe_mode = gfx_v9_0_exit_rlc_safe_mode
+};
+
+static int gfx_v9_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	bool enable = (state == AMD_PG_STATE_GATE) ? true : false;
+
+	switch (adev->asic_type) {
+	case CHIP_RAVEN:
+		if (adev->pg_flags & AMD_PG_SUPPORT_RLC_SMU_HS) {
+			gfx_v9_0_enable_sck_slow_down_on_power_up(adev, true);
+			gfx_v9_0_enable_sck_slow_down_on_power_down(adev, true);
+		} else {
+			gfx_v9_0_enable_sck_slow_down_on_power_up(adev, false);
+			gfx_v9_0_enable_sck_slow_down_on_power_down(adev, false);
+		}
+
+		if (adev->pg_flags & AMD_PG_SUPPORT_CP)
+			gfx_v9_0_enable_cp_power_gating(adev, true);
+		else
+			gfx_v9_0_enable_cp_power_gating(adev, false);
+
+		/* update gfx cgpg state */
+		gfx_v9_0_update_gfx_cg_power_gating(adev, enable);
+
+		/* update mgcg state */
+		gfx_v9_0_update_gfx_mg_power_gating(adev, enable);
+
+		/* set gfx off through smu */
+		if (enable && adev->powerplay.pp_funcs->set_powergating_by_smu)
+			amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_GFX, true);
+		break;
+	case CHIP_VEGA12:
+		/* set gfx off through smu */
+		if (enable && adev->powerplay.pp_funcs->set_powergating_by_smu)
+			amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_GFX, true);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v9_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+	case CHIP_RAVEN:
+		gfx_v9_0_update_gfx_clock_gating(adev,
+						 state == AMD_CG_STATE_GATE ? true : false);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static void gfx_v9_0_get_clockgating_state(void *handle, u32 *flags)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_GFX_MGCG */
+	data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_MGCG;
+
+	/* AMD_CG_SUPPORT_GFX_CGCG */
+	data = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL);
+	if (data & RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CGCG;
+
+	/* AMD_CG_SUPPORT_GFX_CGLS */
+	if (data & RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CGLS;
+
+	/* AMD_CG_SUPPORT_GFX_RLC_LS */
+	data = RREG32_SOC15(GC, 0, mmRLC_MEM_SLP_CNTL);
+	if (data & RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_RLC_LS | AMD_CG_SUPPORT_GFX_MGLS;
+
+	/* AMD_CG_SUPPORT_GFX_CP_LS */
+	data = RREG32_SOC15(GC, 0, mmCP_MEM_SLP_CNTL);
+	if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_CP_LS | AMD_CG_SUPPORT_GFX_MGLS;
+
+	/* AMD_CG_SUPPORT_GFX_3D_CGCG */
+	data = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D);
+	if (data & RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_3D_CGCG;
+
+	/* AMD_CG_SUPPORT_GFX_3D_CGLS */
+	if (data & RLC_CGCG_CGLS_CTRL_3D__CGLS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_GFX_3D_CGLS;
+}
+
+static u64 gfx_v9_0_ring_get_rptr_gfx(struct amdgpu_ring *ring)
+{
+	return ring->adev->wb.wb[ring->rptr_offs]; /* gfx9 is 32bit rptr*/
+}
+
+static u64 gfx_v9_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u64 wptr;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell) {
+		wptr = atomic64_read((atomic64_t *)&adev->wb.wb[ring->wptr_offs]);
+	} else {
+		wptr = RREG32_SOC15(GC, 0, mmCP_RB0_WPTR);
+		wptr += (u64)RREG32_SOC15(GC, 0, mmCP_RB0_WPTR_HI) << 32;
+	}
+
+	return wptr;
+}
+
+static void gfx_v9_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		atomic64_set((atomic64_t*)&adev->wb.wb[ring->wptr_offs], ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
+	} else {
+		WREG32_SOC15(GC, 0, mmCP_RB0_WPTR, lower_32_bits(ring->wptr));
+		WREG32_SOC15(GC, 0, mmCP_RB0_WPTR_HI, upper_32_bits(ring->wptr));
+	}
+}
+
+static void gfx_v9_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 ref_and_mask, reg_mem_engine;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio_funcs->hdp_flush_reg;
+
+	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) {
+		switch (ring->me) {
+		case 1:
+			ref_and_mask = nbio_hf_reg->ref_and_mask_cp2 << ring->pipe;
+			break;
+		case 2:
+			ref_and_mask = nbio_hf_reg->ref_and_mask_cp6 << ring->pipe;
+			break;
+		default:
+			return;
+		}
+		reg_mem_engine = 0;
+	} else {
+		ref_and_mask = nbio_hf_reg->ref_and_mask_cp0;
+		reg_mem_engine = 1; /* pfp */
+	}
+
+	gfx_v9_0_wait_reg_mem(ring, reg_mem_engine, 0, 1,
+			      adev->nbio_funcs->get_hdp_flush_req_offset(adev),
+			      adev->nbio_funcs->get_hdp_flush_done_offset(adev),
+			      ref_and_mask, ref_and_mask, 0x20);
+}
+
+static void gfx_v9_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
+                                      struct amdgpu_ib *ib,
+                                      unsigned vmid, bool ctx_switch)
+{
+	u32 header, control = 0;
+
+	if (ib->flags & AMDGPU_IB_FLAG_CE)
+		header = PACKET3(PACKET3_INDIRECT_BUFFER_CONST, 2);
+	else
+		header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
+
+	control |= ib->length_dw | (vmid << 24);
+
+	if (amdgpu_sriov_vf(ring->adev) && (ib->flags & AMDGPU_IB_FLAG_PREEMPT)) {
+		control |= INDIRECT_BUFFER_PRE_ENB(1);
+
+		if (!(ib->flags & AMDGPU_IB_FLAG_CE))
+			gfx_v9_0_ring_emit_de_meta(ring);
+	}
+
+	amdgpu_ring_write(ring, header);
+	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+		(2 << 0) |
+#endif
+		lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v9_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
+                                          struct amdgpu_ib *ib,
+                                          unsigned vmid, bool ctx_switch)
+{
+        u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
+
+        amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
+	BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
+        amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+                                (2 << 0) |
+#endif
+                                lower_32_bits(ib->gpu_addr));
+        amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+        amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v9_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+				     u64 seq, unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+	bool writeback = flags & AMDGPU_FENCE_FLAG_TC_WB_ONLY;
+
+	/* RELEASE_MEM - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RELEASE_MEM, 6));
+	amdgpu_ring_write(ring, ((writeback ? (EOP_TC_WB_ACTION_EN |
+					       EOP_TC_NC_ACTION_EN) :
+					      (EOP_TCL1_ACTION_EN |
+					       EOP_TC_ACTION_EN |
+					       EOP_TC_WB_ACTION_EN |
+					       EOP_TC_MD_ACTION_EN)) |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
+
+	/*
+	 * the address should be Qword aligned if 64bit write, Dword
+	 * aligned if only send 32bit data low (discard data high)
+	 */
+	if (write64bit)
+		BUG_ON(addr & 0x7);
+	else
+		BUG_ON(addr & 0x3);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+	amdgpu_ring_write(ring, 0);
+}
+
+static void gfx_v9_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	gfx_v9_0_wait_reg_mem(ring, usepfp, 1, 0,
+			      lower_32_bits(addr), upper_32_bits(addr),
+			      seq, 0xffffffff, 4);
+}
+
+static void gfx_v9_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* compute doesn't have PFP */
+	if (ring->funcs->type == AMDGPU_RING_TYPE_GFX) {
+		/* sync PFP to ME, otherwise we might get invalid PFP reads */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
+		amdgpu_ring_write(ring, 0x0);
+	}
+}
+
+static u64 gfx_v9_0_ring_get_rptr_compute(struct amdgpu_ring *ring)
+{
+	return ring->adev->wb.wb[ring->rptr_offs]; /* gfx9 hardware is 32bit rptr */
+}
+
+static u64 gfx_v9_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
+{
+	u64 wptr;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell)
+		wptr = atomic64_read((atomic64_t *)&ring->adev->wb.wb[ring->wptr_offs]);
+	else
+		BUG();
+	return wptr;
+}
+
+static void gfx_v9_0_ring_set_pipe_percent(struct amdgpu_ring *ring,
+					   bool acquire)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int pipe_num, tmp, reg;
+	int pipe_percent = acquire ? SPI_WCL_PIPE_PERCENT_GFX__VALUE_MASK : 0x1;
+
+	pipe_num = ring->me * adev->gfx.mec.num_pipe_per_mec + ring->pipe;
+
+	/* first me only has 2 entries, GFX and HP3D */
+	if (ring->me > 0)
+		pipe_num -= 2;
+
+	reg = SOC15_REG_OFFSET(GC, 0, mmSPI_WCL_PIPE_PERCENT_GFX) + pipe_num;
+	tmp = RREG32(reg);
+	tmp = REG_SET_FIELD(tmp, SPI_WCL_PIPE_PERCENT_GFX, VALUE, pipe_percent);
+	WREG32(reg, tmp);
+}
+
+static void gfx_v9_0_pipe_reserve_resources(struct amdgpu_device *adev,
+					    struct amdgpu_ring *ring,
+					    bool acquire)
+{
+	int i, pipe;
+	bool reserve;
+	struct amdgpu_ring *iring;
+
+	mutex_lock(&adev->gfx.pipe_reserve_mutex);
+	pipe = amdgpu_gfx_queue_to_bit(adev, ring->me, ring->pipe, 0);
+	if (acquire)
+		set_bit(pipe, adev->gfx.pipe_reserve_bitmap);
+	else
+		clear_bit(pipe, adev->gfx.pipe_reserve_bitmap);
+
+	if (!bitmap_weight(adev->gfx.pipe_reserve_bitmap, AMDGPU_MAX_COMPUTE_QUEUES)) {
+		/* Clear all reservations - everyone reacquires all resources */
+		for (i = 0; i < adev->gfx.num_gfx_rings; ++i)
+			gfx_v9_0_ring_set_pipe_percent(&adev->gfx.gfx_ring[i],
+						       true);
+
+		for (i = 0; i < adev->gfx.num_compute_rings; ++i)
+			gfx_v9_0_ring_set_pipe_percent(&adev->gfx.compute_ring[i],
+						       true);
+	} else {
+		/* Lower all pipes without a current reservation */
+		for (i = 0; i < adev->gfx.num_gfx_rings; ++i) {
+			iring = &adev->gfx.gfx_ring[i];
+			pipe = amdgpu_gfx_queue_to_bit(adev,
+						       iring->me,
+						       iring->pipe,
+						       0);
+			reserve = test_bit(pipe, adev->gfx.pipe_reserve_bitmap);
+			gfx_v9_0_ring_set_pipe_percent(iring, reserve);
+		}
+
+		for (i = 0; i < adev->gfx.num_compute_rings; ++i) {
+			iring = &adev->gfx.compute_ring[i];
+			pipe = amdgpu_gfx_queue_to_bit(adev,
+						       iring->me,
+						       iring->pipe,
+						       0);
+			reserve = test_bit(pipe, adev->gfx.pipe_reserve_bitmap);
+			gfx_v9_0_ring_set_pipe_percent(iring, reserve);
+		}
+	}
+
+	mutex_unlock(&adev->gfx.pipe_reserve_mutex);
+}
+
+static void gfx_v9_0_hqd_set_priority(struct amdgpu_device *adev,
+				      struct amdgpu_ring *ring,
+				      bool acquire)
+{
+	uint32_t pipe_priority = acquire ? 0x2 : 0x0;
+	uint32_t queue_priority = acquire ? 0xf : 0x0;
+
+	mutex_lock(&adev->srbm_mutex);
+	soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
+
+	WREG32_SOC15(GC, 0, mmCP_HQD_PIPE_PRIORITY, pipe_priority);
+	WREG32_SOC15(GC, 0, mmCP_HQD_QUEUE_PRIORITY, queue_priority);
+
+	soc15_grbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void gfx_v9_0_ring_set_priority_compute(struct amdgpu_ring *ring,
+					       enum drm_sched_priority priority)
+{
+	struct amdgpu_device *adev = ring->adev;
+	bool acquire = priority == DRM_SCHED_PRIORITY_HIGH_HW;
+
+	if (ring->funcs->type != AMDGPU_RING_TYPE_COMPUTE)
+		return;
+
+	gfx_v9_0_hqd_set_priority(adev, ring, acquire);
+	gfx_v9_0_pipe_reserve_resources(adev, ring, acquire);
+}
+
+static void gfx_v9_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* XXX check if swapping is necessary on BE */
+	if (ring->use_doorbell) {
+		atomic64_set((atomic64_t*)&adev->wb.wb[ring->wptr_offs], ring->wptr);
+		WDOORBELL64(ring->doorbell_index, ring->wptr);
+	} else{
+		BUG(); /* only DOORBELL method supported on gfx9 now */
+	}
+}
+
+static void gfx_v9_0_ring_emit_fence_kiq(struct amdgpu_ring *ring, u64 addr,
+					 u64 seq, unsigned int flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* we only allocate 32bit for each seq wb address */
+	BUG_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	/* write fence seq to the "addr" */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				 WRITE_DATA_DST_SEL(5) | WR_CONFIRM));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	if (flags & AMDGPU_FENCE_FLAG_INT) {
+		/* set register to trigger INT */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+		amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+					 WRITE_DATA_DST_SEL(0) | WR_CONFIRM));
+		amdgpu_ring_write(ring, SOC15_REG_OFFSET(GC, 0, mmCPC_INT_STATUS));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, 0x20000000); /* src_id is 178 */
+	}
+}
+
+static void gfx_v9_ring_emit_sb(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+	amdgpu_ring_write(ring, 0);
+}
+
+static void gfx_v9_0_ring_emit_ce_meta(struct amdgpu_ring *ring)
+{
+	struct v9_ce_ib_state ce_payload = {0};
+	uint64_t csa_addr;
+	int cnt;
+
+	cnt = (sizeof(ce_payload) >> 2) + 4 - 2;
+	csa_addr = amdgpu_csa_vaddr(ring->adev);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, cnt));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(2) |
+				 WRITE_DATA_DST_SEL(8) |
+				 WR_CONFIRM) |
+				 WRITE_DATA_CACHE_POLICY(0));
+	amdgpu_ring_write(ring, lower_32_bits(csa_addr + offsetof(struct v9_gfx_meta_data, ce_payload)));
+	amdgpu_ring_write(ring, upper_32_bits(csa_addr + offsetof(struct v9_gfx_meta_data, ce_payload)));
+	amdgpu_ring_write_multiple(ring, (void *)&ce_payload, sizeof(ce_payload) >> 2);
+}
+
+static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring)
+{
+	struct v9_de_ib_state de_payload = {0};
+	uint64_t csa_addr, gds_addr;
+	int cnt;
+
+	csa_addr = amdgpu_csa_vaddr(ring->adev);
+	gds_addr = csa_addr + 4096;
+	de_payload.gds_backup_addrlo = lower_32_bits(gds_addr);
+	de_payload.gds_backup_addrhi = upper_32_bits(gds_addr);
+
+	cnt = (sizeof(de_payload) >> 2) + 4 - 2;
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, cnt));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(1) |
+				 WRITE_DATA_DST_SEL(8) |
+				 WR_CONFIRM) |
+				 WRITE_DATA_CACHE_POLICY(0));
+	amdgpu_ring_write(ring, lower_32_bits(csa_addr + offsetof(struct v9_gfx_meta_data, de_payload)));
+	amdgpu_ring_write(ring, upper_32_bits(csa_addr + offsetof(struct v9_gfx_meta_data, de_payload)));
+	amdgpu_ring_write_multiple(ring, (void *)&de_payload, sizeof(de_payload) >> 2);
+}
+
+static void gfx_v9_0_ring_emit_tmz(struct amdgpu_ring *ring, bool start)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_FRAME_CONTROL, 0));
+	amdgpu_ring_write(ring, FRAME_CMD(start ? 0 : 1)); /* frame_end */
+}
+
+static void gfx_v9_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags)
+{
+	uint32_t dw2 = 0;
+
+	if (amdgpu_sriov_vf(ring->adev))
+		gfx_v9_0_ring_emit_ce_meta(ring);
+
+	gfx_v9_0_ring_emit_tmz(ring, true);
+
+	dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */
+	if (flags & AMDGPU_HAVE_CTX_SWITCH) {
+		/* set load_global_config & load_global_uconfig */
+		dw2 |= 0x8001;
+		/* set load_cs_sh_regs */
+		dw2 |= 0x01000000;
+		/* set load_per_context_state & load_gfx_sh_regs for GFX */
+		dw2 |= 0x10002;
+
+		/* set load_ce_ram if preamble presented */
+		if (AMDGPU_PREAMBLE_IB_PRESENT & flags)
+			dw2 |= 0x10000000;
+	} else {
+		/* still load_ce_ram if this is the first time preamble presented
+		 * although there is no context switch happens.
+		 */
+		if (AMDGPU_PREAMBLE_IB_PRESENT_FIRST & flags)
+			dw2 |= 0x10000000;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, dw2);
+	amdgpu_ring_write(ring, 0);
+}
+
+static unsigned gfx_v9_0_ring_emit_init_cond_exec(struct amdgpu_ring *ring)
+{
+	unsigned ret;
+	amdgpu_ring_write(ring, PACKET3(PACKET3_COND_EXEC, 3));
+	amdgpu_ring_write(ring, lower_32_bits(ring->cond_exe_gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ring->cond_exe_gpu_addr));
+	amdgpu_ring_write(ring, 0); /* discard following DWs if *cond_exec_gpu_addr==0 */
+	ret = ring->wptr & ring->buf_mask;
+	amdgpu_ring_write(ring, 0x55aa55aa); /* patch dummy value later */
+	return ret;
+}
+
+static void gfx_v9_0_ring_emit_patch_cond_exec(struct amdgpu_ring *ring, unsigned offset)
+{
+	unsigned cur;
+	BUG_ON(offset > ring->buf_mask);
+	BUG_ON(ring->ring[offset] != 0x55aa55aa);
+
+	cur = (ring->wptr & ring->buf_mask) - 1;
+	if (likely(cur > offset))
+		ring->ring[offset] = cur - offset;
+	else
+		ring->ring[offset] = (ring->ring_size>>2) - offset + cur;
+}
+
+static void gfx_v9_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_COPY_DATA, 4));
+	amdgpu_ring_write(ring, 0 |	/* src: register*/
+				(5 << 8) |	/* dst: memory */
+				(1 << 20));	/* write confirm */
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, lower_32_bits(adev->wb.gpu_addr +
+				adev->virt.reg_val_offs * 4));
+	amdgpu_ring_write(ring, upper_32_bits(adev->wb.gpu_addr +
+				adev->virt.reg_val_offs * 4));
+}
+
+static void gfx_v9_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg,
+				    uint32_t val)
+{
+	uint32_t cmd = 0;
+
+	switch (ring->funcs->type) {
+	case AMDGPU_RING_TYPE_GFX:
+		cmd = WRITE_DATA_ENGINE_SEL(1) | WR_CONFIRM;
+		break;
+	case AMDGPU_RING_TYPE_KIQ:
+		cmd = (1 << 16); /* no inc addr */
+		break;
+	default:
+		cmd = WR_CONFIRM;
+		break;
+	}
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, cmd);
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, val);
+}
+
+static void gfx_v9_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					uint32_t val, uint32_t mask)
+{
+	gfx_v9_0_wait_reg_mem(ring, 0, 0, 0, reg, 0, val, mask, 0x20);
+}
+
+static void gfx_v9_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
+						  uint32_t reg0, uint32_t reg1,
+						  uint32_t ref, uint32_t mask)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+
+	if (amdgpu_sriov_vf(ring->adev))
+		gfx_v9_0_wait_reg_mem(ring, usepfp, 0, 1, reg0, reg1,
+				      ref, mask, 0x20);
+	else
+		amdgpu_ring_emit_reg_write_reg_wait_helper(ring, reg0, reg1,
+							   ref, mask);
+}
+
+static void gfx_v9_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
+						 enum amdgpu_interrupt_state state)
+{
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		WREG32_FIELD15(GC, 0, CP_INT_CNTL_RING0,
+			       TIME_STAMP_INT_ENABLE,
+			       state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v9_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev,
+						     int me, int pipe,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 mec_int_cntl, mec_int_cntl_reg;
+
+	/*
+	 * amdgpu controls only the first MEC. That's why this function only
+	 * handles the setting of interrupts for this specific MEC. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+
+	if (me == 1) {
+		switch (pipe) {
+		case 0:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE0_INT_CNTL);
+			break;
+		case 1:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE1_INT_CNTL);
+			break;
+		case 2:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE2_INT_CNTL);
+			break;
+		case 3:
+			mec_int_cntl_reg = SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE3_INT_CNTL);
+			break;
+		default:
+			DRM_DEBUG("invalid pipe %d\n", pipe);
+			return;
+		}
+	} else {
+		DRM_DEBUG("invalid me %d\n", me);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		mec_int_cntl = RREG32(mec_int_cntl_reg);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     TIME_STAMP_INT_ENABLE, 0);
+		WREG32(mec_int_cntl_reg, mec_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		mec_int_cntl = RREG32(mec_int_cntl_reg);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     TIME_STAMP_INT_ENABLE, 1);
+		WREG32(mec_int_cntl_reg, mec_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static int gfx_v9_0_set_priv_reg_fault_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *source,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		WREG32_FIELD15(GC, 0, CP_INT_CNTL_RING0,
+			       PRIV_REG_INT_ENABLE,
+			       state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v9_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      unsigned type,
+					      enum amdgpu_interrupt_state state)
+{
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+	case AMDGPU_IRQ_STATE_ENABLE:
+		WREG32_FIELD15(GC, 0, CP_INT_CNTL_RING0,
+			       PRIV_INSTR_INT_ENABLE,
+			       state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v9_0_set_eop_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CP_IRQ_GFX_EOP:
+		gfx_v9_0_set_gfx_eop_interrupt_state(adev, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
+		gfx_v9_0_set_compute_eop_interrupt_state(adev, 1, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP:
+		gfx_v9_0_set_compute_eop_interrupt_state(adev, 1, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP:
+		gfx_v9_0_set_compute_eop_interrupt_state(adev, 1, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE3_EOP:
+		gfx_v9_0_set_compute_eop_interrupt_state(adev, 1, 3, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE0_EOP:
+		gfx_v9_0_set_compute_eop_interrupt_state(adev, 2, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE1_EOP:
+		gfx_v9_0_set_compute_eop_interrupt_state(adev, 2, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE2_EOP:
+		gfx_v9_0_set_compute_eop_interrupt_state(adev, 2, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE3_EOP:
+		gfx_v9_0_set_compute_eop_interrupt_state(adev, 2, 3, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v9_0_eop_irq(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *source,
+			    struct amdgpu_iv_entry *entry)
+{
+	int i;
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+
+	DRM_DEBUG("IH: CP EOP\n");
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+
+	switch (me_id) {
+	case 0:
+		amdgpu_fence_process(&adev->gfx.gfx_ring[0]);
+		break;
+	case 1:
+	case 2:
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring[i];
+			/* Per-queue interrupt is supported for MEC starting from VI.
+			  * The interrupt can only be enabled/disabled per pipe instead of per queue.
+			  */
+			if ((ring->me == me_id) && (ring->pipe == pipe_id) && (ring->queue == queue_id))
+				amdgpu_fence_process(ring);
+		}
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v9_0_priv_reg_irq(struct amdgpu_device *adev,
+				 struct amdgpu_irq_src *source,
+				 struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal register access in command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int gfx_v9_0_priv_inst_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int gfx_v9_0_kiq_set_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned int type,
+					    enum amdgpu_interrupt_state state)
+{
+	uint32_t tmp, target;
+	struct amdgpu_ring *ring = &(adev->gfx.kiq.ring);
+
+	if (ring->me == 1)
+		target = SOC15_REG_OFFSET(GC, 0, mmCP_ME1_PIPE0_INT_CNTL);
+	else
+		target = SOC15_REG_OFFSET(GC, 0, mmCP_ME2_PIPE0_INT_CNTL);
+	target += ring->pipe;
+
+	switch (type) {
+	case AMDGPU_CP_KIQ_IRQ_DRIVER0:
+		if (state == AMDGPU_IRQ_STATE_DISABLE) {
+			tmp = RREG32_SOC15(GC, 0, mmCPC_INT_CNTL);
+			tmp = REG_SET_FIELD(tmp, CPC_INT_CNTL,
+						 GENERIC2_INT_ENABLE, 0);
+			WREG32_SOC15(GC, 0, mmCPC_INT_CNTL, tmp);
+
+			tmp = RREG32(target);
+			tmp = REG_SET_FIELD(tmp, CP_ME2_PIPE0_INT_CNTL,
+						 GENERIC2_INT_ENABLE, 0);
+			WREG32(target, tmp);
+		} else {
+			tmp = RREG32_SOC15(GC, 0, mmCPC_INT_CNTL);
+			tmp = REG_SET_FIELD(tmp, CPC_INT_CNTL,
+						 GENERIC2_INT_ENABLE, 1);
+			WREG32_SOC15(GC, 0, mmCPC_INT_CNTL, tmp);
+
+			tmp = RREG32(target);
+			tmp = REG_SET_FIELD(tmp, CP_ME2_PIPE0_INT_CNTL,
+						 GENERIC2_INT_ENABLE, 1);
+			WREG32(target, tmp);
+		}
+		break;
+	default:
+		BUG(); /* kiq only support GENERIC2_INT now */
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v9_0_kiq_irq(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *source,
+			    struct amdgpu_iv_entry *entry)
+{
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring = &(adev->gfx.kiq.ring);
+
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+	DRM_DEBUG("IH: CPC GENERIC2_INT, me:%d, pipe:%d, queue:%d\n",
+		   me_id, pipe_id, queue_id);
+
+	amdgpu_fence_process(ring);
+	return 0;
+}
+
+static const struct amd_ip_funcs gfx_v9_0_ip_funcs = {
+	.name = "gfx_v9_0",
+	.early_init = gfx_v9_0_early_init,
+	.late_init = gfx_v9_0_late_init,
+	.sw_init = gfx_v9_0_sw_init,
+	.sw_fini = gfx_v9_0_sw_fini,
+	.hw_init = gfx_v9_0_hw_init,
+	.hw_fini = gfx_v9_0_hw_fini,
+	.suspend = gfx_v9_0_suspend,
+	.resume = gfx_v9_0_resume,
+	.is_idle = gfx_v9_0_is_idle,
+	.wait_for_idle = gfx_v9_0_wait_for_idle,
+	.soft_reset = gfx_v9_0_soft_reset,
+	.set_clockgating_state = gfx_v9_0_set_clockgating_state,
+	.set_powergating_state = gfx_v9_0_set_powergating_state,
+	.get_clockgating_state = gfx_v9_0_get_clockgating_state,
+};
+
+static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_gfx = {
+	.type = AMDGPU_RING_TYPE_GFX,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.vmhub = AMDGPU_GFXHUB,
+	.get_rptr = gfx_v9_0_ring_get_rptr_gfx,
+	.get_wptr = gfx_v9_0_ring_get_wptr_gfx,
+	.set_wptr = gfx_v9_0_ring_set_wptr_gfx,
+	.emit_frame_size = /* totally 242 maximum if 16 IBs */
+		5 +  /* COND_EXEC */
+		7 +  /* PIPELINE_SYNC */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		2 + /* VM_FLUSH */
+		8 +  /* FENCE for VM_FLUSH */
+		20 + /* GDS switch */
+		4 + /* double SWITCH_BUFFER,
+		       the first COND_EXEC jump to the place just
+			   prior to this double SWITCH_BUFFER  */
+		5 + /* COND_EXEC */
+		7 +	 /*	HDP_flush */
+		4 +	 /*	VGT_flush */
+		14 + /*	CE_META */
+		31 + /*	DE_META */
+		3 + /* CNTX_CTRL */
+		5 + /* HDP_INVL */
+		8 + 8 + /* FENCE x2 */
+		2, /* SWITCH_BUFFER */
+	.emit_ib_size =	4, /* gfx_v9_0_ring_emit_ib_gfx */
+	.emit_ib = gfx_v9_0_ring_emit_ib_gfx,
+	.emit_fence = gfx_v9_0_ring_emit_fence,
+	.emit_pipeline_sync = gfx_v9_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v9_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v9_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v9_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v9_0_ring_test_ring,
+	.test_ib = gfx_v9_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_switch_buffer = gfx_v9_ring_emit_sb,
+	.emit_cntxcntl = gfx_v9_ring_emit_cntxcntl,
+	.init_cond_exec = gfx_v9_0_ring_emit_init_cond_exec,
+	.patch_cond_exec = gfx_v9_0_ring_emit_patch_cond_exec,
+	.emit_tmz = gfx_v9_0_ring_emit_tmz,
+	.emit_wreg = gfx_v9_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v9_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v9_0_ring_emit_reg_write_reg_wait,
+};
+
+static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_compute = {
+	.type = AMDGPU_RING_TYPE_COMPUTE,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.vmhub = AMDGPU_GFXHUB,
+	.get_rptr = gfx_v9_0_ring_get_rptr_compute,
+	.get_wptr = gfx_v9_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v9_0_ring_set_wptr_compute,
+	.emit_frame_size =
+		20 + /* gfx_v9_0_ring_emit_gds_switch */
+		7 + /* gfx_v9_0_ring_emit_hdp_flush */
+		5 + /* hdp invalidate */
+		7 + /* gfx_v9_0_ring_emit_pipeline_sync */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		2 + /* gfx_v9_0_ring_emit_vm_flush */
+		8 + 8 + 8, /* gfx_v9_0_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size =	4, /* gfx_v9_0_ring_emit_ib_compute */
+	.emit_ib = gfx_v9_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v9_0_ring_emit_fence,
+	.emit_pipeline_sync = gfx_v9_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = gfx_v9_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v9_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v9_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v9_0_ring_test_ring,
+	.test_ib = gfx_v9_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.set_priority = gfx_v9_0_ring_set_priority_compute,
+	.emit_wreg = gfx_v9_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v9_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v9_0_ring_emit_reg_write_reg_wait,
+};
+
+static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_kiq = {
+	.type = AMDGPU_RING_TYPE_KIQ,
+	.align_mask = 0xff,
+	.nop = PACKET3(PACKET3_NOP, 0x3FFF),
+	.support_64bit_ptrs = true,
+	.vmhub = AMDGPU_GFXHUB,
+	.get_rptr = gfx_v9_0_ring_get_rptr_compute,
+	.get_wptr = gfx_v9_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v9_0_ring_set_wptr_compute,
+	.emit_frame_size =
+		20 + /* gfx_v9_0_ring_emit_gds_switch */
+		7 + /* gfx_v9_0_ring_emit_hdp_flush */
+		5 + /* hdp invalidate */
+		7 + /* gfx_v9_0_ring_emit_pipeline_sync */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 5 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
+		2 + /* gfx_v9_0_ring_emit_vm_flush */
+		8 + 8 + 8, /* gfx_v9_0_ring_emit_fence_kiq x3 for user fence, vm fence */
+	.emit_ib_size =	4, /* gfx_v9_0_ring_emit_ib_compute */
+	.emit_ib = gfx_v9_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v9_0_ring_emit_fence_kiq,
+	.test_ring = gfx_v9_0_ring_test_ring,
+	.test_ib = gfx_v9_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_rreg = gfx_v9_0_ring_emit_rreg,
+	.emit_wreg = gfx_v9_0_ring_emit_wreg,
+	.emit_reg_wait = gfx_v9_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v9_0_ring_emit_reg_write_reg_wait,
+};
+
+static void gfx_v9_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	adev->gfx.kiq.ring.funcs = &gfx_v9_0_ring_funcs_kiq;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		adev->gfx.gfx_ring[i].funcs = &gfx_v9_0_ring_funcs_gfx;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		adev->gfx.compute_ring[i].funcs = &gfx_v9_0_ring_funcs_compute;
+}
+
+static const struct amdgpu_irq_src_funcs gfx_v9_0_kiq_irq_funcs = {
+	.set = gfx_v9_0_kiq_set_interrupt_state,
+	.process = gfx_v9_0_kiq_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v9_0_eop_irq_funcs = {
+	.set = gfx_v9_0_set_eop_interrupt_state,
+	.process = gfx_v9_0_eop_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v9_0_priv_reg_irq_funcs = {
+	.set = gfx_v9_0_set_priv_reg_fault_state,
+	.process = gfx_v9_0_priv_reg_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v9_0_priv_inst_irq_funcs = {
+	.set = gfx_v9_0_set_priv_inst_fault_state,
+	.process = gfx_v9_0_priv_inst_irq,
+};
+
+static void gfx_v9_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.eop_irq.num_types = AMDGPU_CP_IRQ_LAST;
+	adev->gfx.eop_irq.funcs = &gfx_v9_0_eop_irq_funcs;
+
+	adev->gfx.priv_reg_irq.num_types = 1;
+	adev->gfx.priv_reg_irq.funcs = &gfx_v9_0_priv_reg_irq_funcs;
+
+	adev->gfx.priv_inst_irq.num_types = 1;
+	adev->gfx.priv_inst_irq.funcs = &gfx_v9_0_priv_inst_irq_funcs;
+
+	adev->gfx.kiq.irq.num_types = AMDGPU_CP_KIQ_IRQ_LAST;
+	adev->gfx.kiq.irq.funcs = &gfx_v9_0_kiq_irq_funcs;
+}
+
+static void gfx_v9_0_set_rlc_funcs(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+	case CHIP_RAVEN:
+		adev->gfx.rlc.funcs = &gfx_v9_0_rlc_funcs;
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v9_0_set_gds_init(struct amdgpu_device *adev)
+{
+	/* init asci gds info */
+	adev->gds.mem.total_size = RREG32_SOC15(GC, 0, mmGDS_VMID0_SIZE);
+	adev->gds.gws.total_size = 64;
+	adev->gds.oa.total_size = 16;
+
+	if (adev->gds.mem.total_size == 64 * 1024) {
+		adev->gds.mem.gfx_partition_size = 4096;
+		adev->gds.mem.cs_partition_size = 4096;
+
+		adev->gds.gws.gfx_partition_size = 4;
+		adev->gds.gws.cs_partition_size = 4;
+
+		adev->gds.oa.gfx_partition_size = 4;
+		adev->gds.oa.cs_partition_size = 1;
+	} else {
+		adev->gds.mem.gfx_partition_size = 1024;
+		adev->gds.mem.cs_partition_size = 1024;
+
+		adev->gds.gws.gfx_partition_size = 16;
+		adev->gds.gws.cs_partition_size = 16;
+
+		adev->gds.oa.gfx_partition_size = 4;
+		adev->gds.oa.cs_partition_size = 4;
+	}
+}
+
+static void gfx_v9_0_set_user_cu_inactive_bitmap(struct amdgpu_device *adev,
+						 u32 bitmap)
+{
+	u32 data;
+
+	if (!bitmap)
+		return;
+
+	data = bitmap << GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT;
+	data &= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK;
+
+	WREG32_SOC15(GC, 0, mmGC_USER_SHADER_ARRAY_CONFIG, data);
+}
+
+static u32 gfx_v9_0_get_cu_active_bitmap(struct amdgpu_device *adev)
+{
+	u32 data, mask;
+
+	data = RREG32_SOC15(GC, 0, mmCC_GC_SHADER_ARRAY_CONFIG);
+	data |= RREG32_SOC15(GC, 0, mmGC_USER_SHADER_ARRAY_CONFIG);
+
+	data &= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK;
+	data >>= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT;
+
+	mask = amdgpu_gfx_create_bitmask(adev->gfx.config.max_cu_per_sh);
+
+	return (~data) & mask;
+}
+
+static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev,
+				 struct amdgpu_cu_info *cu_info)
+{
+	int i, j, k, counter, active_cu_number = 0;
+	u32 mask, bitmap, ao_bitmap, ao_cu_mask = 0;
+	unsigned disable_masks[4 * 2];
+
+	if (!adev || !cu_info)
+		return -EINVAL;
+
+	amdgpu_gfx_parse_disable_cu(disable_masks, 4, 2);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			mask = 1;
+			ao_bitmap = 0;
+			counter = 0;
+			gfx_v9_0_select_se_sh(adev, i, j, 0xffffffff);
+			if (i < 4 && j < 2)
+				gfx_v9_0_set_user_cu_inactive_bitmap(
+					adev, disable_masks[i * 2 + j]);
+			bitmap = gfx_v9_0_get_cu_active_bitmap(adev);
+			cu_info->bitmap[i][j] = bitmap;
+
+			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) {
+				if (bitmap & mask) {
+					if (counter < adev->gfx.config.max_cu_per_sh)
+						ao_bitmap |= mask;
+					counter ++;
+				}
+				mask <<= 1;
+			}
+			active_cu_number += counter;
+			if (i < 2 && j < 2)
+				ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8));
+			cu_info->ao_cu_bitmap[i][j] = ao_bitmap;
+		}
+	}
+	gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	cu_info->number = active_cu_number;
+	cu_info->ao_cu_mask = ao_cu_mask;
+	cu_info->simd_per_cu = NUM_SIMD_PER_CU;
+
+	return 0;
+}
+
+const struct amdgpu_ip_block_version gfx_v9_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GFX,
+	.major = 9,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gfx_v9_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/gfx_v9_0.h b/sys/dev/pci/drm/amd/amdgpu/gfx_v9_0.h
new file mode 100644
index 00000000000..fa5a3fbaf6a
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gfx_v9_0.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V9_0_H__
+#define __GFX_V9_0_H__
+
+extern const struct amdgpu_ip_block_version gfx_v9_0_ip_block;
+
+void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num);
+
+uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev);
+int gfx_v9_0_get_cu_info(struct amdgpu_device *adev, struct amdgpu_cu_info *cu_info);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/gfxhub_v1_0.c b/sys/dev/pci/drm/amd/amdgpu/gfxhub_v1_0.c
new file mode 100644
index 00000000000..acfbd2d749c
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gfxhub_v1_0.c
@@ -0,0 +1,366 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "gfxhub_v1_0.h"
+
+#include "gc/gc_9_0_offset.h"
+#include "gc/gc_9_0_sh_mask.h"
+#include "gc/gc_9_0_default.h"
+#include "vega10_enum.h"
+
+#include "soc15_common.h"
+
+u64 gfxhub_v1_0_get_mc_fb_offset(struct amdgpu_device *adev)
+{
+	return (u64)RREG32_SOC15(GC, 0, mmMC_VM_FB_OFFSET) << 24;
+}
+
+static void gfxhub_v1_0_init_gart_pt_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+
+	BUG_ON(adev->gart.table_addr & (~0x0000FFFFFFFFF000ULL));
+	value = adev->gart.table_addr - adev->gmc.vram_start
+		+ adev->vm_manager.vram_base_offset;
+	value &= 0x0000FFFFFFFFF000ULL;
+	value |= 0x1; /*valid bit*/
+
+	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+		     lower_32_bits(value));
+
+	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+		     upper_32_bits(value));
+}
+
+static void gfxhub_v1_0_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	gfxhub_v1_0_init_gart_pt_regs(adev);
+
+	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void gfxhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+
+	/* Disable AGP. */
+	WREG32_SOC15(GC, 0, mmMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(GC, 0, mmMC_VM_AGP_TOP, 0);
+	WREG32_SOC15(GC, 0, mmMC_VM_AGP_BOT, 0xFFFFFFFF);
+
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     adev->gmc.vram_start >> 18);
+	WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     adev->gmc.vram_end >> 18);
+
+	/* Set default page address. */
+	value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start
+		+ adev->vm_manager.vram_base_offset;
+	WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(GC, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	WREG32_FIELD15(GC, 0, VM_L2_PROTECTION_FAULT_CNTL2,
+		       ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+}
+
+static void gfxhub_v1_0_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, mmMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC);/* XXX for emulation. */
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ATC_EN, 1);
+
+	WREG32_SOC15(GC, 0, mmMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void gfxhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(GC, 0, mmVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
+			    0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(GC, 0, mmVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(GC, 0, mmVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(GC, 0, mmVM_L2_CNTL2, tmp);
+
+	tmp = mmVM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(GC, 0, mmVM_L2_CNTL3, tmp);
+
+	tmp = mmVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(GC, 0, mmVM_L2_CNTL4, tmp);
+}
+
+static void gfxhub_v1_0_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(GC, 0, mmVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	WREG32_SOC15(GC, 0, mmVM_CONTEXT0_CNTL, tmp);
+}
+
+static void gfxhub_v1_0_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(GC, 0, mmVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0XFFFFFFFF);
+	WREG32_SOC15(GC, 0, mmVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(GC, 0, mmVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32,
+		     0);
+	WREG32_SOC15(GC, 0, mmVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32,
+		     0);
+
+	WREG32_SOC15(GC, 0, mmVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32, 0);
+	WREG32_SOC15(GC, 0, mmVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32, 0);
+
+}
+
+static void gfxhub_v1_0_setup_vmid_config(struct amdgpu_device *adev)
+{
+	unsigned num_level, block_size;
+	uint32_t tmp;
+	int i;
+
+	num_level = adev->vm_manager.num_level;
+	block_size = adev->vm_manager.block_size;
+	if (adev->gmc.translate_further)
+		num_level -= 1;
+	else
+		block_size -= 9;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_CNTL, i);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    num_level);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    PAGE_TABLE_BLOCK_SIZE,
+				    block_size);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_CNTL, i, tmp);
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, i*2, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32, i*2, 0);
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32,  i*2,
+			lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32, i*2,
+			upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+}
+
+static void gfxhub_v1_0_program_invalidation(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	for (i = 0 ; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    2 * i, 0xffffffff);
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    2 * i, 0x1f);
+	}
+}
+
+int gfxhub_v1_0_gart_enable(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev)) {
+		/*
+		 * MC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are
+		 * VF copy registers so vbios post doesn't program them, for
+		 * SRIOV driver need to program them
+		 */
+		WREG32_SOC15(GC, 0, mmMC_VM_FB_LOCATION_BASE,
+			     adev->gmc.vram_start >> 24);
+		WREG32_SOC15(GC, 0, mmMC_VM_FB_LOCATION_TOP,
+			     adev->gmc.vram_end >> 24);
+	}
+
+	/* GART Enable. */
+	gfxhub_v1_0_init_gart_aperture_regs(adev);
+	gfxhub_v1_0_init_system_aperture_regs(adev);
+	gfxhub_v1_0_init_tlb_regs(adev);
+	gfxhub_v1_0_init_cache_regs(adev);
+
+	gfxhub_v1_0_enable_system_domain(adev);
+	gfxhub_v1_0_disable_identity_aperture(adev);
+	gfxhub_v1_0_setup_vmid_config(adev);
+	gfxhub_v1_0_program_invalidation(adev);
+
+	return 0;
+}
+
+void gfxhub_v1_0_gart_disable(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT0_CNTL, i, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(GC, 0, mmMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp,
+				MC_VM_MX_L1_TLB_CNTL,
+				ENABLE_ADVANCED_DRIVER_MODEL,
+				0);
+	WREG32_SOC15(GC, 0, mmMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	WREG32_FIELD15(GC, 0, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(GC, 0, mmVM_L2_CNTL3, 0);
+}
+
+/**
+ * gfxhub_v1_0_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+void gfxhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev,
+					  bool value)
+{
+	u32 tmp;
+	tmp = RREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp,
+			VM_L2_PROTECTION_FAULT_CNTL,
+			TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+    }
+	WREG32_SOC15(GC, 0, mmVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+void gfxhub_v1_0_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(GC, 0,
+				 mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(GC, 0, mmVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(GC, 0, mmVM_L2_PROTECTION_FAULT_CNTL);
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/gfxhub_v1_0.h b/sys/dev/pci/drm/amd/amdgpu/gfxhub_v1_0.h
new file mode 100644
index 00000000000..206e29cad75
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gfxhub_v1_0.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFXHUB_V1_0_H__
+#define __GFXHUB_V1_0_H__
+
+int gfxhub_v1_0_gart_enable(struct amdgpu_device *adev);
+void gfxhub_v1_0_gart_disable(struct amdgpu_device *adev);
+void gfxhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev,
+					  bool value);
+void gfxhub_v1_0_init(struct amdgpu_device *adev);
+u64 gfxhub_v1_0_get_mc_fb_offset(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/gmc_v6_0.c b/sys/dev/pci/drm/amd/amdgpu/gmc_v6_0.c
new file mode 100644
index 00000000000..a0eab1dc9d5
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gmc_v6_0.c
@@ -0,0 +1,1200 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include <drm/drm_cache.h>
+#include "amdgpu.h"
+#include "gmc_v6_0.h"
+#include "amdgpu_ucode.h"
+
+#include "bif/bif_3_0_d.h"
+#include "bif/bif_3_0_sh_mask.h"
+#include "oss/oss_1_0_d.h"
+#include "oss/oss_1_0_sh_mask.h"
+#include "gmc/gmc_6_0_d.h"
+#include "gmc/gmc_6_0_sh_mask.h"
+#include "dce/dce_6_0_d.h"
+#include "dce/dce_6_0_sh_mask.h"
+#include "si_enums.h"
+
+static void gmc_v6_0_set_gmc_funcs(struct amdgpu_device *adev);
+static void gmc_v6_0_set_irq_funcs(struct amdgpu_device *adev);
+static int gmc_v6_0_wait_for_idle(void *handle);
+
+MODULE_FIRMWARE("amdgpu/tahiti_mc.bin");
+MODULE_FIRMWARE("amdgpu/pitcairn_mc.bin");
+MODULE_FIRMWARE("amdgpu/verde_mc.bin");
+MODULE_FIRMWARE("amdgpu/oland_mc.bin");
+MODULE_FIRMWARE("amdgpu/hainan_mc.bin");
+MODULE_FIRMWARE("amdgpu/si58_mc.bin");
+
+#define MC_SEQ_MISC0__MT__MASK   0xf0000000
+#define MC_SEQ_MISC0__MT__GDDR1  0x10000000
+#define MC_SEQ_MISC0__MT__DDR2   0x20000000
+#define MC_SEQ_MISC0__MT__GDDR3  0x30000000
+#define MC_SEQ_MISC0__MT__GDDR4  0x40000000
+#define MC_SEQ_MISC0__MT__GDDR5  0x50000000
+#define MC_SEQ_MISC0__MT__HBM    0x60000000
+#define MC_SEQ_MISC0__MT__DDR3   0xB0000000
+
+
+static const u32 crtc_offsets[6] =
+{
+	SI_CRTC0_REGISTER_OFFSET,
+	SI_CRTC1_REGISTER_OFFSET,
+	SI_CRTC2_REGISTER_OFFSET,
+	SI_CRTC3_REGISTER_OFFSET,
+	SI_CRTC4_REGISTER_OFFSET,
+	SI_CRTC5_REGISTER_OFFSET
+};
+
+static void gmc_v6_0_mc_stop(struct amdgpu_device *adev)
+{
+	u32 blackout;
+
+	gmc_v6_0_wait_for_idle((void *)adev);
+
+	blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
+	if (REG_GET_FIELD(blackout, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE) != 1) {
+		/* Block CPU access */
+		WREG32(mmBIF_FB_EN, 0);
+		/* blackout the MC */
+		blackout = REG_SET_FIELD(blackout,
+					 MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 0);
+		WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout | 1);
+	}
+	/* wait for the MC to settle */
+	udelay(100);
+
+}
+
+static void gmc_v6_0_mc_resume(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	/* unblackout the MC */
+	tmp = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 0);
+	WREG32(mmMC_SHARED_BLACKOUT_CNTL, tmp);
+	/* allow CPU access */
+	tmp = REG_SET_FIELD(0, BIF_FB_EN, FB_READ_EN, 1);
+	tmp = REG_SET_FIELD(tmp, BIF_FB_EN, FB_WRITE_EN, 1);
+	WREG32(mmBIF_FB_EN, tmp);
+}
+
+static int gmc_v6_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+	bool is_58_fw = false;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_TAHITI:
+		chip_name = "tahiti";
+		break;
+	case CHIP_PITCAIRN:
+		chip_name = "pitcairn";
+		break;
+	case CHIP_VERDE:
+		chip_name = "verde";
+		break;
+	case CHIP_OLAND:
+		chip_name = "oland";
+		break;
+	case CHIP_HAINAN:
+		chip_name = "hainan";
+		break;
+	default: BUG();
+	}
+
+	/* this memory configuration requires special firmware */
+	if (((RREG32(mmMC_SEQ_MISC0) & 0xff000000) >> 24) == 0x58)
+		is_58_fw = true;
+
+	if (is_58_fw)
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/si58_mc.bin");
+	else
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mc.bin", chip_name);
+	err = request_firmware(&adev->gmc.fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+
+	err = amdgpu_ucode_validate(adev->gmc.fw);
+
+out:
+	if (err) {
+		dev_err(adev->dev,
+		       "si_mc: Failed to load firmware \"%s\"\n",
+		       fw_name);
+		release_firmware(adev->gmc.fw);
+		adev->gmc.fw = NULL;
+	}
+	return err;
+}
+
+static int gmc_v6_0_mc_load_microcode(struct amdgpu_device *adev)
+{
+	const __le32 *new_fw_data = NULL;
+	u32 running;
+	const __le32 *new_io_mc_regs = NULL;
+	int i, regs_size, ucode_size;
+	const struct mc_firmware_header_v1_0 *hdr;
+
+	if (!adev->gmc.fw)
+		return -EINVAL;
+
+	hdr = (const struct mc_firmware_header_v1_0 *)adev->gmc.fw->data;
+
+	amdgpu_ucode_print_mc_hdr(&hdr->header);
+
+	adev->gmc.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	regs_size = le32_to_cpu(hdr->io_debug_size_bytes) / (4 * 2);
+	new_io_mc_regs = (const __le32 *)
+		(adev->gmc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
+	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+	new_fw_data = (const __le32 *)
+		(adev->gmc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	running = RREG32(mmMC_SEQ_SUP_CNTL) & MC_SEQ_SUP_CNTL__RUN_MASK;
+
+	if (running == 0) {
+
+		/* reset the engine and set to writable */
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000010);
+
+		/* load mc io regs */
+		for (i = 0; i < regs_size; i++) {
+			WREG32(mmMC_SEQ_IO_DEBUG_INDEX, le32_to_cpup(new_io_mc_regs++));
+			WREG32(mmMC_SEQ_IO_DEBUG_DATA, le32_to_cpup(new_io_mc_regs++));
+		}
+		/* load the MC ucode */
+		for (i = 0; i < ucode_size; i++) {
+			WREG32(mmMC_SEQ_SUP_PGM, le32_to_cpup(new_fw_data++));
+		}
+
+		/* put the engine back into the active state */
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000004);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000001);
+
+		/* wait for training to complete */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (RREG32(mmMC_SEQ_TRAIN_WAKEUP_CNTL) & MC_SEQ_TRAIN_WAKEUP_CNTL__TRAIN_DONE_D0_MASK)
+				break;
+			udelay(1);
+		}
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (RREG32(mmMC_SEQ_TRAIN_WAKEUP_CNTL) & MC_SEQ_TRAIN_WAKEUP_CNTL__TRAIN_DONE_D1_MASK)
+				break;
+			udelay(1);
+		}
+
+	}
+
+	return 0;
+}
+
+static void gmc_v6_0_vram_gtt_location(struct amdgpu_device *adev,
+				       struct amdgpu_gmc *mc)
+{
+	u64 base = RREG32(mmMC_VM_FB_LOCATION) & 0xFFFF;
+	base <<= 24;
+
+	amdgpu_device_vram_location(adev, &adev->gmc, base);
+	amdgpu_device_gart_location(adev, mc);
+}
+
+static void gmc_v6_0_mc_program(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	/* Initialize HDP */
+	for (i = 0, j = 0; i < 32; i++, j += 0x6) {
+		WREG32((0xb05 + j), 0x00000000);
+		WREG32((0xb06 + j), 0x00000000);
+		WREG32((0xb07 + j), 0x00000000);
+		WREG32((0xb08 + j), 0x00000000);
+		WREG32((0xb09 + j), 0x00000000);
+	}
+	WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
+
+	if (gmc_v6_0_wait_for_idle((void *)adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
+	}
+
+	if (adev->mode_info.num_crtc) {
+		u32 tmp;
+
+		/* Lockout access through VGA aperture*/
+		tmp = RREG32(mmVGA_HDP_CONTROL);
+		tmp |= VGA_HDP_CONTROL__VGA_MEMORY_DISABLE_MASK;
+		WREG32(mmVGA_HDP_CONTROL, tmp);
+
+		/* disable VGA render */
+		tmp = RREG32(mmVGA_RENDER_CONTROL);
+		tmp &= ~VGA_VSTATUS_CNTL;
+		WREG32(mmVGA_RENDER_CONTROL, tmp);
+	}
+	/* Update configuration */
+	WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+	       adev->gmc.vram_start >> 12);
+	WREG32(mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+	       adev->gmc.vram_end >> 12);
+	WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
+	       adev->vram_scratch.gpu_addr >> 12);
+	WREG32(mmMC_VM_AGP_BASE, 0);
+	WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
+	WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
+
+	if (gmc_v6_0_wait_for_idle((void *)adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
+	}
+}
+
+static int gmc_v6_0_mc_init(struct amdgpu_device *adev)
+{
+
+	u32 tmp;
+	int chansize, numchan;
+	int r;
+
+	tmp = RREG32(mmMC_ARB_RAMCFG);
+	if (tmp & (1 << 11)) {
+		chansize = 16;
+	} else if (tmp & MC_ARB_RAMCFG__CHANSIZE_MASK) {
+		chansize = 64;
+	} else {
+		chansize = 32;
+	}
+	tmp = RREG32(mmMC_SHARED_CHMAP);
+	switch ((tmp & MC_SHARED_CHMAP__NOOFCHAN_MASK) >> MC_SHARED_CHMAP__NOOFCHAN__SHIFT) {
+	case 0:
+	default:
+		numchan = 1;
+		break;
+	case 1:
+		numchan = 2;
+		break;
+	case 2:
+		numchan = 4;
+		break;
+	case 3:
+		numchan = 8;
+		break;
+	case 4:
+		numchan = 3;
+		break;
+	case 5:
+		numchan = 6;
+		break;
+	case 6:
+		numchan = 10;
+		break;
+	case 7:
+		numchan = 12;
+		break;
+	case 8:
+		numchan = 16;
+		break;
+	}
+	adev->gmc.vram_width = numchan * chansize;
+	/* size in MB on si */
+	adev->gmc.mc_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+	adev->gmc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_device_resize_fb_bar(adev);
+		if (r)
+			return r;
+	}
+	adev->gmc.aper_base = adev->fb_aper_offset;
+	adev->gmc.aper_size = adev->fb_aper_size;
+	adev->gmc.visible_vram_size = adev->gmc.aper_size;
+
+	/* set the gart size */
+	if (amdgpu_gart_size == -1) {
+		switch (adev->asic_type) {
+		case CHIP_HAINAN:    /* no MM engines */
+		default:
+			adev->gmc.gart_size = 256ULL << 20;
+			break;
+		case CHIP_VERDE:    /* UVD, VCE do not support GPUVM */
+		case CHIP_TAHITI:   /* UVD, VCE do not support GPUVM */
+		case CHIP_PITCAIRN: /* UVD, VCE do not support GPUVM */
+		case CHIP_OLAND:    /* UVD, VCE do not support GPUVM */
+			adev->gmc.gart_size = 1024ULL << 20;
+			break;
+		}
+	} else {
+		adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
+	}
+
+	gmc_v6_0_vram_gtt_location(adev, &adev->gmc);
+
+	return 0;
+}
+
+static void gmc_v6_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid)
+{
+	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+}
+
+static uint64_t gmc_v6_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
+					    unsigned vmid, uint64_t pd_addr)
+{
+	uint32_t reg;
+
+	/* write new base address */
+	if (vmid < 8)
+		reg = mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vmid;
+	else
+		reg = mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + (vmid - 8);
+	amdgpu_ring_emit_wreg(ring, reg, pd_addr >> 12);
+
+	/* bits 0-15 are the VM contexts0-15 */
+	amdgpu_ring_emit_wreg(ring, mmVM_INVALIDATE_REQUEST, 1 << vmid);
+
+	return pd_addr;
+}
+
+static int gmc_v6_0_set_pte_pde(struct amdgpu_device *adev, void *cpu_pt_addr,
+				uint32_t gpu_page_idx, uint64_t addr,
+				uint64_t flags)
+{
+	void __iomem *ptr = (void *)cpu_pt_addr;
+	uint64_t value;
+
+	value = addr & 0xFFFFFFFFFFFFF000ULL;
+	value |= flags;
+	writeq(value, ptr + (gpu_page_idx * 8));
+
+	return 0;
+}
+
+static uint64_t gmc_v6_0_get_vm_pte_flags(struct amdgpu_device *adev,
+					  uint32_t flags)
+{
+	uint64_t pte_flag = 0;
+
+	if (flags & AMDGPU_VM_PAGE_READABLE)
+		pte_flag |= AMDGPU_PTE_READABLE;
+	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
+		pte_flag |= AMDGPU_PTE_WRITEABLE;
+	if (flags & AMDGPU_VM_PAGE_PRT)
+		pte_flag |= AMDGPU_PTE_PRT;
+
+	return pte_flag;
+}
+
+static void gmc_v6_0_get_vm_pde(struct amdgpu_device *adev, int level,
+				uint64_t *addr, uint64_t *flags)
+{
+	BUG_ON(*addr & 0xFFFFFF0000000FFFULL);
+}
+
+static void gmc_v6_0_set_fault_enable_default(struct amdgpu_device *adev,
+					      bool value)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmVM_CONTEXT1_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	WREG32(mmVM_CONTEXT1_CNTL, tmp);
+}
+
+ /**
+   + * gmc_v8_0_set_prt - set PRT VM fault
+   + *
+   + * @adev: amdgpu_device pointer
+   + * @enable: enable/disable VM fault handling for PRT
+   +*/
+static void gmc_v6_0_set_prt(struct amdgpu_device *adev, bool enable)
+{
+	u32 tmp;
+
+	if (enable && !adev->gmc.prt_warning) {
+		dev_warn(adev->dev, "Disabling VM faults because of PRT request!\n");
+		adev->gmc.prt_warning = true;
+	}
+
+	tmp = RREG32(mmVM_PRT_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    CB_DISABLE_FAULT_ON_UNMAPPED_ACCESS,
+			    enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    TC_DISABLE_FAULT_ON_UNMAPPED_ACCESS,
+			    enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    L2_CACHE_STORE_INVALID_ENTRIES,
+			    enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    L1_TLB_STORE_INVALID_ENTRIES,
+			    enable);
+	WREG32(mmVM_PRT_CNTL, tmp);
+
+	if (enable) {
+		uint32_t low = AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT;
+		uint32_t high = adev->vm_manager.max_pfn -
+			(AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT);
+
+		WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, low);
+		WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, low);
+		WREG32(mmVM_PRT_APERTURE2_LOW_ADDR, low);
+		WREG32(mmVM_PRT_APERTURE3_LOW_ADDR, low);
+		WREG32(mmVM_PRT_APERTURE0_HIGH_ADDR, high);
+		WREG32(mmVM_PRT_APERTURE1_HIGH_ADDR, high);
+		WREG32(mmVM_PRT_APERTURE2_HIGH_ADDR, high);
+		WREG32(mmVM_PRT_APERTURE3_HIGH_ADDR, high);
+	} else {
+		WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, 0xfffffff);
+		WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, 0xfffffff);
+		WREG32(mmVM_PRT_APERTURE2_LOW_ADDR, 0xfffffff);
+		WREG32(mmVM_PRT_APERTURE3_LOW_ADDR, 0xfffffff);
+		WREG32(mmVM_PRT_APERTURE0_HIGH_ADDR, 0x0);
+		WREG32(mmVM_PRT_APERTURE1_HIGH_ADDR, 0x0);
+		WREG32(mmVM_PRT_APERTURE2_HIGH_ADDR, 0x0);
+		WREG32(mmVM_PRT_APERTURE3_HIGH_ADDR, 0x0);
+	}
+}
+
+static int gmc_v6_0_gart_enable(struct amdgpu_device *adev)
+{
+	int r, i;
+	u32 field;
+
+	if (adev->gart.robj == NULL) {
+		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
+		return -EINVAL;
+	}
+	r = amdgpu_gart_table_vram_pin(adev);
+	if (r)
+		return r;
+	/* Setup TLB control */
+	WREG32(mmMC_VM_MX_L1_TLB_CNTL,
+	       (0xA << 7) |
+	       MC_VM_MX_L1_TLB_CNTL__ENABLE_L1_TLB_MASK |
+	       MC_VM_MX_L1_TLB_CNTL__ENABLE_L1_FRAGMENT_PROCESSING_MASK |
+	       MC_VM_MX_L1_TLB_CNTL__SYSTEM_ACCESS_MODE_MASK |
+	       MC_VM_MX_L1_TLB_CNTL__ENABLE_ADVANCED_DRIVER_MODEL_MASK |
+	       (0UL << MC_VM_MX_L1_TLB_CNTL__SYSTEM_APERTURE_UNMAPPED_ACCESS__SHIFT));
+	/* Setup L2 cache */
+	WREG32(mmVM_L2_CNTL,
+	       VM_L2_CNTL__ENABLE_L2_CACHE_MASK |
+	       VM_L2_CNTL__ENABLE_L2_FRAGMENT_PROCESSING_MASK |
+	       VM_L2_CNTL__ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE_MASK |
+	       VM_L2_CNTL__ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE_MASK |
+	       (7UL << VM_L2_CNTL__EFFECTIVE_L2_QUEUE_SIZE__SHIFT) |
+	       (1UL << VM_L2_CNTL__CONTEXT1_IDENTITY_ACCESS_MODE__SHIFT));
+	WREG32(mmVM_L2_CNTL2,
+	       VM_L2_CNTL2__INVALIDATE_ALL_L1_TLBS_MASK |
+	       VM_L2_CNTL2__INVALIDATE_L2_CACHE_MASK);
+
+	field = adev->vm_manager.fragment_size;
+	WREG32(mmVM_L2_CNTL3,
+	       VM_L2_CNTL3__L2_CACHE_BIGK_ASSOCIATIVITY_MASK |
+	       (field << VM_L2_CNTL3__BANK_SELECT__SHIFT) |
+	       (field << VM_L2_CNTL3__L2_CACHE_BIGK_FRAGMENT_SIZE__SHIFT));
+	/* setup context0 */
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->gmc.gart_start >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->gmc.gart_end >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
+	WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
+			(u32)(adev->dummy_page_addr >> 12));
+	WREG32(mmVM_CONTEXT0_CNTL2, 0);
+	WREG32(mmVM_CONTEXT0_CNTL,
+	       VM_CONTEXT0_CNTL__ENABLE_CONTEXT_MASK |
+	       (0UL << VM_CONTEXT0_CNTL__PAGE_TABLE_DEPTH__SHIFT) |
+	       VM_CONTEXT0_CNTL__RANGE_PROTECTION_FAULT_ENABLE_DEFAULT_MASK);
+
+	WREG32(0x575, 0);
+	WREG32(0x576, 0);
+	WREG32(0x577, 0);
+
+	/* empty context1-15 */
+	/* set vm size, must be a multiple of 4 */
+	WREG32(mmVM_CONTEXT1_PAGE_TABLE_START_ADDR, 0);
+	WREG32(mmVM_CONTEXT1_PAGE_TABLE_END_ADDR, adev->vm_manager.max_pfn - 1);
+	/* Assign the pt base to something valid for now; the pts used for
+	 * the VMs are determined by the application and setup and assigned
+	 * on the fly in the vm part of radeon_gart.c
+	 */
+	for (i = 1; i < 16; i++) {
+		if (i < 8)
+			WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i,
+			       adev->gart.table_addr >> 12);
+		else
+			WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + i - 8,
+			       adev->gart.table_addr >> 12);
+	}
+
+	/* enable context1-15 */
+	WREG32(mmVM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR,
+	       (u32)(adev->dummy_page_addr >> 12));
+	WREG32(mmVM_CONTEXT1_CNTL2, 4);
+	WREG32(mmVM_CONTEXT1_CNTL,
+	       VM_CONTEXT1_CNTL__ENABLE_CONTEXT_MASK |
+	       (1UL << VM_CONTEXT1_CNTL__PAGE_TABLE_DEPTH__SHIFT) |
+	       ((adev->vm_manager.block_size - 9)
+	       << VM_CONTEXT1_CNTL__PAGE_TABLE_BLOCK_SIZE__SHIFT));
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS)
+		gmc_v6_0_set_fault_enable_default(adev, false);
+	else
+		gmc_v6_0_set_fault_enable_default(adev, true);
+
+	gmc_v6_0_flush_gpu_tlb(adev, 0);
+	dev_info(adev->dev, "PCIE GART of %uM enabled (table at 0x%016llX).\n",
+		 (unsigned)(adev->gmc.gart_size >> 20),
+		 (unsigned long long)adev->gart.table_addr);
+	adev->gart.ready = true;
+	return 0;
+}
+
+static int gmc_v6_0_gart_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.robj) {
+		dev_warn(adev->dev, "gmc_v6_0 PCIE GART already initialized\n");
+		return 0;
+	}
+	r = amdgpu_gart_init(adev);
+	if (r)
+		return r;
+	adev->gart.table_size = adev->gart.num_gpu_pages * 8;
+	adev->gart.gart_pte_flags = 0;
+	return amdgpu_gart_table_vram_alloc(adev);
+}
+
+static void gmc_v6_0_gart_disable(struct amdgpu_device *adev)
+{
+	/*unsigned i;
+
+	for (i = 1; i < 16; ++i) {
+		uint32_t reg;
+		if (i < 8)
+			reg = VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i ;
+		else
+			reg = VM_CONTEXT8_PAGE_TABLE_BASE_ADDR + (i - 8);
+		adev->vm_manager.saved_table_addr[i] = RREG32(reg);
+	}*/
+
+	/* Disable all tables */
+	WREG32(mmVM_CONTEXT0_CNTL, 0);
+	WREG32(mmVM_CONTEXT1_CNTL, 0);
+	/* Setup TLB control */
+	WREG32(mmMC_VM_MX_L1_TLB_CNTL,
+	       MC_VM_MX_L1_TLB_CNTL__SYSTEM_ACCESS_MODE_MASK |
+	       (0UL << MC_VM_MX_L1_TLB_CNTL__SYSTEM_APERTURE_UNMAPPED_ACCESS__SHIFT));
+	/* Setup L2 cache */
+	WREG32(mmVM_L2_CNTL,
+	       VM_L2_CNTL__ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE_MASK |
+	       VM_L2_CNTL__ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE_MASK |
+	       (7UL << VM_L2_CNTL__EFFECTIVE_L2_QUEUE_SIZE__SHIFT) |
+	       (1UL << VM_L2_CNTL__CONTEXT1_IDENTITY_ACCESS_MODE__SHIFT));
+	WREG32(mmVM_L2_CNTL2, 0);
+	WREG32(mmVM_L2_CNTL3,
+	       VM_L2_CNTL3__L2_CACHE_BIGK_ASSOCIATIVITY_MASK |
+	       (0UL << VM_L2_CNTL3__L2_CACHE_BIGK_FRAGMENT_SIZE__SHIFT));
+	amdgpu_gart_table_vram_unpin(adev);
+}
+
+static void gmc_v6_0_vm_decode_fault(struct amdgpu_device *adev,
+				     u32 status, u32 addr, u32 mc_client)
+{
+	u32 mc_id;
+	u32 vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
+	u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					PROTECTIONS);
+	char block[5] = { mc_client >> 24, (mc_client >> 16) & 0xff,
+		(mc_client >> 8) & 0xff, mc_client & 0xff, 0 };
+
+	mc_id = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			      MEMORY_CLIENT_ID);
+
+	dev_err(adev->dev, "VM fault (0x%02x, vmid %d) at page %u, %s from '%s' (0x%08x) (%d)\n",
+	       protections, vmid, addr,
+	       REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			     MEMORY_CLIENT_RW) ?
+	       "write" : "read", block, mc_client, mc_id);
+}
+
+/*
+static const u32 mc_cg_registers[] = {
+	MC_HUB_MISC_HUB_CG,
+	MC_HUB_MISC_SIP_CG,
+	MC_HUB_MISC_VM_CG,
+	MC_XPB_CLK_GAT,
+	ATC_MISC_CG,
+	MC_CITF_MISC_WR_CG,
+	MC_CITF_MISC_RD_CG,
+	MC_CITF_MISC_VM_CG,
+	VM_L2_CG,
+};
+
+static const u32 mc_cg_ls_en[] = {
+	MC_HUB_MISC_HUB_CG__MEM_LS_ENABLE_MASK,
+	MC_HUB_MISC_SIP_CG__MEM_LS_ENABLE_MASK,
+	MC_HUB_MISC_VM_CG__MEM_LS_ENABLE_MASK,
+	MC_XPB_CLK_GAT__MEM_LS_ENABLE_MASK,
+	ATC_MISC_CG__MEM_LS_ENABLE_MASK,
+	MC_CITF_MISC_WR_CG__MEM_LS_ENABLE_MASK,
+	MC_CITF_MISC_RD_CG__MEM_LS_ENABLE_MASK,
+	MC_CITF_MISC_VM_CG__MEM_LS_ENABLE_MASK,
+	VM_L2_CG__MEM_LS_ENABLE_MASK,
+};
+
+static const u32 mc_cg_en[] = {
+	MC_HUB_MISC_HUB_CG__ENABLE_MASK,
+	MC_HUB_MISC_SIP_CG__ENABLE_MASK,
+	MC_HUB_MISC_VM_CG__ENABLE_MASK,
+	MC_XPB_CLK_GAT__ENABLE_MASK,
+	ATC_MISC_CG__ENABLE_MASK,
+	MC_CITF_MISC_WR_CG__ENABLE_MASK,
+	MC_CITF_MISC_RD_CG__ENABLE_MASK,
+	MC_CITF_MISC_VM_CG__ENABLE_MASK,
+	VM_L2_CG__ENABLE_MASK,
+};
+
+static void gmc_v6_0_enable_mc_ls(struct amdgpu_device *adev,
+				  bool enable)
+{
+	int i;
+	u32 orig, data;
+
+	for (i = 0; i < ARRAY_SIZE(mc_cg_registers); i++) {
+		orig = data = RREG32(mc_cg_registers[i]);
+		if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_MC_LS))
+			data |= mc_cg_ls_en[i];
+		else
+			data &= ~mc_cg_ls_en[i];
+		if (data != orig)
+			WREG32(mc_cg_registers[i], data);
+	}
+}
+
+static void gmc_v6_0_enable_mc_mgcg(struct amdgpu_device *adev,
+				    bool enable)
+{
+	int i;
+	u32 orig, data;
+
+	for (i = 0; i < ARRAY_SIZE(mc_cg_registers); i++) {
+		orig = data = RREG32(mc_cg_registers[i]);
+		if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_MC_MGCG))
+			data |= mc_cg_en[i];
+		else
+			data &= ~mc_cg_en[i];
+		if (data != orig)
+			WREG32(mc_cg_registers[i], data);
+	}
+}
+
+static void gmc_v6_0_enable_bif_mgls(struct amdgpu_device *adev,
+				     bool enable)
+{
+	u32 orig, data;
+
+	orig = data = RREG32_PCIE(ixPCIE_CNTL2);
+
+	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_BIF_LS)) {
+		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_LS_EN, 1);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, MST_MEM_LS_EN, 1);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, REPLAY_MEM_LS_EN, 1);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_AGGRESSIVE_LS_EN, 1);
+	} else {
+		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_LS_EN, 0);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, MST_MEM_LS_EN, 0);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, REPLAY_MEM_LS_EN, 0);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_AGGRESSIVE_LS_EN, 0);
+	}
+
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_CNTL2, data);
+}
+
+static void gmc_v6_0_enable_hdp_mgcg(struct amdgpu_device *adev,
+				     bool enable)
+{
+	u32 orig, data;
+
+	orig = data = RREG32(mmHDP_HOST_PATH_CNTL);
+
+	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_HDP_MGCG))
+		data = REG_SET_FIELD(data, HDP_HOST_PATH_CNTL, CLOCK_GATING_DIS, 0);
+	else
+		data = REG_SET_FIELD(data, HDP_HOST_PATH_CNTL, CLOCK_GATING_DIS, 1);
+
+	if (orig != data)
+		WREG32(mmHDP_HOST_PATH_CNTL, data);
+}
+
+static void gmc_v6_0_enable_hdp_ls(struct amdgpu_device *adev,
+				   bool enable)
+{
+	u32 orig, data;
+
+	orig = data = RREG32(mmHDP_MEM_POWER_LS);
+
+	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_HDP_LS))
+		data = REG_SET_FIELD(data, HDP_MEM_POWER_LS, LS_ENABLE, 1);
+	else
+		data = REG_SET_FIELD(data, HDP_MEM_POWER_LS, LS_ENABLE, 0);
+
+	if (orig != data)
+		WREG32(mmHDP_MEM_POWER_LS, data);
+}
+*/
+
+static int gmc_v6_0_convert_vram_type(int mc_seq_vram_type)
+{
+	switch (mc_seq_vram_type) {
+	case MC_SEQ_MISC0__MT__GDDR1:
+		return AMDGPU_VRAM_TYPE_GDDR1;
+	case MC_SEQ_MISC0__MT__DDR2:
+		return AMDGPU_VRAM_TYPE_DDR2;
+	case MC_SEQ_MISC0__MT__GDDR3:
+		return AMDGPU_VRAM_TYPE_GDDR3;
+	case MC_SEQ_MISC0__MT__GDDR4:
+		return AMDGPU_VRAM_TYPE_GDDR4;
+	case MC_SEQ_MISC0__MT__GDDR5:
+		return AMDGPU_VRAM_TYPE_GDDR5;
+	case MC_SEQ_MISC0__MT__DDR3:
+		return AMDGPU_VRAM_TYPE_DDR3;
+	default:
+		return AMDGPU_VRAM_TYPE_UNKNOWN;
+	}
+}
+
+static int gmc_v6_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v6_0_set_gmc_funcs(adev);
+	gmc_v6_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int gmc_v6_0_late_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_bo_late_init(adev);
+
+	if (amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS)
+		return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
+	else
+		return 0;
+}
+
+static unsigned gmc_v6_0_get_vbios_fb_size(struct amdgpu_device *adev)
+{
+	u32 d1vga_control = RREG32(mmD1VGA_CONTROL);
+	unsigned size;
+
+	if (REG_GET_FIELD(d1vga_control, D1VGA_CONTROL, D1VGA_MODE_ENABLE)) {
+		size = 9 * 1024 * 1024; /* reserve 8MB for vga emulator and 1 MB for FB */
+	} else {
+		u32 viewport = RREG32(mmVIEWPORT_SIZE);
+		size = (REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_HEIGHT) *
+			REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_WIDTH) *
+			4);
+	}
+	/* return 0 if the pre-OS buffer uses up most of vram */
+	if ((adev->gmc.real_vram_size - size) < (8 * 1024 * 1024))
+		return 0;
+	return size;
+}
+
+static int gmc_v6_0_sw_init(void *handle)
+{
+	int r;
+	int dma_bits;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->flags & AMD_IS_APU) {
+		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+	} else {
+		u32 tmp = RREG32(mmMC_SEQ_MISC0);
+		tmp &= MC_SEQ_MISC0__MT__MASK;
+		adev->gmc.vram_type = gmc_v6_0_convert_vram_type(tmp);
+	}
+
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 146, &adev->gmc.vm_fault);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 147, &adev->gmc.vm_fault);
+	if (r)
+		return r;
+
+	amdgpu_vm_adjust_size(adev, 64, 9, 1, 40);
+
+	adev->gmc.mc_mask = 0xffffffffffULL;
+
+	adev->need_dma32 = false;
+#ifdef notyet
+	dma_bits = adev->need_dma32 ? 32 : 40;
+	r = pci_set_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	if (r) {
+		adev->need_dma32 = true;
+		dma_bits = 32;
+		dev_warn(adev->dev, "amdgpu: No suitable DMA available.\n");
+	}
+	r = pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	if (r) {
+		pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(32));
+		dev_warn(adev->dev, "amdgpu: No coherent DMA available.\n");
+	}
+	adev->need_swiotlb = drm_get_max_iomem() > ((u64)1 << dma_bits);
+#endif
+
+	r = gmc_v6_0_init_microcode(adev);
+	if (r) {
+		dev_err(adev->dev, "Failed to load mc firmware!\n");
+		return r;
+	}
+
+	r = gmc_v6_0_mc_init(adev);
+	if (r)
+		return r;
+
+	adev->gmc.stolen_size = gmc_v6_0_get_vbios_fb_size(adev);
+
+	r = amdgpu_bo_init(adev);
+	if (r)
+		return r;
+
+	r = gmc_v6_0_gart_init(adev);
+	if (r)
+		return r;
+
+	/*
+	 * number of VMs
+	 * VMID 0 is reserved for System
+	 * amdgpu graphics/compute will use VMIDs 1-7
+	 * amdkfd will use VMIDs 8-15
+	 */
+	adev->vm_manager.id_mgr[0].num_ids = AMDGPU_NUM_OF_VMIDS;
+	amdgpu_vm_manager_init(adev);
+
+	/* base offset of vram pages */
+	if (adev->flags & AMD_IS_APU) {
+		u64 tmp = RREG32(mmMC_VM_FB_OFFSET);
+
+		tmp <<= 22;
+		adev->vm_manager.vram_base_offset = tmp;
+	} else {
+		adev->vm_manager.vram_base_offset = 0;
+	}
+
+	return 0;
+}
+
+static int gmc_v6_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_gem_force_release(adev);
+	amdgpu_vm_manager_fini(adev);
+	amdgpu_gart_table_vram_free(adev);
+	amdgpu_bo_fini(adev);
+	amdgpu_gart_fini(adev);
+	release_firmware(adev->gmc.fw);
+	adev->gmc.fw = NULL;
+
+	return 0;
+}
+
+static int gmc_v6_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v6_0_mc_program(adev);
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = gmc_v6_0_mc_load_microcode(adev);
+		if (r) {
+			dev_err(adev->dev, "Failed to load MC firmware!\n");
+			return r;
+		}
+	}
+
+	r = gmc_v6_0_gart_enable(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int gmc_v6_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
+	gmc_v6_0_gart_disable(adev);
+
+	return 0;
+}
+
+static int gmc_v6_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v6_0_hw_fini(adev);
+
+	return 0;
+}
+
+static int gmc_v6_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = gmc_v6_0_hw_init(adev);
+	if (r)
+		return r;
+
+	amdgpu_vmid_reset_all(adev);
+
+	return 0;
+}
+
+static bool gmc_v6_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK | SRBM_STATUS__VMC_BUSY_MASK))
+		return false;
+
+	return true;
+}
+
+static int gmc_v6_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (gmc_v6_0_is_idle(handle))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+
+}
+
+static int gmc_v6_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset = 0;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__VMC_BUSY_MASK)
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+						SRBM_SOFT_RESET, SOFT_RESET_VMC, 1);
+
+	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK)) {
+		if (!(adev->flags & AMD_IS_APU))
+			srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+							SRBM_SOFT_RESET, SOFT_RESET_MC, 1);
+	}
+
+	if (srbm_soft_reset) {
+		gmc_v6_0_mc_stop(adev);
+		if (gmc_v6_0_wait_for_idle(adev)) {
+			dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
+		}
+
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		gmc_v6_0_mc_resume(adev);
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int gmc_v6_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	u32 tmp;
+	u32 bits = (VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK);
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		tmp = RREG32(mmVM_CONTEXT0_CNTL);
+		tmp &= ~bits;
+		WREG32(mmVM_CONTEXT0_CNTL, tmp);
+		tmp = RREG32(mmVM_CONTEXT1_CNTL);
+		tmp &= ~bits;
+		WREG32(mmVM_CONTEXT1_CNTL, tmp);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		tmp = RREG32(mmVM_CONTEXT0_CNTL);
+		tmp |= bits;
+		WREG32(mmVM_CONTEXT0_CNTL, tmp);
+		tmp = RREG32(mmVM_CONTEXT1_CNTL);
+		tmp |= bits;
+		WREG32(mmVM_CONTEXT1_CNTL, tmp);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gmc_v6_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	u32 addr, status;
+
+	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
+	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
+	WREG32_P(mmVM_CONTEXT1_CNTL2, 1, ~1);
+
+	if (!addr && !status)
+		return 0;
+
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_FIRST)
+		gmc_v6_0_set_fault_enable_default(adev, false);
+
+#ifdef notyet
+	if (printk_ratelimit()) {
+		dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
+			entry->src_id, entry->src_data[0]);
+		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x%08X\n",
+			addr);
+		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
+			status);
+		gmc_v6_0_vm_decode_fault(adev, status, addr, 0);
+	}
+#endif
+
+	return 0;
+}
+
+static int gmc_v6_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int gmc_v6_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs gmc_v6_0_ip_funcs = {
+	.name = "gmc_v6_0",
+	.early_init = gmc_v6_0_early_init,
+	.late_init = gmc_v6_0_late_init,
+	.sw_init = gmc_v6_0_sw_init,
+	.sw_fini = gmc_v6_0_sw_fini,
+	.hw_init = gmc_v6_0_hw_init,
+	.hw_fini = gmc_v6_0_hw_fini,
+	.suspend = gmc_v6_0_suspend,
+	.resume = gmc_v6_0_resume,
+	.is_idle = gmc_v6_0_is_idle,
+	.wait_for_idle = gmc_v6_0_wait_for_idle,
+	.soft_reset = gmc_v6_0_soft_reset,
+	.set_clockgating_state = gmc_v6_0_set_clockgating_state,
+	.set_powergating_state = gmc_v6_0_set_powergating_state,
+};
+
+static const struct amdgpu_gmc_funcs gmc_v6_0_gmc_funcs = {
+	.flush_gpu_tlb = gmc_v6_0_flush_gpu_tlb,
+	.emit_flush_gpu_tlb = gmc_v6_0_emit_flush_gpu_tlb,
+	.set_pte_pde = gmc_v6_0_set_pte_pde,
+	.set_prt = gmc_v6_0_set_prt,
+	.get_vm_pde = gmc_v6_0_get_vm_pde,
+	.get_vm_pte_flags = gmc_v6_0_get_vm_pte_flags
+};
+
+static const struct amdgpu_irq_src_funcs gmc_v6_0_irq_funcs = {
+	.set = gmc_v6_0_vm_fault_interrupt_state,
+	.process = gmc_v6_0_process_interrupt,
+};
+
+static void gmc_v6_0_set_gmc_funcs(struct amdgpu_device *adev)
+{
+	if (adev->gmc.gmc_funcs == NULL)
+		adev->gmc.gmc_funcs = &gmc_v6_0_gmc_funcs;
+}
+
+static void gmc_v6_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gmc.vm_fault.num_types = 1;
+	adev->gmc.vm_fault.funcs = &gmc_v6_0_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version gmc_v6_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GMC,
+	.major = 6,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gmc_v6_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/gmc_v6_0.h b/sys/dev/pci/drm/amd/amdgpu/gmc_v6_0.h
new file mode 100644
index 00000000000..ed2f64dec47
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gmc_v6_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GMC_V6_0_H__
+#define __GMC_V6_0_H__
+
+extern const struct amdgpu_ip_block_version gmc_v6_0_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/gmc_v7_0.c b/sys/dev/pci/drm/amd/amdgpu/gmc_v7_0.c
new file mode 100644
index 00000000000..2685032dbd3
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gmc_v7_0.c
@@ -0,0 +1,1416 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include <drm/drm_cache.h>
+#include "amdgpu.h"
+#include "cikd.h"
+#include "cik.h"
+#include "gmc_v7_0.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_amdkfd.h"
+
+#include "bif/bif_4_1_d.h"
+#include "bif/bif_4_1_sh_mask.h"
+
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+#include "dce/dce_8_0_d.h"
+#include "dce/dce_8_0_sh_mask.h"
+
+#include "amdgpu_atombios.h"
+
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+static void gmc_v7_0_set_gmc_funcs(struct amdgpu_device *adev);
+static void gmc_v7_0_set_irq_funcs(struct amdgpu_device *adev);
+static int gmc_v7_0_wait_for_idle(void *handle);
+
+MODULE_FIRMWARE("amdgpu/bonaire_mc.bin");
+MODULE_FIRMWARE("amdgpu/hawaii_mc.bin");
+MODULE_FIRMWARE("amdgpu/topaz_mc.bin");
+
+static const u32 golden_settings_iceland_a11[] =
+{
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff
+};
+
+static const u32 iceland_mgcg_cgcg_init[] =
+{
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
+};
+
+static void gmc_v7_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		amdgpu_device_program_register_sequence(adev,
+							iceland_mgcg_cgcg_init,
+							ARRAY_SIZE(iceland_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_iceland_a11,
+							ARRAY_SIZE(golden_settings_iceland_a11));
+		break;
+	default:
+		break;
+	}
+}
+
+static void gmc_v7_0_mc_stop(struct amdgpu_device *adev)
+{
+	u32 blackout;
+
+	gmc_v7_0_wait_for_idle((void *)adev);
+
+	blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
+	if (REG_GET_FIELD(blackout, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE) != 1) {
+		/* Block CPU access */
+		WREG32(mmBIF_FB_EN, 0);
+		/* blackout the MC */
+		blackout = REG_SET_FIELD(blackout,
+					 MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 0);
+		WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout | 1);
+	}
+	/* wait for the MC to settle */
+	udelay(100);
+}
+
+static void gmc_v7_0_mc_resume(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	/* unblackout the MC */
+	tmp = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 0);
+	WREG32(mmMC_SHARED_BLACKOUT_CNTL, tmp);
+	/* allow CPU access */
+	tmp = REG_SET_FIELD(0, BIF_FB_EN, FB_READ_EN, 1);
+	tmp = REG_SET_FIELD(tmp, BIF_FB_EN, FB_WRITE_EN, 1);
+	WREG32(mmBIF_FB_EN, tmp);
+}
+
+/**
+ * gmc_v7_0_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int gmc_v7_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		chip_name = "bonaire";
+		break;
+	case CHIP_HAWAII:
+		chip_name = "hawaii";
+		break;
+	case CHIP_TOPAZ:
+		chip_name = "topaz";
+		break;
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		return 0;
+	default: BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mc.bin", chip_name);
+
+	err = request_firmware(&adev->gmc.fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gmc.fw);
+
+out:
+	if (err) {
+		pr_err("cik_mc: Failed to load firmware \"%s\"\n", fw_name);
+		release_firmware(adev->gmc.fw);
+		adev->gmc.fw = NULL;
+	}
+	return err;
+}
+
+/**
+ * gmc_v7_0_mc_load_microcode - load MC ucode into the hw
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Load the GDDR MC ucode into the hw (CIK).
+ * Returns 0 on success, error on failure.
+ */
+static int gmc_v7_0_mc_load_microcode(struct amdgpu_device *adev)
+{
+	const struct mc_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data = NULL;
+	const __le32 *io_mc_regs = NULL;
+	u32 running;
+	int i, ucode_size, regs_size;
+
+	if (!adev->gmc.fw)
+		return -EINVAL;
+
+	hdr = (const struct mc_firmware_header_v1_0 *)adev->gmc.fw->data;
+	amdgpu_ucode_print_mc_hdr(&hdr->header);
+
+	adev->gmc.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	regs_size = le32_to_cpu(hdr->io_debug_size_bytes) / (4 * 2);
+	io_mc_regs = (const __le32 *)
+		(adev->gmc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
+	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+	fw_data = (const __le32 *)
+		(adev->gmc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	running = REG_GET_FIELD(RREG32(mmMC_SEQ_SUP_CNTL), MC_SEQ_SUP_CNTL, RUN);
+
+	if (running == 0) {
+		/* reset the engine and set to writable */
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000010);
+
+		/* load mc io regs */
+		for (i = 0; i < regs_size; i++) {
+			WREG32(mmMC_SEQ_IO_DEBUG_INDEX, le32_to_cpup(io_mc_regs++));
+			WREG32(mmMC_SEQ_IO_DEBUG_DATA, le32_to_cpup(io_mc_regs++));
+		}
+		/* load the MC ucode */
+		for (i = 0; i < ucode_size; i++)
+			WREG32(mmMC_SEQ_SUP_PGM, le32_to_cpup(fw_data++));
+
+		/* put the engine back into the active state */
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000004);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000001);
+
+		/* wait for training to complete */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (REG_GET_FIELD(RREG32(mmMC_SEQ_TRAIN_WAKEUP_CNTL),
+					  MC_SEQ_TRAIN_WAKEUP_CNTL, TRAIN_DONE_D0))
+				break;
+			udelay(1);
+		}
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (REG_GET_FIELD(RREG32(mmMC_SEQ_TRAIN_WAKEUP_CNTL),
+					  MC_SEQ_TRAIN_WAKEUP_CNTL, TRAIN_DONE_D1))
+				break;
+			udelay(1);
+		}
+	}
+
+	return 0;
+}
+
+static void gmc_v7_0_vram_gtt_location(struct amdgpu_device *adev,
+				       struct amdgpu_gmc *mc)
+{
+	u64 base = RREG32(mmMC_VM_FB_LOCATION) & 0xFFFF;
+	base <<= 24;
+
+	amdgpu_device_vram_location(adev, &adev->gmc, base);
+	amdgpu_device_gart_location(adev, mc);
+}
+
+/**
+ * gmc_v7_0_mc_program - program the GPU memory controller
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the location of vram, gart, and AGP in the GPU's
+ * physical address space (CIK).
+ */
+static void gmc_v7_0_mc_program(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	int i, j;
+
+	/* Initialize HDP */
+	for (i = 0, j = 0; i < 32; i++, j += 0x6) {
+		WREG32((0xb05 + j), 0x00000000);
+		WREG32((0xb06 + j), 0x00000000);
+		WREG32((0xb07 + j), 0x00000000);
+		WREG32((0xb08 + j), 0x00000000);
+		WREG32((0xb09 + j), 0x00000000);
+	}
+	WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
+
+	if (gmc_v7_0_wait_for_idle((void *)adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
+	}
+	if (adev->mode_info.num_crtc) {
+		/* Lockout access through VGA aperture*/
+		tmp = RREG32(mmVGA_HDP_CONTROL);
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
+		WREG32(mmVGA_HDP_CONTROL, tmp);
+
+		/* disable VGA render */
+		tmp = RREG32(mmVGA_RENDER_CONTROL);
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
+		WREG32(mmVGA_RENDER_CONTROL, tmp);
+	}
+	/* Update configuration */
+	WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+	       adev->gmc.vram_start >> 12);
+	WREG32(mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+	       adev->gmc.vram_end >> 12);
+	WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
+	       adev->vram_scratch.gpu_addr >> 12);
+	WREG32(mmMC_VM_AGP_BASE, 0);
+	WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
+	WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
+	if (gmc_v7_0_wait_for_idle((void *)adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
+	}
+
+	WREG32(mmBIF_FB_EN, BIF_FB_EN__FB_READ_EN_MASK | BIF_FB_EN__FB_WRITE_EN_MASK);
+
+	tmp = RREG32(mmHDP_MISC_CNTL);
+	tmp = REG_SET_FIELD(tmp, HDP_MISC_CNTL, FLUSH_INVALIDATE_CACHE, 0);
+	WREG32(mmHDP_MISC_CNTL, tmp);
+
+	tmp = RREG32(mmHDP_HOST_PATH_CNTL);
+	WREG32(mmHDP_HOST_PATH_CNTL, tmp);
+}
+
+/**
+ * gmc_v7_0_mc_init - initialize the memory controller driver params
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the amount of vram, vram width, and decide how to place
+ * vram and gart within the GPU's physical address space (CIK).
+ * Returns 0 for success.
+ */
+static int gmc_v7_0_mc_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	adev->gmc.vram_width = amdgpu_atombios_get_vram_width(adev);
+	if (!adev->gmc.vram_width) {
+		u32 tmp;
+		int chansize, numchan;
+
+		/* Get VRAM informations */
+		tmp = RREG32(mmMC_ARB_RAMCFG);
+		if (REG_GET_FIELD(tmp, MC_ARB_RAMCFG, CHANSIZE)) {
+			chansize = 64;
+		} else {
+			chansize = 32;
+		}
+		tmp = RREG32(mmMC_SHARED_CHMAP);
+		switch (REG_GET_FIELD(tmp, MC_SHARED_CHMAP, NOOFCHAN)) {
+		case 0:
+		default:
+			numchan = 1;
+			break;
+		case 1:
+			numchan = 2;
+			break;
+		case 2:
+			numchan = 4;
+			break;
+		case 3:
+			numchan = 8;
+			break;
+		case 4:
+			numchan = 3;
+			break;
+		case 5:
+			numchan = 6;
+			break;
+		case 6:
+			numchan = 10;
+			break;
+		case 7:
+			numchan = 12;
+			break;
+		case 8:
+			numchan = 16;
+			break;
+		}
+		adev->gmc.vram_width = numchan * chansize;
+	}
+	/* size in MB on si */
+	adev->gmc.mc_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+	adev->gmc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_device_resize_fb_bar(adev);
+		if (r)
+			return r;
+	}
+	adev->gmc.aper_base = adev->fb_aper_offset;
+	adev->gmc.aper_size = adev->fb_aper_size;
+
+#ifdef __amd64__
+	if (adev->flags & AMD_IS_APU) {
+		adev->gmc.aper_base = ((u64)RREG32(mmMC_VM_FB_OFFSET)) << 22;
+		adev->gmc.aper_size = adev->gmc.real_vram_size;
+	}
+#endif
+
+	/* In case the PCI BAR is larger than the actual amount of vram */
+	adev->gmc.visible_vram_size = adev->gmc.aper_size;
+	if (adev->gmc.visible_vram_size > adev->gmc.real_vram_size)
+		adev->gmc.visible_vram_size = adev->gmc.real_vram_size;
+
+	/* set the gart size */
+	if (amdgpu_gart_size == -1) {
+		switch (adev->asic_type) {
+		case CHIP_TOPAZ:     /* no MM engines */
+		default:
+			adev->gmc.gart_size = 256ULL << 20;
+			break;
+#ifdef CONFIG_DRM_AMDGPU_CIK
+		case CHIP_BONAIRE: /* UVD, VCE do not support GPUVM */
+		case CHIP_HAWAII:  /* UVD, VCE do not support GPUVM */
+		case CHIP_KAVERI:  /* UVD, VCE do not support GPUVM */
+		case CHIP_KABINI:  /* UVD, VCE do not support GPUVM */
+		case CHIP_MULLINS: /* UVD, VCE do not support GPUVM */
+			adev->gmc.gart_size = 1024ULL << 20;
+			break;
+#endif
+		}
+	} else {
+		adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
+	}
+
+	gmc_v7_0_vram_gtt_location(adev, &adev->gmc);
+
+	return 0;
+}
+
+/*
+ * GART
+ * VMID 0 is the physical GPU addresses as used by the kernel.
+ * VMIDs 1-15 are used for userspace clients and are handled
+ * by the amdgpu vm/hsa code.
+ */
+
+/**
+ * gmc_v7_0_flush_gpu_tlb - gart tlb flush callback
+ *
+ * @adev: amdgpu_device pointer
+ * @vmid: vm instance to flush
+ *
+ * Flush the TLB for the requested page table (CIK).
+ */
+static void gmc_v7_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid)
+{
+	/* bits 0-15 are the VM contexts0-15 */
+	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+}
+
+static uint64_t gmc_v7_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
+					    unsigned vmid, uint64_t pd_addr)
+{
+	uint32_t reg;
+
+	if (vmid < 8)
+		reg = mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vmid;
+	else
+		reg = mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8;
+	amdgpu_ring_emit_wreg(ring, reg, pd_addr >> 12);
+
+	/* bits 0-15 are the VM contexts0-15 */
+	amdgpu_ring_emit_wreg(ring, mmVM_INVALIDATE_REQUEST, 1 << vmid);
+
+	return pd_addr;
+}
+
+static void gmc_v7_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned vmid,
+					unsigned pasid)
+{
+	amdgpu_ring_emit_wreg(ring, mmIH_VMID_0_LUT + vmid, pasid);
+}
+
+/**
+ * gmc_v7_0_set_pte_pde - update the page tables using MMIO
+ *
+ * @adev: amdgpu_device pointer
+ * @cpu_pt_addr: cpu address of the page table
+ * @gpu_page_idx: entry in the page table to update
+ * @addr: dst addr to write into pte/pde
+ * @flags: access flags
+ *
+ * Update the page tables using the CPU.
+ */
+static int gmc_v7_0_set_pte_pde(struct amdgpu_device *adev, void *cpu_pt_addr,
+				 uint32_t gpu_page_idx, uint64_t addr,
+				 uint64_t flags)
+{
+	void __iomem *ptr = (void *)cpu_pt_addr;
+	uint64_t value;
+
+	value = addr & 0xFFFFFFFFFFFFF000ULL;
+	value |= flags;
+	writeq(value, ptr + (gpu_page_idx * 8));
+
+	return 0;
+}
+
+static uint64_t gmc_v7_0_get_vm_pte_flags(struct amdgpu_device *adev,
+					  uint32_t flags)
+{
+	uint64_t pte_flag = 0;
+
+	if (flags & AMDGPU_VM_PAGE_READABLE)
+		pte_flag |= AMDGPU_PTE_READABLE;
+	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
+		pte_flag |= AMDGPU_PTE_WRITEABLE;
+	if (flags & AMDGPU_VM_PAGE_PRT)
+		pte_flag |= AMDGPU_PTE_PRT;
+
+	return pte_flag;
+}
+
+static void gmc_v7_0_get_vm_pde(struct amdgpu_device *adev, int level,
+				uint64_t *addr, uint64_t *flags)
+{
+	BUG_ON(*addr & 0xFFFFFF0000000FFFULL);
+}
+
+/**
+ * gmc_v8_0_set_fault_enable_default - update VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void gmc_v7_0_set_fault_enable_default(struct amdgpu_device *adev,
+					      bool value)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmVM_CONTEXT1_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	WREG32(mmVM_CONTEXT1_CNTL, tmp);
+}
+
+/**
+ * gmc_v7_0_set_prt - set PRT VM fault
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable VM fault handling for PRT
+ */
+static void gmc_v7_0_set_prt(struct amdgpu_device *adev, bool enable)
+{
+	uint32_t tmp;
+
+	if (enable && !adev->gmc.prt_warning) {
+		dev_warn(adev->dev, "Disabling VM faults because of PRT request!\n");
+		adev->gmc.prt_warning = true;
+	}
+
+	tmp = RREG32(mmVM_PRT_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    CB_DISABLE_READ_FAULT_ON_UNMAPPED_ACCESS, enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    CB_DISABLE_WRITE_FAULT_ON_UNMAPPED_ACCESS, enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    TC_DISABLE_READ_FAULT_ON_UNMAPPED_ACCESS, enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    TC_DISABLE_WRITE_FAULT_ON_UNMAPPED_ACCESS, enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    L2_CACHE_STORE_INVALID_ENTRIES, enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    L1_TLB_STORE_INVALID_ENTRIES, enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    MASK_PDE0_FAULT, enable);
+	WREG32(mmVM_PRT_CNTL, tmp);
+
+	if (enable) {
+		uint32_t low = AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT;
+		uint32_t high = adev->vm_manager.max_pfn -
+			(AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT);
+
+		WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, low);
+		WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, low);
+		WREG32(mmVM_PRT_APERTURE2_LOW_ADDR, low);
+		WREG32(mmVM_PRT_APERTURE3_LOW_ADDR, low);
+		WREG32(mmVM_PRT_APERTURE0_HIGH_ADDR, high);
+		WREG32(mmVM_PRT_APERTURE1_HIGH_ADDR, high);
+		WREG32(mmVM_PRT_APERTURE2_HIGH_ADDR, high);
+		WREG32(mmVM_PRT_APERTURE3_HIGH_ADDR, high);
+	} else {
+		WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, 0xfffffff);
+		WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, 0xfffffff);
+		WREG32(mmVM_PRT_APERTURE2_LOW_ADDR, 0xfffffff);
+		WREG32(mmVM_PRT_APERTURE3_LOW_ADDR, 0xfffffff);
+		WREG32(mmVM_PRT_APERTURE0_HIGH_ADDR, 0x0);
+		WREG32(mmVM_PRT_APERTURE1_HIGH_ADDR, 0x0);
+		WREG32(mmVM_PRT_APERTURE2_HIGH_ADDR, 0x0);
+		WREG32(mmVM_PRT_APERTURE3_HIGH_ADDR, 0x0);
+	}
+}
+
+/**
+ * gmc_v7_0_gart_enable - gart enable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This sets up the TLBs, programs the page tables for VMID0,
+ * sets up the hw for VMIDs 1-15 which are allocated on
+ * demand, and sets up the global locations for the LDS, GDS,
+ * and GPUVM for FSA64 clients (CIK).
+ * Returns 0 for success, errors for failure.
+ */
+static int gmc_v7_0_gart_enable(struct amdgpu_device *adev)
+{
+	int r, i;
+	u32 tmp, field;
+
+	if (adev->gart.robj == NULL) {
+		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
+		return -EINVAL;
+	}
+	r = amdgpu_gart_table_vram_pin(adev);
+	if (r)
+		return r;
+	/* Setup TLB control */
+	tmp = RREG32(mmMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_FRAGMENT_PROCESSING, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	WREG32(mmMC_VM_MX_L1_TLB_CNTL, tmp);
+	/* Setup L2 cache */
+	tmp = RREG32(mmVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, EFFECTIVE_L2_QUEUE_SIZE, 7);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	WREG32(mmVM_L2_CNTL, tmp);
+	tmp = REG_SET_FIELD(0, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32(mmVM_L2_CNTL2, tmp);
+
+	field = adev->vm_manager.fragment_size;
+	tmp = RREG32(mmVM_L2_CNTL3);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_ASSOCIATIVITY, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, field);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, field);
+	WREG32(mmVM_L2_CNTL3, tmp);
+	/* setup context0 */
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->gmc.gart_start >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->gmc.gart_end >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
+	WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
+			(u32)(adev->dummy_page_addr >> 12));
+	WREG32(mmVM_CONTEXT0_CNTL2, 0);
+	tmp = RREG32(mmVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	WREG32(mmVM_CONTEXT0_CNTL, tmp);
+
+	WREG32(0x575, 0);
+	WREG32(0x576, 0);
+	WREG32(0x577, 0);
+
+	/* empty context1-15 */
+	/* FIXME start with 4G, once using 2 level pt switch to full
+	 * vm size space
+	 */
+	/* set vm size, must be a multiple of 4 */
+	WREG32(mmVM_CONTEXT1_PAGE_TABLE_START_ADDR, 0);
+	WREG32(mmVM_CONTEXT1_PAGE_TABLE_END_ADDR, adev->vm_manager.max_pfn - 1);
+	for (i = 1; i < 16; i++) {
+		if (i < 8)
+			WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i,
+			       adev->gart.table_addr >> 12);
+		else
+			WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + i - 8,
+			       adev->gart.table_addr >> 12);
+	}
+
+	/* enable context1-15 */
+	WREG32(mmVM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR,
+	       (u32)(adev->dummy_page_addr >> 12));
+	WREG32(mmVM_CONTEXT1_CNTL2, 4);
+	tmp = RREG32(mmVM_CONTEXT1_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_BLOCK_SIZE,
+			    adev->vm_manager.block_size - 9);
+	WREG32(mmVM_CONTEXT1_CNTL, tmp);
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS)
+		gmc_v7_0_set_fault_enable_default(adev, false);
+	else
+		gmc_v7_0_set_fault_enable_default(adev, true);
+
+	if (adev->asic_type == CHIP_KAVERI) {
+		tmp = RREG32(mmCHUB_CONTROL);
+		tmp &= ~BYPASS_VM;
+		WREG32(mmCHUB_CONTROL, tmp);
+	}
+
+	gmc_v7_0_flush_gpu_tlb(adev, 0);
+	DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
+		 (unsigned)(adev->gmc.gart_size >> 20),
+		 (unsigned long long)adev->gart.table_addr);
+	adev->gart.ready = true;
+	return 0;
+}
+
+static int gmc_v7_0_gart_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.robj) {
+		WARN(1, "R600 PCIE GART already initialized\n");
+		return 0;
+	}
+	/* Initialize common gart structure */
+	r = amdgpu_gart_init(adev);
+	if (r)
+		return r;
+	adev->gart.table_size = adev->gart.num_gpu_pages * 8;
+	adev->gart.gart_pte_flags = 0;
+	return amdgpu_gart_table_vram_alloc(adev);
+}
+
+/**
+ * gmc_v7_0_gart_disable - gart disable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This disables all VM page table (CIK).
+ */
+static void gmc_v7_0_gart_disable(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	/* Disable all tables */
+	WREG32(mmVM_CONTEXT0_CNTL, 0);
+	WREG32(mmVM_CONTEXT1_CNTL, 0);
+	/* Setup TLB control */
+	tmp = RREG32(mmMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32(mmMC_VM_MX_L1_TLB_CNTL, tmp);
+	/* Setup L2 cache */
+	tmp = RREG32(mmVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32(mmVM_L2_CNTL, tmp);
+	WREG32(mmVM_L2_CNTL2, 0);
+	amdgpu_gart_table_vram_unpin(adev);
+}
+
+/**
+ * gmc_v7_0_vm_decode_fault - print human readable fault info
+ *
+ * @adev: amdgpu_device pointer
+ * @status: VM_CONTEXT1_PROTECTION_FAULT_STATUS register value
+ * @addr: VM_CONTEXT1_PROTECTION_FAULT_ADDR register value
+ *
+ * Print human readable fault information (CIK).
+ */
+static void gmc_v7_0_vm_decode_fault(struct amdgpu_device *adev, u32 status,
+				     u32 addr, u32 mc_client, unsigned pasid)
+{
+	u32 vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
+	u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					PROTECTIONS);
+	char block[5] = { mc_client >> 24, (mc_client >> 16) & 0xff,
+		(mc_client >> 8) & 0xff, mc_client & 0xff, 0 };
+	u32 mc_id;
+
+	mc_id = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			      MEMORY_CLIENT_ID);
+
+	dev_err(adev->dev, "VM fault (0x%02x, vmid %d, pasid %d) at page %u, %s from '%s' (0x%08x) (%d)\n",
+	       protections, vmid, pasid, addr,
+	       REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			     MEMORY_CLIENT_RW) ?
+	       "write" : "read", block, mc_client, mc_id);
+}
+
+
+static const u32 mc_cg_registers[] = {
+	mmMC_HUB_MISC_HUB_CG,
+	mmMC_HUB_MISC_SIP_CG,
+	mmMC_HUB_MISC_VM_CG,
+	mmMC_XPB_CLK_GAT,
+	mmATC_MISC_CG,
+	mmMC_CITF_MISC_WR_CG,
+	mmMC_CITF_MISC_RD_CG,
+	mmMC_CITF_MISC_VM_CG,
+	mmVM_L2_CG,
+};
+
+static const u32 mc_cg_ls_en[] = {
+	MC_HUB_MISC_HUB_CG__MEM_LS_ENABLE_MASK,
+	MC_HUB_MISC_SIP_CG__MEM_LS_ENABLE_MASK,
+	MC_HUB_MISC_VM_CG__MEM_LS_ENABLE_MASK,
+	MC_XPB_CLK_GAT__MEM_LS_ENABLE_MASK,
+	ATC_MISC_CG__MEM_LS_ENABLE_MASK,
+	MC_CITF_MISC_WR_CG__MEM_LS_ENABLE_MASK,
+	MC_CITF_MISC_RD_CG__MEM_LS_ENABLE_MASK,
+	MC_CITF_MISC_VM_CG__MEM_LS_ENABLE_MASK,
+	VM_L2_CG__MEM_LS_ENABLE_MASK,
+};
+
+static const u32 mc_cg_en[] = {
+	MC_HUB_MISC_HUB_CG__ENABLE_MASK,
+	MC_HUB_MISC_SIP_CG__ENABLE_MASK,
+	MC_HUB_MISC_VM_CG__ENABLE_MASK,
+	MC_XPB_CLK_GAT__ENABLE_MASK,
+	ATC_MISC_CG__ENABLE_MASK,
+	MC_CITF_MISC_WR_CG__ENABLE_MASK,
+	MC_CITF_MISC_RD_CG__ENABLE_MASK,
+	MC_CITF_MISC_VM_CG__ENABLE_MASK,
+	VM_L2_CG__ENABLE_MASK,
+};
+
+static void gmc_v7_0_enable_mc_ls(struct amdgpu_device *adev,
+				  bool enable)
+{
+	int i;
+	u32 orig, data;
+
+	for (i = 0; i < ARRAY_SIZE(mc_cg_registers); i++) {
+		orig = data = RREG32(mc_cg_registers[i]);
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS))
+			data |= mc_cg_ls_en[i];
+		else
+			data &= ~mc_cg_ls_en[i];
+		if (data != orig)
+			WREG32(mc_cg_registers[i], data);
+	}
+}
+
+static void gmc_v7_0_enable_mc_mgcg(struct amdgpu_device *adev,
+				    bool enable)
+{
+	int i;
+	u32 orig, data;
+
+	for (i = 0; i < ARRAY_SIZE(mc_cg_registers); i++) {
+		orig = data = RREG32(mc_cg_registers[i]);
+		if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG))
+			data |= mc_cg_en[i];
+		else
+			data &= ~mc_cg_en[i];
+		if (data != orig)
+			WREG32(mc_cg_registers[i], data);
+	}
+}
+
+static void gmc_v7_0_enable_bif_mgls(struct amdgpu_device *adev,
+				     bool enable)
+{
+	u32 orig, data;
+
+	orig = data = RREG32_PCIE(ixPCIE_CNTL2);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS)) {
+		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_LS_EN, 1);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, MST_MEM_LS_EN, 1);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, REPLAY_MEM_LS_EN, 1);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_AGGRESSIVE_LS_EN, 1);
+	} else {
+		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_LS_EN, 0);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, MST_MEM_LS_EN, 0);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, REPLAY_MEM_LS_EN, 0);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_AGGRESSIVE_LS_EN, 0);
+	}
+
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_CNTL2, data);
+}
+
+static void gmc_v7_0_enable_hdp_mgcg(struct amdgpu_device *adev,
+				     bool enable)
+{
+	u32 orig, data;
+
+	orig = data = RREG32(mmHDP_HOST_PATH_CNTL);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_MGCG))
+		data = REG_SET_FIELD(data, HDP_HOST_PATH_CNTL, CLOCK_GATING_DIS, 0);
+	else
+		data = REG_SET_FIELD(data, HDP_HOST_PATH_CNTL, CLOCK_GATING_DIS, 1);
+
+	if (orig != data)
+		WREG32(mmHDP_HOST_PATH_CNTL, data);
+}
+
+static void gmc_v7_0_enable_hdp_ls(struct amdgpu_device *adev,
+				   bool enable)
+{
+	u32 orig, data;
+
+	orig = data = RREG32(mmHDP_MEM_POWER_LS);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
+		data = REG_SET_FIELD(data, HDP_MEM_POWER_LS, LS_ENABLE, 1);
+	else
+		data = REG_SET_FIELD(data, HDP_MEM_POWER_LS, LS_ENABLE, 0);
+
+	if (orig != data)
+		WREG32(mmHDP_MEM_POWER_LS, data);
+}
+
+static int gmc_v7_0_convert_vram_type(int mc_seq_vram_type)
+{
+	switch (mc_seq_vram_type) {
+	case MC_SEQ_MISC0__MT__GDDR1:
+		return AMDGPU_VRAM_TYPE_GDDR1;
+	case MC_SEQ_MISC0__MT__DDR2:
+		return AMDGPU_VRAM_TYPE_DDR2;
+	case MC_SEQ_MISC0__MT__GDDR3:
+		return AMDGPU_VRAM_TYPE_GDDR3;
+	case MC_SEQ_MISC0__MT__GDDR4:
+		return AMDGPU_VRAM_TYPE_GDDR4;
+	case MC_SEQ_MISC0__MT__GDDR5:
+		return AMDGPU_VRAM_TYPE_GDDR5;
+	case MC_SEQ_MISC0__MT__HBM:
+		return AMDGPU_VRAM_TYPE_HBM;
+	case MC_SEQ_MISC0__MT__DDR3:
+		return AMDGPU_VRAM_TYPE_DDR3;
+	default:
+		return AMDGPU_VRAM_TYPE_UNKNOWN;
+	}
+}
+
+static int gmc_v7_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v7_0_set_gmc_funcs(adev);
+	gmc_v7_0_set_irq_funcs(adev);
+
+	adev->gmc.shared_aperture_start = 0x2000000000000000ULL;
+	adev->gmc.shared_aperture_end =
+		adev->gmc.shared_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.private_aperture_start =
+		adev->gmc.shared_aperture_end + 1;
+	adev->gmc.private_aperture_end =
+		adev->gmc.private_aperture_start + (4ULL << 30) - 1;
+
+	return 0;
+}
+
+static int gmc_v7_0_late_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_bo_late_init(adev);
+
+	if (amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS)
+		return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
+	else
+		return 0;
+}
+
+static unsigned gmc_v7_0_get_vbios_fb_size(struct amdgpu_device *adev)
+{
+	u32 d1vga_control = RREG32(mmD1VGA_CONTROL);
+	unsigned size;
+
+	if (REG_GET_FIELD(d1vga_control, D1VGA_CONTROL, D1VGA_MODE_ENABLE)) {
+		size = 9 * 1024 * 1024; /* reserve 8MB for vga emulator and 1 MB for FB */
+	} else {
+		u32 viewport = RREG32(mmVIEWPORT_SIZE);
+		size = (REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_HEIGHT) *
+			REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_WIDTH) *
+			4);
+	}
+	/* return 0 if the pre-OS buffer uses up most of vram */
+	if ((adev->gmc.real_vram_size - size) < (8 * 1024 * 1024))
+		return 0;
+	return size;
+}
+
+static int gmc_v7_0_sw_init(void *handle)
+{
+	int r;
+	int dma_bits;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->flags & AMD_IS_APU) {
+		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+	} else {
+		u32 tmp = RREG32(mmMC_SEQ_MISC0);
+		tmp &= MC_SEQ_MISC0__MT__MASK;
+		adev->gmc.vram_type = gmc_v7_0_convert_vram_type(tmp);
+	}
+
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_PAGE_INV_FAULT, &adev->gmc.vm_fault);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_MEM_PROT_FAULT, &adev->gmc.vm_fault);
+	if (r)
+		return r;
+
+	/* Adjust VM size here.
+	 * Currently set to 4GB ((1 << 20) 4k pages).
+	 * Max GPUVM size for cayman and SI is 40 bits.
+	 */
+	amdgpu_vm_adjust_size(adev, 64, 9, 1, 40);
+
+	/* Set the internal MC address mask
+	 * This is the max address of the GPU's
+	 * internal address space.
+	 */
+	adev->gmc.mc_mask = 0xffffffffffULL; /* 40 bit MC */
+
+	/* set DMA mask + need_dma32 flags.
+	 * PCIE - can handle 40-bits.
+	 * IGP - can handle 40-bits
+	 * PCI - dma32 for legacy pci gart, 40 bits on newer asics
+	 */
+	adev->need_dma32 = false;
+#ifdef notyet
+	dma_bits = adev->need_dma32 ? 32 : 40;
+	r = pci_set_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	if (r) {
+		adev->need_dma32 = true;
+		dma_bits = 32;
+		pr_warn("amdgpu: No suitable DMA available\n");
+	}
+	r = pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	if (r) {
+		pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(32));
+		pr_warn("amdgpu: No coherent DMA available\n");
+	}
+	adev->need_swiotlb = drm_get_max_iomem() > ((u64)1 << dma_bits);
+#endif
+
+	r = gmc_v7_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load mc firmware!\n");
+		return r;
+	}
+
+	r = gmc_v7_0_mc_init(adev);
+	if (r)
+		return r;
+
+	adev->gmc.stolen_size = gmc_v7_0_get_vbios_fb_size(adev);
+
+	/* Memory manager */
+	r = amdgpu_bo_init(adev);
+	if (r)
+		return r;
+
+	r = gmc_v7_0_gart_init(adev);
+	if (r)
+		return r;
+
+	/*
+	 * number of VMs
+	 * VMID 0 is reserved for System
+	 * amdgpu graphics/compute will use VMIDs 1-7
+	 * amdkfd will use VMIDs 8-15
+	 */
+	adev->vm_manager.id_mgr[0].num_ids = AMDGPU_NUM_OF_VMIDS;
+	amdgpu_vm_manager_init(adev);
+
+	/* base offset of vram pages */
+	if (adev->flags & AMD_IS_APU) {
+		u64 tmp = RREG32(mmMC_VM_FB_OFFSET);
+
+		tmp <<= 22;
+		adev->vm_manager.vram_base_offset = tmp;
+	} else {
+		adev->vm_manager.vram_base_offset = 0;
+	}
+
+	adev->gmc.vm_fault_info = kmalloc(sizeof(struct kfd_vm_fault_info),
+					GFP_KERNEL);
+	if (!adev->gmc.vm_fault_info)
+		return -ENOMEM;
+	atomic_set(&adev->gmc.vm_fault_info_updated, 0);
+
+	return 0;
+}
+
+static int gmc_v7_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_gem_force_release(adev);
+	amdgpu_vm_manager_fini(adev);
+	kfree(adev->gmc.vm_fault_info);
+	amdgpu_gart_table_vram_free(adev);
+	amdgpu_bo_fini(adev);
+	amdgpu_gart_fini(adev);
+	release_firmware(adev->gmc.fw);
+	adev->gmc.fw = NULL;
+
+	return 0;
+}
+
+static int gmc_v7_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v7_0_init_golden_registers(adev);
+
+	gmc_v7_0_mc_program(adev);
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = gmc_v7_0_mc_load_microcode(adev);
+		if (r) {
+			DRM_ERROR("Failed to load MC firmware!\n");
+			return r;
+		}
+	}
+
+	r = gmc_v7_0_gart_enable(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int gmc_v7_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
+	gmc_v7_0_gart_disable(adev);
+
+	return 0;
+}
+
+static int gmc_v7_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v7_0_hw_fini(adev);
+
+	return 0;
+}
+
+static int gmc_v7_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = gmc_v7_0_hw_init(adev);
+	if (r)
+		return r;
+
+	amdgpu_vmid_reset_all(adev);
+
+	return 0;
+}
+
+static bool gmc_v7_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK | SRBM_STATUS__VMC_BUSY_MASK))
+		return false;
+
+	return true;
+}
+
+static int gmc_v7_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS) & (SRBM_STATUS__MCB_BUSY_MASK |
+					       SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+					       SRBM_STATUS__MCC_BUSY_MASK |
+					       SRBM_STATUS__MCD_BUSY_MASK |
+					       SRBM_STATUS__VMC_BUSY_MASK);
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+
+}
+
+static int gmc_v7_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset = 0;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__VMC_BUSY_MASK)
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+						SRBM_SOFT_RESET, SOFT_RESET_VMC, 1);
+
+	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK)) {
+		if (!(adev->flags & AMD_IS_APU))
+			srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+							SRBM_SOFT_RESET, SOFT_RESET_MC, 1);
+	}
+
+	if (srbm_soft_reset) {
+		gmc_v7_0_mc_stop(adev);
+		if (gmc_v7_0_wait_for_idle((void *)adev)) {
+			dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
+		}
+
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+
+		gmc_v7_0_mc_resume(adev);
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int gmc_v7_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	u32 tmp;
+	u32 bits = (VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK);
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		/* system context */
+		tmp = RREG32(mmVM_CONTEXT0_CNTL);
+		tmp &= ~bits;
+		WREG32(mmVM_CONTEXT0_CNTL, tmp);
+		/* VMs */
+		tmp = RREG32(mmVM_CONTEXT1_CNTL);
+		tmp &= ~bits;
+		WREG32(mmVM_CONTEXT1_CNTL, tmp);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		/* system context */
+		tmp = RREG32(mmVM_CONTEXT0_CNTL);
+		tmp |= bits;
+		WREG32(mmVM_CONTEXT0_CNTL, tmp);
+		/* VMs */
+		tmp = RREG32(mmVM_CONTEXT1_CNTL);
+		tmp |= bits;
+		WREG32(mmVM_CONTEXT1_CNTL, tmp);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gmc_v7_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	u32 addr, status, mc_client, vmid;
+
+	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
+	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
+	mc_client = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_MCCLIENT);
+	/* reset addr and status */
+	WREG32_P(mmVM_CONTEXT1_CNTL2, 1, ~1);
+
+	if (!addr && !status)
+		return 0;
+
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_FIRST)
+		gmc_v7_0_set_fault_enable_default(adev, false);
+
+#ifdef notyet
+	if (printk_ratelimit()) {
+		dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
+			entry->src_id, entry->src_data[0]);
+		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x%08X\n",
+			addr);
+		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
+			status);
+		gmc_v7_0_vm_decode_fault(adev, status, addr, mc_client,
+					 entry->pasid);
+	}
+#endif
+
+	vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			     VMID);
+	if (amdgpu_amdkfd_is_kfd_vmid(adev, vmid)
+		&& !atomic_read(&adev->gmc.vm_fault_info_updated)) {
+		struct kfd_vm_fault_info *info = adev->gmc.vm_fault_info;
+		u32 protections = REG_GET_FIELD(status,
+					VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					PROTECTIONS);
+
+		info->vmid = vmid;
+		info->mc_id = REG_GET_FIELD(status,
+					    VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					    MEMORY_CLIENT_ID);
+		info->status = status;
+		info->page_addr = addr;
+		info->prot_valid = protections & 0x7 ? true : false;
+		info->prot_read = protections & 0x8 ? true : false;
+		info->prot_write = protections & 0x10 ? true : false;
+		info->prot_exec = protections & 0x20 ? true : false;
+		mb();
+		atomic_set(&adev->gmc.vm_fault_info_updated, 1);
+	}
+
+	return 0;
+}
+
+static int gmc_v7_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	bool gate = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_CG_STATE_GATE)
+		gate = true;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		gmc_v7_0_enable_mc_mgcg(adev, gate);
+		gmc_v7_0_enable_mc_ls(adev, gate);
+	}
+	gmc_v7_0_enable_bif_mgls(adev, gate);
+	gmc_v7_0_enable_hdp_mgcg(adev, gate);
+	gmc_v7_0_enable_hdp_ls(adev, gate);
+
+	return 0;
+}
+
+static int gmc_v7_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs gmc_v7_0_ip_funcs = {
+	.name = "gmc_v7_0",
+	.early_init = gmc_v7_0_early_init,
+	.late_init = gmc_v7_0_late_init,
+	.sw_init = gmc_v7_0_sw_init,
+	.sw_fini = gmc_v7_0_sw_fini,
+	.hw_init = gmc_v7_0_hw_init,
+	.hw_fini = gmc_v7_0_hw_fini,
+	.suspend = gmc_v7_0_suspend,
+	.resume = gmc_v7_0_resume,
+	.is_idle = gmc_v7_0_is_idle,
+	.wait_for_idle = gmc_v7_0_wait_for_idle,
+	.soft_reset = gmc_v7_0_soft_reset,
+	.set_clockgating_state = gmc_v7_0_set_clockgating_state,
+	.set_powergating_state = gmc_v7_0_set_powergating_state,
+};
+
+static const struct amdgpu_gmc_funcs gmc_v7_0_gmc_funcs = {
+	.flush_gpu_tlb = gmc_v7_0_flush_gpu_tlb,
+	.emit_flush_gpu_tlb = gmc_v7_0_emit_flush_gpu_tlb,
+	.emit_pasid_mapping = gmc_v7_0_emit_pasid_mapping,
+	.set_pte_pde = gmc_v7_0_set_pte_pde,
+	.set_prt = gmc_v7_0_set_prt,
+	.get_vm_pte_flags = gmc_v7_0_get_vm_pte_flags,
+	.get_vm_pde = gmc_v7_0_get_vm_pde
+};
+
+static const struct amdgpu_irq_src_funcs gmc_v7_0_irq_funcs = {
+	.set = gmc_v7_0_vm_fault_interrupt_state,
+	.process = gmc_v7_0_process_interrupt,
+};
+
+static void gmc_v7_0_set_gmc_funcs(struct amdgpu_device *adev)
+{
+	if (adev->gmc.gmc_funcs == NULL)
+		adev->gmc.gmc_funcs = &gmc_v7_0_gmc_funcs;
+}
+
+static void gmc_v7_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gmc.vm_fault.num_types = 1;
+	adev->gmc.vm_fault.funcs = &gmc_v7_0_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version gmc_v7_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GMC,
+	.major = 7,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gmc_v7_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version gmc_v7_4_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GMC,
+	.major = 7,
+	.minor = 4,
+	.rev = 0,
+	.funcs = &gmc_v7_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/gmc_v7_0.h b/sys/dev/pci/drm/amd/amdgpu/gmc_v7_0.h
new file mode 100644
index 00000000000..ebce2966c1c
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gmc_v7_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GMC_V7_0_H__
+#define __GMC_V7_0_H__
+
+extern const struct amdgpu_ip_block_version gmc_v7_0_ip_block;
+extern const struct amdgpu_ip_block_version gmc_v7_4_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/gmc_v8_0.c b/sys/dev/pci/drm/amd/amdgpu/gmc_v8_0.c
new file mode 100644
index 00000000000..9aeb237b674
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gmc_v8_0.c
@@ -0,0 +1,1792 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include <drm/drm_cache.h>
+#include "amdgpu.h"
+#include "gmc_v8_0.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_amdkfd.h"
+
+#include "gmc/gmc_8_1_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+
+#include "bif/bif_5_0_d.h"
+#include "bif/bif_5_0_sh_mask.h"
+
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+
+#include "dce/dce_10_0_d.h"
+#include "dce/dce_10_0_sh_mask.h"
+
+#include "vid.h"
+#include "vi.h"
+
+#include "amdgpu_atombios.h"
+
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+static void gmc_v8_0_set_gmc_funcs(struct amdgpu_device *adev);
+static void gmc_v8_0_set_irq_funcs(struct amdgpu_device *adev);
+static int gmc_v8_0_wait_for_idle(void *handle);
+
+MODULE_FIRMWARE("amdgpu/tonga_mc.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_mc.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_mc.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_mc.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_k_mc.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_k_mc.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_k_mc.bin");
+
+static const u32 golden_settings_tonga_a11[] =
+{
+	mmMC_ARB_WTM_GRPWT_RD, 0x00000003, 0x00000000,
+	mmMC_HUB_RDREQ_DMIF_LIMIT, 0x0000007f, 0x00000028,
+	mmMC_HUB_WDP_UMC, 0x00007fb6, 0x00000991,
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+};
+
+static const u32 tonga_mgcg_cgcg_init[] =
+{
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
+};
+
+static const u32 golden_settings_fiji_a10[] =
+{
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+};
+
+static const u32 fiji_mgcg_cgcg_init[] =
+{
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
+};
+
+static const u32 golden_settings_polaris11_a11[] =
+{
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff
+};
+
+static const u32 golden_settings_polaris10_a11[] =
+{
+	mmMC_ARB_WTM_GRPWT_RD, 0x00000003, 0x00000000,
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff
+};
+
+static const u32 cz_mgcg_cgcg_init[] =
+{
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
+};
+
+static const u32 stoney_mgcg_cgcg_init[] =
+{
+	mmATC_MISC_CG, 0xffffffff, 0x000c0200,
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
+};
+
+static const u32 golden_settings_stoney_common[] =
+{
+	mmMC_HUB_RDREQ_UVD, MC_HUB_RDREQ_UVD__PRESCALE_MASK, 0x00000004,
+	mmMC_RD_GRP_OTH, MC_RD_GRP_OTH__UVD_MASK, 0x00600000
+};
+
+static void gmc_v8_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+		amdgpu_device_program_register_sequence(adev,
+							fiji_mgcg_cgcg_init,
+							ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_fiji_a10,
+							ARRAY_SIZE(golden_settings_fiji_a10));
+		break;
+	case CHIP_TONGA:
+		amdgpu_device_program_register_sequence(adev,
+							tonga_mgcg_cgcg_init,
+							ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_tonga_a11,
+							ARRAY_SIZE(golden_settings_tonga_a11));
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_polaris11_a11,
+							ARRAY_SIZE(golden_settings_polaris11_a11));
+		break;
+	case CHIP_POLARIS10:
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_polaris10_a11,
+							ARRAY_SIZE(golden_settings_polaris10_a11));
+		break;
+	case CHIP_CARRIZO:
+		amdgpu_device_program_register_sequence(adev,
+							cz_mgcg_cgcg_init,
+							ARRAY_SIZE(cz_mgcg_cgcg_init));
+		break;
+	case CHIP_STONEY:
+		amdgpu_device_program_register_sequence(adev,
+							stoney_mgcg_cgcg_init,
+							ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_stoney_common,
+							ARRAY_SIZE(golden_settings_stoney_common));
+		break;
+	default:
+		break;
+	}
+}
+
+static void gmc_v8_0_mc_stop(struct amdgpu_device *adev)
+{
+	u32 blackout;
+
+	gmc_v8_0_wait_for_idle(adev);
+
+	blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
+	if (REG_GET_FIELD(blackout, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE) != 1) {
+		/* Block CPU access */
+		WREG32(mmBIF_FB_EN, 0);
+		/* blackout the MC */
+		blackout = REG_SET_FIELD(blackout,
+					 MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 1);
+		WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout);
+	}
+	/* wait for the MC to settle */
+	udelay(100);
+}
+
+static void gmc_v8_0_mc_resume(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	/* unblackout the MC */
+	tmp = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 0);
+	WREG32(mmMC_SHARED_BLACKOUT_CNTL, tmp);
+	/* allow CPU access */
+	tmp = REG_SET_FIELD(0, BIF_FB_EN, FB_READ_EN, 1);
+	tmp = REG_SET_FIELD(tmp, BIF_FB_EN, FB_WRITE_EN, 1);
+	WREG32(mmBIF_FB_EN, tmp);
+}
+
+/**
+ * gmc_v8_0_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int gmc_v8_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_TONGA:
+		chip_name = "tonga";
+		break;
+	case CHIP_POLARIS11:
+		if (((adev->pdev->device == 0x67ef) &&
+		     ((adev->pdev->revision == 0xe0) ||
+		      (adev->pdev->revision == 0xe5))) ||
+		    ((adev->pdev->device == 0x67ff) &&
+		     ((adev->pdev->revision == 0xcf) ||
+		      (adev->pdev->revision == 0xef) ||
+		      (adev->pdev->revision == 0xff))))
+			chip_name = "polaris11_k";
+		else if ((adev->pdev->device == 0x67ef) &&
+			 (adev->pdev->revision == 0xe2))
+			chip_name = "polaris11_k";
+		else
+			chip_name = "polaris11";
+		break;
+	case CHIP_POLARIS10:
+		if ((adev->pdev->device == 0x67df) &&
+		    ((adev->pdev->revision == 0xe1) ||
+		     (adev->pdev->revision == 0xf7)))
+			chip_name = "polaris10_k";
+		else
+			chip_name = "polaris10";
+		break;
+	case CHIP_POLARIS12:
+		if (((adev->pdev->device == 0x6987) &&
+		     ((adev->pdev->revision == 0xc0) ||
+		      (adev->pdev->revision == 0xc3))) ||
+		    ((adev->pdev->device == 0x6981) &&
+		     ((adev->pdev->revision == 0x00) ||
+		      (adev->pdev->revision == 0x01) ||
+		      (adev->pdev->revision == 0x10))))
+			chip_name = "polaris12_k";
+		else
+			chip_name = "polaris12";
+		break;
+	case CHIP_FIJI:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+	case CHIP_VEGAM:
+		return 0;
+	default: BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mc.bin", chip_name);
+	err = request_firmware(&adev->gmc.fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gmc.fw);
+
+out:
+	if (err) {
+		pr_err("mc: Failed to load firmware \"%s\"\n", fw_name);
+		release_firmware(adev->gmc.fw);
+		adev->gmc.fw = NULL;
+	}
+	return err;
+}
+
+/**
+ * gmc_v8_0_tonga_mc_load_microcode - load tonga MC ucode into the hw
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Load the GDDR MC ucode into the hw (CIK).
+ * Returns 0 on success, error on failure.
+ */
+static int gmc_v8_0_tonga_mc_load_microcode(struct amdgpu_device *adev)
+{
+	const struct mc_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data = NULL;
+	const __le32 *io_mc_regs = NULL;
+	u32 running;
+	int i, ucode_size, regs_size;
+
+	/* Skip MC ucode loading on SR-IOV capable boards.
+	 * vbios does this for us in asic_init in that case.
+	 * Skip MC ucode loading on VF, because hypervisor will do that
+	 * for this adaptor.
+	 */
+	if (amdgpu_sriov_bios(adev))
+		return 0;
+
+	if (!adev->gmc.fw)
+		return -EINVAL;
+
+	hdr = (const struct mc_firmware_header_v1_0 *)adev->gmc.fw->data;
+	amdgpu_ucode_print_mc_hdr(&hdr->header);
+
+	adev->gmc.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	regs_size = le32_to_cpu(hdr->io_debug_size_bytes) / (4 * 2);
+	io_mc_regs = (const __le32 *)
+		(adev->gmc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
+	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+	fw_data = (const __le32 *)
+		(adev->gmc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	running = REG_GET_FIELD(RREG32(mmMC_SEQ_SUP_CNTL), MC_SEQ_SUP_CNTL, RUN);
+
+	if (running == 0) {
+		/* reset the engine and set to writable */
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000010);
+
+		/* load mc io regs */
+		for (i = 0; i < regs_size; i++) {
+			WREG32(mmMC_SEQ_IO_DEBUG_INDEX, le32_to_cpup(io_mc_regs++));
+			WREG32(mmMC_SEQ_IO_DEBUG_DATA, le32_to_cpup(io_mc_regs++));
+		}
+		/* load the MC ucode */
+		for (i = 0; i < ucode_size; i++)
+			WREG32(mmMC_SEQ_SUP_PGM, le32_to_cpup(fw_data++));
+
+		/* put the engine back into the active state */
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000004);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000001);
+
+		/* wait for training to complete */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (REG_GET_FIELD(RREG32(mmMC_SEQ_TRAIN_WAKEUP_CNTL),
+					  MC_SEQ_TRAIN_WAKEUP_CNTL, TRAIN_DONE_D0))
+				break;
+			udelay(1);
+		}
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (REG_GET_FIELD(RREG32(mmMC_SEQ_TRAIN_WAKEUP_CNTL),
+					  MC_SEQ_TRAIN_WAKEUP_CNTL, TRAIN_DONE_D1))
+				break;
+			udelay(1);
+		}
+	}
+
+	return 0;
+}
+
+static int gmc_v8_0_polaris_mc_load_microcode(struct amdgpu_device *adev)
+{
+	const struct mc_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data = NULL;
+	const __le32 *io_mc_regs = NULL;
+	u32 data;
+	int i, ucode_size, regs_size;
+
+	/* Skip MC ucode loading on SR-IOV capable boards.
+	 * vbios does this for us in asic_init in that case.
+	 * Skip MC ucode loading on VF, because hypervisor will do that
+	 * for this adaptor.
+	 */
+	if (amdgpu_sriov_bios(adev))
+		return 0;
+
+	if (!adev->gmc.fw)
+		return -EINVAL;
+
+	hdr = (const struct mc_firmware_header_v1_0 *)adev->gmc.fw->data;
+	amdgpu_ucode_print_mc_hdr(&hdr->header);
+
+	adev->gmc.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	regs_size = le32_to_cpu(hdr->io_debug_size_bytes) / (4 * 2);
+	io_mc_regs = (const __le32 *)
+		(adev->gmc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
+	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+	fw_data = (const __le32 *)
+		(adev->gmc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	data = RREG32(mmMC_SEQ_MISC0);
+	data &= ~(0x40);
+	WREG32(mmMC_SEQ_MISC0, data);
+
+	/* load mc io regs */
+	for (i = 0; i < regs_size; i++) {
+		WREG32(mmMC_SEQ_IO_DEBUG_INDEX, le32_to_cpup(io_mc_regs++));
+		WREG32(mmMC_SEQ_IO_DEBUG_DATA, le32_to_cpup(io_mc_regs++));
+	}
+
+	WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
+	WREG32(mmMC_SEQ_SUP_CNTL, 0x00000010);
+
+	/* load the MC ucode */
+	for (i = 0; i < ucode_size; i++)
+		WREG32(mmMC_SEQ_SUP_PGM, le32_to_cpup(fw_data++));
+
+	/* put the engine back into the active state */
+	WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
+	WREG32(mmMC_SEQ_SUP_CNTL, 0x00000004);
+	WREG32(mmMC_SEQ_SUP_CNTL, 0x00000001);
+
+	/* wait for training to complete */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		data = RREG32(mmMC_SEQ_MISC0);
+		if (data & 0x80)
+			break;
+		udelay(1);
+	}
+
+	return 0;
+}
+
+static void gmc_v8_0_vram_gtt_location(struct amdgpu_device *adev,
+				       struct amdgpu_gmc *mc)
+{
+	u64 base = 0;
+
+	if (!amdgpu_sriov_vf(adev))
+		base = RREG32(mmMC_VM_FB_LOCATION) & 0xFFFF;
+	base <<= 24;
+
+	amdgpu_device_vram_location(adev, &adev->gmc, base);
+	amdgpu_device_gart_location(adev, mc);
+}
+
+/**
+ * gmc_v8_0_mc_program - program the GPU memory controller
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the location of vram, gart, and AGP in the GPU's
+ * physical address space (CIK).
+ */
+static void gmc_v8_0_mc_program(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	int i, j;
+
+	/* Initialize HDP */
+	for (i = 0, j = 0; i < 32; i++, j += 0x6) {
+		WREG32((0xb05 + j), 0x00000000);
+		WREG32((0xb06 + j), 0x00000000);
+		WREG32((0xb07 + j), 0x00000000);
+		WREG32((0xb08 + j), 0x00000000);
+		WREG32((0xb09 + j), 0x00000000);
+	}
+	WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
+
+	if (gmc_v8_0_wait_for_idle((void *)adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
+	}
+	if (adev->mode_info.num_crtc) {
+		/* Lockout access through VGA aperture*/
+		tmp = RREG32(mmVGA_HDP_CONTROL);
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
+		WREG32(mmVGA_HDP_CONTROL, tmp);
+
+		/* disable VGA render */
+		tmp = RREG32(mmVGA_RENDER_CONTROL);
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
+		WREG32(mmVGA_RENDER_CONTROL, tmp);
+	}
+	/* Update configuration */
+	WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+	       adev->gmc.vram_start >> 12);
+	WREG32(mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+	       adev->gmc.vram_end >> 12);
+	WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
+	       adev->vram_scratch.gpu_addr >> 12);
+
+	if (amdgpu_sriov_vf(adev)) {
+		tmp = ((adev->gmc.vram_end >> 24) & 0xFFFF) << 16;
+		tmp |= ((adev->gmc.vram_start >> 24) & 0xFFFF);
+		WREG32(mmMC_VM_FB_LOCATION, tmp);
+		/* XXX double check these! */
+		WREG32(mmHDP_NONSURFACE_BASE, (adev->gmc.vram_start >> 8));
+		WREG32(mmHDP_NONSURFACE_INFO, (2 << 7) | (1 << 30));
+		WREG32(mmHDP_NONSURFACE_SIZE, 0x3FFFFFFF);
+	}
+
+	WREG32(mmMC_VM_AGP_BASE, 0);
+	WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
+	WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
+	if (gmc_v8_0_wait_for_idle((void *)adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
+	}
+
+	WREG32(mmBIF_FB_EN, BIF_FB_EN__FB_READ_EN_MASK | BIF_FB_EN__FB_WRITE_EN_MASK);
+
+	tmp = RREG32(mmHDP_MISC_CNTL);
+	tmp = REG_SET_FIELD(tmp, HDP_MISC_CNTL, FLUSH_INVALIDATE_CACHE, 0);
+	WREG32(mmHDP_MISC_CNTL, tmp);
+
+	tmp = RREG32(mmHDP_HOST_PATH_CNTL);
+	WREG32(mmHDP_HOST_PATH_CNTL, tmp);
+}
+
+/**
+ * gmc_v8_0_mc_init - initialize the memory controller driver params
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the amount of vram, vram width, and decide how to place
+ * vram and gart within the GPU's physical address space (CIK).
+ * Returns 0 for success.
+ */
+static int gmc_v8_0_mc_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	adev->gmc.vram_width = amdgpu_atombios_get_vram_width(adev);
+	if (!adev->gmc.vram_width) {
+		u32 tmp;
+		int chansize, numchan;
+
+		/* Get VRAM informations */
+		tmp = RREG32(mmMC_ARB_RAMCFG);
+		if (REG_GET_FIELD(tmp, MC_ARB_RAMCFG, CHANSIZE)) {
+			chansize = 64;
+		} else {
+			chansize = 32;
+		}
+		tmp = RREG32(mmMC_SHARED_CHMAP);
+		switch (REG_GET_FIELD(tmp, MC_SHARED_CHMAP, NOOFCHAN)) {
+		case 0:
+		default:
+			numchan = 1;
+			break;
+		case 1:
+			numchan = 2;
+			break;
+		case 2:
+			numchan = 4;
+			break;
+		case 3:
+			numchan = 8;
+			break;
+		case 4:
+			numchan = 3;
+			break;
+		case 5:
+			numchan = 6;
+			break;
+		case 6:
+			numchan = 10;
+			break;
+		case 7:
+			numchan = 12;
+			break;
+		case 8:
+			numchan = 16;
+			break;
+		}
+		adev->gmc.vram_width = numchan * chansize;
+	}
+	/* size in MB on si */
+	adev->gmc.mc_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+	adev->gmc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_device_resize_fb_bar(adev);
+		if (r)
+			return r;
+	}
+	adev->gmc.aper_base = adev->fb_aper_offset;
+	adev->gmc.aper_size = adev->fb_aper_size;
+
+#ifdef __amd64__
+	if (adev->flags & AMD_IS_APU) {
+		adev->gmc.aper_base = ((u64)RREG32(mmMC_VM_FB_OFFSET)) << 22;
+		adev->gmc.aper_size = adev->gmc.real_vram_size;
+	}
+#endif
+
+	/* In case the PCI BAR is larger than the actual amount of vram */
+	adev->gmc.visible_vram_size = adev->gmc.aper_size;
+	if (adev->gmc.visible_vram_size > adev->gmc.real_vram_size)
+		adev->gmc.visible_vram_size = adev->gmc.real_vram_size;
+
+	/* set the gart size */
+	if (amdgpu_gart_size == -1) {
+		switch (adev->asic_type) {
+		case CHIP_POLARIS10: /* all engines support GPUVM */
+		case CHIP_POLARIS11: /* all engines support GPUVM */
+		case CHIP_POLARIS12: /* all engines support GPUVM */
+		case CHIP_VEGAM:     /* all engines support GPUVM */
+		default:
+			adev->gmc.gart_size = 256ULL << 20;
+			break;
+		case CHIP_TONGA:   /* UVD, VCE do not support GPUVM */
+		case CHIP_FIJI:    /* UVD, VCE do not support GPUVM */
+		case CHIP_CARRIZO: /* UVD, VCE do not support GPUVM, DCE SG support */
+		case CHIP_STONEY:  /* UVD does not support GPUVM, DCE SG support */
+			adev->gmc.gart_size = 1024ULL << 20;
+			break;
+		}
+	} else {
+		adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
+	}
+
+	gmc_v8_0_vram_gtt_location(adev, &adev->gmc);
+
+	return 0;
+}
+
+/*
+ * GART
+ * VMID 0 is the physical GPU addresses as used by the kernel.
+ * VMIDs 1-15 are used for userspace clients and are handled
+ * by the amdgpu vm/hsa code.
+ */
+
+/**
+ * gmc_v8_0_flush_gpu_tlb - gart tlb flush callback
+ *
+ * @adev: amdgpu_device pointer
+ * @vmid: vm instance to flush
+ *
+ * Flush the TLB for the requested page table (CIK).
+ */
+static void gmc_v8_0_flush_gpu_tlb(struct amdgpu_device *adev,
+					uint32_t vmid)
+{
+	/* bits 0-15 are the VM contexts0-15 */
+	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+}
+
+static uint64_t gmc_v8_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
+					    unsigned vmid, uint64_t pd_addr)
+{
+	uint32_t reg;
+
+	if (vmid < 8)
+		reg = mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vmid;
+	else
+		reg = mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8;
+	amdgpu_ring_emit_wreg(ring, reg, pd_addr >> 12);
+
+	/* bits 0-15 are the VM contexts0-15 */
+	amdgpu_ring_emit_wreg(ring, mmVM_INVALIDATE_REQUEST, 1 << vmid);
+
+	return pd_addr;
+}
+
+static void gmc_v8_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned vmid,
+					unsigned pasid)
+{
+	amdgpu_ring_emit_wreg(ring, mmIH_VMID_0_LUT + vmid, pasid);
+}
+
+/**
+ * gmc_v8_0_set_pte_pde - update the page tables using MMIO
+ *
+ * @adev: amdgpu_device pointer
+ * @cpu_pt_addr: cpu address of the page table
+ * @gpu_page_idx: entry in the page table to update
+ * @addr: dst addr to write into pte/pde
+ * @flags: access flags
+ *
+ * Update the page tables using the CPU.
+ */
+static int gmc_v8_0_set_pte_pde(struct amdgpu_device *adev, void *cpu_pt_addr,
+				uint32_t gpu_page_idx, uint64_t addr,
+				uint64_t flags)
+{
+	void __iomem *ptr = (void *)cpu_pt_addr;
+	uint64_t value;
+
+	/*
+	 * PTE format on VI:
+	 * 63:40 reserved
+	 * 39:12 4k physical page base address
+	 * 11:7 fragment
+	 * 6 write
+	 * 5 read
+	 * 4 exe
+	 * 3 reserved
+	 * 2 snooped
+	 * 1 system
+	 * 0 valid
+	 *
+	 * PDE format on VI:
+	 * 63:59 block fragment size
+	 * 58:40 reserved
+	 * 39:1 physical base address of PTE
+	 * bits 5:1 must be 0.
+	 * 0 valid
+	 */
+	value = addr & 0x000000FFFFFFF000ULL;
+	value |= flags;
+	writeq(value, ptr + (gpu_page_idx * 8));
+
+	return 0;
+}
+
+static uint64_t gmc_v8_0_get_vm_pte_flags(struct amdgpu_device *adev,
+					  uint32_t flags)
+{
+	uint64_t pte_flag = 0;
+
+	if (flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		pte_flag |= AMDGPU_PTE_EXECUTABLE;
+	if (flags & AMDGPU_VM_PAGE_READABLE)
+		pte_flag |= AMDGPU_PTE_READABLE;
+	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
+		pte_flag |= AMDGPU_PTE_WRITEABLE;
+	if (flags & AMDGPU_VM_PAGE_PRT)
+		pte_flag |= AMDGPU_PTE_PRT;
+
+	return pte_flag;
+}
+
+static void gmc_v8_0_get_vm_pde(struct amdgpu_device *adev, int level,
+				uint64_t *addr, uint64_t *flags)
+{
+	BUG_ON(*addr & 0xFFFFFF0000000FFFULL);
+}
+
+/**
+ * gmc_v8_0_set_fault_enable_default - update VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void gmc_v8_0_set_fault_enable_default(struct amdgpu_device *adev,
+					      bool value)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmVM_CONTEXT1_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	WREG32(mmVM_CONTEXT1_CNTL, tmp);
+}
+
+/**
+ * gmc_v8_0_set_prt - set PRT VM fault
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable VM fault handling for PRT
+*/
+static void gmc_v8_0_set_prt(struct amdgpu_device *adev, bool enable)
+{
+	u32 tmp;
+
+	if (enable && !adev->gmc.prt_warning) {
+		dev_warn(adev->dev, "Disabling VM faults because of PRT request!\n");
+		adev->gmc.prt_warning = true;
+	}
+
+	tmp = RREG32(mmVM_PRT_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    CB_DISABLE_READ_FAULT_ON_UNMAPPED_ACCESS, enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    CB_DISABLE_WRITE_FAULT_ON_UNMAPPED_ACCESS, enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    TC_DISABLE_READ_FAULT_ON_UNMAPPED_ACCESS, enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    TC_DISABLE_WRITE_FAULT_ON_UNMAPPED_ACCESS, enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    L2_CACHE_STORE_INVALID_ENTRIES, enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    L1_TLB_STORE_INVALID_ENTRIES, enable);
+	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
+			    MASK_PDE0_FAULT, enable);
+	WREG32(mmVM_PRT_CNTL, tmp);
+
+	if (enable) {
+		uint32_t low = AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT;
+		uint32_t high = adev->vm_manager.max_pfn -
+			(AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT);
+
+		WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, low);
+		WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, low);
+		WREG32(mmVM_PRT_APERTURE2_LOW_ADDR, low);
+		WREG32(mmVM_PRT_APERTURE3_LOW_ADDR, low);
+		WREG32(mmVM_PRT_APERTURE0_HIGH_ADDR, high);
+		WREG32(mmVM_PRT_APERTURE1_HIGH_ADDR, high);
+		WREG32(mmVM_PRT_APERTURE2_HIGH_ADDR, high);
+		WREG32(mmVM_PRT_APERTURE3_HIGH_ADDR, high);
+	} else {
+		WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, 0xfffffff);
+		WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, 0xfffffff);
+		WREG32(mmVM_PRT_APERTURE2_LOW_ADDR, 0xfffffff);
+		WREG32(mmVM_PRT_APERTURE3_LOW_ADDR, 0xfffffff);
+		WREG32(mmVM_PRT_APERTURE0_HIGH_ADDR, 0x0);
+		WREG32(mmVM_PRT_APERTURE1_HIGH_ADDR, 0x0);
+		WREG32(mmVM_PRT_APERTURE2_HIGH_ADDR, 0x0);
+		WREG32(mmVM_PRT_APERTURE3_HIGH_ADDR, 0x0);
+	}
+}
+
+/**
+ * gmc_v8_0_gart_enable - gart enable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This sets up the TLBs, programs the page tables for VMID0,
+ * sets up the hw for VMIDs 1-15 which are allocated on
+ * demand, and sets up the global locations for the LDS, GDS,
+ * and GPUVM for FSA64 clients (CIK).
+ * Returns 0 for success, errors for failure.
+ */
+static int gmc_v8_0_gart_enable(struct amdgpu_device *adev)
+{
+	int r, i;
+	u32 tmp, field;
+
+	if (adev->gart.robj == NULL) {
+		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
+		return -EINVAL;
+	}
+	r = amdgpu_gart_table_vram_pin(adev);
+	if (r)
+		return r;
+	/* Setup TLB control */
+	tmp = RREG32(mmMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_FRAGMENT_PROCESSING, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	WREG32(mmMC_VM_MX_L1_TLB_CNTL, tmp);
+	/* Setup L2 cache */
+	tmp = RREG32(mmVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, EFFECTIVE_L2_QUEUE_SIZE, 7);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	WREG32(mmVM_L2_CNTL, tmp);
+	tmp = RREG32(mmVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32(mmVM_L2_CNTL2, tmp);
+
+	field = adev->vm_manager.fragment_size;
+	tmp = RREG32(mmVM_L2_CNTL3);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_ASSOCIATIVITY, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, field);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, field);
+	WREG32(mmVM_L2_CNTL3, tmp);
+	/* XXX: set to enable PTE/PDE in system memory */
+	tmp = RREG32(mmVM_L2_CNTL4);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT0_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT0_PDE_REQUEST_SHARED, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT0_PDE_REQUEST_SNOOP, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT0_PTE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT0_PTE_REQUEST_SHARED, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT0_PTE_REQUEST_SNOOP, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PDE_REQUEST_SHARED, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PDE_REQUEST_SNOOP, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PTE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PTE_REQUEST_SHARED, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PTE_REQUEST_SNOOP, 0);
+	WREG32(mmVM_L2_CNTL4, tmp);
+	/* setup context0 */
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->gmc.gart_start >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->gmc.gart_end >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
+	WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
+			(u32)(adev->dummy_page_addr >> 12));
+	WREG32(mmVM_CONTEXT0_CNTL2, 0);
+	tmp = RREG32(mmVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	WREG32(mmVM_CONTEXT0_CNTL, tmp);
+
+	WREG32(mmVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR, 0);
+	WREG32(mmVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR, 0);
+	WREG32(mmVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET, 0);
+
+	/* empty context1-15 */
+	/* FIXME start with 4G, once using 2 level pt switch to full
+	 * vm size space
+	 */
+	/* set vm size, must be a multiple of 4 */
+	WREG32(mmVM_CONTEXT1_PAGE_TABLE_START_ADDR, 0);
+	WREG32(mmVM_CONTEXT1_PAGE_TABLE_END_ADDR, adev->vm_manager.max_pfn - 1);
+	for (i = 1; i < 16; i++) {
+		if (i < 8)
+			WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i,
+			       adev->gart.table_addr >> 12);
+		else
+			WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + i - 8,
+			       adev->gart.table_addr >> 12);
+	}
+
+	/* enable context1-15 */
+	WREG32(mmVM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR,
+	       (u32)(adev->dummy_page_addr >> 12));
+	WREG32(mmVM_CONTEXT1_CNTL2, 4);
+	tmp = RREG32(mmVM_CONTEXT1_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_BLOCK_SIZE,
+			    adev->vm_manager.block_size - 9);
+	WREG32(mmVM_CONTEXT1_CNTL, tmp);
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS)
+		gmc_v8_0_set_fault_enable_default(adev, false);
+	else
+		gmc_v8_0_set_fault_enable_default(adev, true);
+
+	gmc_v8_0_flush_gpu_tlb(adev, 0);
+	DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
+		 (unsigned)(adev->gmc.gart_size >> 20),
+		 (unsigned long long)adev->gart.table_addr);
+	adev->gart.ready = true;
+	return 0;
+}
+
+static int gmc_v8_0_gart_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.robj) {
+		WARN(1, "R600 PCIE GART already initialized\n");
+		return 0;
+	}
+	/* Initialize common gart structure */
+	r = amdgpu_gart_init(adev);
+	if (r)
+		return r;
+	adev->gart.table_size = adev->gart.num_gpu_pages * 8;
+	adev->gart.gart_pte_flags = AMDGPU_PTE_EXECUTABLE;
+	return amdgpu_gart_table_vram_alloc(adev);
+}
+
+/**
+ * gmc_v8_0_gart_disable - gart disable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This disables all VM page table (CIK).
+ */
+static void gmc_v8_0_gart_disable(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	/* Disable all tables */
+	WREG32(mmVM_CONTEXT0_CNTL, 0);
+	WREG32(mmVM_CONTEXT1_CNTL, 0);
+	/* Setup TLB control */
+	tmp = RREG32(mmMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32(mmMC_VM_MX_L1_TLB_CNTL, tmp);
+	/* Setup L2 cache */
+	tmp = RREG32(mmVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32(mmVM_L2_CNTL, tmp);
+	WREG32(mmVM_L2_CNTL2, 0);
+	amdgpu_gart_table_vram_unpin(adev);
+}
+
+/**
+ * gmc_v8_0_vm_decode_fault - print human readable fault info
+ *
+ * @adev: amdgpu_device pointer
+ * @status: VM_CONTEXT1_PROTECTION_FAULT_STATUS register value
+ * @addr: VM_CONTEXT1_PROTECTION_FAULT_ADDR register value
+ *
+ * Print human readable fault information (CIK).
+ */
+static void gmc_v8_0_vm_decode_fault(struct amdgpu_device *adev, u32 status,
+				     u32 addr, u32 mc_client, unsigned pasid)
+{
+	u32 vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
+	u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					PROTECTIONS);
+	char block[5] = { mc_client >> 24, (mc_client >> 16) & 0xff,
+		(mc_client >> 8) & 0xff, mc_client & 0xff, 0 };
+	u32 mc_id;
+
+	mc_id = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			      MEMORY_CLIENT_ID);
+
+	dev_err(adev->dev, "VM fault (0x%02x, vmid %d, pasid %d) at page %u, %s from '%s' (0x%08x) (%d)\n",
+	       protections, vmid, pasid, addr,
+	       REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			     MEMORY_CLIENT_RW) ?
+	       "write" : "read", block, mc_client, mc_id);
+}
+
+static int gmc_v8_0_convert_vram_type(int mc_seq_vram_type)
+{
+	switch (mc_seq_vram_type) {
+	case MC_SEQ_MISC0__MT__GDDR1:
+		return AMDGPU_VRAM_TYPE_GDDR1;
+	case MC_SEQ_MISC0__MT__DDR2:
+		return AMDGPU_VRAM_TYPE_DDR2;
+	case MC_SEQ_MISC0__MT__GDDR3:
+		return AMDGPU_VRAM_TYPE_GDDR3;
+	case MC_SEQ_MISC0__MT__GDDR4:
+		return AMDGPU_VRAM_TYPE_GDDR4;
+	case MC_SEQ_MISC0__MT__GDDR5:
+		return AMDGPU_VRAM_TYPE_GDDR5;
+	case MC_SEQ_MISC0__MT__HBM:
+		return AMDGPU_VRAM_TYPE_HBM;
+	case MC_SEQ_MISC0__MT__DDR3:
+		return AMDGPU_VRAM_TYPE_DDR3;
+	default:
+		return AMDGPU_VRAM_TYPE_UNKNOWN;
+	}
+}
+
+static int gmc_v8_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v8_0_set_gmc_funcs(adev);
+	gmc_v8_0_set_irq_funcs(adev);
+
+	adev->gmc.shared_aperture_start = 0x2000000000000000ULL;
+	adev->gmc.shared_aperture_end =
+		adev->gmc.shared_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.private_aperture_start =
+		adev->gmc.shared_aperture_end + 1;
+	adev->gmc.private_aperture_end =
+		adev->gmc.private_aperture_start + (4ULL << 30) - 1;
+
+	return 0;
+}
+
+static int gmc_v8_0_late_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_bo_late_init(adev);
+
+	if (amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS)
+		return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
+	else
+		return 0;
+}
+
+static unsigned gmc_v8_0_get_vbios_fb_size(struct amdgpu_device *adev)
+{
+	u32 d1vga_control = RREG32(mmD1VGA_CONTROL);
+	unsigned size;
+
+	if (REG_GET_FIELD(d1vga_control, D1VGA_CONTROL, D1VGA_MODE_ENABLE)) {
+		size = 9 * 1024 * 1024; /* reserve 8MB for vga emulator and 1 MB for FB */
+	} else {
+		u32 viewport = RREG32(mmVIEWPORT_SIZE);
+		size = (REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_HEIGHT) *
+			REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_WIDTH) *
+			4);
+	}
+	/* return 0 if the pre-OS buffer uses up most of vram */
+	if ((adev->gmc.real_vram_size - size) < (8 * 1024 * 1024))
+		return 0;
+	return size;
+}
+
+#define mmMC_SEQ_MISC0_FIJI 0xA71
+
+static int gmc_v8_0_sw_init(void *handle)
+{
+	int r;
+	int dma_bits;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->flags & AMD_IS_APU) {
+		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+	} else {
+		u32 tmp;
+
+		if ((adev->asic_type == CHIP_FIJI) ||
+		    (adev->asic_type == CHIP_VEGAM))
+			tmp = RREG32(mmMC_SEQ_MISC0_FIJI);
+		else
+			tmp = RREG32(mmMC_SEQ_MISC0);
+		tmp &= MC_SEQ_MISC0__MT__MASK;
+		adev->gmc.vram_type = gmc_v8_0_convert_vram_type(tmp);
+	}
+
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_PAGE_INV_FAULT, &adev->gmc.vm_fault);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_GFX_MEM_PROT_FAULT, &adev->gmc.vm_fault);
+	if (r)
+		return r;
+
+	/* Adjust VM size here.
+	 * Currently set to 4GB ((1 << 20) 4k pages).
+	 * Max GPUVM size for cayman and SI is 40 bits.
+	 */
+	amdgpu_vm_adjust_size(adev, 64, 9, 1, 40);
+
+	/* Set the internal MC address mask
+	 * This is the max address of the GPU's
+	 * internal address space.
+	 */
+	adev->gmc.mc_mask = 0xffffffffffULL; /* 40 bit MC */
+
+	/* set DMA mask + need_dma32 flags.
+	 * PCIE - can handle 40-bits.
+	 * IGP - can handle 40-bits
+	 * PCI - dma32 for legacy pci gart, 40 bits on newer asics
+	 */
+	adev->need_dma32 = false;
+#ifdef notyet
+	dma_bits = adev->need_dma32 ? 32 : 40;
+	r = pci_set_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	if (r) {
+		adev->need_dma32 = true;
+		dma_bits = 32;
+		pr_warn("amdgpu: No suitable DMA available\n");
+	}
+	r = pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	if (r) {
+		pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(32));
+		pr_warn("amdgpu: No coherent DMA available\n");
+	}
+	adev->need_swiotlb = drm_get_max_iomem() > ((u64)1 << dma_bits);
+#endif
+
+	r = gmc_v8_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load mc firmware!\n");
+		return r;
+	}
+
+	r = gmc_v8_0_mc_init(adev);
+	if (r)
+		return r;
+
+	adev->gmc.stolen_size = gmc_v8_0_get_vbios_fb_size(adev);
+
+	/* Memory manager */
+	r = amdgpu_bo_init(adev);
+	if (r)
+		return r;
+
+	r = gmc_v8_0_gart_init(adev);
+	if (r)
+		return r;
+
+	/*
+	 * number of VMs
+	 * VMID 0 is reserved for System
+	 * amdgpu graphics/compute will use VMIDs 1-7
+	 * amdkfd will use VMIDs 8-15
+	 */
+	adev->vm_manager.id_mgr[0].num_ids = AMDGPU_NUM_OF_VMIDS;
+	amdgpu_vm_manager_init(adev);
+
+	/* base offset of vram pages */
+	if (adev->flags & AMD_IS_APU) {
+		u64 tmp = RREG32(mmMC_VM_FB_OFFSET);
+
+		tmp <<= 22;
+		adev->vm_manager.vram_base_offset = tmp;
+	} else {
+		adev->vm_manager.vram_base_offset = 0;
+	}
+
+	adev->gmc.vm_fault_info = kmalloc(sizeof(struct kfd_vm_fault_info),
+					GFP_KERNEL);
+	if (!adev->gmc.vm_fault_info)
+		return -ENOMEM;
+	atomic_set(&adev->gmc.vm_fault_info_updated, 0);
+
+	return 0;
+}
+
+static int gmc_v8_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_gem_force_release(adev);
+	amdgpu_vm_manager_fini(adev);
+	kfree(adev->gmc.vm_fault_info);
+	amdgpu_gart_table_vram_free(adev);
+	amdgpu_bo_fini(adev);
+	amdgpu_gart_fini(adev);
+	release_firmware(adev->gmc.fw);
+	adev->gmc.fw = NULL;
+
+	return 0;
+}
+
+static int gmc_v8_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v8_0_init_golden_registers(adev);
+
+	gmc_v8_0_mc_program(adev);
+
+	if (adev->asic_type == CHIP_TONGA) {
+		r = gmc_v8_0_tonga_mc_load_microcode(adev);
+		if (r) {
+			DRM_ERROR("Failed to load MC firmware!\n");
+			return r;
+		}
+	} else if (adev->asic_type == CHIP_POLARIS11 ||
+			adev->asic_type == CHIP_POLARIS10 ||
+			adev->asic_type == CHIP_POLARIS12) {
+		r = gmc_v8_0_polaris_mc_load_microcode(adev);
+		if (r) {
+			DRM_ERROR("Failed to load MC firmware!\n");
+			return r;
+		}
+	}
+
+	r = gmc_v8_0_gart_enable(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int gmc_v8_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
+	gmc_v8_0_gart_disable(adev);
+
+	return 0;
+}
+
+static int gmc_v8_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v8_0_hw_fini(adev);
+
+	return 0;
+}
+
+static int gmc_v8_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = gmc_v8_0_hw_init(adev);
+	if (r)
+		return r;
+
+	amdgpu_vmid_reset_all(adev);
+
+	return 0;
+}
+
+static bool gmc_v8_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK | SRBM_STATUS__VMC_BUSY_MASK))
+		return false;
+
+	return true;
+}
+
+static int gmc_v8_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS) & (SRBM_STATUS__MCB_BUSY_MASK |
+					       SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+					       SRBM_STATUS__MCC_BUSY_MASK |
+					       SRBM_STATUS__MCD_BUSY_MASK |
+					       SRBM_STATUS__VMC_BUSY_MASK |
+					       SRBM_STATUS__VMC1_BUSY_MASK);
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+
+}
+
+static bool gmc_v8_0_check_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__VMC_BUSY_MASK)
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+						SRBM_SOFT_RESET, SOFT_RESET_VMC, 1);
+
+	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK)) {
+		if (!(adev->flags & AMD_IS_APU))
+			srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+							SRBM_SOFT_RESET, SOFT_RESET_MC, 1);
+	}
+	if (srbm_soft_reset) {
+		adev->gmc.srbm_soft_reset = srbm_soft_reset;
+		return true;
+	} else {
+		adev->gmc.srbm_soft_reset = 0;
+		return false;
+	}
+}
+
+static int gmc_v8_0_pre_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->gmc.srbm_soft_reset)
+		return 0;
+
+	gmc_v8_0_mc_stop(adev);
+	if (gmc_v8_0_wait_for_idle(adev)) {
+		dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
+	}
+
+	return 0;
+}
+
+static int gmc_v8_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset;
+
+	if (!adev->gmc.srbm_soft_reset)
+		return 0;
+	srbm_soft_reset = adev->gmc.srbm_soft_reset;
+
+	if (srbm_soft_reset) {
+		u32 tmp;
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int gmc_v8_0_post_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->gmc.srbm_soft_reset)
+		return 0;
+
+	gmc_v8_0_mc_resume(adev);
+	return 0;
+}
+
+static int gmc_v8_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	u32 tmp;
+	u32 bits = (VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK);
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		/* system context */
+		tmp = RREG32(mmVM_CONTEXT0_CNTL);
+		tmp &= ~bits;
+		WREG32(mmVM_CONTEXT0_CNTL, tmp);
+		/* VMs */
+		tmp = RREG32(mmVM_CONTEXT1_CNTL);
+		tmp &= ~bits;
+		WREG32(mmVM_CONTEXT1_CNTL, tmp);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		/* system context */
+		tmp = RREG32(mmVM_CONTEXT0_CNTL);
+		tmp |= bits;
+		WREG32(mmVM_CONTEXT0_CNTL, tmp);
+		/* VMs */
+		tmp = RREG32(mmVM_CONTEXT1_CNTL);
+		tmp |= bits;
+		WREG32(mmVM_CONTEXT1_CNTL, tmp);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	u32 addr, status, mc_client, vmid;
+
+	if (amdgpu_sriov_vf(adev)) {
+		dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
+			entry->src_id, entry->src_data[0]);
+		dev_err(adev->dev, " Can't decode VM fault info here on SRIOV VF\n");
+		return 0;
+	}
+
+	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
+	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
+	mc_client = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_MCCLIENT);
+	/* reset addr and status */
+	WREG32_P(mmVM_CONTEXT1_CNTL2, 1, ~1);
+
+	if (!addr && !status)
+		return 0;
+
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_FIRST)
+		gmc_v8_0_set_fault_enable_default(adev, false);
+
+#ifdef notyet
+	if (printk_ratelimit()) {
+		struct amdgpu_task_info task_info = { 0 };
+
+		amdgpu_vm_get_task_info(adev, entry->pasid, &task_info);
+
+		dev_err(adev->dev, "GPU fault detected: %d 0x%08x for process %s pid %d thread %s pid %d\n",
+			entry->src_id, entry->src_data[0], task_info.process_name,
+			task_info.tgid, task_info.task_name, task_info.pid);
+		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x%08X\n",
+			addr);
+		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
+			status);
+		gmc_v8_0_vm_decode_fault(adev, status, addr, mc_client,
+					 entry->pasid);
+	}
+#endif
+
+	vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			     VMID);
+	if (amdgpu_amdkfd_is_kfd_vmid(adev, vmid)
+		&& !atomic_read(&adev->gmc.vm_fault_info_updated)) {
+		struct kfd_vm_fault_info *info = adev->gmc.vm_fault_info;
+		u32 protections = REG_GET_FIELD(status,
+					VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					PROTECTIONS);
+
+		info->vmid = vmid;
+		info->mc_id = REG_GET_FIELD(status,
+					    VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					    MEMORY_CLIENT_ID);
+		info->status = status;
+		info->page_addr = addr;
+		info->prot_valid = protections & 0x7 ? true : false;
+		info->prot_read = protections & 0x8 ? true : false;
+		info->prot_write = protections & 0x10 ? true : false;
+		info->prot_exec = protections & 0x20 ? true : false;
+		mb();
+		atomic_set(&adev->gmc.vm_fault_info_updated, 1);
+	}
+
+	return 0;
+}
+
+static void fiji_update_mc_medium_grain_clock_gating(struct amdgpu_device *adev,
+						     bool enable)
+{
+	uint32_t data;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG)) {
+		data = RREG32(mmMC_HUB_MISC_HUB_CG);
+		data |= MC_HUB_MISC_HUB_CG__ENABLE_MASK;
+		WREG32(mmMC_HUB_MISC_HUB_CG, data);
+
+		data = RREG32(mmMC_HUB_MISC_SIP_CG);
+		data |= MC_HUB_MISC_SIP_CG__ENABLE_MASK;
+		WREG32(mmMC_HUB_MISC_SIP_CG, data);
+
+		data = RREG32(mmMC_HUB_MISC_VM_CG);
+		data |= MC_HUB_MISC_VM_CG__ENABLE_MASK;
+		WREG32(mmMC_HUB_MISC_VM_CG, data);
+
+		data = RREG32(mmMC_XPB_CLK_GAT);
+		data |= MC_XPB_CLK_GAT__ENABLE_MASK;
+		WREG32(mmMC_XPB_CLK_GAT, data);
+
+		data = RREG32(mmATC_MISC_CG);
+		data |= ATC_MISC_CG__ENABLE_MASK;
+		WREG32(mmATC_MISC_CG, data);
+
+		data = RREG32(mmMC_CITF_MISC_WR_CG);
+		data |= MC_CITF_MISC_WR_CG__ENABLE_MASK;
+		WREG32(mmMC_CITF_MISC_WR_CG, data);
+
+		data = RREG32(mmMC_CITF_MISC_RD_CG);
+		data |= MC_CITF_MISC_RD_CG__ENABLE_MASK;
+		WREG32(mmMC_CITF_MISC_RD_CG, data);
+
+		data = RREG32(mmMC_CITF_MISC_VM_CG);
+		data |= MC_CITF_MISC_VM_CG__ENABLE_MASK;
+		WREG32(mmMC_CITF_MISC_VM_CG, data);
+
+		data = RREG32(mmVM_L2_CG);
+		data |= VM_L2_CG__ENABLE_MASK;
+		WREG32(mmVM_L2_CG, data);
+	} else {
+		data = RREG32(mmMC_HUB_MISC_HUB_CG);
+		data &= ~MC_HUB_MISC_HUB_CG__ENABLE_MASK;
+		WREG32(mmMC_HUB_MISC_HUB_CG, data);
+
+		data = RREG32(mmMC_HUB_MISC_SIP_CG);
+		data &= ~MC_HUB_MISC_SIP_CG__ENABLE_MASK;
+		WREG32(mmMC_HUB_MISC_SIP_CG, data);
+
+		data = RREG32(mmMC_HUB_MISC_VM_CG);
+		data &= ~MC_HUB_MISC_VM_CG__ENABLE_MASK;
+		WREG32(mmMC_HUB_MISC_VM_CG, data);
+
+		data = RREG32(mmMC_XPB_CLK_GAT);
+		data &= ~MC_XPB_CLK_GAT__ENABLE_MASK;
+		WREG32(mmMC_XPB_CLK_GAT, data);
+
+		data = RREG32(mmATC_MISC_CG);
+		data &= ~ATC_MISC_CG__ENABLE_MASK;
+		WREG32(mmATC_MISC_CG, data);
+
+		data = RREG32(mmMC_CITF_MISC_WR_CG);
+		data &= ~MC_CITF_MISC_WR_CG__ENABLE_MASK;
+		WREG32(mmMC_CITF_MISC_WR_CG, data);
+
+		data = RREG32(mmMC_CITF_MISC_RD_CG);
+		data &= ~MC_CITF_MISC_RD_CG__ENABLE_MASK;
+		WREG32(mmMC_CITF_MISC_RD_CG, data);
+
+		data = RREG32(mmMC_CITF_MISC_VM_CG);
+		data &= ~MC_CITF_MISC_VM_CG__ENABLE_MASK;
+		WREG32(mmMC_CITF_MISC_VM_CG, data);
+
+		data = RREG32(mmVM_L2_CG);
+		data &= ~VM_L2_CG__ENABLE_MASK;
+		WREG32(mmVM_L2_CG, data);
+	}
+}
+
+static void fiji_update_mc_light_sleep(struct amdgpu_device *adev,
+				       bool enable)
+{
+	uint32_t data;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS)) {
+		data = RREG32(mmMC_HUB_MISC_HUB_CG);
+		data |= MC_HUB_MISC_HUB_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_HUB_MISC_HUB_CG, data);
+
+		data = RREG32(mmMC_HUB_MISC_SIP_CG);
+		data |= MC_HUB_MISC_SIP_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_HUB_MISC_SIP_CG, data);
+
+		data = RREG32(mmMC_HUB_MISC_VM_CG);
+		data |= MC_HUB_MISC_VM_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_HUB_MISC_VM_CG, data);
+
+		data = RREG32(mmMC_XPB_CLK_GAT);
+		data |= MC_XPB_CLK_GAT__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_XPB_CLK_GAT, data);
+
+		data = RREG32(mmATC_MISC_CG);
+		data |= ATC_MISC_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmATC_MISC_CG, data);
+
+		data = RREG32(mmMC_CITF_MISC_WR_CG);
+		data |= MC_CITF_MISC_WR_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_CITF_MISC_WR_CG, data);
+
+		data = RREG32(mmMC_CITF_MISC_RD_CG);
+		data |= MC_CITF_MISC_RD_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_CITF_MISC_RD_CG, data);
+
+		data = RREG32(mmMC_CITF_MISC_VM_CG);
+		data |= MC_CITF_MISC_VM_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_CITF_MISC_VM_CG, data);
+
+		data = RREG32(mmVM_L2_CG);
+		data |= VM_L2_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmVM_L2_CG, data);
+	} else {
+		data = RREG32(mmMC_HUB_MISC_HUB_CG);
+		data &= ~MC_HUB_MISC_HUB_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_HUB_MISC_HUB_CG, data);
+
+		data = RREG32(mmMC_HUB_MISC_SIP_CG);
+		data &= ~MC_HUB_MISC_SIP_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_HUB_MISC_SIP_CG, data);
+
+		data = RREG32(mmMC_HUB_MISC_VM_CG);
+		data &= ~MC_HUB_MISC_VM_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_HUB_MISC_VM_CG, data);
+
+		data = RREG32(mmMC_XPB_CLK_GAT);
+		data &= ~MC_XPB_CLK_GAT__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_XPB_CLK_GAT, data);
+
+		data = RREG32(mmATC_MISC_CG);
+		data &= ~ATC_MISC_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmATC_MISC_CG, data);
+
+		data = RREG32(mmMC_CITF_MISC_WR_CG);
+		data &= ~MC_CITF_MISC_WR_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_CITF_MISC_WR_CG, data);
+
+		data = RREG32(mmMC_CITF_MISC_RD_CG);
+		data &= ~MC_CITF_MISC_RD_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_CITF_MISC_RD_CG, data);
+
+		data = RREG32(mmMC_CITF_MISC_VM_CG);
+		data &= ~MC_CITF_MISC_VM_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmMC_CITF_MISC_VM_CG, data);
+
+		data = RREG32(mmVM_L2_CG);
+		data &= ~VM_L2_CG__MEM_LS_ENABLE_MASK;
+		WREG32(mmVM_L2_CG, data);
+	}
+}
+
+static int gmc_v8_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+		fiji_update_mc_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		fiji_update_mc_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gmc_v8_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static void gmc_v8_0_get_clockgating_state(void *handle, u32 *flags)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_MC_MGCG */
+	data = RREG32(mmMC_HUB_MISC_HUB_CG);
+	if (data & MC_HUB_MISC_HUB_CG__ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_MGCG;
+
+	/* AMD_CG_SUPPORT_MC_LS */
+	if (data & MC_HUB_MISC_HUB_CG__MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_LS;
+}
+
+static const struct amd_ip_funcs gmc_v8_0_ip_funcs = {
+	.name = "gmc_v8_0",
+	.early_init = gmc_v8_0_early_init,
+	.late_init = gmc_v8_0_late_init,
+	.sw_init = gmc_v8_0_sw_init,
+	.sw_fini = gmc_v8_0_sw_fini,
+	.hw_init = gmc_v8_0_hw_init,
+	.hw_fini = gmc_v8_0_hw_fini,
+	.suspend = gmc_v8_0_suspend,
+	.resume = gmc_v8_0_resume,
+	.is_idle = gmc_v8_0_is_idle,
+	.wait_for_idle = gmc_v8_0_wait_for_idle,
+	.check_soft_reset = gmc_v8_0_check_soft_reset,
+	.pre_soft_reset = gmc_v8_0_pre_soft_reset,
+	.soft_reset = gmc_v8_0_soft_reset,
+	.post_soft_reset = gmc_v8_0_post_soft_reset,
+	.set_clockgating_state = gmc_v8_0_set_clockgating_state,
+	.set_powergating_state = gmc_v8_0_set_powergating_state,
+	.get_clockgating_state = gmc_v8_0_get_clockgating_state,
+};
+
+static const struct amdgpu_gmc_funcs gmc_v8_0_gmc_funcs = {
+	.flush_gpu_tlb = gmc_v8_0_flush_gpu_tlb,
+	.emit_flush_gpu_tlb = gmc_v8_0_emit_flush_gpu_tlb,
+	.emit_pasid_mapping = gmc_v8_0_emit_pasid_mapping,
+	.set_pte_pde = gmc_v8_0_set_pte_pde,
+	.set_prt = gmc_v8_0_set_prt,
+	.get_vm_pte_flags = gmc_v8_0_get_vm_pte_flags,
+	.get_vm_pde = gmc_v8_0_get_vm_pde
+};
+
+static const struct amdgpu_irq_src_funcs gmc_v8_0_irq_funcs = {
+	.set = gmc_v8_0_vm_fault_interrupt_state,
+	.process = gmc_v8_0_process_interrupt,
+};
+
+static void gmc_v8_0_set_gmc_funcs(struct amdgpu_device *adev)
+{
+	if (adev->gmc.gmc_funcs == NULL)
+		adev->gmc.gmc_funcs = &gmc_v8_0_gmc_funcs;
+}
+
+static void gmc_v8_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gmc.vm_fault.num_types = 1;
+	adev->gmc.vm_fault.funcs = &gmc_v8_0_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version gmc_v8_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GMC,
+	.major = 8,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gmc_v8_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version gmc_v8_1_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GMC,
+	.major = 8,
+	.minor = 1,
+	.rev = 0,
+	.funcs = &gmc_v8_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version gmc_v8_5_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GMC,
+	.major = 8,
+	.minor = 5,
+	.rev = 0,
+	.funcs = &gmc_v8_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/gmc_v8_0.h b/sys/dev/pci/drm/amd/amdgpu/gmc_v8_0.h
new file mode 100644
index 00000000000..19b8a8aed20
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gmc_v8_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GMC_V8_0_H__
+#define __GMC_V8_0_H__
+
+extern const struct amdgpu_ip_block_version gmc_v8_0_ip_block;
+extern const struct amdgpu_ip_block_version gmc_v8_1_ip_block;
+extern const struct amdgpu_ip_block_version gmc_v8_5_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/gmc_v9_0.c b/sys/dev/pci/drm/amd/amdgpu/gmc_v9_0.c
new file mode 100644
index 00000000000..836227031b8
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gmc_v9_0.c
@@ -0,0 +1,1199 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <drm/drm_cache.h>
+#include "amdgpu.h"
+#include "gmc_v9_0.h"
+#include "amdgpu_atomfirmware.h"
+
+#include "hdp/hdp_4_0_offset.h"
+#include "hdp/hdp_4_0_sh_mask.h"
+#include "gc/gc_9_0_sh_mask.h"
+#include "dce/dce_12_0_offset.h"
+#include "dce/dce_12_0_sh_mask.h"
+#include "vega10_enum.h"
+#include "mmhub/mmhub_1_0_offset.h"
+#include "athub/athub_1_0_offset.h"
+#include "oss/osssys_4_0_offset.h"
+
+#include "soc15.h"
+#include "soc15_common.h"
+#include "umc/umc_6_0_sh_mask.h"
+
+#include "gfxhub_v1_0.h"
+#include "mmhub_v1_0.h"
+
+#include "ivsrcid/vmc/irqsrcs_vmc_1_0.h"
+
+/* add these here since we already include dce12 headers and these are for DCN */
+#define mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION                                                          0x055d
+#define mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION_BASE_IDX                                                 2
+#define HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION__PRI_VIEWPORT_WIDTH__SHIFT                                        0x0
+#define HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION__PRI_VIEWPORT_HEIGHT__SHIFT                                       0x10
+#define HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION__PRI_VIEWPORT_WIDTH_MASK                                          0x00003FFFL
+#define HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION__PRI_VIEWPORT_HEIGHT_MASK                                         0x3FFF0000L
+
+/* XXX Move this macro to VEGA10 header file, which is like vid.h for VI.*/
+#define AMDGPU_NUM_OF_VMIDS			8
+
+static const u32 golden_settings_vega10_hdp[] =
+{
+	0xf64, 0x0fffffff, 0x00000000,
+	0xf65, 0x0fffffff, 0x00000000,
+	0xf66, 0x0fffffff, 0x00000000,
+	0xf67, 0x0fffffff, 0x00000000,
+	0xf68, 0x0fffffff, 0x00000000,
+	0xf6a, 0x0fffffff, 0x00000000,
+	0xf6b, 0x0fffffff, 0x00000000,
+	0xf6c, 0x0fffffff, 0x00000000,
+	0xf6d, 0x0fffffff, 0x00000000,
+	0xf6e, 0x0fffffff, 0x00000000,
+};
+
+static const struct soc15_reg_golden golden_settings_mmhub_1_0_0[] =
+{
+	SOC15_REG_GOLDEN_VALUE(MMHUB, 0, mmDAGB1_WRCLI2, 0x00000007, 0xfe5fe0fa),
+	SOC15_REG_GOLDEN_VALUE(MMHUB, 0, mmMMEA1_DRAM_WR_CLI2GRP_MAP0, 0x00000030, 0x55555565)
+};
+
+static const struct soc15_reg_golden golden_settings_athub_1_0_0[] =
+{
+	SOC15_REG_GOLDEN_VALUE(ATHUB, 0, mmRPB_ARB_CNTL, 0x0000ff00, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(ATHUB, 0, mmRPB_ARB_CNTL2, 0x00ff00ff, 0x00080008)
+};
+
+/* Ecc related register addresses, (BASE + reg offset) */
+/* Universal Memory Controller caps (may be fused). */
+/* UMCCH:UmcLocalCap */
+#define UMCLOCALCAPS_ADDR0	(0x00014306 + 0x00000000)
+#define UMCLOCALCAPS_ADDR1	(0x00014306 + 0x00000800)
+#define UMCLOCALCAPS_ADDR2	(0x00014306 + 0x00001000)
+#define UMCLOCALCAPS_ADDR3	(0x00014306 + 0x00001800)
+#define UMCLOCALCAPS_ADDR4	(0x00054306 + 0x00000000)
+#define UMCLOCALCAPS_ADDR5	(0x00054306 + 0x00000800)
+#define UMCLOCALCAPS_ADDR6	(0x00054306 + 0x00001000)
+#define UMCLOCALCAPS_ADDR7	(0x00054306 + 0x00001800)
+#define UMCLOCALCAPS_ADDR8	(0x00094306 + 0x00000000)
+#define UMCLOCALCAPS_ADDR9	(0x00094306 + 0x00000800)
+#define UMCLOCALCAPS_ADDR10	(0x00094306 + 0x00001000)
+#define UMCLOCALCAPS_ADDR11	(0x00094306 + 0x00001800)
+#define UMCLOCALCAPS_ADDR12	(0x000d4306 + 0x00000000)
+#define UMCLOCALCAPS_ADDR13	(0x000d4306 + 0x00000800)
+#define UMCLOCALCAPS_ADDR14	(0x000d4306 + 0x00001000)
+#define UMCLOCALCAPS_ADDR15	(0x000d4306 + 0x00001800)
+
+/* Universal Memory Controller Channel config. */
+/* UMCCH:UMC_CONFIG */
+#define UMCCH_UMC_CONFIG_ADDR0	(0x00014040 + 0x00000000)
+#define UMCCH_UMC_CONFIG_ADDR1	(0x00014040 + 0x00000800)
+#define UMCCH_UMC_CONFIG_ADDR2	(0x00014040 + 0x00001000)
+#define UMCCH_UMC_CONFIG_ADDR3	(0x00014040 + 0x00001800)
+#define UMCCH_UMC_CONFIG_ADDR4	(0x00054040 + 0x00000000)
+#define UMCCH_UMC_CONFIG_ADDR5	(0x00054040 + 0x00000800)
+#define UMCCH_UMC_CONFIG_ADDR6	(0x00054040 + 0x00001000)
+#define UMCCH_UMC_CONFIG_ADDR7	(0x00054040 + 0x00001800)
+#define UMCCH_UMC_CONFIG_ADDR8	(0x00094040 + 0x00000000)
+#define UMCCH_UMC_CONFIG_ADDR9	(0x00094040 + 0x00000800)
+#define UMCCH_UMC_CONFIG_ADDR10	(0x00094040 + 0x00001000)
+#define UMCCH_UMC_CONFIG_ADDR11	(0x00094040 + 0x00001800)
+#define UMCCH_UMC_CONFIG_ADDR12	(0x000d4040 + 0x00000000)
+#define UMCCH_UMC_CONFIG_ADDR13	(0x000d4040 + 0x00000800)
+#define UMCCH_UMC_CONFIG_ADDR14	(0x000d4040 + 0x00001000)
+#define UMCCH_UMC_CONFIG_ADDR15	(0x000d4040 + 0x00001800)
+
+/* Universal Memory Controller Channel Ecc config. */
+/* UMCCH:EccCtrl */
+#define UMCCH_ECCCTRL_ADDR0	(0x00014053 + 0x00000000)
+#define UMCCH_ECCCTRL_ADDR1	(0x00014053 + 0x00000800)
+#define UMCCH_ECCCTRL_ADDR2	(0x00014053 + 0x00001000)
+#define UMCCH_ECCCTRL_ADDR3	(0x00014053 + 0x00001800)
+#define UMCCH_ECCCTRL_ADDR4	(0x00054053 + 0x00000000)
+#define UMCCH_ECCCTRL_ADDR5	(0x00054053 + 0x00000800)
+#define UMCCH_ECCCTRL_ADDR6	(0x00054053 + 0x00001000)
+#define UMCCH_ECCCTRL_ADDR7	(0x00054053 + 0x00001800)
+#define UMCCH_ECCCTRL_ADDR8	(0x00094053 + 0x00000000)
+#define UMCCH_ECCCTRL_ADDR9	(0x00094053 + 0x00000800)
+#define UMCCH_ECCCTRL_ADDR10	(0x00094053 + 0x00001000)
+#define UMCCH_ECCCTRL_ADDR11	(0x00094053 + 0x00001800)
+#define UMCCH_ECCCTRL_ADDR12	(0x000d4053 + 0x00000000)
+#define UMCCH_ECCCTRL_ADDR13	(0x000d4053 + 0x00000800)
+#define UMCCH_ECCCTRL_ADDR14	(0x000d4053 + 0x00001000)
+#define UMCCH_ECCCTRL_ADDR15	(0x000d4053 + 0x00001800)
+
+static const uint32_t ecc_umclocalcap_addrs[] = {
+	UMCLOCALCAPS_ADDR0,
+	UMCLOCALCAPS_ADDR1,
+	UMCLOCALCAPS_ADDR2,
+	UMCLOCALCAPS_ADDR3,
+	UMCLOCALCAPS_ADDR4,
+	UMCLOCALCAPS_ADDR5,
+	UMCLOCALCAPS_ADDR6,
+	UMCLOCALCAPS_ADDR7,
+	UMCLOCALCAPS_ADDR8,
+	UMCLOCALCAPS_ADDR9,
+	UMCLOCALCAPS_ADDR10,
+	UMCLOCALCAPS_ADDR11,
+	UMCLOCALCAPS_ADDR12,
+	UMCLOCALCAPS_ADDR13,
+	UMCLOCALCAPS_ADDR14,
+	UMCLOCALCAPS_ADDR15,
+};
+
+static const uint32_t ecc_umcch_umc_config_addrs[] = {
+	UMCCH_UMC_CONFIG_ADDR0,
+	UMCCH_UMC_CONFIG_ADDR1,
+	UMCCH_UMC_CONFIG_ADDR2,
+	UMCCH_UMC_CONFIG_ADDR3,
+	UMCCH_UMC_CONFIG_ADDR4,
+	UMCCH_UMC_CONFIG_ADDR5,
+	UMCCH_UMC_CONFIG_ADDR6,
+	UMCCH_UMC_CONFIG_ADDR7,
+	UMCCH_UMC_CONFIG_ADDR8,
+	UMCCH_UMC_CONFIG_ADDR9,
+	UMCCH_UMC_CONFIG_ADDR10,
+	UMCCH_UMC_CONFIG_ADDR11,
+	UMCCH_UMC_CONFIG_ADDR12,
+	UMCCH_UMC_CONFIG_ADDR13,
+	UMCCH_UMC_CONFIG_ADDR14,
+	UMCCH_UMC_CONFIG_ADDR15,
+};
+
+static const uint32_t ecc_umcch_eccctrl_addrs[] = {
+	UMCCH_ECCCTRL_ADDR0,
+	UMCCH_ECCCTRL_ADDR1,
+	UMCCH_ECCCTRL_ADDR2,
+	UMCCH_ECCCTRL_ADDR3,
+	UMCCH_ECCCTRL_ADDR4,
+	UMCCH_ECCCTRL_ADDR5,
+	UMCCH_ECCCTRL_ADDR6,
+	UMCCH_ECCCTRL_ADDR7,
+	UMCCH_ECCCTRL_ADDR8,
+	UMCCH_ECCCTRL_ADDR9,
+	UMCCH_ECCCTRL_ADDR10,
+	UMCCH_ECCCTRL_ADDR11,
+	UMCCH_ECCCTRL_ADDR12,
+	UMCCH_ECCCTRL_ADDR13,
+	UMCCH_ECCCTRL_ADDR14,
+	UMCCH_ECCCTRL_ADDR15,
+};
+
+static int gmc_v9_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *src,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	struct amdgpu_vmhub *hub;
+	u32 tmp, reg, bits, i, j;
+
+	bits = VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		VM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		for (j = 0; j < AMDGPU_MAX_VMHUBS; j++) {
+			hub = &adev->vmhub[j];
+			for (i = 0; i < 16; i++) {
+				reg = hub->vm_context0_cntl + i;
+				tmp = RREG32(reg);
+				tmp &= ~bits;
+				WREG32(reg, tmp);
+			}
+		}
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		for (j = 0; j < AMDGPU_MAX_VMHUBS; j++) {
+			hub = &adev->vmhub[j];
+			for (i = 0; i < 16; i++) {
+				reg = hub->vm_context0_cntl + i;
+				tmp = RREG32(reg);
+				tmp |= bits;
+				WREG32(reg, tmp);
+			}
+		}
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
+				struct amdgpu_irq_src *source,
+				struct amdgpu_iv_entry *entry)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[entry->vmid_src];
+	uint32_t status = 0;
+	u64 addr;
+
+	addr = (u64)entry->src_data[0] << 12;
+	addr |= ((u64)entry->src_data[1] & 0xf) << 44;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		status = RREG32(hub->vm_l2_pro_fault_status);
+		WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1);
+	}
+
+#ifdef notyet
+	if (printk_ratelimit()) {
+		struct amdgpu_task_info task_info = { 0 };
+
+		amdgpu_vm_get_task_info(adev, entry->pasid, &task_info);
+
+		dev_err(adev->dev,
+			"[%s] VMC page fault (src_id:%u ring:%u vmid:%u pasid:%u, for process %s pid %d thread %s pid %d\n)\n",
+			entry->vmid_src ? "mmhub" : "gfxhub",
+			entry->src_id, entry->ring_id, entry->vmid,
+			entry->pasid, task_info.process_name, task_info.tgid,
+			task_info.task_name, task_info.pid);
+		dev_err(adev->dev, "  at address 0x%016llx from %d\n",
+			addr, entry->client_id);
+		if (!amdgpu_sriov_vf(adev))
+			dev_err(adev->dev,
+				"VM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
+				status);
+	}
+#endif
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs gmc_v9_0_irq_funcs = {
+	.set = gmc_v9_0_vm_fault_interrupt_state,
+	.process = gmc_v9_0_process_interrupt,
+};
+
+static void gmc_v9_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gmc.vm_fault.num_types = 1;
+	adev->gmc.vm_fault.funcs = &gmc_v9_0_irq_funcs;
+}
+
+static uint32_t gmc_v9_0_get_invalidate_req(unsigned int vmid)
+{
+	u32 req = 0;
+
+	/* invalidate using legacy mode on vmid*/
+	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ,
+			    PER_VMID_INVALIDATE_REQ, 1 << vmid);
+	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, 0);
+	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
+	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
+	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
+	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
+	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
+	req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ,
+			    CLEAR_PROTECTION_FAULT_STATUS_ADDR,	0);
+
+	return req;
+}
+
+/*
+ * GART
+ * VMID 0 is the physical GPU addresses as used by the kernel.
+ * VMIDs 1-15 are used for userspace clients and are handled
+ * by the amdgpu vm/hsa code.
+ */
+
+/**
+ * gmc_v9_0_flush_gpu_tlb - gart tlb flush callback
+ *
+ * @adev: amdgpu_device pointer
+ * @vmid: vm instance to flush
+ *
+ * Flush the TLB for the requested page table.
+ */
+static void gmc_v9_0_flush_gpu_tlb(struct amdgpu_device *adev,
+					uint32_t vmid)
+{
+	/* Use register 17 for GART */
+	const unsigned eng = 17;
+	unsigned i, j;
+
+	spin_lock(&adev->gmc.invalidate_lock);
+
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) {
+		struct amdgpu_vmhub *hub = &adev->vmhub[i];
+		u32 tmp = gmc_v9_0_get_invalidate_req(vmid);
+
+		WREG32_NO_KIQ(hub->vm_inv_eng0_req + eng, tmp);
+
+		/* Busy wait for ACK.*/
+		for (j = 0; j < 100; j++) {
+			tmp = RREG32_NO_KIQ(hub->vm_inv_eng0_ack + eng);
+			tmp &= 1 << vmid;
+			if (tmp)
+				break;
+			cpu_relax();
+		}
+		if (j < 100)
+			continue;
+
+		/* Wait for ACK with a delay.*/
+		for (j = 0; j < adev->usec_timeout; j++) {
+			tmp = RREG32_NO_KIQ(hub->vm_inv_eng0_ack + eng);
+			tmp &= 1 << vmid;
+			if (tmp)
+				break;
+			udelay(1);
+		}
+		if (j < adev->usec_timeout)
+			continue;
+
+		DRM_ERROR("Timeout waiting for VM flush ACK!\n");
+	}
+
+	spin_unlock(&adev->gmc.invalidate_lock);
+}
+
+static uint64_t gmc_v9_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
+					    unsigned vmid, uint64_t pd_addr)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_vmhub *hub = &adev->vmhub[ring->funcs->vmhub];
+	uint32_t req = gmc_v9_0_get_invalidate_req(vmid);
+	uint64_t flags = AMDGPU_PTE_VALID;
+	unsigned eng = ring->vm_inv_eng;
+
+	amdgpu_gmc_get_vm_pde(adev, -1, &pd_addr, &flags);
+	pd_addr |= flags;
+
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_lo32 + (2 * vmid),
+			      lower_32_bits(pd_addr));
+
+	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_hi32 + (2 * vmid),
+			      upper_32_bits(pd_addr));
+
+	amdgpu_ring_emit_reg_write_reg_wait(ring, hub->vm_inv_eng0_req + eng,
+					    hub->vm_inv_eng0_ack + eng,
+					    req, 1 << vmid);
+
+	return pd_addr;
+}
+
+static void gmc_v9_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned vmid,
+					unsigned pasid)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reg;
+
+	if (ring->funcs->vmhub == AMDGPU_GFXHUB)
+		reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT) + vmid;
+	else
+		reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT_MM) + vmid;
+
+	amdgpu_ring_emit_wreg(ring, reg, pasid);
+}
+
+/**
+ * gmc_v9_0_set_pte_pde - update the page tables using MMIO
+ *
+ * @adev: amdgpu_device pointer
+ * @cpu_pt_addr: cpu address of the page table
+ * @gpu_page_idx: entry in the page table to update
+ * @addr: dst addr to write into pte/pde
+ * @flags: access flags
+ *
+ * Update the page tables using the CPU.
+ */
+static int gmc_v9_0_set_pte_pde(struct amdgpu_device *adev, void *cpu_pt_addr,
+				uint32_t gpu_page_idx, uint64_t addr,
+				uint64_t flags)
+{
+	void __iomem *ptr = (void *)cpu_pt_addr;
+	uint64_t value;
+
+	/*
+	 * PTE format on VEGA 10:
+	 * 63:59 reserved
+	 * 58:57 mtype
+	 * 56 F
+	 * 55 L
+	 * 54 P
+	 * 53 SW
+	 * 52 T
+	 * 50:48 reserved
+	 * 47:12 4k physical page base address
+	 * 11:7 fragment
+	 * 6 write
+	 * 5 read
+	 * 4 exe
+	 * 3 Z
+	 * 2 snooped
+	 * 1 system
+	 * 0 valid
+	 *
+	 * PDE format on VEGA 10:
+	 * 63:59 block fragment size
+	 * 58:55 reserved
+	 * 54 P
+	 * 53:48 reserved
+	 * 47:6 physical base address of PD or PTE
+	 * 5:3 reserved
+	 * 2 C
+	 * 1 system
+	 * 0 valid
+	 */
+
+	/*
+	 * The following is for PTE only. GART does not have PDEs.
+	*/
+	value = addr & 0x0000FFFFFFFFF000ULL;
+	value |= flags;
+	writeq(value, ptr + (gpu_page_idx * 8));
+	return 0;
+}
+
+static uint64_t gmc_v9_0_get_vm_pte_flags(struct amdgpu_device *adev,
+						uint32_t flags)
+
+{
+	uint64_t pte_flag = 0;
+
+	if (flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		pte_flag |= AMDGPU_PTE_EXECUTABLE;
+	if (flags & AMDGPU_VM_PAGE_READABLE)
+		pte_flag |= AMDGPU_PTE_READABLE;
+	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
+		pte_flag |= AMDGPU_PTE_WRITEABLE;
+
+	switch (flags & AMDGPU_VM_MTYPE_MASK) {
+	case AMDGPU_VM_MTYPE_DEFAULT:
+		pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_NC);
+		break;
+	case AMDGPU_VM_MTYPE_NC:
+		pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_NC);
+		break;
+	case AMDGPU_VM_MTYPE_WC:
+		pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_WC);
+		break;
+	case AMDGPU_VM_MTYPE_CC:
+		pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_CC);
+		break;
+	case AMDGPU_VM_MTYPE_UC:
+		pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_UC);
+		break;
+	default:
+		pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_NC);
+		break;
+	}
+
+	if (flags & AMDGPU_VM_PAGE_PRT)
+		pte_flag |= AMDGPU_PTE_PRT;
+
+	return pte_flag;
+}
+
+static void gmc_v9_0_get_vm_pde(struct amdgpu_device *adev, int level,
+				uint64_t *addr, uint64_t *flags)
+{
+	if (!(*flags & AMDGPU_PDE_PTE))
+		*addr = adev->vm_manager.vram_base_offset + *addr -
+			adev->gmc.vram_start;
+	BUG_ON(*addr & 0xFFFF00000000003FULL);
+
+	if (!adev->gmc.translate_further)
+		return;
+
+	if (level == AMDGPU_VM_PDB1) {
+		/* Set the block fragment size */
+		if (!(*flags & AMDGPU_PDE_PTE))
+			*flags |= AMDGPU_PDE_BFS(0x9);
+
+	} else if (level == AMDGPU_VM_PDB0) {
+		if (*flags & AMDGPU_PDE_PTE)
+			*flags &= ~AMDGPU_PDE_PTE;
+		else
+			*flags |= AMDGPU_PTE_TF;
+	}
+}
+
+static const struct amdgpu_gmc_funcs gmc_v9_0_gmc_funcs = {
+	.flush_gpu_tlb = gmc_v9_0_flush_gpu_tlb,
+	.emit_flush_gpu_tlb = gmc_v9_0_emit_flush_gpu_tlb,
+	.emit_pasid_mapping = gmc_v9_0_emit_pasid_mapping,
+	.set_pte_pde = gmc_v9_0_set_pte_pde,
+	.get_vm_pte_flags = gmc_v9_0_get_vm_pte_flags,
+	.get_vm_pde = gmc_v9_0_get_vm_pde
+};
+
+static void gmc_v9_0_set_gmc_funcs(struct amdgpu_device *adev)
+{
+	if (adev->gmc.gmc_funcs == NULL)
+		adev->gmc.gmc_funcs = &gmc_v9_0_gmc_funcs;
+}
+
+static int gmc_v9_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v9_0_set_gmc_funcs(adev);
+	gmc_v9_0_set_irq_funcs(adev);
+
+	adev->gmc.shared_aperture_start = 0x2000000000000000ULL;
+	adev->gmc.shared_aperture_end =
+		adev->gmc.shared_aperture_start + (4ULL << 30) - 1;
+	adev->gmc.private_aperture_start = 0x1000000000000000ULL;
+	adev->gmc.private_aperture_end =
+		adev->gmc.private_aperture_start + (4ULL << 30) - 1;
+
+	return 0;
+}
+
+static int gmc_v9_0_ecc_available(struct amdgpu_device *adev)
+{
+	uint32_t reg_val;
+	uint32_t reg_addr;
+	uint32_t field_val;
+	size_t i;
+	uint32_t fv2;
+	size_t lost_sheep;
+
+	DRM_DEBUG("ecc: gmc_v9_0_ecc_available()\n");
+
+	lost_sheep = 0;
+	for (i = 0; i < ARRAY_SIZE(ecc_umclocalcap_addrs); ++i) {
+		reg_addr = ecc_umclocalcap_addrs[i];
+		DRM_DEBUG("ecc: "
+			  "UMCCH_UmcLocalCap[%zu]: reg_addr: 0x%08x\n",
+			  i, reg_addr);
+		reg_val = RREG32(reg_addr);
+		field_val = REG_GET_FIELD(reg_val, UMCCH0_0_UmcLocalCap,
+					  EccDis);
+		DRM_DEBUG("ecc: "
+			  "reg_val: 0x%08x, "
+			  "EccDis: 0x%08x, ",
+			  reg_val, field_val);
+		if (field_val) {
+			DRM_ERROR("ecc: UmcLocalCap:EccDis is set.\n");
+			++lost_sheep;
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(ecc_umcch_umc_config_addrs); ++i) {
+		reg_addr = ecc_umcch_umc_config_addrs[i];
+		DRM_DEBUG("ecc: "
+			  "UMCCH0_0_UMC_CONFIG[%zu]: reg_addr: 0x%08x",
+			  i, reg_addr);
+		reg_val = RREG32(reg_addr);
+		field_val = REG_GET_FIELD(reg_val, UMCCH0_0_UMC_CONFIG,
+					  DramReady);
+		DRM_DEBUG("ecc: "
+			  "reg_val: 0x%08x, "
+			  "DramReady: 0x%08x\n",
+			  reg_val, field_val);
+
+		if (!field_val) {
+			DRM_ERROR("ecc: UMC_CONFIG:DramReady is not set.\n");
+			++lost_sheep;
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(ecc_umcch_eccctrl_addrs); ++i) {
+		reg_addr = ecc_umcch_eccctrl_addrs[i];
+		DRM_DEBUG("ecc: "
+			  "UMCCH_EccCtrl[%zu]: reg_addr: 0x%08x, ",
+			  i, reg_addr);
+		reg_val = RREG32(reg_addr);
+		field_val = REG_GET_FIELD(reg_val, UMCCH0_0_EccCtrl,
+					  WrEccEn);
+		fv2 = REG_GET_FIELD(reg_val, UMCCH0_0_EccCtrl,
+				    RdEccEn);
+		DRM_DEBUG("ecc: "
+			  "reg_val: 0x%08x, "
+			  "WrEccEn: 0x%08x, "
+			  "RdEccEn: 0x%08x\n",
+			  reg_val, field_val, fv2);
+
+		if (!field_val) {
+			DRM_DEBUG("ecc: WrEccEn is not set\n");
+			++lost_sheep;
+		}
+		if (!fv2) {
+			DRM_DEBUG("ecc: RdEccEn is not set\n");
+			++lost_sheep;
+		}
+	}
+
+	DRM_DEBUG("ecc: lost_sheep: %zu\n", lost_sheep);
+	return lost_sheep == 0;
+}
+
+static int gmc_v9_0_late_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	/*
+	 * The latest engine allocation on gfx9 is:
+	 * Engine 0, 1: idle
+	 * Engine 2, 3: firmware
+	 * Engine 4~13: amdgpu ring, subject to change when ring number changes
+	 * Engine 14~15: idle
+	 * Engine 16: kfd tlb invalidation
+	 * Engine 17: Gart flushes
+	 */
+	unsigned vm_inv_eng[AMDGPU_MAX_VMHUBS] = { 4, 4 };
+	unsigned i;
+	int r;
+
+	/*
+	 * TODO - Uncomment once GART corruption issue is fixed.
+	 */
+	/* amdgpu_bo_late_init(adev); */
+
+	for(i = 0; i < adev->num_rings; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		unsigned vmhub = ring->funcs->vmhub;
+
+		ring->vm_inv_eng = vm_inv_eng[vmhub]++;
+		dev_info(adev->dev, "ring %u(%s) uses VM inv eng %u on hub %u\n",
+			 ring->idx, ring->name, ring->vm_inv_eng,
+			 ring->funcs->vmhub);
+	}
+
+	/* Engine 16 is used for KFD and 17 for GART flushes */
+	for(i = 0; i < AMDGPU_MAX_VMHUBS; ++i)
+		BUG_ON(vm_inv_eng[i] > 16);
+
+	if (adev->asic_type == CHIP_VEGA10 && !amdgpu_sriov_vf(adev)) {
+		r = gmc_v9_0_ecc_available(adev);
+		if (r == 1) {
+			DRM_INFO("ECC is active.\n");
+		} else if (r == 0) {
+			DRM_INFO("ECC is not present.\n");
+			adev->df_funcs->enable_ecc_force_par_wr_rmw(adev, false);
+		} else {
+			DRM_ERROR("gmc_v9_0_ecc_available() failed. r: %d\n", r);
+			return r;
+		}
+	}
+
+	return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
+}
+
+static void gmc_v9_0_vram_gtt_location(struct amdgpu_device *adev,
+					struct amdgpu_gmc *mc)
+{
+	u64 base = 0;
+	if (!amdgpu_sriov_vf(adev))
+		base = mmhub_v1_0_get_fb_location(adev);
+	amdgpu_device_vram_location(adev, &adev->gmc, base);
+	amdgpu_device_gart_location(adev, mc);
+	/* base offset of vram pages */
+	adev->vm_manager.vram_base_offset = gfxhub_v1_0_get_mc_fb_offset(adev);
+}
+
+/**
+ * gmc_v9_0_mc_init - initialize the memory controller driver params
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the amount of vram, vram width, and decide how to place
+ * vram and gart within the GPU's physical address space.
+ * Returns 0 for success.
+ */
+static int gmc_v9_0_mc_init(struct amdgpu_device *adev)
+{
+	int chansize, numchan;
+	int r;
+
+	if (amdgpu_emu_mode != 1)
+		adev->gmc.vram_width = amdgpu_atomfirmware_get_vram_width(adev);
+	if (!adev->gmc.vram_width) {
+		/* hbm memory channel size */
+		if (adev->flags & AMD_IS_APU)
+			chansize = 64;
+		else
+			chansize = 128;
+
+		numchan = adev->df_funcs->get_hbm_channel_number(adev);
+		adev->gmc.vram_width = numchan * chansize;
+	}
+
+	/* size in MB on si */
+	adev->gmc.mc_vram_size =
+		adev->nbio_funcs->get_memsize(adev) * 1024ULL * 1024ULL;
+	adev->gmc.real_vram_size = adev->gmc.mc_vram_size;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_device_resize_fb_bar(adev);
+		if (r)
+			return r;
+	}
+	adev->gmc.aper_base = adev->fb_aper_offset;
+	adev->gmc.aper_size = adev->fb_aper_size;
+
+#ifdef __amd64__
+	if (adev->flags & AMD_IS_APU) {
+		adev->gmc.aper_base = gfxhub_v1_0_get_mc_fb_offset(adev);
+		adev->gmc.aper_size = adev->gmc.real_vram_size;
+	}
+#endif
+	/* In case the PCI BAR is larger than the actual amount of vram */
+	adev->gmc.visible_vram_size = adev->gmc.aper_size;
+	if (adev->gmc.visible_vram_size > adev->gmc.real_vram_size)
+		adev->gmc.visible_vram_size = adev->gmc.real_vram_size;
+
+	/* set the gart size */
+	if (amdgpu_gart_size == -1) {
+		switch (adev->asic_type) {
+		case CHIP_VEGA10:  /* all engines support GPUVM */
+		case CHIP_VEGA12:  /* all engines support GPUVM */
+		case CHIP_VEGA20:
+		default:
+			adev->gmc.gart_size = 512ULL << 20;
+			break;
+		case CHIP_RAVEN:   /* DCE SG support */
+			adev->gmc.gart_size = 1024ULL << 20;
+			break;
+		}
+	} else {
+		adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
+	}
+
+	gmc_v9_0_vram_gtt_location(adev, &adev->gmc);
+
+	return 0;
+}
+
+static int gmc_v9_0_gart_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.robj) {
+		WARN(1, "VEGA10 PCIE GART already initialized\n");
+		return 0;
+	}
+	/* Initialize common gart structure */
+	r = amdgpu_gart_init(adev);
+	if (r)
+		return r;
+	adev->gart.table_size = adev->gart.num_gpu_pages * 8;
+	adev->gart.gart_pte_flags = AMDGPU_PTE_MTYPE(MTYPE_UC) |
+				 AMDGPU_PTE_EXECUTABLE;
+	return amdgpu_gart_table_vram_alloc(adev);
+}
+
+static unsigned gmc_v9_0_get_vbios_fb_size(struct amdgpu_device *adev)
+{
+#if 0
+	u32 d1vga_control = RREG32_SOC15(DCE, 0, mmD1VGA_CONTROL);
+#endif
+	unsigned size;
+
+	/*
+	 * TODO Remove once GART corruption is resolved
+	 * Check related code in gmc_v9_0_sw_fini
+	 * */
+	size = 9 * 1024 * 1024;
+
+#if 0
+	if (REG_GET_FIELD(d1vga_control, D1VGA_CONTROL, D1VGA_MODE_ENABLE)) {
+		size = 9 * 1024 * 1024; /* reserve 8MB for vga emulator and 1 MB for FB */
+	} else {
+		u32 viewport;
+
+		switch (adev->asic_type) {
+		case CHIP_RAVEN:
+			viewport = RREG32_SOC15(DCE, 0, mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION);
+			size = (REG_GET_FIELD(viewport,
+					      HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION, PRI_VIEWPORT_HEIGHT) *
+				REG_GET_FIELD(viewport,
+					      HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION, PRI_VIEWPORT_WIDTH) *
+				4);
+			break;
+		case CHIP_VEGA10:
+		case CHIP_VEGA12:
+		default:
+			viewport = RREG32_SOC15(DCE, 0, mmSCL0_VIEWPORT_SIZE);
+			size = (REG_GET_FIELD(viewport, SCL0_VIEWPORT_SIZE, VIEWPORT_HEIGHT) *
+				REG_GET_FIELD(viewport, SCL0_VIEWPORT_SIZE, VIEWPORT_WIDTH) *
+				4);
+			break;
+		}
+	}
+	/* return 0 if the pre-OS buffer uses up most of vram */
+	if ((adev->gmc.real_vram_size - size) < (8 * 1024 * 1024))
+		return 0;
+
+#endif
+	return size;
+}
+
+static int gmc_v9_0_sw_init(void *handle)
+{
+	int r;
+	int dma_bits;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gfxhub_v1_0_init(adev);
+	mmhub_v1_0_init(adev);
+
+	mtx_init(&adev->gmc.invalidate_lock, IPL_TTY);
+
+	adev->gmc.vram_type = amdgpu_atomfirmware_get_vram_type(adev);
+	switch (adev->asic_type) {
+	case CHIP_RAVEN:
+		if (adev->rev_id == 0x0 || adev->rev_id == 0x1) {
+			amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
+		} else {
+			/* vm_size is 128TB + 512GB for legacy 3-level page support */
+			amdgpu_vm_adjust_size(adev, 128 * 1024 + 512, 9, 2, 48);
+			adev->gmc.translate_further =
+				adev->vm_manager.num_level > 1;
+		}
+		break;
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+		/*
+		 * To fulfill 4-level page support,
+		 * vm size is 256TB (48bit), maximum size of Vega10,
+		 * block size 512 (9bit)
+		 */
+		amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
+		break;
+	default:
+		break;
+	}
+
+	/* This interrupt is VMC page fault.*/
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VMC, VMC_1_0__SRCID__VM_FAULT,
+				&adev->gmc.vm_fault);
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_UTCL2, UTCL2_1_0__SRCID__FAULT,
+				&adev->gmc.vm_fault);
+
+	if (r)
+		return r;
+
+	/* Set the internal MC address mask
+	 * This is the max address of the GPU's
+	 * internal address space.
+	 */
+	adev->gmc.mc_mask = 0xffffffffffffULL; /* 48 bit MC */
+
+	/* set DMA mask + need_dma32 flags.
+	 * PCIE - can handle 44-bits.
+	 * IGP - can handle 44-bits
+	 * PCI - dma32 for legacy pci gart, 44 bits on vega10
+	 */
+	adev->need_dma32 = false;
+#ifdef notyet
+	dma_bits = adev->need_dma32 ? 32 : 44;
+	r = pci_set_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	if (r) {
+		adev->need_dma32 = true;
+		dma_bits = 32;
+		printk(KERN_WARNING "amdgpu: No suitable DMA available.\n");
+	}
+	r = pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	if (r) {
+		pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(32));
+		printk(KERN_WARNING "amdgpu: No coherent DMA available.\n");
+	}
+	adev->need_swiotlb = drm_get_max_iomem() > ((u64)1 << dma_bits);
+#endif
+
+	r = gmc_v9_0_mc_init(adev);
+	if (r)
+		return r;
+
+	adev->gmc.stolen_size = gmc_v9_0_get_vbios_fb_size(adev);
+
+	/* Memory manager */
+	r = amdgpu_bo_init(adev);
+	if (r)
+		return r;
+
+	r = gmc_v9_0_gart_init(adev);
+	if (r)
+		return r;
+
+	/*
+	 * number of VMs
+	 * VMID 0 is reserved for System
+	 * amdgpu graphics/compute will use VMIDs 1-7
+	 * amdkfd will use VMIDs 8-15
+	 */
+	adev->vm_manager.id_mgr[AMDGPU_GFXHUB].num_ids = AMDGPU_NUM_OF_VMIDS;
+	adev->vm_manager.id_mgr[AMDGPU_MMHUB].num_ids = AMDGPU_NUM_OF_VMIDS;
+
+	amdgpu_vm_manager_init(adev);
+
+	return 0;
+}
+
+static int gmc_v9_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_gem_force_release(adev);
+	amdgpu_vm_manager_fini(adev);
+
+	/*
+	* TODO:
+	* Currently there is a bug where some memory client outside
+	* of the driver writes to first 8M of VRAM on S3 resume,
+	* this overrides GART which by default gets placed in first 8M and
+	* causes VM_FAULTS once GTT is accessed.
+	* Keep the stolen memory reservation until the while this is not solved.
+	* Also check code in gmc_v9_0_get_vbios_fb_size and gmc_v9_0_late_init
+	*/
+	amdgpu_bo_free_kernel(&adev->stolen_vga_memory, NULL, NULL);
+
+	amdgpu_gart_table_vram_free(adev);
+	amdgpu_bo_fini(adev);
+	amdgpu_gart_fini(adev);
+
+	return 0;
+}
+
+static void gmc_v9_0_init_golden_registers(struct amdgpu_device *adev)
+{
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+	case CHIP_VEGA20:
+		soc15_program_register_sequence(adev,
+						golden_settings_mmhub_1_0_0,
+						ARRAY_SIZE(golden_settings_mmhub_1_0_0));
+		soc15_program_register_sequence(adev,
+						golden_settings_athub_1_0_0,
+						ARRAY_SIZE(golden_settings_athub_1_0_0));
+		break;
+	case CHIP_VEGA12:
+		break;
+	case CHIP_RAVEN:
+		soc15_program_register_sequence(adev,
+						golden_settings_athub_1_0_0,
+						ARRAY_SIZE(golden_settings_athub_1_0_0));
+		break;
+	default:
+		break;
+	}
+}
+
+/**
+ * gmc_v9_0_gart_enable - gart enable
+ *
+ * @adev: amdgpu_device pointer
+ */
+static int gmc_v9_0_gart_enable(struct amdgpu_device *adev)
+{
+	int r;
+	bool value;
+	u32 tmp;
+
+	amdgpu_device_program_register_sequence(adev,
+						golden_settings_vega10_hdp,
+						ARRAY_SIZE(golden_settings_vega10_hdp));
+
+	if (adev->gart.robj == NULL) {
+		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
+		return -EINVAL;
+	}
+	r = amdgpu_gart_table_vram_pin(adev);
+	if (r)
+		return r;
+
+	switch (adev->asic_type) {
+	case CHIP_RAVEN:
+		mmhub_v1_0_initialize_power_gating(adev);
+		mmhub_v1_0_update_power_gating(adev, true);
+		break;
+	default:
+		break;
+	}
+
+	r = gfxhub_v1_0_gart_enable(adev);
+	if (r)
+		return r;
+
+	r = mmhub_v1_0_gart_enable(adev);
+	if (r)
+		return r;
+
+	WREG32_FIELD15(HDP, 0, HDP_MISC_CNTL, FLUSH_INVALIDATE_CACHE, 1);
+
+	tmp = RREG32_SOC15(HDP, 0, mmHDP_HOST_PATH_CNTL);
+	WREG32_SOC15(HDP, 0, mmHDP_HOST_PATH_CNTL, tmp);
+
+	/* After HDP is initialized, flush HDP.*/
+	adev->nbio_funcs->hdp_flush(adev, NULL);
+
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS)
+		value = false;
+	else
+		value = true;
+
+	gfxhub_v1_0_set_fault_enable_default(adev, value);
+	mmhub_v1_0_set_fault_enable_default(adev, value);
+	gmc_v9_0_flush_gpu_tlb(adev, 0);
+
+	DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
+		 (unsigned)(adev->gmc.gart_size >> 20),
+		 (unsigned long long)adev->gart.table_addr);
+	adev->gart.ready = true;
+	return 0;
+}
+
+static int gmc_v9_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* The sequence of these two function calls matters.*/
+	gmc_v9_0_init_golden_registers(adev);
+
+	if (adev->mode_info.num_crtc) {
+		/* Lockout access through VGA aperture*/
+		WREG32_FIELD15(DCE, 0, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
+
+		/* disable VGA render */
+		WREG32_FIELD15(DCE, 0, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
+	}
+
+	r = gmc_v9_0_gart_enable(adev);
+
+	return r;
+}
+
+/**
+ * gmc_v9_0_gart_disable - gart disable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This disables all VM page table.
+ */
+static void gmc_v9_0_gart_disable(struct amdgpu_device *adev)
+{
+	gfxhub_v1_0_gart_disable(adev);
+	mmhub_v1_0_gart_disable(adev);
+	amdgpu_gart_table_vram_unpin(adev);
+}
+
+static int gmc_v9_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev)) {
+		/* full access mode, so don't touch any GMC register */
+		DRM_DEBUG("For SRIOV client, shouldn't do anything.\n");
+		return 0;
+	}
+
+	amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
+	gmc_v9_0_gart_disable(adev);
+
+	return 0;
+}
+
+static int gmc_v9_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return gmc_v9_0_hw_fini(adev);
+}
+
+static int gmc_v9_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = gmc_v9_0_hw_init(adev);
+	if (r)
+		return r;
+
+	amdgpu_vmid_reset_all(adev);
+
+	return 0;
+}
+
+static bool gmc_v9_0_is_idle(void *handle)
+{
+	/* MC is always ready in GMC v9.*/
+	return true;
+}
+
+static int gmc_v9_0_wait_for_idle(void *handle)
+{
+	/* There is no need to wait for MC idle in GMC v9.*/
+	return 0;
+}
+
+static int gmc_v9_0_soft_reset(void *handle)
+{
+	/* XXX for emulation.*/
+	return 0;
+}
+
+static int gmc_v9_0_set_clockgating_state(void *handle,
+					enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return mmhub_v1_0_set_clockgating(adev, state);
+}
+
+static void gmc_v9_0_get_clockgating_state(void *handle, u32 *flags)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mmhub_v1_0_get_clockgating(adev, flags);
+}
+
+static int gmc_v9_0_set_powergating_state(void *handle,
+					enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs gmc_v9_0_ip_funcs = {
+	.name = "gmc_v9_0",
+	.early_init = gmc_v9_0_early_init,
+	.late_init = gmc_v9_0_late_init,
+	.sw_init = gmc_v9_0_sw_init,
+	.sw_fini = gmc_v9_0_sw_fini,
+	.hw_init = gmc_v9_0_hw_init,
+	.hw_fini = gmc_v9_0_hw_fini,
+	.suspend = gmc_v9_0_suspend,
+	.resume = gmc_v9_0_resume,
+	.is_idle = gmc_v9_0_is_idle,
+	.wait_for_idle = gmc_v9_0_wait_for_idle,
+	.soft_reset = gmc_v9_0_soft_reset,
+	.set_clockgating_state = gmc_v9_0_set_clockgating_state,
+	.set_powergating_state = gmc_v9_0_set_powergating_state,
+	.get_clockgating_state = gmc_v9_0_get_clockgating_state,
+};
+
+const struct amdgpu_ip_block_version gmc_v9_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_GMC,
+	.major = 9,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &gmc_v9_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/gmc_v9_0.h b/sys/dev/pci/drm/amd/amdgpu/gmc_v9_0.h
new file mode 100644
index 00000000000..b030ca5ea10
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/gmc_v9_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GMC_V9_0_H__
+#define __GMC_V9_0_H__
+
+extern const struct amd_ip_funcs gmc_v9_0_ip_funcs;
+extern const struct amdgpu_ip_block_version gmc_v9_0_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/iceland_ih.c b/sys/dev/pci/drm/amd/amdgpu/iceland_ih.c
new file mode 100644
index 00000000000..842c4b677b4
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/iceland_ih.c
@@ -0,0 +1,461 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "vid.h"
+
+#include "oss/oss_2_4_d.h"
+#include "oss/oss_2_4_sh_mask.h"
+
+#include "bif/bif_5_1_d.h"
+#include "bif/bif_5_1_sh_mask.h"
+
+/*
+ * Interrupts
+ * Starting with r6xx, interrupts are handled via a ring buffer.
+ * Ring buffers are areas of GPU accessible memory that the GPU
+ * writes interrupt vectors into and the host reads vectors out of.
+ * There is a rptr (read pointer) that determines where the
+ * host is currently reading, and a wptr (write pointer)
+ * which determines where the GPU has written.  When the
+ * pointers are equal, the ring is idle.  When the GPU
+ * writes vectors to the ring buffer, it increments the
+ * wptr.  When there is an interrupt, the host then starts
+ * fetching commands and processing them until the pointers are
+ * equal again at which point it updates the rptr.
+ */
+
+static void iceland_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * iceland_ih_enable_interrupts - Enable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enable the interrupt ring buffer (VI).
+ */
+static void iceland_ih_enable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_cntl = RREG32(mmIH_CNTL);
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, ENABLE_INTR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 1);
+	WREG32(mmIH_CNTL, ih_cntl);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	adev->irq.ih.enabled = true;
+}
+
+/**
+ * iceland_ih_disable_interrupts - Disable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable the interrupt ring buffer (VI).
+ */
+static void iceland_ih_disable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+	u32 ih_cntl = RREG32(mmIH_CNTL);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 0);
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, ENABLE_INTR, 0);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	WREG32(mmIH_CNTL, ih_cntl);
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+	adev->irq.ih.enabled = false;
+	adev->irq.ih.rptr = 0;
+}
+
+/**
+ * iceland_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (VI).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int iceland_ih_irq_init(struct amdgpu_device *adev)
+{
+	int rb_bufsz;
+	u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
+	u64 wptr_off;
+
+	/* disable irqs */
+	iceland_ih_disable_interrupts(adev);
+
+	/* setup interrupt control */
+	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
+	interrupt_cntl = RREG32(mmINTERRUPT_CNTL);
+	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_DUMMY_RD_OVERRIDE, 0);
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_REQ_NONSNOOP_EN, 0);
+	WREG32(mmINTERRUPT_CNTL, interrupt_cntl);
+
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	WREG32(mmIH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
+
+	rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
+	ih_rb_cntl = REG_SET_FIELD(0, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register value is written to memory */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
+
+	/* set the writeback address whether it's enabled or not */
+	wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
+	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
+	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+
+	/* Default settings for IH_CNTL (disabled at first) */
+	ih_cntl = RREG32(mmIH_CNTL);
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, MC_VMID, 0);
+
+	if (adev->irq.msi_enabled)
+		ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, RPTR_REARM, 1);
+	WREG32(mmIH_CNTL, ih_cntl);
+
+	pci_set_master(adev->pdev);
+
+	/* enable interrupts */
+	iceland_ih_enable_interrupts(adev);
+
+	return 0;
+}
+
+/**
+ * iceland_ih_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw (VI).
+ */
+static void iceland_ih_irq_disable(struct amdgpu_device *adev)
+{
+	iceland_ih_disable_interrupts(adev);
+
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * iceland_ih_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer (VI).  Also check for
+ * ring buffer overflow and deal with it.
+ * Used by cz_irq_process(VI).
+ * Returns the value of the wptr.
+ */
+static u32 iceland_ih_get_wptr(struct amdgpu_device *adev)
+{
+	u32 wptr, tmp;
+
+	wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
+
+	if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
+		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+		/* When a ring buffer overflow happen start parsing interrupt
+		 * from the last not overwritten vector (wptr + 16). Hopefully
+		 * this should allow us to catchup.
+		 */
+		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
+		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
+		tmp = RREG32(mmIH_RB_CNTL);
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+		WREG32(mmIH_RB_CNTL, tmp);
+	}
+	return (wptr & adev->irq.ih.ptr_mask);
+}
+
+/**
+ * iceland_ih_prescreen_iv - prescreen an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns true if the interrupt vector should be further processed.
+ */
+static bool iceland_ih_prescreen_iv(struct amdgpu_device *adev)
+{
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	u16 pasid;
+
+	switch (le32_to_cpu(adev->irq.ih.ring[ring_index]) & 0xff) {
+	case 146:
+	case 147:
+		pasid = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]) >> 16;
+		if (!pasid || amdgpu_vm_pasid_fault_credit(adev, pasid))
+			return true;
+		break;
+	default:
+		/* Not a VM fault */
+		return true;
+	}
+
+	adev->irq.ih.rptr += 16;
+	return false;
+}
+
+/**
+ * iceland_ih_decode_iv - decode an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Decodes the interrupt vector at the current rptr
+ * position and also advance the position.
+ */
+static void iceland_ih_decode_iv(struct amdgpu_device *adev,
+				 struct amdgpu_iv_entry *entry)
+{
+	/* wptr/rptr are in bytes! */
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	uint32_t dw[4];
+
+	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+
+	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->src_id = dw[0] & 0xff;
+	entry->src_data[0] = dw[1] & 0xfffffff;
+	entry->ring_id = dw[2] & 0xff;
+	entry->vmid = (dw[2] >> 8) & 0xff;
+	entry->pasid = (dw[2] >> 16) & 0xffff;
+
+	/* wptr/rptr are in bytes! */
+	adev->irq.ih.rptr += 16;
+}
+
+/**
+ * iceland_ih_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void iceland_ih_set_rptr(struct amdgpu_device *adev)
+{
+	WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
+}
+
+static int iceland_ih_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret;
+
+	ret = amdgpu_irq_add_domain(adev);
+	if (ret)
+		return ret;
+
+	iceland_ih_set_interrupt_funcs(adev);
+
+	return 0;
+}
+
+static int iceland_ih_sw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int iceland_ih_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_fini(adev);
+	amdgpu_ih_ring_fini(adev);
+	amdgpu_irq_remove_domain(adev);
+
+	return 0;
+}
+
+static int iceland_ih_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = iceland_ih_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int iceland_ih_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	iceland_ih_irq_disable(adev);
+
+	return 0;
+}
+
+static int iceland_ih_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return iceland_ih_hw_fini(adev);
+}
+
+static int iceland_ih_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return iceland_ih_hw_init(adev);
+}
+
+static bool iceland_ih_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
+		return false;
+
+	return true;
+}
+
+static int iceland_ih_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS);
+		if (!REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int iceland_ih_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET,
+						SOFT_RESET_IH, 1);
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int iceland_ih_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int iceland_ih_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs iceland_ih_ip_funcs = {
+	.name = "iceland_ih",
+	.early_init = iceland_ih_early_init,
+	.late_init = NULL,
+	.sw_init = iceland_ih_sw_init,
+	.sw_fini = iceland_ih_sw_fini,
+	.hw_init = iceland_ih_hw_init,
+	.hw_fini = iceland_ih_hw_fini,
+	.suspend = iceland_ih_suspend,
+	.resume = iceland_ih_resume,
+	.is_idle = iceland_ih_is_idle,
+	.wait_for_idle = iceland_ih_wait_for_idle,
+	.soft_reset = iceland_ih_soft_reset,
+	.set_clockgating_state = iceland_ih_set_clockgating_state,
+	.set_powergating_state = iceland_ih_set_powergating_state,
+};
+
+static const struct amdgpu_ih_funcs iceland_ih_funcs = {
+	.get_wptr = iceland_ih_get_wptr,
+	.prescreen_iv = iceland_ih_prescreen_iv,
+	.decode_iv = iceland_ih_decode_iv,
+	.set_rptr = iceland_ih_set_rptr
+};
+
+static void iceland_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih_funcs == NULL)
+		adev->irq.ih_funcs = &iceland_ih_funcs;
+}
+
+const struct amdgpu_ip_block_version iceland_ih_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_IH,
+	.major = 2,
+	.minor = 4,
+	.rev = 0,
+	.funcs = &iceland_ih_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/iceland_ih.h b/sys/dev/pci/drm/amd/amdgpu/iceland_ih.h
new file mode 100644
index 00000000000..3235f427754
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/iceland_ih.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ICELAND_IH_H__
+#define __ICELAND_IH_H__
+
+extern const struct amdgpu_ip_block_version iceland_ih_ip_block;
+
+#endif /* __ICELAND_IH_H__ */
diff --git a/sys/dev/pci/drm/amd/amdgpu/iceland_sdma_pkt_open.h b/sys/dev/pci/drm/amd/amdgpu/iceland_sdma_pkt_open.h
new file mode 100644
index 00000000000..ee6a041cb28
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/iceland_sdma_pkt_open.h
@@ -0,0 +1,2172 @@
+/*
+ * Copyright (C) 2014  Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ICELAND_SDMA_PKT_OPEN_H_
+#define __ICELAND_SDMA_PKT_OPEN_H_
+
+#define SDMA_OP_NOP  0
+#define SDMA_OP_COPY  1
+#define SDMA_OP_WRITE  2
+#define SDMA_OP_INDIRECT  4
+#define SDMA_OP_FENCE  5
+#define SDMA_OP_TRAP  6
+#define SDMA_OP_SEM  7
+#define SDMA_OP_POLL_REGMEM  8
+#define SDMA_OP_COND_EXE  9
+#define SDMA_OP_ATOMIC  10
+#define SDMA_OP_CONST_FILL  11
+#define SDMA_OP_GEN_PTEPDE  12
+#define SDMA_OP_TIMESTAMP  13
+#define SDMA_OP_SRBM_WRITE  14
+#define SDMA_OP_PRE_EXE  15
+#define SDMA_SUBOP_TIMESTAMP_SET  0
+#define SDMA_SUBOP_TIMESTAMP_GET  1
+#define SDMA_SUBOP_TIMESTAMP_GET_GLOBAL  2
+#define SDMA_SUBOP_COPY_LINEAR  0
+#define SDMA_SUBOP_COPY_LINEAR_SUB_WIND  4
+#define SDMA_SUBOP_COPY_TILED  1
+#define SDMA_SUBOP_COPY_TILED_SUB_WIND  5
+#define SDMA_SUBOP_COPY_T2T_SUB_WIND  6
+#define SDMA_SUBOP_COPY_SOA  3
+#define SDMA_SUBOP_WRITE_LINEAR  0
+#define SDMA_SUBOP_WRITE_TILED  1
+
+/*define for op field*/
+#define SDMA_PKT_HEADER_op_offset 0
+#define SDMA_PKT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_HEADER_op_shift  0
+#define SDMA_PKT_HEADER_OP(x) (((x) & SDMA_PKT_HEADER_op_mask) << SDMA_PKT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_HEADER_sub_op_offset 0
+#define SDMA_PKT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_HEADER_sub_op_shift  8
+#define SDMA_PKT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_HEADER_sub_op_mask) << SDMA_PKT_HEADER_sub_op_shift)
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for dst_ha field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_shift  22
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_DST_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_shift  30
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_SRC_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_BROADCAST_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST2_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift)
+
+/*define for dst2_ha field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_shift  14
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST2_HA(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_shift)
+
+/*define for dst1_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift  16
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST1_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift)
+
+/*define for dst1_ha field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_shift  22
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST1_HA(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_shift  30
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_SRC_HA(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST1_ADDR_LO word*/
+/*define for dst1_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_offset 5
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_DST1_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift)
+
+/*define for DST1_ADDR_HI word*/
+/*define for dst1_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_offset 6
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_DST1_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift)
+
+/*define for DST2_ADDR_LO word*/
+/*define for dst2_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_offset 7
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_DST2_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift)
+
+/*define for DST2_ADDR_HI word*/
+/*define for dst2_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_offset 8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_DST2_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift)
+
+/*define for elementsize field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift  29
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_ELEMENTSIZE(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift)
+
+/*define for src_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_offset 6
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_offset 7
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_8 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift)
+
+/*define for DW_9 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift)
+
+/*define for dst_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift)
+
+/*define for DW_10 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_offset 10
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift)
+
+/*define for DW_12 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_RECT_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift)
+
+/*define for dst_ha field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_shift  22
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_DST_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_shift  30
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_SRC_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_op_mask) << SDMA_PKT_COPY_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_offset 3
+#define SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_shift  0
+#define SDMA_PKT_COPY_TILED_DW_3_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_mask) << SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_shift)
+
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_DW_3_height_offset 3
+#define SDMA_PKT_COPY_TILED_DW_3_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_3_height_shift  16
+#define SDMA_PKT_COPY_TILED_DW_3_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_DW_3_height_mask) << SDMA_PKT_COPY_TILED_DW_3_height_shift)
+
+/*define for DW_4 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_DW_4_slice_pitch_offset 4
+#define SDMA_PKT_COPY_TILED_DW_4_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_TILED_DW_4_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_DW_4_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_DW_4_slice_pitch_mask) << SDMA_PKT_COPY_TILED_DW_4_slice_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_element_size_mask) << SDMA_PKT_COPY_TILED_DW_5_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_TILED_DW_5_array_mode_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_DW_5_array_mode_shift  3
+#define SDMA_PKT_COPY_TILED_DW_5_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_array_mode_mask) << SDMA_PKT_COPY_TILED_DW_5_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_TILED_DW_5_mit_mode_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_mit_mode_shift  8
+#define SDMA_PKT_COPY_TILED_DW_5_MIT_MODE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_mit_mode_mask) << SDMA_PKT_COPY_TILED_DW_5_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_TILED_DW_5_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_mask) << SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_TILED_DW_5_bank_w_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_bank_w_shift  15
+#define SDMA_PKT_COPY_TILED_DW_5_BANK_W(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_bank_w_mask) << SDMA_PKT_COPY_TILED_DW_5_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_TILED_DW_5_bank_h_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_bank_h_shift  18
+#define SDMA_PKT_COPY_TILED_DW_5_BANK_H(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_bank_h_mask) << SDMA_PKT_COPY_TILED_DW_5_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_TILED_DW_5_num_bank_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_num_bank_shift  21
+#define SDMA_PKT_COPY_TILED_DW_5_NUM_BANK(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_num_bank_mask) << SDMA_PKT_COPY_TILED_DW_5_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_TILED_DW_5_mat_aspt_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_mat_aspt_shift  24
+#define SDMA_PKT_COPY_TILED_DW_5_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_mat_aspt_mask) << SDMA_PKT_COPY_TILED_DW_5_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_TILED_DW_5_pipe_config_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_DW_5_pipe_config_shift  26
+#define SDMA_PKT_COPY_TILED_DW_5_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_pipe_config_mask) << SDMA_PKT_COPY_TILED_DW_5_pipe_config_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_TILED_DW_6_x_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_x_shift  0
+#define SDMA_PKT_COPY_TILED_DW_6_X(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_x_mask) << SDMA_PKT_COPY_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_TILED_DW_6_y_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_y_shift  16
+#define SDMA_PKT_COPY_TILED_DW_6_Y(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_y_mask) << SDMA_PKT_COPY_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_TILED_DW_7_z_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_z_mask   0x00000FFF
+#define SDMA_PKT_COPY_TILED_DW_7_z_shift  0
+#define SDMA_PKT_COPY_TILED_DW_7_Z(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_z_mask) << SDMA_PKT_COPY_TILED_DW_7_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_DW_7_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_DW_7_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_offset 8
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_offset 9
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_TILED_COUNT_count_offset 11
+#define SDMA_PKT_COPY_TILED_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_COPY_TILED_COUNT_count_shift  0
+#define SDMA_PKT_COPY_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_TILED_COUNT_count_mask) << SDMA_PKT_COPY_TILED_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_L2T_BROADCAST packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift)
+
+/*define for videocopy field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift  26
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_VIDEOCOPY(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift)
+
+/*define for TILED_ADDR_LO_0 word*/
+/*define for tiled_addr0_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_offset 1
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_TILED_ADDR0_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift)
+
+/*define for TILED_ADDR_HI_0 word*/
+/*define for tiled_addr0_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_offset 2
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_TILED_ADDR0_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift)
+
+/*define for TILED_ADDR_LO_1 word*/
+/*define for tiled_addr1_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_offset 3
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_TILED_ADDR1_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift)
+
+/*define for TILED_ADDR_HI_1 word*/
+/*define for tiled_addr1_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_offset 4
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_TILED_ADDR1_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift)
+
+/*define for DW_5 word*/
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_offset 5
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_mask   0x000007FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_shift)
+
+/*define for height field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_offset 5
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_HEIGHT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_shift)
+
+/*define for DW_6 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_offset 6
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_shift)
+
+/*define for DW_7 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_shift  3
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_MIT_MODE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_shift  15
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_BANK_W(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_shift  18
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_BANK_H(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_shift  21
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_NUM_BANK(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_shift  24
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_shift  26
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_shift)
+
+/*define for DW_8 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_X(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_Y(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift)
+
+/*define for DW_9 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_offset 9
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask   0x00000FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_Z(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift)
+
+/*define for DW_10 word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_DST2_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift)
+
+/*define for dst2_ha field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_shift  14
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_DST2_HA(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift  24
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_TILE_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_offset 11
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_offset 12
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_offset 13
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_offset 14
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask) << SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_T2T packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_op_shift  0
+#define SDMA_PKT_COPY_T2T_HEADER_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_op_mask) << SDMA_PKT_COPY_T2T_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_T2T_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_sub_op_mask) << SDMA_PKT_COPY_T2T_HEADER_sub_op_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_x_mask) << SDMA_PKT_COPY_T2T_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_y_mask) << SDMA_PKT_COPY_T2T_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_z_mask) << SDMA_PKT_COPY_T2T_DW_4_src_z_shift)
+
+/*define for src_pitch_in_tile field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_mask   0x00000FFF
+#define SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_shift  16
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_mask) << SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_T2T_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_shift)
+
+/*define for DW_6 word*/
+/*define for src_element_size field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask) << SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift)
+
+/*define for src_array_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_array_mode_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_6_src_array_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_array_mode_mask) << SDMA_PKT_COPY_T2T_DW_6_src_array_mode_shift)
+
+/*define for src_mit_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_shift  8
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_MIT_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_mask) << SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_shift)
+
+/*define for src_tilesplit_size field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_mask) << SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_shift)
+
+/*define for src_bank_w field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_w_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_w_shift  15
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_BANK_W(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_bank_w_mask) << SDMA_PKT_COPY_T2T_DW_6_src_bank_w_shift)
+
+/*define for src_bank_h field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_h_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_h_shift  18
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_BANK_H(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_bank_h_mask) << SDMA_PKT_COPY_T2T_DW_6_src_bank_h_shift)
+
+/*define for src_num_bank field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_num_bank_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_num_bank_shift  21
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_NUM_BANK(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_num_bank_mask) << SDMA_PKT_COPY_T2T_DW_6_src_num_bank_shift)
+
+/*define for src_mat_aspt field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_shift  24
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_mask) << SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_shift)
+
+/*define for src_pipe_config field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_shift  26
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_mask) << SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_offset 7
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_offset 8
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_9_DST_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_x_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_9_DST_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_y_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_y_shift)
+
+/*define for DW_10 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_10_DST_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_z_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_z_shift)
+
+/*define for dst_pitch_in_tile field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_mask   0x00000FFF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_shift  16
+#define SDMA_PKT_COPY_T2T_DW_10_DST_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_shift)
+
+/*define for DW_11 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_offset 11
+#define SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_T2T_DW_11_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_mask) << SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for dst_array_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_12_DST_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_shift)
+
+/*define for dst_mit_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_shift  8
+#define SDMA_PKT_COPY_T2T_DW_12_DST_MIT_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_shift)
+
+/*define for dst_tilesplit_size field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_T2T_DW_12_DST_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_shift)
+
+/*define for dst_bank_w field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_shift  15
+#define SDMA_PKT_COPY_T2T_DW_12_DST_BANK_W(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_shift)
+
+/*define for dst_bank_h field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_shift  18
+#define SDMA_PKT_COPY_T2T_DW_12_DST_BANK_H(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_shift)
+
+/*define for dst_num_bank field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_shift  21
+#define SDMA_PKT_COPY_T2T_DW_12_DST_NUM_BANK(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_shift)
+
+/*define for dst_mat_aspt field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_shift  24
+#define SDMA_PKT_COPY_T2T_DW_12_DST_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_shift)
+
+/*define for dst_pipe_config field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_shift  26
+#define SDMA_PKT_COPY_T2T_DW_12_DST_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_shift)
+
+/*define for DW_13 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_x_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_y_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_y_shift)
+
+/*define for DW_14 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_14_RECT_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_rect_z_mask) << SDMA_PKT_COPY_T2T_DW_14_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift  16
+#define SDMA_PKT_COPY_T2T_DW_14_DST_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_shift  24
+#define SDMA_PKT_COPY_T2T_DW_14_SRC_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_src_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_src_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for tiled_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift)
+
+/*define for tiled_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift)
+
+/*define for DW_4 word*/
+/*define for tiled_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_TILED_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift)
+
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_mask   0x00000FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_shift)
+
+/*define for DW_5 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_shift)
+
+/*define for DW_6 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_shift  3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_MIT_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_shift  15
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_BANK_W(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_shift  18
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_BANK_H(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_shift  21
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_NUM_BANK(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_shift  26
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_offset 7
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_offset 8
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for linear_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift)
+
+/*define for linear_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift)
+
+/*define for DW_10 word*/
+/*define for linear_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift)
+
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift)
+
+/*define for DW_13 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_RECT_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_STRUCT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_shift  0
+#define SDMA_PKT_COPY_STRUCT_HEADER_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_STRUCT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_STRUCT_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_detile_mask) << SDMA_PKT_COPY_STRUCT_HEADER_detile_shift)
+
+/*define for SB_ADDR_LO word*/
+/*define for sb_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_offset 1
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_SB_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift)
+
+/*define for SB_ADDR_HI word*/
+/*define for sb_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_offset 2
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_SB_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift)
+
+/*define for START_INDEX word*/
+/*define for start_index field*/
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_offset 3
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift  0
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_START_INDEX(x) (((x) & SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask) << SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_offset 4
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_shift  0
+#define SDMA_PKT_COPY_STRUCT_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_STRUCT_COUNT_count_mask) << SDMA_PKT_COPY_STRUCT_COUNT_count_shift)
+
+/*define for DW_5 word*/
+/*define for stride field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_mask   0x000007FF
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_shift  0
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRIDE(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_stride_mask) << SDMA_PKT_COPY_STRUCT_DW_5_stride_shift)
+
+/*define for struct_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift  16
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRUCT_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift)
+
+/*define for struct_ha field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_shift  22
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRUCT_HA(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_mask) << SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift  24
+#define SDMA_PKT_COPY_STRUCT_DW_5_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift)
+
+/*define for linear_ha field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_shift  30
+#define SDMA_PKT_COPY_STRUCT_DW_5_LINEAR_HA(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_mask) << SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_offset 6
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_offset 7
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_UNTILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_UNTILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_mask   0x003FFFFF
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_count_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_count_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift  24
+#define SDMA_PKT_WRITE_UNTILED_DW_3_SW(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_offset 4
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask) << SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_TILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_offset 3
+#define SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_mask   0x000007FF
+#define SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_3_PITCH_IN_TILE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_mask) << SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_shift)
+
+/*define for height field*/
+#define SDMA_PKT_WRITE_TILED_DW_3_height_offset 3
+#define SDMA_PKT_WRITE_TILED_DW_3_height_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_3_height_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_3_HEIGHT(x) (((x) & SDMA_PKT_WRITE_TILED_DW_3_height_mask) << SDMA_PKT_WRITE_TILED_DW_3_height_shift)
+
+/*define for DW_4 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_offset 4
+#define SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_4_SLICE_PITCH(x) (((x) & SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_mask) << SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_element_size_mask) << SDMA_PKT_WRITE_TILED_DW_5_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_array_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_array_mode_mask   0x0000000F
+#define SDMA_PKT_WRITE_TILED_DW_5_array_mode_shift  3
+#define SDMA_PKT_WRITE_TILED_DW_5_ARRAY_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_array_mode_mask) << SDMA_PKT_WRITE_TILED_DW_5_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_mit_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_mit_mode_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_mit_mode_shift  8
+#define SDMA_PKT_WRITE_TILED_DW_5_MIT_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_mit_mode_mask) << SDMA_PKT_WRITE_TILED_DW_5_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_shift  11
+#define SDMA_PKT_WRITE_TILED_DW_5_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_mask) << SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_w_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_w_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_w_shift  15
+#define SDMA_PKT_WRITE_TILED_DW_5_BANK_W(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_bank_w_mask) << SDMA_PKT_WRITE_TILED_DW_5_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_h_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_h_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_h_shift  18
+#define SDMA_PKT_WRITE_TILED_DW_5_BANK_H(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_bank_h_mask) << SDMA_PKT_WRITE_TILED_DW_5_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_num_bank_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_num_bank_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_num_bank_shift  21
+#define SDMA_PKT_WRITE_TILED_DW_5_NUM_BANK(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_num_bank_mask) << SDMA_PKT_WRITE_TILED_DW_5_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_shift  24
+#define SDMA_PKT_WRITE_TILED_DW_5_MAT_ASPT(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_mask) << SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_pipe_config_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_pipe_config_mask   0x0000001F
+#define SDMA_PKT_WRITE_TILED_DW_5_pipe_config_shift  26
+#define SDMA_PKT_WRITE_TILED_DW_5_PIPE_CONFIG(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_pipe_config_mask) << SDMA_PKT_WRITE_TILED_DW_5_pipe_config_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_x_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_x_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_6_X(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_x_mask) << SDMA_PKT_WRITE_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_y_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_y_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_6_Y(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_y_mask) << SDMA_PKT_WRITE_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_z_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_z_mask   0x00000FFF
+#define SDMA_PKT_WRITE_TILED_DW_7_z_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_7_Z(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_z_mask) << SDMA_PKT_WRITE_TILED_DW_7_z_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_shift  24
+#define SDMA_PKT_WRITE_TILED_DW_7_SW(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_sw_mask) << SDMA_PKT_WRITE_TILED_DW_7_sw_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_TILED_COUNT_count_offset 8
+#define SDMA_PKT_WRITE_TILED_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_WRITE_TILED_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_TILED_COUNT_count_mask) << SDMA_PKT_WRITE_TILED_COUNT_count_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_offset 9
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_TILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_TILED_DATA0_data0_mask) << SDMA_PKT_WRITE_TILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_INCR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_INCR_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_INCR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_DW0 word*/
+/*define for mask_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_offset 3
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_MASK_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask) << SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift)
+
+/*define for MASK_DW1 word*/
+/*define for mask_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_offset 4
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_MASK_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask) << SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift)
+
+/*define for INIT_DW0 word*/
+/*define for init_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_offset 5
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_INIT_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask) << SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift)
+
+/*define for INIT_DW1 word*/
+/*define for init_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_offset 6
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_INIT_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask) << SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift)
+
+/*define for INCR_DW0 word*/
+/*define for incr_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_offset 7
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_INCR_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask) << SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift)
+
+/*define for INCR_DW1 word*/
+/*define for incr_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_offset 8
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_INCR_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask) << SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_INCR_COUNT_count_offset 9
+#define SDMA_PKT_WRITE_INCR_COUNT_count_mask   0x0007FFFF
+#define SDMA_PKT_WRITE_INCR_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_INCR_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_INCR_COUNT_count_mask) << SDMA_PKT_WRITE_INCR_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_INDIRECT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_INDIRECT_HEADER_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_op_shift  0
+#define SDMA_PKT_INDIRECT_HEADER_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_op_mask) << SDMA_PKT_INDIRECT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_shift  8
+#define SDMA_PKT_INDIRECT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_sub_op_mask) << SDMA_PKT_INDIRECT_HEADER_sub_op_shift)
+
+/*define for vmid field*/
+#define SDMA_PKT_INDIRECT_HEADER_vmid_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_vmid_mask   0x0000000F
+#define SDMA_PKT_INDIRECT_HEADER_vmid_shift  16
+#define SDMA_PKT_INDIRECT_HEADER_VMID(x) (((x) & SDMA_PKT_INDIRECT_HEADER_vmid_mask) << SDMA_PKT_INDIRECT_HEADER_vmid_shift)
+
+/*define for BASE_LO word*/
+/*define for ib_base_31_0 field*/
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_offset 1
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift  0
+#define SDMA_PKT_INDIRECT_BASE_LO_IB_BASE_31_0(x) (((x) & SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask) << SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift)
+
+/*define for BASE_HI word*/
+/*define for ib_base_63_32 field*/
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_offset 2
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift  0
+#define SDMA_PKT_INDIRECT_BASE_HI_IB_BASE_63_32(x) (((x) & SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask) << SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift)
+
+/*define for IB_SIZE word*/
+/*define for ib_size field*/
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_offset 3
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask   0x000FFFFF
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift  0
+#define SDMA_PKT_INDIRECT_IB_SIZE_IB_SIZE(x) (((x) & SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask) << SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift)
+
+/*define for CSA_ADDR_LO word*/
+/*define for csa_addr_31_0 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_offset 4
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_CSA_ADDR_31_0(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift)
+
+/*define for CSA_ADDR_HI word*/
+/*define for csa_addr_63_32 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_offset 5
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_CSA_ADDR_63_32(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SEMAPHORE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_op_shift  0
+#define SDMA_PKT_SEMAPHORE_HEADER_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SEMAPHORE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift)
+
+/*define for write_one field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_shift  29
+#define SDMA_PKT_SEMAPHORE_HEADER_WRITE_ONE(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_write_one_mask) << SDMA_PKT_SEMAPHORE_HEADER_write_one_shift)
+
+/*define for signal field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_shift  30
+#define SDMA_PKT_SEMAPHORE_HEADER_SIGNAL(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_signal_mask) << SDMA_PKT_SEMAPHORE_HEADER_signal_shift)
+
+/*define for mailbox field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift  31
+#define SDMA_PKT_SEMAPHORE_HEADER_MAILBOX(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask) << SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_FENCE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_FENCE_HEADER_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_op_shift  0
+#define SDMA_PKT_FENCE_HEADER_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_op_mask) << SDMA_PKT_FENCE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_FENCE_HEADER_sub_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_sub_op_shift  8
+#define SDMA_PKT_FENCE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_sub_op_mask) << SDMA_PKT_FENCE_HEADER_sub_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_FENCE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_FENCE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_FENCE_DATA_data_offset 3
+#define SDMA_PKT_FENCE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_DATA_data_shift  0
+#define SDMA_PKT_FENCE_DATA_DATA(x) (((x) & SDMA_PKT_FENCE_DATA_data_mask) << SDMA_PKT_FENCE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SRBM_WRITE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_shift  0
+#define SDMA_PKT_SRBM_WRITE_HEADER_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SRBM_WRITE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift)
+
+/*define for byte_en field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask   0x0000000F
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift  28
+#define SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask) << SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift)
+
+/*define for ADDR word*/
+/*define for addr field*/
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_offset 1
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_mask   0x0000FFFF
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_shift  0
+#define SDMA_PKT_SRBM_WRITE_ADDR_ADDR(x) (((x) & SDMA_PKT_SRBM_WRITE_ADDR_addr_mask) << SDMA_PKT_SRBM_WRITE_ADDR_addr_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_SRBM_WRITE_DATA_data_offset 2
+#define SDMA_PKT_SRBM_WRITE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_SRBM_WRITE_DATA_data_shift  0
+#define SDMA_PKT_SRBM_WRITE_DATA_DATA(x) (((x) & SDMA_PKT_SRBM_WRITE_DATA_data_mask) << SDMA_PKT_SRBM_WRITE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PRE_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_op_shift  0
+#define SDMA_PKT_PRE_EXE_HEADER_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_op_mask) << SDMA_PKT_PRE_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_PRE_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_sub_op_mask) << SDMA_PKT_PRE_EXE_HEADER_sub_op_shift)
+
+/*define for dev_sel field*/
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift  16
+#define SDMA_PKT_PRE_EXE_HEADER_DEV_SEL(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask) << SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_offset 1
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COND_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COND_EXE_HEADER_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_op_shift  0
+#define SDMA_PKT_COND_EXE_HEADER_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_op_mask) << SDMA_PKT_COND_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_COND_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_sub_op_mask) << SDMA_PKT_COND_EXE_HEADER_sub_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift)
+
+/*define for REFERENCE word*/
+/*define for reference field*/
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_offset 3
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_shift  0
+#define SDMA_PKT_COND_EXE_REFERENCE_REFERENCE(x) (((x) & SDMA_PKT_COND_EXE_REFERENCE_reference_mask) << SDMA_PKT_COND_EXE_REFERENCE_reference_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_offset 4
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_CONSTANT_FILL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_shift  0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift  8
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift  16
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SW(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift)
+
+/*define for fillsize field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift  30
+#define SDMA_PKT_CONSTANT_FILL_HEADER_FILLSIZE(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for src_data_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_offset 3
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DATA_SRC_DATA_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_offset 4
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_shift  0
+#define SDMA_PKT_CONSTANT_FILL_COUNT_COUNT(x) (((x) & SDMA_PKT_CONSTANT_FILL_COUNT_count_mask) << SDMA_PKT_CONSTANT_FILL_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_REGMEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_REGMEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_REGMEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift)
+
+/*define for hdp_flush field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift  26
+#define SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask) << SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift)
+
+/*define for func field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_mask   0x00000007
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_shift  28
+#define SDMA_PKT_POLL_REGMEM_HEADER_FUNC(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_func_mask) << SDMA_PKT_POLL_REGMEM_HEADER_func_shift)
+
+/*define for mem_poll field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift  31
+#define SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask) << SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift)
+
+/*define for VALUE word*/
+/*define for value field*/
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_offset 3
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_shift  0
+#define SDMA_PKT_POLL_REGMEM_VALUE_VALUE(x) (((x) & SDMA_PKT_POLL_REGMEM_VALUE_value_mask) << SDMA_PKT_POLL_REGMEM_VALUE_value_shift)
+
+/*define for MASK word*/
+/*define for mask field*/
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_offset 4
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_shift  0
+#define SDMA_PKT_POLL_REGMEM_MASK_MASK(x) (((x) & SDMA_PKT_POLL_REGMEM_MASK_mask_mask) << SDMA_PKT_POLL_REGMEM_MASK_mask_shift)
+
+/*define for DW5 word*/
+/*define for interval field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_mask   0x0000FFFF
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_shift  0
+#define SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_interval_mask) << SDMA_PKT_POLL_REGMEM_DW5_interval_shift)
+
+/*define for retry_count field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask   0x00000FFF
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift  16
+#define SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask) << SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_SET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift)
+
+/*define for INIT_DATA_LO word*/
+/*define for init_data_31_0 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_offset 1
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_INIT_DATA_31_0(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift)
+
+/*define for INIT_DATA_HI word*/
+/*define for init_data_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_INIT_DATA_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET_GLOBAL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TRAP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TRAP_HEADER_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_op_shift  0
+#define SDMA_PKT_TRAP_HEADER_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_op_mask) << SDMA_PKT_TRAP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TRAP_HEADER_sub_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_sub_op_shift  8
+#define SDMA_PKT_TRAP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_sub_op_mask) << SDMA_PKT_TRAP_HEADER_sub_op_shift)
+
+/*define for INT_CONTEXT word*/
+/*define for int_context field*/
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_offset 1
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask   0x0FFFFFFF
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift  0
+#define SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(x) (((x) & SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask) << SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift)
+
+
+/*
+** Definitions for SDMA_PKT_NOP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_NOP_HEADER_op_offset 0
+#define SDMA_PKT_NOP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_op_shift  0
+#define SDMA_PKT_NOP_HEADER_OP(x) (((x) & SDMA_PKT_NOP_HEADER_op_mask) << SDMA_PKT_NOP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_NOP_HEADER_sub_op_offset 0
+#define SDMA_PKT_NOP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_sub_op_shift  8
+#define SDMA_PKT_NOP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_NOP_HEADER_sub_op_mask) << SDMA_PKT_NOP_HEADER_sub_op_shift)
+
+/*define for count field*/
+#define SDMA_PKT_NOP_HEADER_count_offset 0
+#define SDMA_PKT_NOP_HEADER_count_mask   0x00003FFF
+#define SDMA_PKT_NOP_HEADER_count_shift  16
+#define SDMA_PKT_NOP_HEADER_COUNT(x) (((x) & SDMA_PKT_NOP_HEADER_count_mask) << SDMA_PKT_NOP_HEADER_count_shift)
+
+#endif /* __ICELAND_SDMA_PKT_OPEN_H_ */
diff --git a/sys/dev/pci/drm/amd/amdgpu/kv_dpm.h b/sys/dev/pci/drm/amd/amdgpu/kv_dpm.h
new file mode 100644
index 00000000000..6df0ed41317
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/kv_dpm.h
@@ -0,0 +1,229 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __KV_DPM_H__
+#define __KV_DPM_H__
+
+#define SMU__NUM_SCLK_DPM_STATE  8
+#define SMU__NUM_MCLK_DPM_LEVELS 4
+#define SMU__NUM_LCLK_DPM_LEVELS 8
+#define SMU__NUM_PCIE_DPM_LEVELS 0 /* ??? */
+#include "smu7_fusion.h"
+#include "ppsmc.h"
+
+#define SUMO_MAX_HARDWARE_POWERLEVELS 5
+
+#define SUMO_MAX_NUMBER_VOLTAGES    4
+
+struct sumo_vid_mapping_entry {
+	u16 vid_2bit;
+	u16 vid_7bit;
+};
+
+struct sumo_vid_mapping_table {
+	u32 num_entries;
+	struct sumo_vid_mapping_entry entries[SUMO_MAX_NUMBER_VOLTAGES];
+};
+
+struct sumo_sclk_voltage_mapping_entry {
+	u32 sclk_frequency;
+	u16 vid_2bit;
+	u16 rsv;
+};
+
+struct sumo_sclk_voltage_mapping_table {
+	u32 num_max_dpm_entries;
+	struct sumo_sclk_voltage_mapping_entry entries[SUMO_MAX_HARDWARE_POWERLEVELS];
+};
+
+#define TRINITY_AT_DFLT            30
+
+#define KV_NUM_NBPSTATES   4
+
+enum kv_pt_config_reg_type {
+	KV_CONFIGREG_MMR = 0,
+	KV_CONFIGREG_SMC_IND,
+	KV_CONFIGREG_DIDT_IND,
+	KV_CONFIGREG_CACHE,
+	KV_CONFIGREG_MAX
+};
+
+struct kv_pt_config_reg {
+	u32 offset;
+	u32 mask;
+	u32 shift;
+	u32 value;
+	enum kv_pt_config_reg_type type;
+};
+
+struct kv_lcac_config_values {
+	u32 block_id;
+	u32 signal_id;
+	u32 t;
+};
+
+struct kv_lcac_config_reg {
+	u32 cntl;
+	u32 block_mask;
+	u32 block_shift;
+	u32 signal_mask;
+	u32 signal_shift;
+	u32 t_mask;
+	u32 t_shift;
+	u32 enable_mask;
+	u32 enable_shift;
+};
+
+struct kv_pl {
+	u32 sclk;
+	u8 vddc_index;
+	u8 ds_divider_index;
+	u8 ss_divider_index;
+	u8 allow_gnb_slow;
+	u8 force_nbp_state;
+	u8 display_wm;
+	u8 vce_wm;
+};
+
+struct kv_ps {
+	struct kv_pl levels[SUMO_MAX_HARDWARE_POWERLEVELS];
+	u32 num_levels;
+	bool need_dfs_bypass;
+	u8 dpm0_pg_nb_ps_lo;
+	u8 dpm0_pg_nb_ps_hi;
+	u8 dpmx_nb_ps_lo;
+	u8 dpmx_nb_ps_hi;
+};
+
+struct kv_sys_info {
+	u32 bootup_uma_clk;
+	u32 bootup_sclk;
+	u32 dentist_vco_freq;
+	u32 nb_dpm_enable;
+	u32 nbp_memory_clock[KV_NUM_NBPSTATES];
+	u32 nbp_n_clock[KV_NUM_NBPSTATES];
+	u16 bootup_nb_voltage_index;
+	u8 htc_tmp_lmt;
+	u8 htc_hyst_lmt;
+	struct sumo_sclk_voltage_mapping_table sclk_voltage_mapping_table;
+	struct sumo_vid_mapping_table vid_mapping_table;
+	u32 uma_channel_number;
+};
+
+struct kv_power_info {
+	u32 at[SUMO_MAX_HARDWARE_POWERLEVELS];
+	u32 voltage_drop_t;
+	struct kv_sys_info sys_info;
+	struct kv_pl boot_pl;
+	bool enable_nb_ps_policy;
+	bool disable_nb_ps3_in_battery;
+	bool video_start;
+	bool battery_state;
+	u32 lowest_valid;
+	u32 highest_valid;
+	u16 high_voltage_t;
+	bool cac_enabled;
+	bool bapm_enable;
+	/* smc offsets */
+	u32 sram_end;
+	u32 dpm_table_start;
+	u32 soft_regs_start;
+	/* dpm SMU tables */
+	u8 graphics_dpm_level_count;
+	u8 uvd_level_count;
+	u8 vce_level_count;
+	u8 acp_level_count;
+	u8 samu_level_count;
+	u16 fps_high_t;
+	SMU7_Fusion_GraphicsLevel graphics_level[SMU__NUM_SCLK_DPM_STATE];
+	SMU7_Fusion_ACPILevel acpi_level;
+	SMU7_Fusion_UvdLevel uvd_level[SMU7_MAX_LEVELS_UVD];
+	SMU7_Fusion_ExtClkLevel vce_level[SMU7_MAX_LEVELS_VCE];
+	SMU7_Fusion_ExtClkLevel acp_level[SMU7_MAX_LEVELS_ACP];
+	SMU7_Fusion_ExtClkLevel samu_level[SMU7_MAX_LEVELS_SAMU];
+	u8 uvd_boot_level;
+	u8 vce_boot_level;
+	u8 acp_boot_level;
+	u8 samu_boot_level;
+	u8 uvd_interval;
+	u8 vce_interval;
+	u8 acp_interval;
+	u8 samu_interval;
+	u8 graphics_boot_level;
+	u8 graphics_interval;
+	u8 graphics_therm_throttle_enable;
+	u8 graphics_voltage_change_enable;
+	u8 graphics_clk_slow_enable;
+	u8 graphics_clk_slow_divider;
+	u8 fps_low_t;
+	u32 low_sclk_interrupt_t;
+	bool uvd_power_gated;
+	bool vce_power_gated;
+	bool acp_power_gated;
+	bool samu_power_gated;
+	bool nb_dpm_enabled;
+	/* flags */
+	bool enable_didt;
+	bool enable_dpm;
+	bool enable_auto_thermal_throttling;
+	bool enable_nb_dpm;
+	/* caps */
+	bool caps_cac;
+	bool caps_power_containment;
+	bool caps_sq_ramping;
+	bool caps_db_ramping;
+	bool caps_td_ramping;
+	bool caps_tcp_ramping;
+	bool caps_sclk_throttle_low_notification;
+	bool caps_fps;
+	bool caps_uvd_dpm;
+	bool caps_uvd_pg;
+	bool caps_vce_pg;
+	bool caps_samu_pg;
+	bool caps_acp_pg;
+	bool caps_stable_p_state;
+	bool caps_enable_dfs_bypass;
+	bool caps_sclk_ds;
+	struct amdgpu_ps current_rps;
+	struct kv_ps current_ps;
+	struct amdgpu_ps requested_rps;
+	struct kv_ps requested_ps;
+};
+
+/* XXX are these ok? */
+#define KV_TEMP_RANGE_MIN (90 * 1000)
+#define KV_TEMP_RANGE_MAX (120 * 1000)
+
+/* kv_smc.c */
+int amdgpu_kv_notify_message_to_smu(struct amdgpu_device *adev, u32 id);
+int amdgpu_kv_dpm_get_enable_mask(struct amdgpu_device *adev, u32 *enable_mask);
+int amdgpu_kv_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+				      PPSMC_Msg msg, u32 parameter);
+int amdgpu_kv_read_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
+			   u32 *value, u32 limit);
+int amdgpu_kv_smc_dpm_enable(struct amdgpu_device *adev, bool enable);
+int amdgpu_kv_smc_bapm_enable(struct amdgpu_device *adev, bool enable);
+int amdgpu_kv_copy_bytes_to_smc(struct amdgpu_device *adev,
+			 u32 smc_start_address,
+			 const u8 *src, u32 byte_count, u32 limit);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/mmhub_v1_0.c b/sys/dev/pci/drm/amd/amdgpu/mmhub_v1_0.c
new file mode 100644
index 00000000000..c963eec58c7
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/mmhub_v1_0.c
@@ -0,0 +1,772 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "mmhub_v1_0.h"
+
+#include "mmhub/mmhub_1_0_offset.h"
+#include "mmhub/mmhub_1_0_sh_mask.h"
+#include "mmhub/mmhub_1_0_default.h"
+#include "athub/athub_1_0_offset.h"
+#include "athub/athub_1_0_sh_mask.h"
+#include "vega10_enum.h"
+
+#include "soc15_common.h"
+
+#define mmDAGB0_CNTL_MISC2_RV 0x008f
+#define mmDAGB0_CNTL_MISC2_RV_BASE_IDX 0
+
+u64 mmhub_v1_0_get_fb_location(struct amdgpu_device *adev)
+{
+	u64 base = RREG32_SOC15(MMHUB, 0, mmMC_VM_FB_LOCATION_BASE);
+
+	base &= MC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
+	base <<= 24;
+
+	return base;
+}
+
+static void mmhub_v1_0_init_gart_pt_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+
+	BUG_ON(adev->gart.table_addr & (~0x0000FFFFFFFFF000ULL));
+	value = adev->gart.table_addr - adev->gmc.vram_start +
+		adev->vm_manager.vram_base_offset;
+	value &= 0x0000FFFFFFFFF000ULL;
+	value |= 0x1; /* valid bit */
+
+	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
+		     lower_32_bits(value));
+
+	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
+		     upper_32_bits(value));
+}
+
+static void mmhub_v1_0_init_gart_aperture_regs(struct amdgpu_device *adev)
+{
+	mmhub_v1_0_init_gart_pt_regs(adev);
+
+	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
+		     (u32)(adev->gmc.gart_start >> 12));
+	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
+		     (u32)(adev->gmc.gart_start >> 44));
+
+	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
+		     (u32)(adev->gmc.gart_end >> 12));
+	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
+		     (u32)(adev->gmc.gart_end >> 44));
+}
+
+static void mmhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev)
+{
+	uint64_t value;
+	uint32_t tmp;
+
+	/* Disable AGP. */
+	WREG32_SOC15(MMHUB, 0, mmMC_VM_AGP_BASE, 0);
+	WREG32_SOC15(MMHUB, 0, mmMC_VM_AGP_TOP, 0);
+	WREG32_SOC15(MMHUB, 0, mmMC_VM_AGP_BOT, 0x00FFFFFF);
+
+	/* Program the system aperture low logical page number. */
+	WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+		     adev->gmc.vram_start >> 18);
+	WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+		     adev->gmc.vram_end >> 18);
+
+	/* Set default page address. */
+	value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start +
+		adev->vm_manager.vram_base_offset;
+	WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
+		     (u32)(value >> 12));
+	WREG32_SOC15(MMHUB, 0, mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
+		     (u32)(value >> 44));
+
+	/* Program "protection fault". */
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
+		     (u32)(adev->dummy_page_addr >> 12));
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
+		     (u32)((u64)adev->dummy_page_addr >> 44));
+
+	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_CNTL2);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL2,
+			    ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_CNTL2, tmp);
+}
+
+static void mmhub_v1_0_init_tlb_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, mmMC_VM_MX_L1_TLB_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+			    ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+			    SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL,
+			    MTYPE, MTYPE_UC);/* XXX for emulation. */
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ATC_EN, 1);
+
+	WREG32_SOC15(MMHUB, 0, mmMC_VM_MX_L1_TLB_CNTL, tmp);
+}
+
+static void mmhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 1);
+	/* XXX for emulation, Refer to closed source code.*/
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
+			    0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL, tmp);
+
+	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL2, tmp);
+
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL3, tmp);
+
+	tmp = mmVM_L2_CNTL4_DEFAULT;
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL4, tmp);
+}
+
+static void mmhub_v1_0_enable_system_domain(struct amdgpu_device *adev)
+{
+	uint32_t tmp;
+
+	tmp = RREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_CNTL, tmp);
+}
+
+static void mmhub_v1_0_disable_identity_aperture(struct amdgpu_device *adev)
+{
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
+		     0XFFFFFFFF);
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
+		     0x0000000F);
+
+	WREG32_SOC15(MMHUB, 0,
+		     mmVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32, 0);
+	WREG32_SOC15(MMHUB, 0,
+		     mmVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32, 0);
+
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32,
+		     0);
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32,
+		     0);
+}
+
+static void mmhub_v1_0_setup_vmid_config(struct amdgpu_device *adev)
+{
+	unsigned num_level, block_size;
+	uint32_t tmp;
+	int i;
+
+	num_level = adev->vm_manager.num_level;
+	block_size = adev->vm_manager.block_size;
+	if (adev->gmc.translate_further)
+		num_level -= 1;
+	else
+		block_size -= 9;
+
+	for (i = 0; i <= 14; i++) {
+		tmp = RREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_CNTL, i);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
+				    num_level);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
+				    1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    PAGE_TABLE_BLOCK_SIZE,
+				    block_size);
+		/* Send no-retry XNACK on fault to suppress VM fault storm. */
+		tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+				    RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_CNTL, i, tmp);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, i*2, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32, i*2, 0);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32, i*2,
+			lower_32_bits(adev->vm_manager.max_pfn - 1));
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32, i*2,
+			upper_32_bits(adev->vm_manager.max_pfn - 1));
+	}
+}
+
+static void mmhub_v1_0_program_invalidation(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	for (i = 0; i < 18; ++i) {
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
+				    2 * i, 0xffffffff);
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
+				    2 * i, 0x1f);
+	}
+}
+
+struct pctl_data {
+	uint32_t index;
+	uint32_t data;
+};
+
+static const struct pctl_data pctl0_data[] = {
+	{0x0, 0x7a640},
+	{0x9, 0x2a64a},
+	{0xd, 0x2a680},
+	{0x11, 0x6a684},
+	{0x19, 0xea68e},
+	{0x29, 0xa69e},
+	{0x2b, 0x0010a6c0},
+	{0x3d, 0x83a707},
+	{0xc2, 0x8a7a4},
+	{0xcc, 0x1a7b8},
+	{0xcf, 0xfa7cc},
+	{0xe0, 0x17a7dd},
+	{0xf9, 0xa7dc},
+	{0xfb, 0x12a7f5},
+	{0x10f, 0xa808},
+	{0x111, 0x12a810},
+	{0x125, 0x7a82c}
+};
+#define PCTL0_DATA_LEN (ARRAY_SIZE(pctl0_data))
+
+#define PCTL0_RENG_EXEC_END_PTR 0x12d
+#define PCTL0_STCTRL_REG_SAVE_RANGE0_BASE  0xa640
+#define PCTL0_STCTRL_REG_SAVE_RANGE0_LIMIT 0xa833
+
+static const struct pctl_data pctl1_data[] = {
+	{0x0, 0x39a000},
+	{0x3b, 0x44a040},
+	{0x81, 0x2a08d},
+	{0x85, 0x6ba094},
+	{0xf2, 0x18a100},
+	{0x10c, 0x4a132},
+	{0x112, 0xca141},
+	{0x120, 0x2fa158},
+	{0x151, 0x17a1d0},
+	{0x16a, 0x1a1e9},
+	{0x16d, 0x13a1ec},
+	{0x182, 0x7a201},
+	{0x18b, 0x3a20a},
+	{0x190, 0x7a580},
+	{0x199, 0xa590},
+	{0x19b, 0x4a594},
+	{0x1a1, 0x1a59c},
+	{0x1a4, 0x7a82c},
+	{0x1ad, 0xfa7cc},
+	{0x1be, 0x17a7dd},
+	{0x1d7, 0x12a810},
+	{0x1eb, 0x4000a7e1},
+	{0x1ec, 0x5000a7f5},
+	{0x1ed, 0x4000a7e2},
+	{0x1ee, 0x5000a7dc},
+	{0x1ef, 0x4000a7e3},
+	{0x1f0, 0x5000a7f6},
+	{0x1f1, 0x5000a7e4}
+};
+#define PCTL1_DATA_LEN (ARRAY_SIZE(pctl1_data))
+
+#define PCTL1_RENG_EXEC_END_PTR 0x1f1
+#define PCTL1_STCTRL_REG_SAVE_RANGE0_BASE  0xa000
+#define PCTL1_STCTRL_REG_SAVE_RANGE0_LIMIT 0xa20d
+#define PCTL1_STCTRL_REG_SAVE_RANGE1_BASE  0xa580
+#define PCTL1_STCTRL_REG_SAVE_RANGE1_LIMIT 0xa59d
+#define PCTL1_STCTRL_REG_SAVE_RANGE2_BASE  0xa82c
+#define PCTL1_STCTRL_REG_SAVE_RANGE2_LIMIT 0xa833
+
+static void mmhub_v1_0_power_gating_write_save_ranges(struct amdgpu_device *adev)
+{
+	uint32_t tmp = 0;
+
+	/* PCTL0_STCTRL_REGISTER_SAVE_RANGE0 */
+	tmp = REG_SET_FIELD(tmp, PCTL0_STCTRL_REGISTER_SAVE_RANGE0,
+			STCTRL_REGISTER_SAVE_BASE,
+			PCTL0_STCTRL_REG_SAVE_RANGE0_BASE);
+	tmp = REG_SET_FIELD(tmp, PCTL0_STCTRL_REGISTER_SAVE_RANGE0,
+			STCTRL_REGISTER_SAVE_LIMIT,
+			PCTL0_STCTRL_REG_SAVE_RANGE0_LIMIT);
+	WREG32_SOC15(MMHUB, 0, mmPCTL0_STCTRL_REGISTER_SAVE_RANGE0, tmp);
+
+	/* PCTL1_STCTRL_REGISTER_SAVE_RANGE0 */
+	tmp = 0;
+	tmp = REG_SET_FIELD(tmp, PCTL1_STCTRL_REGISTER_SAVE_RANGE0,
+			STCTRL_REGISTER_SAVE_BASE,
+			PCTL1_STCTRL_REG_SAVE_RANGE0_BASE);
+	tmp = REG_SET_FIELD(tmp, PCTL1_STCTRL_REGISTER_SAVE_RANGE0,
+			STCTRL_REGISTER_SAVE_LIMIT,
+			PCTL1_STCTRL_REG_SAVE_RANGE0_LIMIT);
+	WREG32_SOC15(MMHUB, 0, mmPCTL1_STCTRL_REGISTER_SAVE_RANGE0, tmp);
+
+	/* PCTL1_STCTRL_REGISTER_SAVE_RANGE1 */
+	tmp = 0;
+	tmp = REG_SET_FIELD(tmp, PCTL1_STCTRL_REGISTER_SAVE_RANGE1,
+			STCTRL_REGISTER_SAVE_BASE,
+			PCTL1_STCTRL_REG_SAVE_RANGE1_BASE);
+	tmp = REG_SET_FIELD(tmp, PCTL1_STCTRL_REGISTER_SAVE_RANGE1,
+			STCTRL_REGISTER_SAVE_LIMIT,
+			PCTL1_STCTRL_REG_SAVE_RANGE1_LIMIT);
+	WREG32_SOC15(MMHUB, 0, mmPCTL1_STCTRL_REGISTER_SAVE_RANGE1, tmp);
+
+	/* PCTL1_STCTRL_REGISTER_SAVE_RANGE2 */
+	tmp = 0;
+	tmp = REG_SET_FIELD(tmp, PCTL1_STCTRL_REGISTER_SAVE_RANGE2,
+			STCTRL_REGISTER_SAVE_BASE,
+			PCTL1_STCTRL_REG_SAVE_RANGE2_BASE);
+	tmp = REG_SET_FIELD(tmp, PCTL1_STCTRL_REGISTER_SAVE_RANGE2,
+			STCTRL_REGISTER_SAVE_LIMIT,
+			PCTL1_STCTRL_REG_SAVE_RANGE2_LIMIT);
+	WREG32_SOC15(MMHUB, 0, mmPCTL1_STCTRL_REGISTER_SAVE_RANGE2, tmp);
+}
+
+void mmhub_v1_0_initialize_power_gating(struct amdgpu_device *adev)
+{
+	uint32_t pctl0_misc = 0;
+	uint32_t pctl0_reng_execute = 0;
+	uint32_t pctl1_misc = 0;
+	uint32_t pctl1_reng_execute = 0;
+	int i = 0;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	/****************** pctl0 **********************/
+	pctl0_misc = RREG32_SOC15(MMHUB, 0, mmPCTL0_MISC);
+	pctl0_reng_execute = RREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_EXECUTE);
+
+	/* Light sleep must be disabled before writing to pctl0 registers */
+	pctl0_misc &= ~PCTL0_MISC__RENG_MEM_LS_ENABLE_MASK;
+	WREG32_SOC15(MMHUB, 0, mmPCTL0_MISC, pctl0_misc);
+
+	/* Write data used to access ram of register engine */
+	for (i = 0; i < PCTL0_DATA_LEN; i++) {
+                WREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_RAM_INDEX,
+			pctl0_data[i].index);
+                WREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_RAM_DATA,
+			pctl0_data[i].data);
+        }
+
+	/* Re-enable light sleep */
+	pctl0_misc |= PCTL0_MISC__RENG_MEM_LS_ENABLE_MASK;
+	WREG32_SOC15(MMHUB, 0, mmPCTL0_MISC, pctl0_misc);
+
+	/****************** pctl1 **********************/
+	pctl1_misc = RREG32_SOC15(MMHUB, 0, mmPCTL1_MISC);
+	pctl1_reng_execute = RREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_EXECUTE);
+
+	/* Light sleep must be disabled before writing to pctl1 registers */
+	pctl1_misc &= ~PCTL1_MISC__RENG_MEM_LS_ENABLE_MASK;
+	WREG32_SOC15(MMHUB, 0, mmPCTL1_MISC, pctl1_misc);
+
+	/* Write data used to access ram of register engine */
+	for (i = 0; i < PCTL1_DATA_LEN; i++) {
+                WREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_RAM_INDEX,
+			pctl1_data[i].index);
+                WREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_RAM_DATA,
+			pctl1_data[i].data);
+        }
+
+	/* Re-enable light sleep */
+	pctl1_misc |= PCTL1_MISC__RENG_MEM_LS_ENABLE_MASK;
+	WREG32_SOC15(MMHUB, 0, mmPCTL1_MISC, pctl1_misc);
+
+	mmhub_v1_0_power_gating_write_save_ranges(adev);
+
+	/* Set the reng execute end ptr for pctl0 */
+	pctl0_reng_execute = REG_SET_FIELD(pctl0_reng_execute,
+					PCTL0_RENG_EXECUTE,
+					RENG_EXECUTE_END_PTR,
+					PCTL0_RENG_EXEC_END_PTR);
+	WREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_EXECUTE, pctl0_reng_execute);
+
+	/* Set the reng execute end ptr for pctl1 */
+	pctl1_reng_execute = REG_SET_FIELD(pctl1_reng_execute,
+					PCTL1_RENG_EXECUTE,
+					RENG_EXECUTE_END_PTR,
+					PCTL1_RENG_EXEC_END_PTR);
+	WREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_EXECUTE, pctl1_reng_execute);
+}
+
+void mmhub_v1_0_update_power_gating(struct amdgpu_device *adev,
+				bool enable)
+{
+	uint32_t pctl0_reng_execute = 0;
+	uint32_t pctl1_reng_execute = 0;
+
+	if (amdgpu_sriov_vf(adev))
+		return;
+
+	pctl0_reng_execute = RREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_EXECUTE);
+	pctl1_reng_execute = RREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_EXECUTE);
+
+	if (enable && adev->pg_flags & AMD_PG_SUPPORT_MMHUB) {
+		pctl0_reng_execute = REG_SET_FIELD(pctl0_reng_execute,
+						PCTL0_RENG_EXECUTE,
+						RENG_EXECUTE_ON_PWR_UP, 1);
+		pctl0_reng_execute = REG_SET_FIELD(pctl0_reng_execute,
+						PCTL0_RENG_EXECUTE,
+						RENG_EXECUTE_ON_REG_UPDATE, 1);
+		WREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_EXECUTE, pctl0_reng_execute);
+
+		pctl1_reng_execute = REG_SET_FIELD(pctl1_reng_execute,
+						PCTL1_RENG_EXECUTE,
+						RENG_EXECUTE_ON_PWR_UP, 1);
+		pctl1_reng_execute = REG_SET_FIELD(pctl1_reng_execute,
+						PCTL1_RENG_EXECUTE,
+						RENG_EXECUTE_ON_REG_UPDATE, 1);
+		WREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_EXECUTE, pctl1_reng_execute);
+
+		if (adev->powerplay.pp_funcs->set_powergating_by_smu)
+			amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_GMC, true);
+
+	} else {
+		pctl0_reng_execute = REG_SET_FIELD(pctl0_reng_execute,
+						PCTL0_RENG_EXECUTE,
+						RENG_EXECUTE_ON_PWR_UP, 0);
+		pctl0_reng_execute = REG_SET_FIELD(pctl0_reng_execute,
+						PCTL0_RENG_EXECUTE,
+						RENG_EXECUTE_ON_REG_UPDATE, 0);
+		WREG32_SOC15(MMHUB, 0, mmPCTL0_RENG_EXECUTE, pctl0_reng_execute);
+
+		pctl1_reng_execute = REG_SET_FIELD(pctl1_reng_execute,
+						PCTL1_RENG_EXECUTE,
+						RENG_EXECUTE_ON_PWR_UP, 0);
+		pctl1_reng_execute = REG_SET_FIELD(pctl1_reng_execute,
+						PCTL1_RENG_EXECUTE,
+						RENG_EXECUTE_ON_REG_UPDATE, 0);
+		WREG32_SOC15(MMHUB, 0, mmPCTL1_RENG_EXECUTE, pctl1_reng_execute);
+	}
+}
+
+int mmhub_v1_0_gart_enable(struct amdgpu_device *adev)
+{
+	if (amdgpu_sriov_vf(adev)) {
+		/*
+		 * MC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are
+		 * VF copy registers so vbios post doesn't program them, for
+		 * SRIOV driver need to program them
+		 */
+		WREG32_SOC15(MMHUB, 0, mmMC_VM_FB_LOCATION_BASE,
+			     adev->gmc.vram_start >> 24);
+		WREG32_SOC15(MMHUB, 0, mmMC_VM_FB_LOCATION_TOP,
+			     adev->gmc.vram_end >> 24);
+	}
+
+	/* GART Enable. */
+	mmhub_v1_0_init_gart_aperture_regs(adev);
+	mmhub_v1_0_init_system_aperture_regs(adev);
+	mmhub_v1_0_init_tlb_regs(adev);
+	mmhub_v1_0_init_cache_regs(adev);
+
+	mmhub_v1_0_enable_system_domain(adev);
+	mmhub_v1_0_disable_identity_aperture(adev);
+	mmhub_v1_0_setup_vmid_config(adev);
+	mmhub_v1_0_program_invalidation(adev);
+
+	return 0;
+}
+
+void mmhub_v1_0_gart_disable(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	u32 i;
+
+	/* Disable all tables */
+	for (i = 0; i < 16; i++)
+		WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT0_CNTL, i, 0);
+
+	/* Setup TLB control */
+	tmp = RREG32_SOC15(MMHUB, 0, mmMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp,
+				MC_VM_MX_L1_TLB_CNTL,
+				ENABLE_ADVANCED_DRIVER_MODEL,
+				0);
+	WREG32_SOC15(MMHUB, 0, mmMC_VM_MX_L1_TLB_CNTL, tmp);
+
+	/* Setup L2 cache */
+	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL, tmp);
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL3, 0);
+}
+
+/**
+ * mmhub_v1_0_set_fault_enable_default - update GART/VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+void mmhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev, bool value)
+{
+	u32 tmp;
+	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp,
+			VM_L2_PROTECTION_FAULT_CNTL,
+			TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
+			value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+			EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	if (!value) {
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_NO_RETRY_FAULT, 1);
+		tmp = REG_SET_FIELD(tmp, VM_L2_PROTECTION_FAULT_CNTL,
+				CRASH_ON_RETRY_FAULT, 1);
+    }
+
+	WREG32_SOC15(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_CNTL, tmp);
+}
+
+void mmhub_v1_0_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB];
+
+	hub->ctx0_ptb_addr_lo32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
+	hub->ctx0_ptb_addr_hi32 =
+		SOC15_REG_OFFSET(MMHUB, 0,
+				 mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
+	hub->vm_inv_eng0_req =
+		SOC15_REG_OFFSET(MMHUB, 0, mmVM_INVALIDATE_ENG0_REQ);
+	hub->vm_inv_eng0_ack =
+		SOC15_REG_OFFSET(MMHUB, 0, mmVM_INVALIDATE_ENG0_ACK);
+	hub->vm_context0_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, mmVM_CONTEXT0_CNTL);
+	hub->vm_l2_pro_fault_status =
+		SOC15_REG_OFFSET(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_STATUS);
+	hub->vm_l2_pro_fault_cntl =
+		SOC15_REG_OFFSET(MMHUB, 0, mmVM_L2_PROTECTION_FAULT_CNTL);
+
+}
+
+static void mmhub_v1_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+							bool enable)
+{
+	uint32_t def, data, def1, data1, def2 = 0, data2 = 0;
+
+	def  = data  = RREG32_SOC15(MMHUB, 0, mmATC_L2_MISC_CG);
+
+	if (adev->asic_type != CHIP_RAVEN) {
+		def1 = data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2);
+		def2 = data2 = RREG32_SOC15(MMHUB, 0, mmDAGB1_CNTL_MISC2);
+	} else
+		def1 = data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2_RV);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG)) {
+		data |= ATC_L2_MISC_CG__ENABLE_MASK;
+
+		data1 &= ~(DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+		           DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+
+		if (adev->asic_type != CHIP_RAVEN)
+			data2 &= ~(DAGB1_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+			           DAGB1_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+			           DAGB1_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+			           DAGB1_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+			           DAGB1_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+			           DAGB1_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+	} else {
+		data &= ~ATC_L2_MISC_CG__ENABLE_MASK;
+
+		data1 |= (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+			  DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+
+		if (adev->asic_type != CHIP_RAVEN)
+			data2 |= (DAGB1_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
+			          DAGB1_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
+			          DAGB1_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
+			          DAGB1_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
+			          DAGB1_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
+			          DAGB1_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
+	}
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, mmATC_L2_MISC_CG, data);
+
+	if (def1 != data1) {
+		if (adev->asic_type != CHIP_RAVEN)
+			WREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2, data1);
+		else
+			WREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2_RV, data1);
+	}
+
+	if (adev->asic_type != CHIP_RAVEN && def2 != data2)
+		WREG32_SOC15(MMHUB, 0, mmDAGB1_CNTL_MISC2, data2);
+}
+
+static void athub_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						   bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG))
+		data |= ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+	else
+		data &= ~ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL, data);
+}
+
+static void mmhub_v1_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(MMHUB, 0, mmATC_L2_MISC_CG);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS))
+		data |= ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+	else
+		data &= ~ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
+
+	if (def != data)
+		WREG32_SOC15(MMHUB, 0, mmATC_L2_MISC_CG, data);
+}
+
+static void athub_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						  bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS) &&
+	    (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
+		data |= ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+	else
+		data &= ~ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+
+	if(def != data)
+		WREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL, data);
+}
+
+int mmhub_v1_0_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state)
+{
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+	case CHIP_RAVEN:
+		mmhub_v1_0_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		athub_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		mmhub_v1_0_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		athub_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+void mmhub_v1_0_get_clockgating(struct amdgpu_device *adev, u32 *flags)
+{
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_MC_MGCG */
+	data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
+	if (data & ATHUB_MISC_CNTL__CG_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_MGCG;
+
+	/* AMD_CG_SUPPORT_MC_LS */
+	data = RREG32_SOC15(MMHUB, 0, mmATC_L2_MISC_CG);
+	if (data & ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_MC_LS;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/mmhub_v1_0.h b/sys/dev/pci/drm/amd/amdgpu/mmhub_v1_0.h
new file mode 100644
index 00000000000..5d38229baf6
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/mmhub_v1_0.h
@@ -0,0 +1,39 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __MMHUB_V1_0_H__
+#define __MMHUB_V1_0_H__
+
+u64 mmhub_v1_0_get_fb_location(struct amdgpu_device *adev);
+int mmhub_v1_0_gart_enable(struct amdgpu_device *adev);
+void mmhub_v1_0_gart_disable(struct amdgpu_device *adev);
+void mmhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev,
+					 bool value);
+void mmhub_v1_0_init(struct amdgpu_device *adev);
+int mmhub_v1_0_set_clockgating(struct amdgpu_device *adev,
+			       enum amd_clockgating_state state);
+void mmhub_v1_0_get_clockgating(struct amdgpu_device *adev, u32 *flags);
+void mmhub_v1_0_initialize_power_gating(struct amdgpu_device *adev);
+void mmhub_v1_0_update_power_gating(struct amdgpu_device *adev,
+                                bool enable);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/mmsch_v1_0.h b/sys/dev/pci/drm/amd/amdgpu/mmsch_v1_0.h
new file mode 100644
index 00000000000..8af0bddf85e
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/mmsch_v1_0.h
@@ -0,0 +1,144 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __MMSCH_V1_0_H__
+#define __MMSCH_V1_0_H__
+
+#define MMSCH_VERSION_MAJOR	1
+#define MMSCH_VERSION_MINOR	0
+#define MMSCH_VERSION	(MMSCH_VERSION_MAJOR << 16 | MMSCH_VERSION_MINOR)
+
+enum mmsch_v1_0_command_type {
+	MMSCH_COMMAND__DIRECT_REG_WRITE = 0,
+	MMSCH_COMMAND__DIRECT_REG_POLLING = 2,
+	MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE = 3,
+	MMSCH_COMMAND__INDIRECT_REG_WRITE = 8,
+	MMSCH_COMMAND__END = 0xf
+};
+
+struct mmsch_v1_0_init_header {
+	uint32_t version;
+	uint32_t header_size;
+	uint32_t vce_init_status;
+	uint32_t uvd_init_status;
+	uint32_t vce_table_offset;
+	uint32_t vce_table_size;
+	uint32_t uvd_table_offset;
+	uint32_t uvd_table_size;
+};
+
+struct mmsch_v1_0_cmd_direct_reg_header {
+	uint32_t reg_offset   : 28;
+	uint32_t command_type : 4;
+};
+
+struct mmsch_v1_0_cmd_indirect_reg_header {
+	uint32_t reg_offset    : 20;
+	uint32_t reg_idx_space : 8;
+	uint32_t command_type  : 4;
+};
+
+struct mmsch_v1_0_cmd_direct_write {
+	struct mmsch_v1_0_cmd_direct_reg_header cmd_header;
+	uint32_t reg_value;
+};
+
+struct mmsch_v1_0_cmd_direct_read_modify_write {
+	struct mmsch_v1_0_cmd_direct_reg_header cmd_header;
+	uint32_t write_data;
+	uint32_t mask_value;
+};
+
+struct mmsch_v1_0_cmd_direct_polling {
+	struct mmsch_v1_0_cmd_direct_reg_header cmd_header;
+	uint32_t mask_value;
+	uint32_t wait_value;
+};
+
+struct mmsch_v1_0_cmd_end {
+	struct mmsch_v1_0_cmd_direct_reg_header cmd_header;
+};
+
+struct mmsch_v1_0_cmd_indirect_write {
+	struct mmsch_v1_0_cmd_indirect_reg_header cmd_header;
+	uint32_t reg_value;
+};
+
+static inline void mmsch_v1_0_insert_direct_wt(struct mmsch_v1_0_cmd_direct_write *direct_wt,
+					       uint32_t *init_table,
+					       uint32_t reg_offset,
+					       uint32_t value)
+{
+	direct_wt->cmd_header.reg_offset = reg_offset;
+	direct_wt->reg_value = value;
+	memcpy((void *)init_table, direct_wt, sizeof(struct mmsch_v1_0_cmd_direct_write));
+}
+
+static inline void mmsch_v1_0_insert_direct_rd_mod_wt(struct mmsch_v1_0_cmd_direct_read_modify_write *direct_rd_mod_wt,
+						      uint32_t *init_table,
+						      uint32_t reg_offset,
+						      uint32_t mask, uint32_t data)
+{
+	direct_rd_mod_wt->cmd_header.reg_offset = reg_offset;
+	direct_rd_mod_wt->mask_value = mask;
+	direct_rd_mod_wt->write_data = data;
+	memcpy((void *)init_table, direct_rd_mod_wt,
+	       sizeof(struct mmsch_v1_0_cmd_direct_read_modify_write));
+}
+
+static inline void mmsch_v1_0_insert_direct_poll(struct mmsch_v1_0_cmd_direct_polling *direct_poll,
+						 uint32_t *init_table,
+						 uint32_t reg_offset,
+						 uint32_t mask, uint32_t wait)
+{
+	direct_poll->cmd_header.reg_offset = reg_offset;
+	direct_poll->mask_value = mask;
+	direct_poll->wait_value = wait;
+	memcpy((void *)init_table, direct_poll, sizeof(struct mmsch_v1_0_cmd_direct_polling));
+}
+
+#define MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(reg, mask, data) { \
+	mmsch_v1_0_insert_direct_rd_mod_wt(&direct_rd_mod_wt, \
+					   init_table, (reg), \
+					   (mask), (data)); \
+	init_table += sizeof(struct mmsch_v1_0_cmd_direct_read_modify_write)/4; \
+	table_size += sizeof(struct mmsch_v1_0_cmd_direct_read_modify_write)/4; \
+}
+
+#define MMSCH_V1_0_INSERT_DIRECT_WT(reg, value) { \
+	mmsch_v1_0_insert_direct_wt(&direct_wt, \
+				    init_table, (reg), \
+				    (value)); \
+	init_table += sizeof(struct mmsch_v1_0_cmd_direct_write)/4; \
+	table_size += sizeof(struct mmsch_v1_0_cmd_direct_write)/4; \
+}
+
+#define MMSCH_V1_0_INSERT_DIRECT_POLL(reg, mask, wait) { \
+	mmsch_v1_0_insert_direct_poll(&direct_poll, \
+				      init_table, (reg), \
+				      (mask), (wait)); \
+	init_table += sizeof(struct mmsch_v1_0_cmd_direct_polling)/4; \
+	table_size += sizeof(struct mmsch_v1_0_cmd_direct_polling)/4; \
+}
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/mxgpu_ai.c b/sys/dev/pci/drm/amd/amdgpu/mxgpu_ai.c
new file mode 100644
index 00000000000..448d19f38a5
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/mxgpu_ai.c
@@ -0,0 +1,377 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "nbio/nbio_6_1_offset.h"
+#include "nbio/nbio_6_1_sh_mask.h"
+#include "gc/gc_9_0_offset.h"
+#include "gc/gc_9_0_sh_mask.h"
+#include "soc15.h"
+#include "vega10_ih.h"
+#include "soc15_common.h"
+#include "mxgpu_ai.h"
+
+static void xgpu_ai_mailbox_send_ack(struct amdgpu_device *adev)
+{
+	WREG8(AI_MAIBOX_CONTROL_RCV_OFFSET_BYTE, 2);
+}
+
+static void xgpu_ai_mailbox_set_valid(struct amdgpu_device *adev, bool val)
+{
+	WREG8(AI_MAIBOX_CONTROL_TRN_OFFSET_BYTE, val ? 1 : 0);
+}
+
+/*
+ * this peek_msg could *only* be called in IRQ routine becuase in IRQ routine
+ * RCV_MSG_VALID filed of BIF_BX_PF0_MAILBOX_CONTROL must already be set to 1
+ * by host.
+ *
+ * if called no in IRQ routine, this peek_msg cannot guaranteed to return the
+ * correct value since it doesn't return the RCV_DW0 under the case that
+ * RCV_MSG_VALID is set by host.
+ */
+static enum idh_event xgpu_ai_mailbox_peek_msg(struct amdgpu_device *adev)
+{
+	return RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
+				mmBIF_BX_PF0_MAILBOX_MSGBUF_RCV_DW0));
+}
+
+
+static int xgpu_ai_mailbox_rcv_msg(struct amdgpu_device *adev,
+				   enum idh_event event)
+{
+	u32 reg;
+
+	reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
+					     mmBIF_BX_PF0_MAILBOX_MSGBUF_RCV_DW0));
+	if (reg != event)
+		return -ENOENT;
+
+	xgpu_ai_mailbox_send_ack(adev);
+
+	return 0;
+}
+
+static uint8_t xgpu_ai_peek_ack(struct amdgpu_device *adev) {
+	return RREG8(AI_MAIBOX_CONTROL_TRN_OFFSET_BYTE) & 2;
+}
+
+static int xgpu_ai_poll_ack(struct amdgpu_device *adev)
+{
+	int timeout  = AI_MAILBOX_POLL_ACK_TIMEDOUT;
+	u8 reg;
+
+	do {
+		reg = RREG8(AI_MAIBOX_CONTROL_TRN_OFFSET_BYTE);
+		if (reg & 2)
+			return 0;
+
+		mdelay(5);
+		timeout -= 5;
+	} while (timeout > 1);
+
+	pr_err("Doesn't get TRN_MSG_ACK from pf in %d msec\n", AI_MAILBOX_POLL_ACK_TIMEDOUT);
+
+	return -ETIME;
+}
+
+static int xgpu_ai_poll_msg(struct amdgpu_device *adev, enum idh_event event)
+{
+	int r, timeout = AI_MAILBOX_POLL_MSG_TIMEDOUT;
+
+	do {
+		r = xgpu_ai_mailbox_rcv_msg(adev, event);
+		if (!r)
+			return 0;
+
+		drm_msleep(10);
+		timeout -= 10;
+	} while (timeout > 1);
+
+	pr_err("Doesn't get msg:%d from pf, error=%d\n", event, r);
+
+	return -ETIME;
+}
+
+static void xgpu_ai_mailbox_trans_msg (struct amdgpu_device *adev,
+	      enum idh_request req, u32 data1, u32 data2, u32 data3) {
+	u32 reg;
+	int r;
+	uint8_t trn;
+
+	/* IMPORTANT:
+	 * clear TRN_MSG_VALID valid to clear host's RCV_MSG_ACK
+	 * and with host's RCV_MSG_ACK cleared hw automatically clear host's RCV_MSG_ACK
+	 * which lead to VF's TRN_MSG_ACK cleared, otherwise below xgpu_ai_poll_ack()
+	 * will return immediatly
+	 */
+	do {
+		xgpu_ai_mailbox_set_valid(adev, false);
+		trn = xgpu_ai_peek_ack(adev);
+		if (trn) {
+			pr_err("trn=%x ACK should not assert! wait again !\n", trn);
+			drm_msleep(1);
+		}
+	} while(trn);
+
+	reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
+					     mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW0));
+	reg = REG_SET_FIELD(reg, BIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW0,
+			    MSGBUF_DATA, req);
+	WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW0),
+		      reg);
+	WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW1),
+				data1);
+	WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW2),
+				data2);
+	WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW3),
+				data3);
+
+	xgpu_ai_mailbox_set_valid(adev, true);
+
+	/* start to poll ack */
+	r = xgpu_ai_poll_ack(adev);
+	if (r)
+		pr_err("Doesn't get ack from pf, continue\n");
+
+	xgpu_ai_mailbox_set_valid(adev, false);
+}
+
+static int xgpu_ai_send_access_requests(struct amdgpu_device *adev,
+					enum idh_request req)
+{
+	int r;
+
+	xgpu_ai_mailbox_trans_msg(adev, req, 0, 0, 0);
+
+	/* start to check msg if request is idh_req_gpu_init_access */
+	if (req == IDH_REQ_GPU_INIT_ACCESS ||
+		req == IDH_REQ_GPU_FINI_ACCESS ||
+		req == IDH_REQ_GPU_RESET_ACCESS) {
+		r = xgpu_ai_poll_msg(adev, IDH_READY_TO_ACCESS_GPU);
+		if (r) {
+			pr_err("Doesn't get READY_TO_ACCESS_GPU from pf, give up\n");
+			return r;
+		}
+		/* Retrieve checksum from mailbox2 */
+		if (req == IDH_REQ_GPU_INIT_ACCESS || req == IDH_REQ_GPU_RESET_ACCESS) {
+			adev->virt.fw_reserve.checksum_key =
+				RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
+					mmBIF_BX_PF0_MAILBOX_MSGBUF_RCV_DW2));
+		}
+	}
+
+	return 0;
+}
+
+static int xgpu_ai_request_reset(struct amdgpu_device *adev)
+{
+	return xgpu_ai_send_access_requests(adev, IDH_REQ_GPU_RESET_ACCESS);
+}
+
+static int xgpu_ai_request_full_gpu_access(struct amdgpu_device *adev,
+					   bool init)
+{
+	enum idh_request req;
+
+	req = init ? IDH_REQ_GPU_INIT_ACCESS : IDH_REQ_GPU_FINI_ACCESS;
+	return xgpu_ai_send_access_requests(adev, req);
+}
+
+static int xgpu_ai_release_full_gpu_access(struct amdgpu_device *adev,
+					   bool init)
+{
+	enum idh_request req;
+	int r = 0;
+
+	req = init ? IDH_REL_GPU_INIT_ACCESS : IDH_REL_GPU_FINI_ACCESS;
+	r = xgpu_ai_send_access_requests(adev, req);
+
+	return r;
+}
+
+static int xgpu_ai_mailbox_ack_irq(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("get ack intr and do nothing.\n");
+	return 0;
+}
+
+static int xgpu_ai_set_mailbox_ack_irq(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 tmp = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_INT_CNTL));
+
+	tmp = REG_SET_FIELD(tmp, BIF_BX_PF0_MAILBOX_INT_CNTL, ACK_INT_EN,
+				(state == AMDGPU_IRQ_STATE_ENABLE) ? 1 : 0);
+	WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_INT_CNTL), tmp);
+
+	return 0;
+}
+
+static void xgpu_ai_mailbox_flr_work(struct work_struct *work)
+{
+	struct amdgpu_virt *virt = container_of(work, struct amdgpu_virt, flr_work);
+	struct amdgpu_device *adev = container_of(virt, struct amdgpu_device, virt);
+	int timeout = AI_MAILBOX_POLL_FLR_TIMEDOUT;
+	int locked;
+
+	/* block amdgpu_gpu_recover till msg FLR COMPLETE received,
+	 * otherwise the mailbox msg will be ruined/reseted by
+	 * the VF FLR.
+	 *
+	 * we can unlock the lock_reset to allow "amdgpu_job_timedout"
+	 * to run gpu_recover() after FLR_NOTIFICATION_CMPL received
+	 * which means host side had finished this VF's FLR.
+	 */
+	locked = mutex_trylock(&adev->lock_reset);
+	if (locked)
+		adev->in_gpu_reset = 1;
+
+	do {
+		if (xgpu_ai_mailbox_peek_msg(adev) == IDH_FLR_NOTIFICATION_CMPL)
+			goto flr_done;
+
+		drm_msleep(10);
+		timeout -= 10;
+	} while (timeout > 1);
+
+flr_done:
+	if (locked) {
+		adev->in_gpu_reset = 0;
+		mutex_unlock(&adev->lock_reset);
+	}
+
+	/* Trigger recovery for world switch failure if no TDR */
+	if (amdgpu_lockup_timeout == 0)
+		amdgpu_device_gpu_recover(adev, NULL, true);
+}
+
+static int xgpu_ai_set_mailbox_rcv_irq(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *src,
+				       unsigned type,
+				       enum amdgpu_interrupt_state state)
+{
+	u32 tmp = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_INT_CNTL));
+
+	tmp = REG_SET_FIELD(tmp, BIF_BX_PF0_MAILBOX_INT_CNTL, VALID_INT_EN,
+			    (state == AMDGPU_IRQ_STATE_ENABLE) ? 1 : 0);
+	WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_INT_CNTL), tmp);
+
+	return 0;
+}
+
+static int xgpu_ai_mailbox_rcv_irq(struct amdgpu_device *adev,
+				   struct amdgpu_irq_src *source,
+				   struct amdgpu_iv_entry *entry)
+{
+	enum idh_event event = xgpu_ai_mailbox_peek_msg(adev);
+
+	switch (event) {
+		case IDH_FLR_NOTIFICATION:
+		if (amdgpu_sriov_runtime(adev))
+			schedule_work(&adev->virt.flr_work);
+		break;
+		/* READY_TO_ACCESS_GPU is fetched by kernel polling, IRQ can ignore
+		 * it byfar since that polling thread will handle it,
+		 * other msg like flr complete is not handled here.
+		 */
+		case IDH_CLR_MSG_BUF:
+		case IDH_FLR_NOTIFICATION_CMPL:
+		case IDH_READY_TO_ACCESS_GPU:
+		default:
+		break;
+	}
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs xgpu_ai_mailbox_ack_irq_funcs = {
+	.set = xgpu_ai_set_mailbox_ack_irq,
+	.process = xgpu_ai_mailbox_ack_irq,
+};
+
+static const struct amdgpu_irq_src_funcs xgpu_ai_mailbox_rcv_irq_funcs = {
+	.set = xgpu_ai_set_mailbox_rcv_irq,
+	.process = xgpu_ai_mailbox_rcv_irq,
+};
+
+void xgpu_ai_mailbox_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->virt.ack_irq.num_types = 1;
+	adev->virt.ack_irq.funcs = &xgpu_ai_mailbox_ack_irq_funcs;
+	adev->virt.rcv_irq.num_types = 1;
+	adev->virt.rcv_irq.funcs = &xgpu_ai_mailbox_rcv_irq_funcs;
+}
+
+int xgpu_ai_mailbox_add_irq_id(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF, 135, &adev->virt.rcv_irq);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF, 138, &adev->virt.ack_irq);
+	if (r) {
+		amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
+		return r;
+	}
+
+	return 0;
+}
+
+int xgpu_ai_mailbox_get_irq(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_irq_get(adev, &adev->virt.rcv_irq, 0);
+	if (r)
+		return r;
+	r = amdgpu_irq_get(adev, &adev->virt.ack_irq, 0);
+	if (r) {
+		amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
+		return r;
+	}
+
+	INIT_WORK(&adev->virt.flr_work, xgpu_ai_mailbox_flr_work);
+
+	return 0;
+}
+
+void xgpu_ai_mailbox_put_irq(struct amdgpu_device *adev)
+{
+	amdgpu_irq_put(adev, &adev->virt.ack_irq, 0);
+	amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
+}
+
+const struct amdgpu_virt_ops xgpu_ai_virt_ops = {
+	.req_full_gpu	= xgpu_ai_request_full_gpu_access,
+	.rel_full_gpu	= xgpu_ai_release_full_gpu_access,
+	.reset_gpu = xgpu_ai_request_reset,
+	.wait_reset = NULL,
+	.trans_msg = xgpu_ai_mailbox_trans_msg,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/mxgpu_ai.h b/sys/dev/pci/drm/amd/amdgpu/mxgpu_ai.h
new file mode 100644
index 00000000000..b4a9ceea334
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/mxgpu_ai.h
@@ -0,0 +1,59 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __MXGPU_AI_H__
+#define __MXGPU_AI_H__
+
+#define AI_MAILBOX_POLL_ACK_TIMEDOUT	500
+#define AI_MAILBOX_POLL_MSG_TIMEDOUT	12000
+#define AI_MAILBOX_POLL_FLR_TIMEDOUT	500
+
+enum idh_request {
+	IDH_REQ_GPU_INIT_ACCESS = 1,
+	IDH_REL_GPU_INIT_ACCESS,
+	IDH_REQ_GPU_FINI_ACCESS,
+	IDH_REL_GPU_FINI_ACCESS,
+	IDH_REQ_GPU_RESET_ACCESS,
+
+	IDH_LOG_VF_ERROR       = 200,
+};
+
+enum idh_event {
+	IDH_CLR_MSG_BUF	= 0,
+	IDH_READY_TO_ACCESS_GPU,
+	IDH_FLR_NOTIFICATION,
+	IDH_FLR_NOTIFICATION_CMPL,
+	IDH_EVENT_MAX
+};
+
+extern const struct amdgpu_virt_ops xgpu_ai_virt_ops;
+
+void xgpu_ai_mailbox_set_irq_funcs(struct amdgpu_device *adev);
+int xgpu_ai_mailbox_add_irq_id(struct amdgpu_device *adev);
+int xgpu_ai_mailbox_get_irq(struct amdgpu_device *adev);
+void xgpu_ai_mailbox_put_irq(struct amdgpu_device *adev);
+
+#define AI_MAIBOX_CONTROL_TRN_OFFSET_BYTE SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_CONTROL) * 4
+#define AI_MAIBOX_CONTROL_RCV_OFFSET_BYTE SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_CONTROL) * 4 + 1
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/mxgpu_vi.c b/sys/dev/pci/drm/amd/amdgpu/mxgpu_vi.c
new file mode 100644
index 00000000000..9fc1c37344c
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/mxgpu_vi.c
@@ -0,0 +1,625 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Xiangliang.Yu@amd.com
+ */
+
+#include "amdgpu.h"
+#include "vi.h"
+#include "bif/bif_5_0_d.h"
+#include "bif/bif_5_0_sh_mask.h"
+#include "vid.h"
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_sh_mask.h"
+#include "gmc_v8_0.h"
+#include "gfx_v8_0.h"
+#include "sdma_v3_0.h"
+#include "tonga_ih.h"
+#include "gmc/gmc_8_2_d.h"
+#include "gmc/gmc_8_2_sh_mask.h"
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+#include "gca/gfx_8_0_sh_mask.h"
+#include "dce/dce_10_0_d.h"
+#include "dce/dce_10_0_sh_mask.h"
+#include "smu/smu_7_1_3_d.h"
+#include "mxgpu_vi.h"
+
+/* VI golden setting */
+static const u32 xgpu_fiji_mgcg_cgcg_init[] = {
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xffffffff,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPF_CLK_CTRL, 0xffffffff, 0x40000100,
+	mmCGTT_DRM_CLK_CTRL0, 0xffffffff, 0x00600100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_WD_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL4, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96e00200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003c,
+	mmPCIE_INDEX, 0xffffffff, 0x0140001c,
+	mmPCIE_DATA, 0x000f0000, 0x00000000,
+	mmSMC_IND_INDEX_4, 0xffffffff, 0xC060000C,
+	mmSMC_IND_DATA_4, 0xc0000fff, 0x00000100,
+	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
+	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104,
+	mmCGTT_DRM_CLK_CTRL0, 0xff000fff, 0x00000100,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
+	mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
+	mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100,
+};
+
+static const u32 xgpu_fiji_golden_settings_a10[] = {
+	mmCB_HW_CONTROL_3, 0x000001ff, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
+	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
+	mmFBC_MISC, 0x1f311fff, 0x12300000,
+	mmHDMI_CONTROL, 0x31000111, 0x00000011,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA0_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA1_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSQ_RANDOM_WAVE_PRI, 0x001fffff, 0x000006fd,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000ff,
+	mmVGT_RESET_DEBUG, 0x00000004, 0x00000004,
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+};
+
+static const u32 xgpu_fiji_golden_common_all[] = {
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG, 0xffffffff, 0x3a00161a,
+	mmPA_SC_RASTER_CONFIG_1, 0xffffffff, 0x0000002e,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22011003,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmSPI_CONFIG_CNTL_1, 0x0000000f, 0x00000009,
+};
+
+static const u32 xgpu_tonga_mgcg_cgcg_init[] = {
+	mmRLC_CGTT_MGCG_OVERRIDE,   0xffffffff, 0xffffffff,
+	mmGRBM_GFX_INDEX,           0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL,        0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL,        0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL,         0xffffffff, 0x00000100,
+	mmCGTT_CPC_CLK_CTRL,        0xffffffff, 0x00000100,
+	mmCGTT_CPF_CLK_CTRL,        0xffffffff, 0x40000100,
+	mmCGTT_DRM_CLK_CTRL0,       0xffffffff, 0x00600100,
+	mmCGTT_GDS_CLK_CTRL,        0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL,         0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL,         0xffffffff, 0x00000100,
+	mmCGTT_WD_CLK_CTRL,         0xffffffff, 0x06000100,
+	mmCGTT_PC_CLK_CTRL,         0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL,        0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL,         0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL,        0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL,         0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL,        0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0,        0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1,        0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2,        0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3,        0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL4,        0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL,        0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL,        0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL,        0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0,       0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL,             0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL,       0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL,       0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL,             0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX,           0xffffffff, 0xe0000000,
+	mmCGTS_CU0_SP0_CTRL_REG,    0xffffffff, 0x00010000,
+	mmCGTS_CU0_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU0_TA_SQC_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU0_SP1_CTRL_REG,    0xffffffff, 0x00060005,
+	mmCGTS_CU0_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU1_SP0_CTRL_REG,    0xffffffff, 0x00010000,
+	mmCGTS_CU1_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU1_TA_CTRL_REG,     0xffffffff, 0x00040007,
+	mmCGTS_CU1_SP1_CTRL_REG,    0xffffffff, 0x00060005,
+	mmCGTS_CU1_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU2_SP0_CTRL_REG,    0xffffffff, 0x00010000,
+	mmCGTS_CU2_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU2_TA_CTRL_REG,     0xffffffff, 0x00040007,
+	mmCGTS_CU2_SP1_CTRL_REG,    0xffffffff, 0x00060005,
+	mmCGTS_CU2_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU3_SP0_CTRL_REG,    0xffffffff, 0x00010000,
+	mmCGTS_CU3_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU3_TA_CTRL_REG,     0xffffffff, 0x00040007,
+	mmCGTS_CU3_SP1_CTRL_REG,    0xffffffff, 0x00060005,
+	mmCGTS_CU3_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU4_SP0_CTRL_REG,    0xffffffff, 0x00010000,
+	mmCGTS_CU4_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU4_TA_SQC_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU4_SP1_CTRL_REG,    0xffffffff, 0x00060005,
+	mmCGTS_CU4_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU5_SP0_CTRL_REG,    0xffffffff, 0x00010000,
+	mmCGTS_CU5_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU5_TA_CTRL_REG,     0xffffffff, 0x00040007,
+	mmCGTS_CU5_SP1_CTRL_REG,    0xffffffff, 0x00060005,
+	mmCGTS_CU5_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU6_SP0_CTRL_REG,    0xffffffff, 0x00010000,
+	mmCGTS_CU6_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU6_TA_CTRL_REG,     0xffffffff, 0x00040007,
+	mmCGTS_CU6_SP1_CTRL_REG,    0xffffffff, 0x00060005,
+	mmCGTS_CU6_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU7_SP0_CTRL_REG,    0xffffffff, 0x00010000,
+	mmCGTS_CU7_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU7_TA_CTRL_REG,     0xffffffff, 0x00040007,
+	mmCGTS_CU7_SP1_CTRL_REG,    0xffffffff, 0x00060005,
+	mmCGTS_CU7_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_SM_CTRL_REG,         0xffffffff, 0x96e00200,
+	mmCP_RB_WPTR_POLL_CNTL,     0xffffffff, 0x00900100,
+	mmRLC_CGCG_CGLS_CTRL,       0xffffffff, 0x0020003c,
+	mmPCIE_INDEX,               0xffffffff, 0x0140001c,
+	mmPCIE_DATA,                0x000f0000, 0x00000000,
+	mmSMC_IND_INDEX_4,          0xffffffff, 0xC060000C,
+	mmSMC_IND_DATA_4,           0xc0000fff, 0x00000100,
+	mmXDMA_CLOCK_GATING_CNTL,   0xffffffff, 0x00000100,
+	mmXDMA_MEM_POWER_CNTL,      0x00000101, 0x00000000,
+	mmMC_MEM_POWER_LS,          0xffffffff, 0x00000104,
+	mmCGTT_DRM_CLK_CTRL0,       0xff000fff, 0x00000100,
+	mmHDP_XDP_CGTT_BLK_CTRL,    0xc0000fff, 0x00000104,
+	mmCP_MEM_SLP_CNTL,          0x00000001, 0x00000001,
+	mmSDMA0_CLK_CTRL,           0xff000ff0, 0x00000100,
+	mmSDMA1_CLK_CTRL,           0xff000ff0, 0x00000100,
+};
+
+static const u32 xgpu_tonga_golden_settings_a11[] = {
+	mmCB_HW_CONTROL, 0xfffdf3cf, 0x00007208,
+	mmCB_HW_CONTROL_3, 0x00000040, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
+	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
+	mmFBC_MISC, 0x1f311fff, 0x12300000,
+	mmGB_GPU_ID, 0x0000000f, 0x00000000,
+	mmHDMI_CONTROL, 0x31000111, 0x00000011,
+	mmMC_ARB_WTM_GRPWT_RD, 0x00000003, 0x00000000,
+	mmMC_HUB_RDREQ_DMIF_LIMIT, 0x0000007f, 0x00000028,
+	mmMC_HUB_WDP_UMC, 0x00007fb6, 0x00000991,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_FIFO_DEPTH_CNTL, 0x000003ff, 0x000000fc,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0000003c,
+	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA0_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA1_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSQ_RANDOM_WAVE_PRI, 0x001fffff, 0x000006fd,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000002fb,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x0000543b,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0xa9210876,
+	mmVGT_RESET_DEBUG, 0x00000004, 0x00000004,
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+};
+
+static const u32 xgpu_tonga_golden_common_all[] = {
+	mmGRBM_GFX_INDEX,               0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG,          0xffffffff, 0x16000012,
+	mmPA_SC_RASTER_CONFIG_1,        0xffffffff, 0x0000002A,
+	mmGB_ADDR_CONFIG,               0xffffffff, 0x22011002,
+	mmSPI_RESOURCE_RESERVE_CU_0,    0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1,    0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
+};
+
+void xgpu_vi_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+		amdgpu_device_program_register_sequence(adev,
+							xgpu_fiji_mgcg_cgcg_init,
+							ARRAY_SIZE(
+								xgpu_fiji_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							xgpu_fiji_golden_settings_a10,
+							ARRAY_SIZE(
+								xgpu_fiji_golden_settings_a10));
+		amdgpu_device_program_register_sequence(adev,
+							xgpu_fiji_golden_common_all,
+							ARRAY_SIZE(
+								xgpu_fiji_golden_common_all));
+		break;
+	case CHIP_TONGA:
+		amdgpu_device_program_register_sequence(adev,
+							xgpu_tonga_mgcg_cgcg_init,
+							ARRAY_SIZE(
+								xgpu_tonga_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							xgpu_tonga_golden_settings_a11,
+							ARRAY_SIZE(
+								xgpu_tonga_golden_settings_a11));
+		amdgpu_device_program_register_sequence(adev,
+							xgpu_tonga_golden_common_all,
+							ARRAY_SIZE(
+								xgpu_tonga_golden_common_all));
+		break;
+	default:
+		BUG_ON("Doesn't support chip type.\n");
+		break;
+	}
+}
+
+/*
+ * Mailbox communication between GPU hypervisor and VFs
+ */
+static void xgpu_vi_mailbox_send_ack(struct amdgpu_device *adev)
+{
+	u32 reg;
+	int timeout = VI_MAILBOX_TIMEDOUT;
+	u32 mask = REG_FIELD_MASK(MAILBOX_CONTROL, RCV_MSG_VALID);
+
+	reg = RREG32_NO_KIQ(mmMAILBOX_CONTROL);
+	reg = REG_SET_FIELD(reg, MAILBOX_CONTROL, RCV_MSG_ACK, 1);
+	WREG32_NO_KIQ(mmMAILBOX_CONTROL, reg);
+
+	/*Wait for RCV_MSG_VALID to be 0*/
+	reg = RREG32_NO_KIQ(mmMAILBOX_CONTROL);
+	while (reg & mask) {
+		if (timeout <= 0) {
+			pr_err("RCV_MSG_VALID is not cleared\n");
+			break;
+		}
+		mdelay(1);
+		timeout -=1;
+
+		reg = RREG32_NO_KIQ(mmMAILBOX_CONTROL);
+	}
+}
+
+static void xgpu_vi_mailbox_set_valid(struct amdgpu_device *adev, bool val)
+{
+	u32 reg;
+
+	reg = RREG32_NO_KIQ(mmMAILBOX_CONTROL);
+	reg = REG_SET_FIELD(reg, MAILBOX_CONTROL,
+			    TRN_MSG_VALID, val ? 1 : 0);
+	WREG32_NO_KIQ(mmMAILBOX_CONTROL, reg);
+}
+
+static void xgpu_vi_mailbox_trans_msg(struct amdgpu_device *adev,
+				      enum idh_request req)
+{
+	u32 reg;
+
+	reg = RREG32_NO_KIQ(mmMAILBOX_MSGBUF_TRN_DW0);
+	reg = REG_SET_FIELD(reg, MAILBOX_MSGBUF_TRN_DW0,
+			    MSGBUF_DATA, req);
+	WREG32_NO_KIQ(mmMAILBOX_MSGBUF_TRN_DW0, reg);
+
+	xgpu_vi_mailbox_set_valid(adev, true);
+}
+
+static int xgpu_vi_mailbox_rcv_msg(struct amdgpu_device *adev,
+				   enum idh_event event)
+{
+	u32 reg;
+	u32 mask = REG_FIELD_MASK(MAILBOX_CONTROL, RCV_MSG_VALID);
+
+	/* workaround: host driver doesn't set VALID for CMPL now */
+	if (event != IDH_FLR_NOTIFICATION_CMPL) {
+		reg = RREG32_NO_KIQ(mmMAILBOX_CONTROL);
+		if (!(reg & mask))
+			return -ENOENT;
+	}
+
+	reg = RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW0);
+	if (reg != event)
+		return -ENOENT;
+
+	/* send ack to PF */
+	xgpu_vi_mailbox_send_ack(adev);
+
+	return 0;
+}
+
+static int xgpu_vi_poll_ack(struct amdgpu_device *adev)
+{
+	int r = 0, timeout = VI_MAILBOX_TIMEDOUT;
+	u32 mask = REG_FIELD_MASK(MAILBOX_CONTROL, TRN_MSG_ACK);
+	u32 reg;
+
+	reg = RREG32_NO_KIQ(mmMAILBOX_CONTROL);
+	while (!(reg & mask)) {
+		if (timeout <= 0) {
+			pr_err("Doesn't get ack from pf.\n");
+			r = -ETIME;
+			break;
+		}
+		mdelay(5);
+		timeout -= 5;
+
+		reg = RREG32_NO_KIQ(mmMAILBOX_CONTROL);
+	}
+
+	return r;
+}
+
+static int xgpu_vi_poll_msg(struct amdgpu_device *adev, enum idh_event event)
+{
+	int r = 0, timeout = VI_MAILBOX_TIMEDOUT;
+
+	r = xgpu_vi_mailbox_rcv_msg(adev, event);
+	while (r) {
+		if (timeout <= 0) {
+			pr_err("Doesn't get ack from pf.\n");
+			r = -ETIME;
+			break;
+		}
+		mdelay(5);
+		timeout -= 5;
+
+		r = xgpu_vi_mailbox_rcv_msg(adev, event);
+	}
+
+	return r;
+}
+
+static int xgpu_vi_send_access_requests(struct amdgpu_device *adev,
+					enum idh_request request)
+{
+	int r;
+
+	xgpu_vi_mailbox_trans_msg(adev, request);
+
+	/* start to poll ack */
+	r = xgpu_vi_poll_ack(adev);
+	if (r)
+		return r;
+
+	xgpu_vi_mailbox_set_valid(adev, false);
+
+	/* start to check msg if request is idh_req_gpu_init_access */
+	if (request == IDH_REQ_GPU_INIT_ACCESS ||
+		request == IDH_REQ_GPU_FINI_ACCESS ||
+		request == IDH_REQ_GPU_RESET_ACCESS) {
+		r = xgpu_vi_poll_msg(adev, IDH_READY_TO_ACCESS_GPU);
+		if (r) {
+			pr_err("Doesn't get ack from pf, give up\n");
+			return r;
+		}
+	}
+
+	return 0;
+}
+
+static int xgpu_vi_request_reset(struct amdgpu_device *adev)
+{
+	return xgpu_vi_send_access_requests(adev, IDH_REQ_GPU_RESET_ACCESS);
+}
+
+static int xgpu_vi_wait_reset_cmpl(struct amdgpu_device *adev)
+{
+	return xgpu_vi_poll_msg(adev, IDH_FLR_NOTIFICATION_CMPL);
+}
+
+static int xgpu_vi_request_full_gpu_access(struct amdgpu_device *adev,
+					   bool init)
+{
+	enum idh_request req;
+
+	req = init ? IDH_REQ_GPU_INIT_ACCESS : IDH_REQ_GPU_FINI_ACCESS;
+	return xgpu_vi_send_access_requests(adev, req);
+}
+
+static int xgpu_vi_release_full_gpu_access(struct amdgpu_device *adev,
+					   bool init)
+{
+	enum idh_request req;
+	int r = 0;
+
+	req = init ? IDH_REL_GPU_INIT_ACCESS : IDH_REL_GPU_FINI_ACCESS;
+	r = xgpu_vi_send_access_requests(adev, req);
+
+	return r;
+}
+
+/* add support mailbox interrupts */
+static int xgpu_vi_mailbox_ack_irq(struct amdgpu_device *adev,
+				   struct amdgpu_irq_src *source,
+				   struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("get ack intr and do nothing.\n");
+	return 0;
+}
+
+static int xgpu_vi_set_mailbox_ack_irq(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *src,
+				       unsigned type,
+				       enum amdgpu_interrupt_state state)
+{
+	u32 tmp = RREG32_NO_KIQ(mmMAILBOX_INT_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, MAILBOX_INT_CNTL, ACK_INT_EN,
+			    (state == AMDGPU_IRQ_STATE_ENABLE) ? 1 : 0);
+	WREG32_NO_KIQ(mmMAILBOX_INT_CNTL, tmp);
+
+	return 0;
+}
+
+static void xgpu_vi_mailbox_flr_work(struct work_struct *work)
+{
+	struct amdgpu_virt *virt = container_of(work, struct amdgpu_virt, flr_work);
+	struct amdgpu_device *adev = container_of(virt, struct amdgpu_device, virt);
+
+	/* wait until RCV_MSG become 3 */
+	if (xgpu_vi_poll_msg(adev, IDH_FLR_NOTIFICATION_CMPL)) {
+		pr_err("failed to recieve FLR_CMPL\n");
+		return;
+	}
+
+	/* Trigger recovery due to world switch failure */
+	amdgpu_device_gpu_recover(adev, NULL, false);
+}
+
+static int xgpu_vi_set_mailbox_rcv_irq(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *src,
+				       unsigned type,
+				       enum amdgpu_interrupt_state state)
+{
+	u32 tmp = RREG32_NO_KIQ(mmMAILBOX_INT_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, MAILBOX_INT_CNTL, VALID_INT_EN,
+			    (state == AMDGPU_IRQ_STATE_ENABLE) ? 1 : 0);
+	WREG32_NO_KIQ(mmMAILBOX_INT_CNTL, tmp);
+
+	return 0;
+}
+
+static int xgpu_vi_mailbox_rcv_irq(struct amdgpu_device *adev,
+				   struct amdgpu_irq_src *source,
+				   struct amdgpu_iv_entry *entry)
+{
+	int r;
+
+	/* trigger gpu-reset by hypervisor only if TDR disbaled */
+	if (!amdgpu_gpu_recovery) {
+		/* see what event we get */
+		r = xgpu_vi_mailbox_rcv_msg(adev, IDH_FLR_NOTIFICATION);
+
+		/* only handle FLR_NOTIFY now */
+		if (!r)
+			schedule_work(&adev->virt.flr_work);
+	}
+
+	return 0;
+}
+
+static const struct amdgpu_irq_src_funcs xgpu_vi_mailbox_ack_irq_funcs = {
+	.set = xgpu_vi_set_mailbox_ack_irq,
+	.process = xgpu_vi_mailbox_ack_irq,
+};
+
+static const struct amdgpu_irq_src_funcs xgpu_vi_mailbox_rcv_irq_funcs = {
+	.set = xgpu_vi_set_mailbox_rcv_irq,
+	.process = xgpu_vi_mailbox_rcv_irq,
+};
+
+void xgpu_vi_mailbox_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->virt.ack_irq.num_types = 1;
+	adev->virt.ack_irq.funcs = &xgpu_vi_mailbox_ack_irq_funcs;
+	adev->virt.rcv_irq.num_types = 1;
+	adev->virt.rcv_irq.funcs = &xgpu_vi_mailbox_rcv_irq_funcs;
+}
+
+int xgpu_vi_mailbox_add_irq_id(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 135, &adev->virt.rcv_irq);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 138, &adev->virt.ack_irq);
+	if (r) {
+		amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
+		return r;
+	}
+
+	return 0;
+}
+
+int xgpu_vi_mailbox_get_irq(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_irq_get(adev, &adev->virt.rcv_irq, 0);
+	if (r)
+		return r;
+	r = amdgpu_irq_get(adev, &adev->virt.ack_irq, 0);
+	if (r) {
+		amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
+		return r;
+	}
+
+	INIT_WORK(&adev->virt.flr_work, xgpu_vi_mailbox_flr_work);
+
+	return 0;
+}
+
+void xgpu_vi_mailbox_put_irq(struct amdgpu_device *adev)
+{
+	amdgpu_irq_put(adev, &adev->virt.ack_irq, 0);
+	amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
+}
+
+const struct amdgpu_virt_ops xgpu_vi_virt_ops = {
+	.req_full_gpu		= xgpu_vi_request_full_gpu_access,
+	.rel_full_gpu		= xgpu_vi_release_full_gpu_access,
+	.reset_gpu		= xgpu_vi_request_reset,
+	.wait_reset             = xgpu_vi_wait_reset_cmpl,
+	.trans_msg		= NULL, /* Does not need to trans VF errors to host. */
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/mxgpu_vi.h b/sys/dev/pci/drm/amd/amdgpu/mxgpu_vi.h
new file mode 100644
index 00000000000..f13dc6cc158
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/mxgpu_vi.h
@@ -0,0 +1,57 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#ifndef __MXGPU_VI_H__
+#define __MXGPU_VI_H__
+
+#define VI_MAILBOX_TIMEDOUT	12000
+#define VI_MAILBOX_RESET_TIME	12
+
+/* VI mailbox messages request */
+enum idh_request {
+	IDH_REQ_GPU_INIT_ACCESS	= 1,
+	IDH_REL_GPU_INIT_ACCESS,
+	IDH_REQ_GPU_FINI_ACCESS,
+	IDH_REL_GPU_FINI_ACCESS,
+	IDH_REQ_GPU_RESET_ACCESS,
+
+	IDH_LOG_VF_ERROR       = 200,
+};
+
+/* VI mailbox messages data */
+enum idh_event {
+	IDH_CLR_MSG_BUF = 0,
+	IDH_READY_TO_ACCESS_GPU,
+	IDH_FLR_NOTIFICATION,
+	IDH_FLR_NOTIFICATION_CMPL,
+	IDH_EVENT_MAX
+};
+
+extern const struct amdgpu_virt_ops xgpu_vi_virt_ops;
+
+void xgpu_vi_init_golden_registers(struct amdgpu_device *adev);
+void xgpu_vi_mailbox_set_irq_funcs(struct amdgpu_device *adev);
+int xgpu_vi_mailbox_add_irq_id(struct amdgpu_device *adev);
+int xgpu_vi_mailbox_get_irq(struct amdgpu_device *adev);
+void xgpu_vi_mailbox_put_irq(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/nbio_v6_1.c b/sys/dev/pci/drm/amd/amdgpu/nbio_v6_1.c
new file mode 100644
index 00000000000..6f9c54978cc
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/nbio_v6_1.c
@@ -0,0 +1,295 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "nbio_v6_1.h"
+
+#include "nbio/nbio_6_1_default.h"
+#include "nbio/nbio_6_1_offset.h"
+#include "nbio/nbio_6_1_sh_mask.h"
+#include "vega10_enum.h"
+
+#define smnCPM_CONTROL                                                                                  0x11180460
+#define smnPCIE_CNTL2                                                                                   0x11180070
+#define smnPCIE_CONFIG_CNTL                                                                             0x11180044
+
+static u32 nbio_v6_1_get_rev_id(struct amdgpu_device *adev)
+{
+        u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
+
+	tmp &= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK;
+	tmp >>= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT;
+
+	return tmp;
+}
+
+static void nbio_v6_1_mc_access_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		WREG32_SOC15(NBIO, 0, mmBIF_FB_EN,
+			     BIF_FB_EN__FB_READ_EN_MASK |
+			     BIF_FB_EN__FB_WRITE_EN_MASK);
+	else
+		WREG32_SOC15(NBIO, 0, mmBIF_FB_EN, 0);
+}
+
+static void nbio_v6_1_hdp_flush(struct amdgpu_device *adev,
+				struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg)
+		WREG32_SOC15_NO_KIQ(NBIO, 0,
+				    mmBIF_BX_PF0_HDP_MEM_COHERENCY_FLUSH_CNTL,
+				    0);
+	else
+		amdgpu_ring_emit_wreg(ring, SOC15_REG_OFFSET(
+			NBIO, 0, mmBIF_BX_PF0_HDP_MEM_COHERENCY_FLUSH_CNTL), 0);
+}
+
+static u32 nbio_v6_1_get_memsize(struct amdgpu_device *adev)
+{
+	return RREG32_SOC15(NBIO, 0, mmRCC_PF_0_0_RCC_CONFIG_MEMSIZE);
+}
+
+static void nbio_v6_1_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
+				  bool use_doorbell, int doorbell_index)
+{
+	u32 reg = instance == 0 ? SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE) :
+			SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA1_DOORBELL_RANGE);
+
+	u32 doorbell_range = RREG32(reg);
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, OFFSET, doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 2);
+	} else
+		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 0);
+
+	WREG32(reg, doorbell_range);
+
+}
+
+static void nbio_v6_1_enable_doorbell_aperture(struct amdgpu_device *adev,
+					       bool enable)
+{
+	WREG32_FIELD15(NBIO, 0, RCC_PF_0_0_RCC_DOORBELL_APER_EN, BIF_DOORBELL_APER_EN, enable ? 1 : 0);
+}
+
+static void nbio_v6_1_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
+							bool enable)
+{
+	u32 tmp = 0;
+
+	if (enable) {
+		tmp = REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL, DOORBELL_SELFRING_GPA_APER_EN, 1) |
+		      REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL, DOORBELL_SELFRING_GPA_APER_MODE, 1) |
+		      REG_SET_FIELD(tmp, BIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL, DOORBELL_SELFRING_GPA_APER_SIZE, 0);
+
+		WREG32_SOC15(NBIO, 0, mmBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_LOW,
+			     lower_32_bits(adev->doorbell.base));
+		WREG32_SOC15(NBIO, 0, mmBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_BASE_HIGH,
+			     upper_32_bits(adev->doorbell.base));
+	}
+
+	WREG32_SOC15(NBIO, 0, mmBIF_BX_PF0_DOORBELL_SELFRING_GPA_APER_CNTL, tmp);
+}
+
+
+static void nbio_v6_1_ih_doorbell_range(struct amdgpu_device *adev,
+					bool use_doorbell, int doorbell_index)
+{
+	u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0 , mmBIF_IH_DOORBELL_RANGE);
+
+	if (use_doorbell) {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, OFFSET, doorbell_index);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 2);
+	} else
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 0);
+
+	WREG32_SOC15(NBIO, 0, mmBIF_IH_DOORBELL_RANGE, ih_doorbell_range);
+}
+
+static void nbio_v6_1_ih_control(struct amdgpu_device *adev)
+{
+	u32 interrupt_cntl;
+
+	/* setup interrupt control */
+	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
+	interrupt_cntl = RREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL);
+	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_DUMMY_RD_OVERRIDE, 0);
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_REQ_NONSNOOP_EN, 0);
+	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL, interrupt_cntl);
+}
+
+static void nbio_v6_1_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_PCIE(smnCPM_CONTROL);
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG)) {
+		data |= (CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_PERM_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
+			 CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
+			 CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
+	} else {
+		data &= ~(CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_PERM_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
+			  CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
+			  CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
+	}
+
+	if (def != data)
+		WREG32_PCIE(smnCPM_CONTROL, data);
+}
+
+static void nbio_v6_1_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_PCIE(smnPCIE_CNTL2);
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS)) {
+		data |= (PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+			 PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+			 PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+	} else {
+		data &= ~(PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+			  PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+			  PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+	}
+
+	if (def != data)
+		WREG32_PCIE(smnPCIE_CNTL2, data);
+}
+
+static void nbio_v6_1_get_clockgating_state(struct amdgpu_device *adev,
+					    u32 *flags)
+{
+	int data;
+
+	/* AMD_CG_SUPPORT_BIF_MGCG */
+	data = RREG32_PCIE(smnCPM_CONTROL);
+	if (data & CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_MGCG;
+
+	/* AMD_CG_SUPPORT_BIF_LS */
+	data = RREG32_PCIE(smnPCIE_CNTL2);
+	if (data & PCIE_CNTL2__SLV_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_LS;
+}
+
+static u32 nbio_v6_1_get_hdp_flush_req_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_GPU_HDP_FLUSH_REQ);
+}
+
+static u32 nbio_v6_1_get_hdp_flush_done_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_GPU_HDP_FLUSH_DONE);
+}
+
+static u32 nbio_v6_1_get_pcie_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_INDEX2);
+}
+
+static u32 nbio_v6_1_get_pcie_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_DATA2);
+}
+
+static const struct nbio_hdp_flush_reg nbio_v6_1_hdp_flush_reg = {
+	.ref_and_mask_cp0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP0_MASK,
+	.ref_and_mask_cp1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP1_MASK,
+	.ref_and_mask_cp2 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP2_MASK,
+	.ref_and_mask_cp3 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP3_MASK,
+	.ref_and_mask_cp4 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP4_MASK,
+	.ref_and_mask_cp5 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP5_MASK,
+	.ref_and_mask_cp6 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP6_MASK,
+	.ref_and_mask_cp7 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP7_MASK,
+	.ref_and_mask_cp8 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP8_MASK,
+	.ref_and_mask_cp9 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__CP9_MASK,
+	.ref_and_mask_sdma0 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA0_MASK,
+	.ref_and_mask_sdma1 = BIF_BX_PF0_GPU_HDP_FLUSH_DONE__SDMA1_MASK
+};
+
+static void nbio_v6_1_detect_hw_virt(struct amdgpu_device *adev)
+{
+	uint32_t reg;
+
+	reg = RREG32_SOC15(NBIO, 0, mmRCC_PF_0_0_RCC_IOV_FUNC_IDENTIFIER);
+	if (reg & 1)
+		adev->virt.caps |= AMDGPU_SRIOV_CAPS_IS_VF;
+
+	if (reg & 0x80000000)
+		adev->virt.caps |= AMDGPU_SRIOV_CAPS_ENABLE_IOV;
+
+	if (!reg) {
+		if (is_virtual_machine())	/* passthrough mode exclus sriov mod */
+			adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
+	}
+}
+
+static void nbio_v6_1_init_registers(struct amdgpu_device *adev)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_PCIE(smnPCIE_CONFIG_CNTL);
+	data = REG_SET_FIELD(data, PCIE_CONFIG_CNTL, CI_SWUS_MAX_READ_REQUEST_SIZE_MODE, 1);
+	data = REG_SET_FIELD(data, PCIE_CONFIG_CNTL, CI_SWUS_MAX_READ_REQUEST_SIZE_PRIV, 1);
+
+	if (def != data)
+		WREG32_PCIE(smnPCIE_CONFIG_CNTL, data);
+}
+
+const struct amdgpu_nbio_funcs nbio_v6_1_funcs = {
+	.hdp_flush_reg = &nbio_v6_1_hdp_flush_reg,
+	.get_hdp_flush_req_offset = nbio_v6_1_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbio_v6_1_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbio_v6_1_get_pcie_index_offset,
+	.get_pcie_data_offset = nbio_v6_1_get_pcie_data_offset,
+	.get_rev_id = nbio_v6_1_get_rev_id,
+	.mc_access_enable = nbio_v6_1_mc_access_enable,
+	.hdp_flush = nbio_v6_1_hdp_flush,
+	.get_memsize = nbio_v6_1_get_memsize,
+	.sdma_doorbell_range = nbio_v6_1_sdma_doorbell_range,
+	.enable_doorbell_aperture = nbio_v6_1_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbio_v6_1_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbio_v6_1_ih_doorbell_range,
+	.update_medium_grain_clock_gating = nbio_v6_1_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbio_v6_1_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbio_v6_1_get_clockgating_state,
+	.ih_control = nbio_v6_1_ih_control,
+	.init_registers = nbio_v6_1_init_registers,
+	.detect_hw_virt = nbio_v6_1_detect_hw_virt,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/nbio_v6_1.h b/sys/dev/pci/drm/amd/amdgpu/nbio_v6_1.h
new file mode 100644
index 00000000000..0743a6f016f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/nbio_v6_1.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NBIO_V6_1_H__
+#define __NBIO_V6_1_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_nbio_funcs nbio_v6_1_funcs;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/nbio_v7_0.c b/sys/dev/pci/drm/amd/amdgpu/nbio_v7_0.c
new file mode 100644
index 00000000000..365517c0121
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/nbio_v7_0.c
@@ -0,0 +1,304 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "nbio_v7_0.h"
+
+#include "nbio/nbio_7_0_default.h"
+#include "nbio/nbio_7_0_offset.h"
+#include "nbio/nbio_7_0_sh_mask.h"
+#include "vega10_enum.h"
+
+#define smnNBIF_MGCG_CTRL_LCLK	0x1013a05c
+
+#define smnCPM_CONTROL                                                                                  0x11180460
+#define smnPCIE_CNTL2                                                                                   0x11180070
+
+/* vega20 */
+#define mmRCC_DEV0_EPF0_STRAP0_VG20                                                                         0x0011
+#define mmRCC_DEV0_EPF0_STRAP0_VG20_BASE_IDX                                                                2
+
+static u32 nbio_v7_0_get_rev_id(struct amdgpu_device *adev)
+{
+        u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
+
+	if (adev->asic_type == CHIP_VEGA20)
+		tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0_VG20);
+	else
+		tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
+
+	tmp &= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK;
+	tmp >>= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT;
+
+	return tmp;
+}
+
+static void nbio_v7_0_mc_access_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		WREG32_SOC15(NBIO, 0, mmBIF_FB_EN,
+			BIF_FB_EN__FB_READ_EN_MASK | BIF_FB_EN__FB_WRITE_EN_MASK);
+	else
+		WREG32_SOC15(NBIO, 0, mmBIF_FB_EN, 0);
+}
+
+static void nbio_v7_0_hdp_flush(struct amdgpu_device *adev,
+				struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg)
+		WREG32_SOC15_NO_KIQ(NBIO, 0, mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0);
+	else
+		amdgpu_ring_emit_wreg(ring, SOC15_REG_OFFSET(
+			NBIO, 0, mmHDP_MEM_COHERENCY_FLUSH_CNTL), 0);
+}
+
+static u32 nbio_v7_0_get_memsize(struct amdgpu_device *adev)
+{
+	return RREG32_SOC15(NBIO, 0, mmRCC_CONFIG_MEMSIZE);
+}
+
+static void nbio_v7_0_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
+					  bool use_doorbell, int doorbell_index)
+{
+	u32 reg = instance == 0 ? SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE) :
+			SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA1_DOORBELL_RANGE);
+
+	u32 doorbell_range = RREG32(reg);
+	u32 range = 2;
+
+	if (adev->asic_type == CHIP_VEGA20)
+		range = 8;
+
+	if (use_doorbell) {
+		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, OFFSET, doorbell_index);
+		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, range);
+	} else
+		doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 0);
+
+	WREG32(reg, doorbell_range);
+}
+
+static void nbio_v7_0_enable_doorbell_aperture(struct amdgpu_device *adev,
+					       bool enable)
+{
+	WREG32_FIELD15(NBIO, 0, RCC_DOORBELL_APER_EN, BIF_DOORBELL_APER_EN, enable ? 1 : 0);
+}
+
+static void nbio_v7_0_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
+							bool enable)
+{
+
+}
+
+static void nbio_v7_0_ih_doorbell_range(struct amdgpu_device *adev,
+					bool use_doorbell, int doorbell_index)
+{
+	u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0 , mmBIF_IH_DOORBELL_RANGE);
+
+	if (use_doorbell) {
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, OFFSET, doorbell_index);
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 2);
+	} else
+		ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range, BIF_IH_DOORBELL_RANGE, SIZE, 0);
+
+	WREG32_SOC15(NBIO, 0, mmBIF_IH_DOORBELL_RANGE, ih_doorbell_range);
+}
+
+static uint32_t nbio_7_0_read_syshub_ind_mmr(struct amdgpu_device *adev, uint32_t offset)
+{
+	uint32_t data;
+
+	WREG32_SOC15(NBIO, 0, mmSYSHUB_INDEX, offset);
+	data = RREG32_SOC15(NBIO, 0, mmSYSHUB_DATA);
+
+	return data;
+}
+
+static void nbio_7_0_write_syshub_ind_mmr(struct amdgpu_device *adev, uint32_t offset,
+				       uint32_t data)
+{
+	WREG32_SOC15(NBIO, 0, mmSYSHUB_INDEX, offset);
+	WREG32_SOC15(NBIO, 0, mmSYSHUB_DATA, data);
+}
+
+static void nbio_v7_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	if (adev->asic_type == CHIP_VEGA20)
+		return;
+
+	/* NBIF_MGCG_CTRL_LCLK */
+	def = data = RREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG))
+		data |= NBIF_MGCG_CTRL_LCLK__NBIF_MGCG_EN_LCLK_MASK;
+	else
+		data &= ~NBIF_MGCG_CTRL_LCLK__NBIF_MGCG_EN_LCLK_MASK;
+
+	if (def != data)
+		WREG32_PCIE(smnNBIF_MGCG_CTRL_LCLK, data);
+
+	/* SYSHUB_MGCG_CTRL_SOCCLK */
+	def = data = nbio_7_0_read_syshub_ind_mmr(adev, ixSYSHUB_MMREG_IND_SYSHUB_MGCG_CTRL_SOCCLK);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG))
+		data |= SYSHUB_MMREG_DIRECT_SYSHUB_MGCG_CTRL_SOCCLK__SYSHUB_MGCG_EN_SOCCLK_MASK;
+	else
+		data &= ~SYSHUB_MMREG_DIRECT_SYSHUB_MGCG_CTRL_SOCCLK__SYSHUB_MGCG_EN_SOCCLK_MASK;
+
+	if (def != data)
+		nbio_7_0_write_syshub_ind_mmr(adev, ixSYSHUB_MMREG_IND_SYSHUB_MGCG_CTRL_SOCCLK, data);
+
+	/* SYSHUB_MGCG_CTRL_SHUBCLK */
+	def = data = nbio_7_0_read_syshub_ind_mmr(adev, ixSYSHUB_MMREG_IND_SYSHUB_MGCG_CTRL_SHUBCLK);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG))
+		data |= SYSHUB_MMREG_DIRECT_SYSHUB_MGCG_CTRL_SHUBCLK__SYSHUB_MGCG_EN_SHUBCLK_MASK;
+	else
+		data &= ~SYSHUB_MMREG_DIRECT_SYSHUB_MGCG_CTRL_SHUBCLK__SYSHUB_MGCG_EN_SHUBCLK_MASK;
+
+	if (def != data)
+		nbio_7_0_write_syshub_ind_mmr(adev, ixSYSHUB_MMREG_IND_SYSHUB_MGCG_CTRL_SHUBCLK, data);
+}
+
+static void nbio_v7_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32_PCIE(smnPCIE_CNTL2);
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS)) {
+		data |= (PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+			 PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+			 PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+	} else {
+		data &= ~(PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+			  PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+			  PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+	}
+
+	if (def != data)
+		WREG32_PCIE(smnPCIE_CNTL2, data);
+}
+
+static void nbio_v7_0_get_clockgating_state(struct amdgpu_device *adev,
+					    u32 *flags)
+{
+	int data;
+
+	/* AMD_CG_SUPPORT_BIF_MGCG */
+	data = RREG32_PCIE(smnCPM_CONTROL);
+	if (data & CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_MGCG;
+
+	/* AMD_CG_SUPPORT_BIF_LS */
+	data = RREG32_PCIE(smnPCIE_CNTL2);
+	if (data & PCIE_CNTL2__SLV_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_LS;
+}
+
+static void nbio_v7_0_ih_control(struct amdgpu_device *adev)
+{
+	u32 interrupt_cntl;
+
+	/* setup interrupt control */
+	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
+	interrupt_cntl = RREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL);
+	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_DUMMY_RD_OVERRIDE, 0);
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_REQ_NONSNOOP_EN, 0);
+	WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL, interrupt_cntl);
+}
+
+static u32 nbio_v7_0_get_hdp_flush_req_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmGPU_HDP_FLUSH_REQ);
+}
+
+static u32 nbio_v7_0_get_hdp_flush_done_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmGPU_HDP_FLUSH_DONE);
+}
+
+static u32 nbio_v7_0_get_pcie_index_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_INDEX2);
+}
+
+static u32 nbio_v7_0_get_pcie_data_offset(struct amdgpu_device *adev)
+{
+	return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_DATA2);
+}
+
+const struct nbio_hdp_flush_reg nbio_v7_0_hdp_flush_reg = {
+	.ref_and_mask_cp0 = GPU_HDP_FLUSH_DONE__CP0_MASK,
+	.ref_and_mask_cp1 = GPU_HDP_FLUSH_DONE__CP1_MASK,
+	.ref_and_mask_cp2 = GPU_HDP_FLUSH_DONE__CP2_MASK,
+	.ref_and_mask_cp3 = GPU_HDP_FLUSH_DONE__CP3_MASK,
+	.ref_and_mask_cp4 = GPU_HDP_FLUSH_DONE__CP4_MASK,
+	.ref_and_mask_cp5 = GPU_HDP_FLUSH_DONE__CP5_MASK,
+	.ref_and_mask_cp6 = GPU_HDP_FLUSH_DONE__CP6_MASK,
+	.ref_and_mask_cp7 = GPU_HDP_FLUSH_DONE__CP7_MASK,
+	.ref_and_mask_cp8 = GPU_HDP_FLUSH_DONE__CP8_MASK,
+	.ref_and_mask_cp9 = GPU_HDP_FLUSH_DONE__CP9_MASK,
+	.ref_and_mask_sdma0 = GPU_HDP_FLUSH_DONE__SDMA0_MASK,
+	.ref_and_mask_sdma1 = GPU_HDP_FLUSH_DONE__SDMA1_MASK,
+};
+
+static void nbio_v7_0_detect_hw_virt(struct amdgpu_device *adev)
+{
+	if (is_virtual_machine())	/* passthrough mode exclus sriov mod */
+		adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
+}
+
+static void nbio_v7_0_init_registers(struct amdgpu_device *adev)
+{
+
+}
+
+const struct amdgpu_nbio_funcs nbio_v7_0_funcs = {
+	.hdp_flush_reg = &nbio_v7_0_hdp_flush_reg,
+	.get_hdp_flush_req_offset = nbio_v7_0_get_hdp_flush_req_offset,
+	.get_hdp_flush_done_offset = nbio_v7_0_get_hdp_flush_done_offset,
+	.get_pcie_index_offset = nbio_v7_0_get_pcie_index_offset,
+	.get_pcie_data_offset = nbio_v7_0_get_pcie_data_offset,
+	.get_rev_id = nbio_v7_0_get_rev_id,
+	.mc_access_enable = nbio_v7_0_mc_access_enable,
+	.hdp_flush = nbio_v7_0_hdp_flush,
+	.get_memsize = nbio_v7_0_get_memsize,
+	.sdma_doorbell_range = nbio_v7_0_sdma_doorbell_range,
+	.enable_doorbell_aperture = nbio_v7_0_enable_doorbell_aperture,
+	.enable_doorbell_selfring_aperture = nbio_v7_0_enable_doorbell_selfring_aperture,
+	.ih_doorbell_range = nbio_v7_0_ih_doorbell_range,
+	.update_medium_grain_clock_gating = nbio_v7_0_update_medium_grain_clock_gating,
+	.update_medium_grain_light_sleep = nbio_v7_0_update_medium_grain_light_sleep,
+	.get_clockgating_state = nbio_v7_0_get_clockgating_state,
+	.ih_control = nbio_v7_0_ih_control,
+	.init_registers = nbio_v7_0_init_registers,
+	.detect_hw_virt = nbio_v7_0_detect_hw_virt,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/nbio_v7_0.h b/sys/dev/pci/drm/amd/amdgpu/nbio_v7_0.h
new file mode 100644
index 00000000000..508d549c502
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/nbio_v7_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NBIO_V7_0_H__
+#define __NBIO_V7_0_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_nbio_funcs nbio_v7_0_funcs;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/ppsmc.h b/sys/dev/pci/drm/amd/amdgpu/ppsmc.h
new file mode 100644
index 00000000000..8463245f424
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/ppsmc.h
@@ -0,0 +1,200 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef PP_SMC_H
+#define PP_SMC_H
+
+#pragma pack(push, 1)
+
+#define PPSMC_SWSTATE_FLAG_DC                           0x01
+#define PPSMC_SWSTATE_FLAG_UVD                          0x02
+#define PPSMC_SWSTATE_FLAG_VCE                          0x04
+#define PPSMC_SWSTATE_FLAG_PCIE_X1                      0x08
+
+#define PPSMC_THERMAL_PROTECT_TYPE_INTERNAL             0x00
+#define PPSMC_THERMAL_PROTECT_TYPE_EXTERNAL             0x01
+#define PPSMC_THERMAL_PROTECT_TYPE_NONE                 0xff
+
+#define PPSMC_SYSTEMFLAG_GPIO_DC                        0x01
+#define PPSMC_SYSTEMFLAG_STEPVDDC                       0x02
+#define PPSMC_SYSTEMFLAG_GDDR5                          0x04
+#define PPSMC_SYSTEMFLAG_DISABLE_BABYSTEP               0x08
+#define PPSMC_SYSTEMFLAG_REGULATOR_HOT                  0x10
+#define PPSMC_SYSTEMFLAG_REGULATOR_HOT_ANALOG           0x20
+#define PPSMC_SYSTEMFLAG_REGULATOR_HOT_PROG_GPIO        0x40
+
+#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_MASK              0x07
+#define PPSMC_EXTRAFLAGS_AC2DC_DONT_WAIT_FOR_VBLANK     0x08
+#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_GOTODPMLOWSTATE   0x00
+#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_GOTOINITIALSTATE  0x01
+#define PPSMC_EXTRAFLAGS_AC2DC_GPIO5_POLARITY_HIGH      0x02
+
+#define PPSMC_DISPLAY_WATERMARK_LOW                     0
+#define PPSMC_DISPLAY_WATERMARK_HIGH                    1
+
+#define PPSMC_STATEFLAG_AUTO_PULSE_SKIP    0x01
+#define PPSMC_STATEFLAG_POWERBOOST         0x02
+#define PPSMC_STATEFLAG_DEEPSLEEP_THROTTLE 0x20
+#define PPSMC_STATEFLAG_DEEPSLEEP_BYPASS   0x40
+
+#define FDO_MODE_HARDWARE 0
+#define FDO_MODE_PIECE_WISE_LINEAR 1
+
+enum FAN_CONTROL {
+	FAN_CONTROL_FUZZY,
+	FAN_CONTROL_TABLE
+};
+
+#define PPSMC_Result_OK             ((uint8_t)0x01)
+#define PPSMC_Result_Failed         ((uint8_t)0xFF)
+
+typedef uint8_t PPSMC_Result;
+
+#define PPSMC_MSG_Halt                      ((uint8_t)0x10)
+#define PPSMC_MSG_Resume                    ((uint8_t)0x11)
+#define PPSMC_MSG_ZeroLevelsDisabled        ((uint8_t)0x13)
+#define PPSMC_MSG_OneLevelsDisabled         ((uint8_t)0x14)
+#define PPSMC_MSG_TwoLevelsDisabled         ((uint8_t)0x15)
+#define PPSMC_MSG_EnableThermalInterrupt    ((uint8_t)0x16)
+#define PPSMC_MSG_RunningOnAC               ((uint8_t)0x17)
+#define PPSMC_MSG_SwitchToSwState           ((uint8_t)0x20)
+#define PPSMC_MSG_SwitchToInitialState      ((uint8_t)0x40)
+#define PPSMC_MSG_NoForcedLevel             ((uint8_t)0x41)
+#define PPSMC_MSG_ForceHigh                 ((uint8_t)0x42)
+#define PPSMC_MSG_ForceMediumOrHigh         ((uint8_t)0x43)
+#define PPSMC_MSG_SwitchToMinimumPower      ((uint8_t)0x51)
+#define PPSMC_MSG_ResumeFromMinimumPower    ((uint8_t)0x52)
+#define PPSMC_MSG_EnableCac                 ((uint8_t)0x53)
+#define PPSMC_MSG_DisableCac                ((uint8_t)0x54)
+#define PPSMC_TDPClampingActive             ((uint8_t)0x59)
+#define PPSMC_TDPClampingInactive           ((uint8_t)0x5A)
+#define PPSMC_StartFanControl               ((uint8_t)0x5B)
+#define PPSMC_StopFanControl                ((uint8_t)0x5C)
+#define PPSMC_MSG_NoDisplay                 ((uint8_t)0x5D)
+#define PPSMC_NoDisplay                     ((uint8_t)0x5D)
+#define PPSMC_MSG_HasDisplay                ((uint8_t)0x5E)
+#define PPSMC_HasDisplay                    ((uint8_t)0x5E)
+#define PPSMC_MSG_UVDPowerOFF               ((uint8_t)0x60)
+#define PPSMC_MSG_UVDPowerON                ((uint8_t)0x61)
+#define PPSMC_MSG_EnableULV                 ((uint8_t)0x62)
+#define PPSMC_MSG_DisableULV                ((uint8_t)0x63)
+#define PPSMC_MSG_EnterULV                  ((uint8_t)0x64)
+#define PPSMC_MSG_ExitULV                   ((uint8_t)0x65)
+#define PPSMC_CACLongTermAvgEnable          ((uint8_t)0x6E)
+#define PPSMC_CACLongTermAvgDisable         ((uint8_t)0x6F)
+#define PPSMC_MSG_CollectCAC_PowerCorreln   ((uint8_t)0x7A)
+#define PPSMC_FlushDataCache                ((uint8_t)0x80)
+#define PPSMC_MSG_SetEnabledLevels          ((uint8_t)0x82)
+#define PPSMC_MSG_SetForcedLevels           ((uint8_t)0x83)
+#define PPSMC_MSG_ResetToDefaults           ((uint8_t)0x84)
+#define PPSMC_MSG_EnableDTE                 ((uint8_t)0x87)
+#define PPSMC_MSG_DisableDTE                ((uint8_t)0x88)
+#define PPSMC_MSG_ThrottleOVRDSCLKDS        ((uint8_t)0x96)
+#define PPSMC_MSG_CancelThrottleOVRDSCLKDS  ((uint8_t)0x97)
+#define PPSMC_MSG_EnableACDCGPIOInterrupt   ((uint16_t) 0x149)
+
+/* CI/KV/KB */
+#define PPSMC_MSG_UVDDPM_SetEnabledMask       ((uint16_t) 0x12D)
+#define PPSMC_MSG_VCEDPM_SetEnabledMask       ((uint16_t) 0x12E)
+#define PPSMC_MSG_ACPDPM_SetEnabledMask       ((uint16_t) 0x12F)
+#define PPSMC_MSG_SAMUDPM_SetEnabledMask      ((uint16_t) 0x130)
+#define PPSMC_MSG_MCLKDPM_ForceState          ((uint16_t) 0x131)
+#define PPSMC_MSG_MCLKDPM_NoForcedLevel       ((uint16_t) 0x132)
+#define PPSMC_MSG_Thermal_Cntl_Disable        ((uint16_t) 0x133)
+#define PPSMC_MSG_Voltage_Cntl_Disable        ((uint16_t) 0x135)
+#define PPSMC_MSG_PCIeDPM_Enable              ((uint16_t) 0x136)
+#define PPSMC_MSG_PCIeDPM_Disable             ((uint16_t) 0x13d)
+#define PPSMC_MSG_ACPPowerOFF                 ((uint16_t) 0x137)
+#define PPSMC_MSG_ACPPowerON                  ((uint16_t) 0x138)
+#define PPSMC_MSG_SAMPowerOFF                 ((uint16_t) 0x139)
+#define PPSMC_MSG_SAMPowerON                  ((uint16_t) 0x13a)
+#define PPSMC_MSG_PCIeDPM_Disable             ((uint16_t) 0x13d)
+#define PPSMC_MSG_NBDPM_Enable                ((uint16_t) 0x140)
+#define PPSMC_MSG_NBDPM_Disable               ((uint16_t) 0x141)
+#define PPSMC_MSG_SCLKDPM_SetEnabledMask      ((uint16_t) 0x145)
+#define PPSMC_MSG_MCLKDPM_SetEnabledMask      ((uint16_t) 0x146)
+#define PPSMC_MSG_PCIeDPM_ForceLevel          ((uint16_t) 0x147)
+#define PPSMC_MSG_PCIeDPM_UnForceLevel        ((uint16_t) 0x148)
+#define PPSMC_MSG_EnableVRHotGPIOInterrupt    ((uint16_t) 0x14a)
+#define PPSMC_MSG_DPM_Enable                  ((uint16_t) 0x14e)
+#define PPSMC_MSG_DPM_Disable                 ((uint16_t) 0x14f)
+#define PPSMC_MSG_MCLKDPM_Enable              ((uint16_t) 0x150)
+#define PPSMC_MSG_MCLKDPM_Disable             ((uint16_t) 0x151)
+#define PPSMC_MSG_UVDDPM_Enable               ((uint16_t) 0x154)
+#define PPSMC_MSG_UVDDPM_Disable              ((uint16_t) 0x155)
+#define PPSMC_MSG_SAMUDPM_Enable              ((uint16_t) 0x156)
+#define PPSMC_MSG_SAMUDPM_Disable             ((uint16_t) 0x157)
+#define PPSMC_MSG_ACPDPM_Enable               ((uint16_t) 0x158)
+#define PPSMC_MSG_ACPDPM_Disable              ((uint16_t) 0x159)
+#define PPSMC_MSG_VCEDPM_Enable               ((uint16_t) 0x15a)
+#define PPSMC_MSG_VCEDPM_Disable              ((uint16_t) 0x15b)
+#define PPSMC_MSG_VddC_Request                ((uint16_t) 0x15f)
+#define PPSMC_MSG_SCLKDPM_GetEnabledMask      ((uint16_t) 0x162)
+#define PPSMC_MSG_PCIeDPM_SetEnabledMask      ((uint16_t) 0x167)
+#define PPSMC_MSG_TDCLimitEnable              ((uint16_t) 0x169)
+#define PPSMC_MSG_TDCLimitDisable             ((uint16_t) 0x16a)
+#define PPSMC_MSG_PkgPwrLimitEnable           ((uint16_t) 0x185)
+#define PPSMC_MSG_PkgPwrLimitDisable          ((uint16_t) 0x186)
+#define PPSMC_MSG_PkgPwrSetLimit              ((uint16_t) 0x187)
+#define PPSMC_MSG_OverDriveSetTargetTdp       ((uint16_t) 0x188)
+#define PPSMC_MSG_SCLKDPM_FreezeLevel         ((uint16_t) 0x189)
+#define PPSMC_MSG_SCLKDPM_UnfreezeLevel       ((uint16_t) 0x18A)
+#define PPSMC_MSG_MCLKDPM_FreezeLevel         ((uint16_t) 0x18B)
+#define PPSMC_MSG_MCLKDPM_UnfreezeLevel       ((uint16_t) 0x18C)
+#define PPSMC_MSG_MASTER_DeepSleep_ON         ((uint16_t) 0x18F)
+#define PPSMC_MSG_MASTER_DeepSleep_OFF        ((uint16_t) 0x190)
+#define PPSMC_MSG_Remove_DC_Clamp             ((uint16_t) 0x191)
+#define PPSMC_MSG_SetFanPwmMax                ((uint16_t) 0x19A)
+#define PPSMC_MSG_SetFanRpmMax                ((uint16_t) 0x205)
+
+#define PPSMC_MSG_ENABLE_THERMAL_DPM          ((uint16_t) 0x19C)
+#define PPSMC_MSG_DISABLE_THERMAL_DPM         ((uint16_t) 0x19D)
+
+#define PPSMC_MSG_API_GetSclkFrequency        ((uint16_t) 0x200)
+#define PPSMC_MSG_API_GetMclkFrequency        ((uint16_t) 0x201)
+
+/* TN */
+#define PPSMC_MSG_DPM_Config                ((uint32_t) 0x102)
+#define PPSMC_MSG_DPM_ForceState            ((uint32_t) 0x104)
+#define PPSMC_MSG_PG_SIMD_Config            ((uint32_t) 0x108)
+#define PPSMC_MSG_Voltage_Cntl_Enable       ((uint32_t) 0x109)
+#define PPSMC_MSG_Thermal_Cntl_Enable       ((uint32_t) 0x10a)
+#define PPSMC_MSG_VCEPowerOFF               ((uint32_t) 0x10e)
+#define PPSMC_MSG_VCEPowerON                ((uint32_t) 0x10f)
+#define PPSMC_MSG_DPM_N_LevelsDisabled      ((uint32_t) 0x112)
+#define PPSMC_MSG_DCE_RemoveVoltageAdjustment   ((uint32_t) 0x11d)
+#define PPSMC_MSG_DCE_AllowVoltageAdjustment    ((uint32_t) 0x11e)
+#define PPSMC_MSG_EnableBAPM                ((uint32_t) 0x120)
+#define PPSMC_MSG_DisableBAPM               ((uint32_t) 0x121)
+#define PPSMC_MSG_UVD_DPM_Config            ((uint32_t) 0x124)
+
+#define PPSMC_MSG_DRV_DRAM_ADDR_HI            ((uint16_t) 0x250)
+#define PPSMC_MSG_DRV_DRAM_ADDR_LO            ((uint16_t) 0x251)
+#define PPSMC_MSG_SMU_DRAM_ADDR_HI            ((uint16_t) 0x252)
+#define PPSMC_MSG_SMU_DRAM_ADDR_LO            ((uint16_t) 0x253)
+#define PPSMC_MSG_LoadUcodes                  ((uint16_t) 0x254)
+
+typedef uint16_t PPSMC_Msg;
+
+#pragma pack(pop)
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/psp_gfx_if.h b/sys/dev/pci/drm/amd/amdgpu/psp_gfx_if.h
new file mode 100644
index 00000000000..0cf48d26c67
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/psp_gfx_if.h
@@ -0,0 +1,294 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _PSP_TEE_GFX_IF_H_
+#define _PSP_TEE_GFX_IF_H_
+
+#define PSP_GFX_CMD_BUF_VERSION     0x00000001
+
+#define GFX_CMD_STATUS_MASK         0x0000FFFF
+#define GFX_CMD_ID_MASK             0x000F0000
+#define GFX_CMD_RESERVED_MASK       0x7FF00000
+#define GFX_CMD_RESPONSE_MASK       0x80000000
+
+/* TEE Gfx Command IDs for the register interface.
+*  Command ID must be between 0x00010000 and 0x000F0000.
+*/
+enum psp_gfx_crtl_cmd_id
+{
+    GFX_CTRL_CMD_ID_INIT_RBI_RING   = 0x00010000,   /* initialize RBI ring */
+    GFX_CTRL_CMD_ID_INIT_GPCOM_RING = 0x00020000,   /* initialize GPCOM ring */
+    GFX_CTRL_CMD_ID_DESTROY_RINGS   = 0x00030000,   /* destroy rings */
+    GFX_CTRL_CMD_ID_CAN_INIT_RINGS  = 0x00040000,   /* is it allowed to initialized the rings */
+    GFX_CTRL_CMD_ID_ENABLE_INT      = 0x00050000,   /* enable PSP-to-Gfx interrupt */
+    GFX_CTRL_CMD_ID_DISABLE_INT     = 0x00060000,   /* disable PSP-to-Gfx interrupt */
+    GFX_CTRL_CMD_ID_MODE1_RST       = 0x00070000,   /* trigger the Mode 1 reset */
+
+    GFX_CTRL_CMD_ID_MAX             = 0x000F0000,   /* max command ID */
+};
+
+
+/*-----------------------------------------------------------------------------
+    NOTE:   All physical addresses used in this interface are actually
+            GPU Virtual Addresses.
+*/
+
+
+/* Control registers of the TEE Gfx interface. These are located in
+*  SRBM-to-PSP mailbox registers (total 8 registers).
+*/
+struct psp_gfx_ctrl
+{
+    volatile uint32_t   cmd_resp;         /* +0   Command/Response register for Gfx commands */
+    volatile uint32_t   rbi_wptr;         /* +4   Write pointer (index) of RBI ring */
+    volatile uint32_t   rbi_rptr;         /* +8   Read pointer (index) of RBI ring */
+    volatile uint32_t   gpcom_wptr;       /* +12  Write pointer (index) of GPCOM ring */
+    volatile uint32_t   gpcom_rptr;       /* +16  Read pointer (index) of GPCOM ring */
+    volatile uint32_t   ring_addr_lo;     /* +20  bits [31:0] of GPU Virtual of ring buffer (VMID=0)*/
+    volatile uint32_t   ring_addr_hi;     /* +24  bits [63:32] of GPU Virtual of ring buffer (VMID=0) */
+    volatile uint32_t   ring_buf_size;    /* +28  Ring buffer size (in bytes) */
+
+};
+
+
+/* Response flag is set in the command when command is completed by PSP.
+*  Used in the GFX_CTRL.CmdResp.
+*  When PSP GFX I/F is initialized, the flag is set.
+*/
+#define GFX_FLAG_RESPONSE               0x80000000
+
+
+/* TEE Gfx Command IDs for the ring buffer interface. */
+enum psp_gfx_cmd_id
+{
+    GFX_CMD_ID_LOAD_TA      = 0x00000001,   /* load TA */
+    GFX_CMD_ID_UNLOAD_TA    = 0x00000002,   /* unload TA */
+    GFX_CMD_ID_INVOKE_CMD   = 0x00000003,   /* send command to TA */
+    GFX_CMD_ID_LOAD_ASD     = 0x00000004,   /* load ASD Driver */
+    GFX_CMD_ID_SETUP_TMR    = 0x00000005,   /* setup TMR region */
+    GFX_CMD_ID_LOAD_IP_FW   = 0x00000006,   /* load HW IP FW */
+    GFX_CMD_ID_DESTROY_TMR  = 0x00000007,   /* destroy TMR region */
+    GFX_CMD_ID_SAVE_RESTORE = 0x00000008,   /* save/restore HW IP FW */
+
+};
+
+
+/* Command to load Trusted Application binary into PSP OS. */
+struct psp_gfx_cmd_load_ta
+{
+    uint32_t        app_phy_addr_lo;        /* bits [31:0] of the GPU Virtual address of the TA binary (must be 4 KB aligned) */
+    uint32_t        app_phy_addr_hi;        /* bits [63:32] of the GPU Virtual address of the TA binary */
+    uint32_t        app_len;                /* length of the TA binary in bytes */
+    uint32_t        cmd_buf_phy_addr_lo;    /* bits [31:0] of the GPU Virtual address of CMD buffer (must be 4 KB aligned) */
+    uint32_t        cmd_buf_phy_addr_hi;    /* bits [63:32] of the GPU Virtual address of CMD buffer */
+    uint32_t        cmd_buf_len;            /* length of the CMD buffer in bytes; must be multiple of 4 KB */
+
+    /* Note: CmdBufLen can be set to 0. In this case no persistent CMD buffer is provided
+    *       for the TA. Each InvokeCommand can have dinamically mapped CMD buffer instead
+    *       of using global persistent buffer.
+    */
+};
+
+
+/* Command to Unload Trusted Application binary from PSP OS. */
+struct psp_gfx_cmd_unload_ta
+{
+    uint32_t        session_id;          /* Session ID of the loaded TA to be unloaded */
+
+};
+
+
+/* Shared buffers for InvokeCommand.
+*/
+struct psp_gfx_buf_desc
+{
+    uint32_t        buf_phy_addr_lo;       /* bits [31:0] of GPU Virtual address of the buffer (must be 4 KB aligned) */
+    uint32_t        buf_phy_addr_hi;       /* bits [63:32] of GPU Virtual address of the buffer */
+    uint32_t        buf_size;              /* buffer size in bytes (must be multiple of 4 KB and no bigger than 64 MB) */
+
+};
+
+/* Max number of descriptors for one shared buffer (in how many different
+*  physical locations one shared buffer can be stored). If buffer is too much
+*  fragmented, error will be returned.
+*/
+#define GFX_BUF_MAX_DESC        64
+
+struct psp_gfx_buf_list
+{
+    uint32_t                num_desc;                    /* number of buffer descriptors in the list */
+    uint32_t                total_size;                  /* total size of all buffers in the list in bytes (must be multiple of 4 KB) */
+    struct psp_gfx_buf_desc buf_desc[GFX_BUF_MAX_DESC];  /* list of buffer descriptors */
+
+    /* total 776 bytes */
+};
+
+/* Command to execute InvokeCommand entry point of the TA. */
+struct psp_gfx_cmd_invoke_cmd
+{
+    uint32_t                session_id;           /* Session ID of the TA to be executed */
+    uint32_t                ta_cmd_id;            /* Command ID to be sent to TA */
+    struct psp_gfx_buf_list buf;                  /* one indirect buffer (scatter/gather list) */
+
+};
+
+
+/* Command to setup TMR region. */
+struct psp_gfx_cmd_setup_tmr
+{
+    uint32_t        buf_phy_addr_lo;       /* bits [31:0] of GPU Virtual address of TMR buffer (must be 4 KB aligned) */
+    uint32_t        buf_phy_addr_hi;       /* bits [63:32] of GPU Virtual address of TMR buffer */
+    uint32_t        buf_size;              /* buffer size in bytes (must be multiple of 4 KB) */
+
+};
+
+
+/* FW types for GFX_CMD_ID_LOAD_IP_FW command. Limit 31. */
+enum psp_gfx_fw_type
+{
+    GFX_FW_TYPE_NONE        = 0,
+    GFX_FW_TYPE_CP_ME       = 1,
+    GFX_FW_TYPE_CP_PFP      = 2,
+    GFX_FW_TYPE_CP_CE       = 3,
+    GFX_FW_TYPE_CP_MEC      = 4,
+    GFX_FW_TYPE_CP_MEC_ME1  = 5,
+    GFX_FW_TYPE_CP_MEC_ME2  = 6,
+    GFX_FW_TYPE_RLC_V       = 7,
+    GFX_FW_TYPE_RLC_G       = 8,
+    GFX_FW_TYPE_SDMA0       = 9,
+    GFX_FW_TYPE_SDMA1       = 10,
+    GFX_FW_TYPE_DMCU_ERAM   = 11,
+    GFX_FW_TYPE_DMCU_ISR    = 12,
+    GFX_FW_TYPE_VCN         = 13,
+    GFX_FW_TYPE_UVD         = 14,
+    GFX_FW_TYPE_VCE         = 15,
+    GFX_FW_TYPE_ISP         = 16,
+    GFX_FW_TYPE_ACP         = 17,
+    GFX_FW_TYPE_SMU         = 18,
+    GFX_FW_TYPE_MMSCH       = 19,
+    GFX_FW_TYPE_RLC_RESTORE_LIST_GPM_MEM        = 20,
+    GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_MEM        = 21,
+    GFX_FW_TYPE_RLC_RESTORE_LIST_CNTL           = 22,
+    GFX_FW_TYPE_MAX         = 23
+};
+
+/* Command to load HW IP FW. */
+struct psp_gfx_cmd_load_ip_fw
+{
+    uint32_t                fw_phy_addr_lo;    /* bits [31:0] of GPU Virtual address of FW location (must be 4 KB aligned) */
+    uint32_t                fw_phy_addr_hi;    /* bits [63:32] of GPU Virtual address of FW location */
+    uint32_t                fw_size;           /* FW buffer size in bytes */
+    enum psp_gfx_fw_type    fw_type;           /* FW type */
+
+};
+
+/* Command to save/restore HW IP FW. */
+struct psp_gfx_cmd_save_restore_ip_fw
+{
+    uint32_t                save_fw;              /* if set, command is used for saving fw otherwise for resetoring*/
+    uint32_t                save_restore_addr_lo; /* bits [31:0] of FB address of GART memory used as save/restore buffer (must be 4 KB aligned) */
+    uint32_t                save_restore_addr_hi; /* bits [63:32] of FB address of GART memory used as save/restore buffer */
+    uint32_t                buf_size;             /* Size of the save/restore buffer in bytes */
+    enum psp_gfx_fw_type    fw_type;              /* FW type */
+};
+
+/* All GFX ring buffer commands. */
+union psp_gfx_commands
+{
+    struct psp_gfx_cmd_load_ta          cmd_load_ta;
+    struct psp_gfx_cmd_unload_ta        cmd_unload_ta;
+    struct psp_gfx_cmd_invoke_cmd       cmd_invoke_cmd;
+    struct psp_gfx_cmd_setup_tmr        cmd_setup_tmr;
+    struct psp_gfx_cmd_load_ip_fw       cmd_load_ip_fw;
+    struct psp_gfx_cmd_save_restore_ip_fw cmd_save_restore_ip_fw;
+};
+
+
+/* Structure of GFX Response buffer.
+* For GPCOM I/F it is part of GFX_CMD_RESP buffer, for RBI
+* it is separate buffer.
+*/
+struct psp_gfx_resp
+{
+    uint32_t    status;             /* +0  status of command execution */
+    uint32_t    session_id;         /* +4  session ID in response to LoadTa command */
+    uint32_t    fw_addr_lo;         /* +8  bits [31:0] of FW address within TMR (in response to cmd_load_ip_fw command) */
+    uint32_t    fw_addr_hi;         /* +12 bits [63:32] of FW address within TMR (in response to cmd_load_ip_fw command) */
+
+    uint32_t    reserved[4];
+
+    /* total 32 bytes */
+};
+
+/* Structure of Command buffer pointed by psp_gfx_rb_frame.cmd_buf_addr_hi
+*  and psp_gfx_rb_frame.cmd_buf_addr_lo.
+*/
+struct psp_gfx_cmd_resp
+{
+    uint32_t        buf_size;           /* +0  total size of the buffer in bytes */
+    uint32_t        buf_version;        /* +4  version of the buffer strusture; must be PSP_GFX_CMD_BUF_VERSION */
+    uint32_t        cmd_id;             /* +8  command ID */
+
+    /* These fields are used for RBI only. They are all 0 in GPCOM commands
+    */
+    uint32_t        resp_buf_addr_lo;   /* +12 bits [31:0] of GPU Virtual address of response buffer (must be 4 KB aligned) */
+    uint32_t        resp_buf_addr_hi;   /* +16 bits [63:32] of GPU Virtual address of response buffer */
+    uint32_t        resp_offset;        /* +20 offset within response buffer */
+    uint32_t        resp_buf_size;      /* +24 total size of the response buffer in bytes */
+
+    union psp_gfx_commands  cmd;        /* +28 command specific structures */
+
+    uint8_t         reserved_1[864 - sizeof(union psp_gfx_commands) - 28];
+
+    /* Note: Resp is part of this buffer for GPCOM ring. For RBI ring the response
+    *        is separate buffer pointed by resp_buf_addr_hi and resp_buf_addr_lo.
+    */
+    struct psp_gfx_resp     resp;       /* +864 response */
+
+    uint8_t         reserved_2[1024 - 864 - sizeof(struct psp_gfx_resp)];
+
+    /* total size 1024 bytes */
+};
+
+
+#define FRAME_TYPE_DESTROY          1   /* frame sent by KMD driver when UMD Scheduler context is destroyed*/
+
+/* Structure of the Ring Buffer Frame */
+struct psp_gfx_rb_frame
+{
+    uint32_t    cmd_buf_addr_lo;    /* +0  bits [31:0] of GPU Virtual address of command buffer (must be 4 KB aligned) */
+    uint32_t    cmd_buf_addr_hi;    /* +4  bits [63:32] of GPU Virtual address of command buffer */
+    uint32_t    cmd_buf_size;       /* +8  command buffer size in bytes */
+    uint32_t    fence_addr_lo;      /* +12 bits [31:0] of GPU Virtual address of Fence for this frame */
+    uint32_t    fence_addr_hi;      /* +16 bits [63:32] of GPU Virtual address of Fence for this frame */
+    uint32_t    fence_value;        /* +20 Fence value */
+    uint32_t    sid_lo;             /* +24 bits [31:0] of SID value (used only for RBI frames) */
+    uint32_t    sid_hi;             /* +28 bits [63:32] of SID value (used only for RBI frames) */
+    uint8_t     vmid;               /* +32 VMID value used for mapping of all addresses for this frame */
+    uint8_t     frame_type;         /* +33 1: destory context frame, 0: all other frames; used only for RBI frames */
+    uint8_t     reserved1[2];       /* +34 reserved, must be 0 */
+    uint32_t    reserved2[7];       /* +36 reserved, must be 0 */
+                /* total 64 bytes */
+};
+
+#endif /* _PSP_TEE_GFX_IF_H_ */
diff --git a/sys/dev/pci/drm/amd/amdgpu/psp_v10_0.c b/sys/dev/pci/drm/amd/amdgpu/psp_v10_0.c
new file mode 100644
index 00000000000..02be34e72ed
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/psp_v10_0.c
@@ -0,0 +1,460 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: Huang Rui
+ *
+ */
+
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_ucode.h"
+#include "soc15_common.h"
+#include "psp_v10_0.h"
+
+#include "mp/mp_10_0_offset.h"
+#include "gc/gc_9_1_offset.h"
+#include "sdma0/sdma0_4_1_offset.h"
+
+MODULE_FIRMWARE("amdgpu/raven_asd.bin");
+
+static int
+psp_v10_0_get_fw_type(struct amdgpu_firmware_info *ucode, enum psp_gfx_fw_type *type)
+{
+	switch(ucode->ucode_id) {
+	case AMDGPU_UCODE_ID_SDMA0:
+		*type = GFX_FW_TYPE_SDMA0;
+		break;
+	case AMDGPU_UCODE_ID_SDMA1:
+		*type = GFX_FW_TYPE_SDMA1;
+		break;
+	case AMDGPU_UCODE_ID_CP_CE:
+		*type = GFX_FW_TYPE_CP_CE;
+		break;
+	case AMDGPU_UCODE_ID_CP_PFP:
+		*type = GFX_FW_TYPE_CP_PFP;
+		break;
+	case AMDGPU_UCODE_ID_CP_ME:
+		*type = GFX_FW_TYPE_CP_ME;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC1:
+		*type = GFX_FW_TYPE_CP_MEC;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC1_JT:
+		*type = GFX_FW_TYPE_CP_MEC_ME1;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC2:
+		*type = GFX_FW_TYPE_CP_MEC;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC2_JT:
+		*type = GFX_FW_TYPE_CP_MEC_ME2;
+		break;
+	case AMDGPU_UCODE_ID_RLC_G:
+		*type = GFX_FW_TYPE_RLC_G;
+		break;
+	case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL:
+		*type = GFX_FW_TYPE_RLC_RESTORE_LIST_CNTL;
+		break;
+	case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM:
+		*type = GFX_FW_TYPE_RLC_RESTORE_LIST_GPM_MEM;
+		break;
+	case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM:
+		*type = GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_MEM;
+		break;
+	case AMDGPU_UCODE_ID_SMC:
+		*type = GFX_FW_TYPE_SMU;
+		break;
+	case AMDGPU_UCODE_ID_UVD:
+		*type = GFX_FW_TYPE_UVD;
+		break;
+	case AMDGPU_UCODE_ID_VCE:
+		*type = GFX_FW_TYPE_VCE;
+		break;
+	case AMDGPU_UCODE_ID_VCN:
+		*type = GFX_FW_TYPE_VCN;
+		break;
+	case AMDGPU_UCODE_ID_MAXIMUM:
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int psp_v10_0_init_microcode(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	const char *chip_name;
+	char fw_name[30];
+	int err = 0;
+	const struct psp_firmware_header_v1_0 *hdr;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_RAVEN:
+		chip_name = "raven";
+		break;
+	default: BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_asd.bin", chip_name);
+	err = request_firmware(&adev->psp.asd_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+
+	err = amdgpu_ucode_validate(adev->psp.asd_fw);
+	if (err)
+		goto out;
+
+	hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.asd_fw->data;
+	adev->psp.asd_fw_version = le32_to_cpu(hdr->header.ucode_version);
+	adev->psp.asd_feature_version = le32_to_cpu(hdr->ucode_feature_version);
+	adev->psp.asd_ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
+	adev->psp.asd_start_addr = (uint8_t *)hdr +
+				le32_to_cpu(hdr->header.ucode_array_offset_bytes);
+
+	return 0;
+out:
+	if (err) {
+		dev_err(adev->dev,
+			"psp v10.0: Failed to load firmware \"%s\"\n",
+			fw_name);
+		release_firmware(adev->psp.asd_fw);
+		adev->psp.asd_fw = NULL;
+	}
+
+	return err;
+}
+
+static int psp_v10_0_prep_cmd_buf(struct amdgpu_firmware_info *ucode,
+				  struct psp_gfx_cmd_resp *cmd)
+{
+	int ret;
+	uint64_t fw_mem_mc_addr = ucode->mc_addr;
+
+	memset(cmd, 0, sizeof(struct psp_gfx_cmd_resp));
+
+	cmd->cmd_id = GFX_CMD_ID_LOAD_IP_FW;
+	cmd->cmd.cmd_load_ip_fw.fw_phy_addr_lo = lower_32_bits(fw_mem_mc_addr);
+	cmd->cmd.cmd_load_ip_fw.fw_phy_addr_hi = upper_32_bits(fw_mem_mc_addr);
+	cmd->cmd.cmd_load_ip_fw.fw_size = ucode->ucode_size;
+
+	ret = psp_v10_0_get_fw_type(ucode, &cmd->cmd.cmd_load_ip_fw.fw_type);
+	if (ret)
+		DRM_ERROR("Unknown firmware type\n");
+
+	return ret;
+}
+
+static int psp_v10_0_ring_init(struct psp_context *psp,
+			       enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct psp_ring *ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	ring = &psp->km_ring;
+
+	ring->ring_type = ring_type;
+
+	/* allocate 4k Page of Local Frame Buffer memory for ring */
+	ring->ring_size = 0x1000;
+	ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->firmware.rbuf,
+				      &ring->ring_mem_mc_addr,
+				      (void **)&ring->ring_mem);
+	if (ret) {
+		ring->ring_size = 0;
+		return ret;
+	}
+
+	return 0;
+}
+
+static int psp_v10_0_ring_create(struct psp_context *psp,
+				 enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	unsigned int psp_ring_reg = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Write low address of the ring to C2PMSG_69 */
+	psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_69, psp_ring_reg);
+	/* Write high address of the ring to C2PMSG_70 */
+	psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_70, psp_ring_reg);
+	/* Write size of ring to C2PMSG_71 */
+	psp_ring_reg = ring->ring_size;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_71, psp_ring_reg);
+	/* Write the ring initialization command to C2PMSG_64 */
+	psp_ring_reg = ring_type;
+	psp_ring_reg = psp_ring_reg << 16;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
+
+	/* There might be handshake issue with hardware which needs delay */
+	mdelay(20);
+
+	/* Wait for response flag (bit 31) in C2PMSG_64 */
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			   0x80000000, 0x8000FFFF, false);
+
+	return ret;
+}
+
+static int psp_v10_0_ring_stop(struct psp_context *psp,
+			       enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct psp_ring *ring;
+	unsigned int psp_ring_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	ring = &psp->km_ring;
+
+	/* Write the ring destroy command to C2PMSG_64 */
+	psp_ring_reg = 3 << 16;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
+
+	/* There might be handshake issue with hardware which needs delay */
+	mdelay(20);
+
+	/* Wait for response flag (bit 31) in C2PMSG_64 */
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			   0x80000000, 0x80000000, false);
+
+	return ret;
+}
+
+static int psp_v10_0_ring_destroy(struct psp_context *psp,
+				  enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	ret = psp_v10_0_ring_stop(psp, ring_type);
+	if (ret)
+		DRM_ERROR("Fail to stop psp ring\n");
+
+	amdgpu_bo_free_kernel(&adev->firmware.rbuf,
+			      &ring->ring_mem_mc_addr,
+			      (void **)&ring->ring_mem);
+
+	return ret;
+}
+
+static int psp_v10_0_cmd_submit(struct psp_context *psp,
+				struct amdgpu_firmware_info *ucode,
+				uint64_t cmd_buf_mc_addr, uint64_t fence_mc_addr,
+				int index)
+{
+	unsigned int psp_write_ptr_reg = 0;
+	struct psp_gfx_rb_frame * write_frame = psp->km_ring.ring_mem;
+	struct psp_ring *ring = &psp->km_ring;
+	struct psp_gfx_rb_frame *ring_buffer_start = ring->ring_mem;
+	struct psp_gfx_rb_frame *ring_buffer_end = ring_buffer_start +
+		ring->ring_size / sizeof(struct psp_gfx_rb_frame) - 1;
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t ring_size_dw = ring->ring_size / 4;
+	uint32_t rb_frame_size_dw = sizeof(struct psp_gfx_rb_frame) / 4;
+
+	/* KM (GPCOM) prepare write pointer */
+	psp_write_ptr_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67);
+
+	/* Update KM RB frame pointer to new frame */
+	if ((psp_write_ptr_reg % ring_size_dw) == 0)
+		write_frame = ring_buffer_start;
+	else
+		write_frame = ring_buffer_start + (psp_write_ptr_reg / rb_frame_size_dw);
+	/* Check invalid write_frame ptr address */
+	if ((write_frame < ring_buffer_start) || (ring_buffer_end < write_frame)) {
+		DRM_ERROR("ring_buffer_start = %p; ring_buffer_end = %p; write_frame = %p\n",
+			  ring_buffer_start, ring_buffer_end, write_frame);
+		DRM_ERROR("write_frame is pointing to address out of bounds\n");
+		return -EINVAL;
+	}
+
+	/* Initialize KM RB frame */
+	memset(write_frame, 0, sizeof(struct psp_gfx_rb_frame));
+
+	/* Update KM RB frame */
+	write_frame->cmd_buf_addr_hi = upper_32_bits(cmd_buf_mc_addr);
+	write_frame->cmd_buf_addr_lo = lower_32_bits(cmd_buf_mc_addr);
+	write_frame->fence_addr_hi = upper_32_bits(fence_mc_addr);
+	write_frame->fence_addr_lo = lower_32_bits(fence_mc_addr);
+	write_frame->fence_value = index;
+
+	/* Update the write Pointer in DWORDs */
+	psp_write_ptr_reg = (psp_write_ptr_reg + rb_frame_size_dw) % ring_size_dw;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67, psp_write_ptr_reg);
+
+	return 0;
+}
+
+static int
+psp_v10_0_sram_map(struct amdgpu_device *adev,
+		   unsigned int *sram_offset, unsigned int *sram_addr_reg_offset,
+		   unsigned int *sram_data_reg_offset,
+		   enum AMDGPU_UCODE_ID ucode_id)
+{
+	int ret = 0;
+
+	switch(ucode_id) {
+/* TODO: needs to confirm */
+#if 0
+	case AMDGPU_UCODE_ID_SMC:
+		*sram_offset = 0;
+		*sram_addr_reg_offset = 0;
+		*sram_data_reg_offset = 0;
+		break;
+#endif
+
+	case AMDGPU_UCODE_ID_CP_CE:
+		*sram_offset = 0x0;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_CE_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_CE_UCODE_DATA);
+		break;
+
+	case AMDGPU_UCODE_ID_CP_PFP:
+		*sram_offset = 0x0;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_PFP_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_PFP_UCODE_DATA);
+		break;
+
+	case AMDGPU_UCODE_ID_CP_ME:
+		*sram_offset = 0x0;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_ME_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_ME_UCODE_DATA);
+		break;
+
+	case AMDGPU_UCODE_ID_CP_MEC1:
+		*sram_offset = 0x10000;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_MEC_ME1_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_MEC_ME1_UCODE_DATA);
+		break;
+
+	case AMDGPU_UCODE_ID_CP_MEC2:
+		*sram_offset = 0x10000;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_MEC2_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_MEC2_UCODE_DATA);
+		break;
+
+	case AMDGPU_UCODE_ID_RLC_G:
+		*sram_offset = 0x2000;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UCODE_DATA);
+		break;
+
+	case AMDGPU_UCODE_ID_SDMA0:
+		*sram_offset = 0x0;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UCODE_DATA);
+		break;
+
+/* TODO: needs to confirm */
+#if 0
+	case AMDGPU_UCODE_ID_SDMA1:
+		*sram_offset = ;
+		*sram_addr_reg_offset = ;
+		break;
+
+	case AMDGPU_UCODE_ID_UVD:
+		*sram_offset = ;
+		*sram_addr_reg_offset = ;
+		break;
+
+	case AMDGPU_UCODE_ID_VCE:
+		*sram_offset = ;
+		*sram_addr_reg_offset = ;
+		break;
+#endif
+
+	case AMDGPU_UCODE_ID_MAXIMUM:
+	default:
+		ret = -EINVAL;
+		break;
+	}
+
+	return ret;
+}
+
+static bool psp_v10_0_compare_sram_data(struct psp_context *psp,
+					struct amdgpu_firmware_info *ucode,
+					enum AMDGPU_UCODE_ID ucode_type)
+{
+	int err = 0;
+	unsigned int fw_sram_reg_val = 0;
+	unsigned int fw_sram_addr_reg_offset = 0;
+	unsigned int fw_sram_data_reg_offset = 0;
+	unsigned int ucode_size;
+	uint32_t *ucode_mem = NULL;
+	struct amdgpu_device *adev = psp->adev;
+
+	err = psp_v10_0_sram_map(adev, &fw_sram_reg_val, &fw_sram_addr_reg_offset,
+				&fw_sram_data_reg_offset, ucode_type);
+	if (err)
+		return false;
+
+	WREG32(fw_sram_addr_reg_offset, fw_sram_reg_val);
+
+	ucode_size = ucode->ucode_size;
+	ucode_mem = (uint32_t *)ucode->kaddr;
+	while (!ucode_size) {
+		fw_sram_reg_val = RREG32(fw_sram_data_reg_offset);
+
+		if (*ucode_mem != fw_sram_reg_val)
+			return false;
+
+		ucode_mem++;
+		/* 4 bytes */
+		ucode_size -= 4;
+	}
+
+	return true;
+}
+
+
+static int psp_v10_0_mode1_reset(struct psp_context *psp)
+{
+	DRM_INFO("psp mode 1 reset not supported now! \n");
+	return -EINVAL;
+}
+
+static const struct psp_funcs psp_v10_0_funcs = {
+	.init_microcode = psp_v10_0_init_microcode,
+	.prep_cmd_buf = psp_v10_0_prep_cmd_buf,
+	.ring_init = psp_v10_0_ring_init,
+	.ring_create = psp_v10_0_ring_create,
+	.ring_stop = psp_v10_0_ring_stop,
+	.ring_destroy = psp_v10_0_ring_destroy,
+	.cmd_submit = psp_v10_0_cmd_submit,
+	.compare_sram_data = psp_v10_0_compare_sram_data,
+	.mode1_reset = psp_v10_0_mode1_reset,
+};
+
+void psp_v10_0_set_psp_funcs(struct psp_context *psp)
+{
+	psp->funcs = &psp_v10_0_funcs;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/psp_v10_0.h b/sys/dev/pci/drm/amd/amdgpu/psp_v10_0.h
new file mode 100644
index 00000000000..20c2a94859d
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/psp_v10_0.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: Huang Rui
+ *
+ */
+#ifndef __PSP_V10_0_H__
+#define __PSP_V10_0_H__
+
+#include "amdgpu_psp.h"
+
+void psp_v10_0_set_psp_funcs(struct psp_context *psp);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/psp_v3_1.c b/sys/dev/pci/drm/amd/amdgpu/psp_v3_1.c
new file mode 100644
index 00000000000..727071fee6f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/psp_v3_1.c
@@ -0,0 +1,634 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: Huang Rui
+ *
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_ucode.h"
+#include "soc15_common.h"
+#include "psp_v3_1.h"
+
+#include "mp/mp_9_0_offset.h"
+#include "mp/mp_9_0_sh_mask.h"
+#include "gc/gc_9_0_offset.h"
+#include "sdma0/sdma0_4_0_offset.h"
+#include "nbio/nbio_6_1_offset.h"
+
+MODULE_FIRMWARE("amdgpu/vega10_sos.bin");
+MODULE_FIRMWARE("amdgpu/vega10_asd.bin");
+MODULE_FIRMWARE("amdgpu/vega12_sos.bin");
+MODULE_FIRMWARE("amdgpu/vega12_asd.bin");
+MODULE_FIRMWARE("amdgpu/vega20_sos.bin");
+MODULE_FIRMWARE("amdgpu/vega20_asd.bin");
+
+
+#define smnMP1_FIRMWARE_FLAGS 0x3010028
+
+static uint32_t sos_old_versions[] = {1517616, 1510592, 1448594, 1446554};
+
+static int
+psp_v3_1_get_fw_type(struct amdgpu_firmware_info *ucode, enum psp_gfx_fw_type *type)
+{
+	switch(ucode->ucode_id) {
+	case AMDGPU_UCODE_ID_SDMA0:
+		*type = GFX_FW_TYPE_SDMA0;
+		break;
+	case AMDGPU_UCODE_ID_SDMA1:
+		*type = GFX_FW_TYPE_SDMA1;
+		break;
+	case AMDGPU_UCODE_ID_CP_CE:
+		*type = GFX_FW_TYPE_CP_CE;
+		break;
+	case AMDGPU_UCODE_ID_CP_PFP:
+		*type = GFX_FW_TYPE_CP_PFP;
+		break;
+	case AMDGPU_UCODE_ID_CP_ME:
+		*type = GFX_FW_TYPE_CP_ME;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC1:
+		*type = GFX_FW_TYPE_CP_MEC;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC1_JT:
+		*type = GFX_FW_TYPE_CP_MEC_ME1;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC2:
+		*type = GFX_FW_TYPE_CP_MEC;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC2_JT:
+		*type = GFX_FW_TYPE_CP_MEC_ME2;
+		break;
+	case AMDGPU_UCODE_ID_RLC_G:
+		*type = GFX_FW_TYPE_RLC_G;
+		break;
+	case AMDGPU_UCODE_ID_SMC:
+		*type = GFX_FW_TYPE_SMU;
+		break;
+	case AMDGPU_UCODE_ID_UVD:
+		*type = GFX_FW_TYPE_UVD;
+		break;
+	case AMDGPU_UCODE_ID_VCE:
+		*type = GFX_FW_TYPE_VCE;
+		break;
+	case AMDGPU_UCODE_ID_MAXIMUM:
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int psp_v3_1_init_microcode(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	const char *chip_name;
+	char fw_name[30];
+	int err = 0;
+	const struct psp_firmware_header_v1_0 *hdr;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+		chip_name = "vega10";
+		break;
+	case CHIP_VEGA12:
+		chip_name = "vega12";
+		break;
+	default: BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sos.bin", chip_name);
+	err = request_firmware(&adev->psp.sos_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+
+	err = amdgpu_ucode_validate(adev->psp.sos_fw);
+	if (err)
+		goto out;
+
+	hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.sos_fw->data;
+	adev->psp.sos_fw_version = le32_to_cpu(hdr->header.ucode_version);
+	adev->psp.sos_feature_version = le32_to_cpu(hdr->ucode_feature_version);
+	adev->psp.sos_bin_size = le32_to_cpu(hdr->sos_size_bytes);
+	adev->psp.sys_bin_size = le32_to_cpu(hdr->header.ucode_size_bytes) -
+					le32_to_cpu(hdr->sos_size_bytes);
+	adev->psp.sys_start_addr = (uint8_t *)hdr +
+				le32_to_cpu(hdr->header.ucode_array_offset_bytes);
+	adev->psp.sos_start_addr = (uint8_t *)adev->psp.sys_start_addr +
+				le32_to_cpu(hdr->sos_offset_bytes);
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_asd.bin", chip_name);
+	err = request_firmware(&adev->psp.asd_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+
+	err = amdgpu_ucode_validate(adev->psp.asd_fw);
+	if (err)
+		goto out;
+
+	hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.asd_fw->data;
+	adev->psp.asd_fw_version = le32_to_cpu(hdr->header.ucode_version);
+	adev->psp.asd_feature_version = le32_to_cpu(hdr->ucode_feature_version);
+	adev->psp.asd_ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
+	adev->psp.asd_start_addr = (uint8_t *)hdr +
+				le32_to_cpu(hdr->header.ucode_array_offset_bytes);
+
+	return 0;
+out:
+	if (err) {
+		dev_err(adev->dev,
+			"psp v3.1: Failed to load firmware \"%s\"\n",
+			fw_name);
+		release_firmware(adev->psp.sos_fw);
+		adev->psp.sos_fw = NULL;
+		release_firmware(adev->psp.asd_fw);
+		adev->psp.asd_fw = NULL;
+	}
+
+	return err;
+}
+
+static int psp_v3_1_bootloader_load_sysdrv(struct psp_context *psp)
+{
+	int ret;
+	uint32_t psp_gfxdrv_command_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t sol_reg;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	sol_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81);
+	if (sol_reg)
+		return 0;
+
+	/* Wait for bootloader to signify that is ready having bit 31 of C2PMSG_35 set to 1 */
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
+			   0x80000000, 0x80000000, false);
+	if (ret)
+		return ret;
+
+	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+
+	/* Copy PSP System Driver binary to memory */
+	memcpy(psp->fw_pri_buf, psp->sys_start_addr, psp->sys_bin_size);
+
+	/* Provide the sys driver to bootrom */
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
+	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
+	psp_gfxdrv_command_reg = 1 << 16;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35,
+	       psp_gfxdrv_command_reg);
+
+	/* there might be handshake issue with hardware which needs delay */
+	mdelay(20);
+
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
+			   0x80000000, 0x80000000, false);
+
+	return ret;
+}
+
+static bool psp_v3_1_match_version(struct amdgpu_device *adev, uint32_t ver)
+{
+	int i;
+
+	if (ver == adev->psp.sos_fw_version)
+		return true;
+
+	/*
+	 * Double check if the latest four legacy versions.
+	 * If yes, it is still the right version.
+	 */
+	for (i = 0; i < sizeof(sos_old_versions) / sizeof(uint32_t); i++) {
+		if (sos_old_versions[i] == adev->psp.sos_fw_version)
+			return true;
+	}
+
+	return false;
+}
+
+static int psp_v3_1_bootloader_load_sos(struct psp_context *psp)
+{
+	int ret;
+	unsigned int psp_gfxdrv_command_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t sol_reg, ver;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	sol_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81);
+	if (sol_reg)
+		return 0;
+
+	/* Wait for bootloader to signify that is ready having bit 31 of C2PMSG_35 set to 1 */
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
+			   0x80000000, 0x80000000, false);
+	if (ret)
+		return ret;
+
+	memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+
+	/* Copy Secure OS binary to PSP memory */
+	memcpy(psp->fw_pri_buf, psp->sos_start_addr, psp->sos_bin_size);
+
+	/* Provide the PSP secure OS to bootrom */
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
+	       (uint32_t)(psp->fw_pri_mc_addr >> 20));
+	psp_gfxdrv_command_reg = 2 << 16;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35,
+	       psp_gfxdrv_command_reg);
+
+	/* there might be handshake issue with hardware which needs delay */
+	mdelay(20);
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_81),
+			   RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81),
+			   0, true);
+
+	ver = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_58);
+	if (!psp_v3_1_match_version(adev, ver))
+		DRM_WARN("SOS version doesn't match\n");
+
+	return ret;
+}
+
+static int psp_v3_1_prep_cmd_buf(struct amdgpu_firmware_info *ucode,
+				 struct psp_gfx_cmd_resp *cmd)
+{
+	int ret;
+	uint64_t fw_mem_mc_addr = ucode->mc_addr;
+
+	memset(cmd, 0, sizeof(struct psp_gfx_cmd_resp));
+
+	cmd->cmd_id = GFX_CMD_ID_LOAD_IP_FW;
+	cmd->cmd.cmd_load_ip_fw.fw_phy_addr_lo = lower_32_bits(fw_mem_mc_addr);
+	cmd->cmd.cmd_load_ip_fw.fw_phy_addr_hi = upper_32_bits(fw_mem_mc_addr);
+	cmd->cmd.cmd_load_ip_fw.fw_size = ucode->ucode_size;
+
+	ret = psp_v3_1_get_fw_type(ucode, &cmd->cmd.cmd_load_ip_fw.fw_type);
+	if (ret)
+		DRM_ERROR("Unknown firmware type\n");
+
+	return ret;
+}
+
+static int psp_v3_1_ring_init(struct psp_context *psp,
+			      enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct psp_ring *ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	ring = &psp->km_ring;
+
+	ring->ring_type = ring_type;
+
+	/* allocate 4k Page of Local Frame Buffer memory for ring */
+	ring->ring_size = 0x1000;
+	ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM,
+				      &adev->firmware.rbuf,
+				      &ring->ring_mem_mc_addr,
+				      (void **)&ring->ring_mem);
+	if (ret) {
+		ring->ring_size = 0;
+		return ret;
+	}
+
+	return 0;
+}
+
+static int psp_v3_1_ring_create(struct psp_context *psp,
+				enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	unsigned int psp_ring_reg = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	/* Write low address of the ring to C2PMSG_69 */
+	psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_69, psp_ring_reg);
+	/* Write high address of the ring to C2PMSG_70 */
+	psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_70, psp_ring_reg);
+	/* Write size of ring to C2PMSG_71 */
+	psp_ring_reg = ring->ring_size;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_71, psp_ring_reg);
+	/* Write the ring initialization command to C2PMSG_64 */
+	psp_ring_reg = ring_type;
+	psp_ring_reg = psp_ring_reg << 16;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
+
+	/* there might be handshake issue with hardware which needs delay */
+	mdelay(20);
+
+	/* Wait for response flag (bit 31) in C2PMSG_64 */
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			   0x80000000, 0x8000FFFF, false);
+
+	return ret;
+}
+
+static int psp_v3_1_ring_stop(struct psp_context *psp,
+			      enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct psp_ring *ring;
+	unsigned int psp_ring_reg = 0;
+	struct amdgpu_device *adev = psp->adev;
+
+	ring = &psp->km_ring;
+
+	/* Write the ring destroy command to C2PMSG_64 */
+	psp_ring_reg = 3 << 16;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
+
+	/* there might be handshake issue with hardware which needs delay */
+	mdelay(20);
+
+	/* Wait for response flag (bit 31) in C2PMSG_64 */
+	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+			   0x80000000, 0x80000000, false);
+
+	return ret;
+}
+
+static int psp_v3_1_ring_destroy(struct psp_context *psp,
+				 enum psp_ring_type ring_type)
+{
+	int ret = 0;
+	struct psp_ring *ring = &psp->km_ring;
+	struct amdgpu_device *adev = psp->adev;
+
+	ret = psp_v3_1_ring_stop(psp, ring_type);
+	if (ret)
+		DRM_ERROR("Fail to stop psp ring\n");
+
+	amdgpu_bo_free_kernel(&adev->firmware.rbuf,
+			      &ring->ring_mem_mc_addr,
+			      (void **)&ring->ring_mem);
+
+	return ret;
+}
+
+static int psp_v3_1_cmd_submit(struct psp_context *psp,
+			       struct amdgpu_firmware_info *ucode,
+			       uint64_t cmd_buf_mc_addr, uint64_t fence_mc_addr,
+			       int index)
+{
+	unsigned int psp_write_ptr_reg = 0;
+	struct psp_gfx_rb_frame * write_frame = psp->km_ring.ring_mem;
+	struct psp_ring *ring = &psp->km_ring;
+	struct psp_gfx_rb_frame *ring_buffer_start = ring->ring_mem;
+	struct psp_gfx_rb_frame *ring_buffer_end = ring_buffer_start +
+		ring->ring_size / sizeof(struct psp_gfx_rb_frame) - 1;
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t ring_size_dw = ring->ring_size / 4;
+	uint32_t rb_frame_size_dw = sizeof(struct psp_gfx_rb_frame) / 4;
+
+	/* KM (GPCOM) prepare write pointer */
+	psp_write_ptr_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67);
+
+	/* Update KM RB frame pointer to new frame */
+	/* write_frame ptr increments by size of rb_frame in bytes */
+	/* psp_write_ptr_reg increments by size of rb_frame in DWORDs */
+	if ((psp_write_ptr_reg % ring_size_dw) == 0)
+		write_frame = ring_buffer_start;
+	else
+		write_frame = ring_buffer_start + (psp_write_ptr_reg / rb_frame_size_dw);
+	/* Check invalid write_frame ptr address */
+	if ((write_frame < ring_buffer_start) || (ring_buffer_end < write_frame)) {
+		DRM_ERROR("ring_buffer_start = %p; ring_buffer_end = %p; write_frame = %p\n",
+			  ring_buffer_start, ring_buffer_end, write_frame);
+		DRM_ERROR("write_frame is pointing to address out of bounds\n");
+		return -EINVAL;
+	}
+
+	/* Initialize KM RB frame */
+	memset(write_frame, 0, sizeof(struct psp_gfx_rb_frame));
+
+	/* Update KM RB frame */
+	write_frame->cmd_buf_addr_hi = upper_32_bits(cmd_buf_mc_addr);
+	write_frame->cmd_buf_addr_lo = lower_32_bits(cmd_buf_mc_addr);
+	write_frame->fence_addr_hi = upper_32_bits(fence_mc_addr);
+	write_frame->fence_addr_lo = lower_32_bits(fence_mc_addr);
+	write_frame->fence_value = index;
+
+	/* Update the write Pointer in DWORDs */
+	psp_write_ptr_reg = (psp_write_ptr_reg + rb_frame_size_dw) % ring_size_dw;
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67, psp_write_ptr_reg);
+
+	return 0;
+}
+
+static int
+psp_v3_1_sram_map(struct amdgpu_device *adev,
+		  unsigned int *sram_offset, unsigned int *sram_addr_reg_offset,
+		  unsigned int *sram_data_reg_offset,
+		  enum AMDGPU_UCODE_ID ucode_id)
+{
+	int ret = 0;
+
+	switch(ucode_id) {
+/* TODO: needs to confirm */
+#if 0
+	case AMDGPU_UCODE_ID_SMC:
+		*sram_offset = 0;
+		*sram_addr_reg_offset = 0;
+		*sram_data_reg_offset = 0;
+		break;
+#endif
+
+	case AMDGPU_UCODE_ID_CP_CE:
+		*sram_offset = 0x0;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_CE_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_CE_UCODE_DATA);
+		break;
+
+	case AMDGPU_UCODE_ID_CP_PFP:
+		*sram_offset = 0x0;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_PFP_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_PFP_UCODE_DATA);
+		break;
+
+	case AMDGPU_UCODE_ID_CP_ME:
+		*sram_offset = 0x0;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_ME_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_ME_UCODE_DATA);
+		break;
+
+	case AMDGPU_UCODE_ID_CP_MEC1:
+		*sram_offset = 0x10000;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_MEC_ME1_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_MEC_ME1_UCODE_DATA);
+		break;
+
+	case AMDGPU_UCODE_ID_CP_MEC2:
+		*sram_offset = 0x10000;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_MEC2_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmCP_HYP_MEC2_UCODE_DATA);
+		break;
+
+	case AMDGPU_UCODE_ID_RLC_G:
+		*sram_offset = 0x2000;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UCODE_DATA);
+		break;
+
+	case AMDGPU_UCODE_ID_SDMA0:
+		*sram_offset = 0x0;
+		*sram_addr_reg_offset = SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UCODE_ADDR);
+		*sram_data_reg_offset = SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UCODE_DATA);
+		break;
+
+/* TODO: needs to confirm */
+#if 0
+	case AMDGPU_UCODE_ID_SDMA1:
+		*sram_offset = ;
+		*sram_addr_reg_offset = ;
+		break;
+
+	case AMDGPU_UCODE_ID_UVD:
+		*sram_offset = ;
+		*sram_addr_reg_offset = ;
+		break;
+
+	case AMDGPU_UCODE_ID_VCE:
+		*sram_offset = ;
+		*sram_addr_reg_offset = ;
+		break;
+#endif
+
+	case AMDGPU_UCODE_ID_MAXIMUM:
+	default:
+		ret = -EINVAL;
+		break;
+	}
+
+	return ret;
+}
+
+static bool psp_v3_1_compare_sram_data(struct psp_context *psp,
+				       struct amdgpu_firmware_info *ucode,
+				       enum AMDGPU_UCODE_ID ucode_type)
+{
+	int err = 0;
+	unsigned int fw_sram_reg_val = 0;
+	unsigned int fw_sram_addr_reg_offset = 0;
+	unsigned int fw_sram_data_reg_offset = 0;
+	unsigned int ucode_size;
+	uint32_t *ucode_mem = NULL;
+	struct amdgpu_device *adev = psp->adev;
+
+	err = psp_v3_1_sram_map(adev, &fw_sram_reg_val, &fw_sram_addr_reg_offset,
+				&fw_sram_data_reg_offset, ucode_type);
+	if (err)
+		return false;
+
+	WREG32(fw_sram_addr_reg_offset, fw_sram_reg_val);
+
+	ucode_size = ucode->ucode_size;
+	ucode_mem = (uint32_t *)ucode->kaddr;
+	while (ucode_size) {
+		fw_sram_reg_val = RREG32(fw_sram_data_reg_offset);
+
+		if (*ucode_mem != fw_sram_reg_val)
+			return false;
+
+		ucode_mem++;
+		/* 4 bytes */
+		ucode_size -= 4;
+	}
+
+	return true;
+}
+
+static bool psp_v3_1_smu_reload_quirk(struct psp_context *psp)
+{
+	struct amdgpu_device *adev = psp->adev;
+	uint32_t reg;
+
+	reg = smnMP1_FIRMWARE_FLAGS | 0x03b00000;
+	WREG32_SOC15(NBIO, 0, mmPCIE_INDEX2, reg);
+	reg = RREG32_SOC15(NBIO, 0, mmPCIE_DATA2);
+	return (reg & MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED_MASK) ? true : false;
+}
+
+static int psp_v3_1_mode1_reset(struct psp_context *psp)
+{
+	int ret;
+	uint32_t offset;
+	struct amdgpu_device *adev = psp->adev;
+
+	offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64);
+
+	ret = psp_wait_for(psp, offset, 0x80000000, 0x8000FFFF, false);
+
+	if (ret) {
+		DRM_INFO("psp is not working correctly before mode1 reset!\n");
+		return -EINVAL;
+	}
+
+	/*send the mode 1 reset command*/
+	WREG32(offset, 0x70000);
+
+	mdelay(1000);
+
+	offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_33);
+
+	ret = psp_wait_for(psp, offset, 0x80000000, 0x80000000, false);
+
+	if (ret) {
+		DRM_INFO("psp mode 1 reset failed!\n");
+		return -EINVAL;
+	}
+
+	DRM_INFO("psp mode1 reset succeed \n");
+
+	return 0;
+}
+
+static const struct psp_funcs psp_v3_1_funcs = {
+	.init_microcode = psp_v3_1_init_microcode,
+	.bootloader_load_sysdrv = psp_v3_1_bootloader_load_sysdrv,
+	.bootloader_load_sos = psp_v3_1_bootloader_load_sos,
+	.prep_cmd_buf = psp_v3_1_prep_cmd_buf,
+	.ring_init = psp_v3_1_ring_init,
+	.ring_create = psp_v3_1_ring_create,
+	.ring_stop = psp_v3_1_ring_stop,
+	.ring_destroy = psp_v3_1_ring_destroy,
+	.cmd_submit = psp_v3_1_cmd_submit,
+	.compare_sram_data = psp_v3_1_compare_sram_data,
+	.smu_reload_quirk = psp_v3_1_smu_reload_quirk,
+	.mode1_reset = psp_v3_1_mode1_reset,
+};
+
+void psp_v3_1_set_psp_funcs(struct psp_context *psp)
+{
+	psp->funcs = &psp_v3_1_funcs;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/psp_v3_1.h b/sys/dev/pci/drm/amd/amdgpu/psp_v3_1.h
new file mode 100644
index 00000000000..e411e31ba45
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/psp_v3_1.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: Huang Rui
+ *
+ */
+#ifndef __PSP_V3_1_H__
+#define __PSP_V3_1_H__
+
+#include "amdgpu_psp.h"
+
+enum { PSP_DIRECTORY_TABLE_ENTRIES = 4 };
+enum { PSP_BINARY_ALIGNMENT = 64 };
+enum { PSP_BOOTLOADER_1_MEG_ALIGNMENT = 0x100000 };
+enum { PSP_BOOTLOADER_8_MEM_ALIGNMENT = 0x800000 };
+
+void psp_v3_1_set_psp_funcs(struct psp_context *psp);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/r600_dpm.h b/sys/dev/pci/drm/amd/amdgpu/r600_dpm.h
new file mode 100644
index 00000000000..055321f61ca
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/r600_dpm.h
@@ -0,0 +1,127 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __R600_DPM_H__
+#define __R600_DPM_H__
+
+#define R600_ASI_DFLT                                10000
+#define R600_BSP_DFLT                                0x41EB
+#define R600_BSU_DFLT                                0x2
+#define R600_AH_DFLT                                 5
+#define R600_RLP_DFLT                                25
+#define R600_RMP_DFLT                                65
+#define R600_LHP_DFLT                                40
+#define R600_LMP_DFLT                                15
+#define R600_TD_DFLT                                 0
+#define R600_UTC_DFLT_00                             0x24
+#define R600_UTC_DFLT_01                             0x22
+#define R600_UTC_DFLT_02                             0x22
+#define R600_UTC_DFLT_03                             0x22
+#define R600_UTC_DFLT_04                             0x22
+#define R600_UTC_DFLT_05                             0x22
+#define R600_UTC_DFLT_06                             0x22
+#define R600_UTC_DFLT_07                             0x22
+#define R600_UTC_DFLT_08                             0x22
+#define R600_UTC_DFLT_09                             0x22
+#define R600_UTC_DFLT_10                             0x22
+#define R600_UTC_DFLT_11                             0x22
+#define R600_UTC_DFLT_12                             0x22
+#define R600_UTC_DFLT_13                             0x22
+#define R600_UTC_DFLT_14                             0x22
+#define R600_DTC_DFLT_00                             0x24
+#define R600_DTC_DFLT_01                             0x22
+#define R600_DTC_DFLT_02                             0x22
+#define R600_DTC_DFLT_03                             0x22
+#define R600_DTC_DFLT_04                             0x22
+#define R600_DTC_DFLT_05                             0x22
+#define R600_DTC_DFLT_06                             0x22
+#define R600_DTC_DFLT_07                             0x22
+#define R600_DTC_DFLT_08                             0x22
+#define R600_DTC_DFLT_09                             0x22
+#define R600_DTC_DFLT_10                             0x22
+#define R600_DTC_DFLT_11                             0x22
+#define R600_DTC_DFLT_12                             0x22
+#define R600_DTC_DFLT_13                             0x22
+#define R600_DTC_DFLT_14                             0x22
+#define R600_VRC_DFLT                                0x0000C003
+#define R600_VOLTAGERESPONSETIME_DFLT                1000
+#define R600_BACKBIASRESPONSETIME_DFLT               1000
+#define R600_VRU_DFLT                                0x3
+#define R600_SPLLSTEPTIME_DFLT                       0x1000
+#define R600_SPLLSTEPUNIT_DFLT                       0x3
+#define R600_TPU_DFLT                                0
+#define R600_TPC_DFLT                                0x200
+#define R600_SSTU_DFLT                               0
+#define R600_SST_DFLT                                0x00C8
+#define R600_GICST_DFLT                              0x200
+#define R600_FCT_DFLT                                0x0400
+#define R600_FCTU_DFLT                               0
+#define R600_CTXCGTT3DRPHC_DFLT                      0x20
+#define R600_CTXCGTT3DRSDC_DFLT                      0x40
+#define R600_VDDC3DOORPHC_DFLT                       0x100
+#define R600_VDDC3DOORSDC_DFLT                       0x7
+#define R600_VDDC3DOORSU_DFLT                        0
+#define R600_MPLLLOCKTIME_DFLT                       100
+#define R600_MPLLRESETTIME_DFLT                      150
+#define R600_VCOSTEPPCT_DFLT                          20
+#define R600_ENDINGVCOSTEPPCT_DFLT                    5
+#define R600_REFERENCEDIVIDER_DFLT                    4
+
+#define R600_PM_NUMBER_OF_TC 15
+#define R600_PM_NUMBER_OF_SCLKS 20
+#define R600_PM_NUMBER_OF_MCLKS 4
+#define R600_PM_NUMBER_OF_VOLTAGE_LEVELS 4
+#define R600_PM_NUMBER_OF_ACTIVITY_LEVELS 3
+
+/* XXX are these ok? */
+#define R600_TEMP_RANGE_MIN (90 * 1000)
+#define R600_TEMP_RANGE_MAX (120 * 1000)
+
+#define FDO_PWM_MODE_STATIC  1
+#define FDO_PWM_MODE_STATIC_RPM 5
+
+enum r600_power_level {
+	R600_POWER_LEVEL_LOW = 0,
+	R600_POWER_LEVEL_MEDIUM = 1,
+	R600_POWER_LEVEL_HIGH = 2,
+	R600_POWER_LEVEL_CTXSW = 3,
+};
+
+enum r600_td {
+	R600_TD_AUTO,
+	R600_TD_UP,
+	R600_TD_DOWN,
+};
+
+enum r600_display_watermark {
+	R600_DISPLAY_WATERMARK_LOW = 0,
+	R600_DISPLAY_WATERMARK_HIGH = 1,
+};
+
+enum r600_display_gap
+{
+    R600_PM_DISPLAY_GAP_VBLANK_OR_WM = 0,
+    R600_PM_DISPLAY_GAP_VBLANK       = 1,
+    R600_PM_DISPLAY_GAP_WATERMARK    = 2,
+    R600_PM_DISPLAY_GAP_IGNORE       = 3,
+};
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/sdma_v2_4.c b/sys/dev/pci/drm/amd/amdgpu/sdma_v2_4.c
new file mode 100644
index 00000000000..3ebda7f91f5
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/sdma_v2_4.c
@@ -0,0 +1,1334 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_trace.h"
+#include "vi.h"
+#include "vid.h"
+
+#include "oss/oss_2_4_d.h"
+#include "oss/oss_2_4_sh_mask.h"
+
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_enum.h"
+#include "gca/gfx_8_0_sh_mask.h"
+
+#include "bif/bif_5_0_d.h"
+#include "bif/bif_5_0_sh_mask.h"
+
+#include "iceland_sdma_pkt_open.h"
+
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev);
+static void sdma_v2_4_set_buffer_funcs(struct amdgpu_device *adev);
+static void sdma_v2_4_set_vm_pte_funcs(struct amdgpu_device *adev);
+static void sdma_v2_4_set_irq_funcs(struct amdgpu_device *adev);
+
+MODULE_FIRMWARE("amdgpu/topaz_sdma.bin");
+MODULE_FIRMWARE("amdgpu/topaz_sdma1.bin");
+
+static const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
+{
+	SDMA0_REGISTER_OFFSET,
+	SDMA1_REGISTER_OFFSET
+};
+
+static const u32 golden_settings_iceland_a11[] =
+{
+	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
+};
+
+static const u32 iceland_mgcg_cgcg_init[] =
+{
+	mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
+	mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100
+};
+
+/*
+ * sDMA - System DMA
+ * Starting with CIK, the GPU has new asynchronous
+ * DMA engines.  These engines are used for compute
+ * and gfx.  There are two DMA engines (SDMA0, SDMA1)
+ * and each one supports 1 ring buffer used for gfx
+ * and 2 queues used for compute.
+ *
+ * The programming model is very similar to the CP
+ * (ring buffer, IBs, etc.), but sDMA has it's own
+ * packet format that is different from the PM4 format
+ * used by the CP. sDMA supports copying data, writing
+ * embedded data, solid fills, and a number of other
+ * things.  It also has support for tiling/detiling of
+ * buffers.
+ */
+
+static void sdma_v2_4_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		amdgpu_device_program_register_sequence(adev,
+							iceland_mgcg_cgcg_init,
+							ARRAY_SIZE(iceland_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_iceland_a11,
+							ARRAY_SIZE(golden_settings_iceland_a11));
+		break;
+	default:
+		break;
+	}
+}
+
+static void sdma_v2_4_free_microcode(struct amdgpu_device *adev)
+{
+	int i;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		release_firmware(adev->sdma.instance[i].fw);
+		adev->sdma.instance[i].fw = NULL;
+	}
+}
+
+/**
+ * sdma_v2_4_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v2_4_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err = 0, i;
+	struct amdgpu_firmware_info *info = NULL;
+	const struct common_firmware_header *header = NULL;
+	const struct sdma_firmware_header_v1_0 *hdr;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		chip_name = "topaz";
+		break;
+	default: BUG();
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (i == 0)
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma.bin", chip_name);
+		else
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma1.bin", chip_name);
+		err = request_firmware(&adev->sdma.instance[i].fw, fw_name, adev->dev);
+		if (err)
+			goto out;
+		err = amdgpu_ucode_validate(adev->sdma.instance[i].fw);
+		if (err)
+			goto out;
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		adev->sdma.instance[i].fw_version = le32_to_cpu(hdr->header.ucode_version);
+		adev->sdma.instance[i].feature_version = le32_to_cpu(hdr->ucode_feature_version);
+		if (adev->sdma.instance[i].feature_version >= 20)
+			adev->sdma.instance[i].burst_nop = true;
+
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_SMU) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA0 + i];
+			info->ucode_id = AMDGPU_UCODE_ID_SDMA0 + i;
+			info->fw = adev->sdma.instance[i].fw;
+			header = (const struct common_firmware_header *)info->fw->data;
+			adev->firmware.fw_size +=
+				roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+		}
+	}
+
+out:
+	if (err) {
+		pr_err("sdma_v2_4: Failed to load firmware \"%s\"\n", fw_name);
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			release_firmware(adev->sdma.instance[i].fw);
+			adev->sdma.instance[i].fw = NULL;
+		}
+	}
+	return err;
+}
+
+/**
+ * sdma_v2_4_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware (VI+).
+ */
+static uint64_t sdma_v2_4_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	/* XXX check if swapping is necessary on BE */
+	return ring->adev->wb.wb[ring->rptr_offs] >> 2;
+}
+
+/**
+ * sdma_v2_4_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware (VI+).
+ */
+static uint64_t sdma_v2_4_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 wptr = RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >> 2;
+
+	return wptr;
+}
+
+/**
+ * sdma_v2_4_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware (VI+).
+ */
+static void sdma_v2_4_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me], lower_32_bits(ring->wptr) << 2);
+}
+
+static void sdma_v2_4_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			amdgpu_ring_write(ring, ring->funcs->nop |
+				SDMA_PKT_NOP_HEADER_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->funcs->nop);
+}
+
+/**
+ * sdma_v2_4_ring_emit_ib - Schedule an IB on the DMA engine
+ *
+ * @ring: amdgpu ring pointer
+ * @ib: IB object to schedule
+ *
+ * Schedule an IB in the DMA ring (VI).
+ */
+static void sdma_v2_4_ring_emit_ib(struct amdgpu_ring *ring,
+				   struct amdgpu_ib *ib,
+				   unsigned vmid, bool ctx_switch)
+{
+	/* IB packet must end on a 8 DW boundary */
+	sdma_v2_4_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8);
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
+	/* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0);
+
+}
+
+/**
+ * sdma_v2_4_hdp_flush_ring_emit - emit an hdp flush on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Emit an hdp flush packet on the requested DMA ring.
+ */
+static void sdma_v2_4_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	u32 ref_and_mask = 0;
+
+	if (ring->me == 0)
+		ref_and_mask = REG_SET_FIELD(ref_and_mask, GPU_HDP_FLUSH_DONE, SDMA0, 1);
+	else
+		ref_and_mask = REG_SET_FIELD(ref_and_mask, GPU_HDP_FLUSH_DONE, SDMA1, 1);
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(1) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* == */
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_DONE << 2);
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_REQ << 2);
+	amdgpu_ring_write(ring, ref_and_mask); /* reference */
+	amdgpu_ring_write(ring, ref_and_mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
+}
+
+/**
+ * sdma_v2_4_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @fence: amdgpu fence object
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed (VI).
+ */
+static void sdma_v2_4_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				      unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE));
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+
+	/* generate an interrupt */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_TRAP));
+	amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0));
+}
+
+/**
+ * sdma_v2_4_gfx_stop - stop the gfx async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the gfx async dma ring buffers (VI).
+ */
+static void sdma_v2_4_gfx_stop(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *sdma0 = &adev->sdma.instance[0].ring;
+	struct amdgpu_ring *sdma1 = &adev->sdma.instance[1].ring;
+	u32 rb_cntl, ib_cntl;
+	int i;
+
+	if ((adev->mman.buffer_funcs_ring == sdma0) ||
+	    (adev->mman.buffer_funcs_ring == sdma1))
+		amdgpu_ttm_set_buffer_funcs_status(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 0);
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+		ib_cntl = RREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i]);
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0);
+		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
+	}
+	sdma0->ready = false;
+	sdma1->ready = false;
+}
+
+/**
+ * sdma_v2_4_rlc_stop - stop the compute async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute async dma queues (VI).
+ */
+static void sdma_v2_4_rlc_stop(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+}
+
+/**
+ * sdma_v2_4_enable - stop the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs.
+ *
+ * Halt or unhalt the async dma engines (VI).
+ */
+static void sdma_v2_4_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl;
+	int i;
+
+	if (!enable) {
+		sdma_v2_4_gfx_stop(adev);
+		sdma_v2_4_rlc_stop(adev);
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		f32_cntl = RREG32(mmSDMA0_F32_CNTL + sdma_offsets[i]);
+		if (enable)
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, 0);
+		else
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, 1);
+		WREG32(mmSDMA0_F32_CNTL + sdma_offsets[i], f32_cntl);
+	}
+}
+
+/**
+ * sdma_v2_4_gfx_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the gfx DMA ring buffers and enable them (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v2_4_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl, ib_cntl;
+	u32 rb_bufsz;
+	u32 wb_offset;
+	int i, j, r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		wb_offset = (ring->rptr_offs * 4);
+
+		mutex_lock(&adev->srbm_mutex);
+		for (j = 0; j < 16; j++) {
+			vi_srbm_select(adev, 0, 0, 0, j);
+			/* SDMA GFX */
+			WREG32(mmSDMA0_GFX_VIRTUAL_ADDR + sdma_offsets[i], 0);
+			WREG32(mmSDMA0_GFX_APE1_CNTL + sdma_offsets[i], 0);
+		}
+		vi_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		WREG32(mmSDMA0_TILING_CONFIG + sdma_offsets[i],
+		       adev->gfx.config.gb_addr_config & 0x70);
+
+		WREG32(mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i], 0);
+
+		/* Set ring buffer size in dwords */
+		rb_bufsz = order_base_2(ring->ring_size / 4);
+		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz);
+#ifdef __BIG_ENDIAN
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL,
+					RPTR_WRITEBACK_SWAP_ENABLE, 1);
+#endif
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+
+		/* Initialize the ring buffer's read and write pointers */
+		WREG32(mmSDMA0_GFX_RB_RPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_IB_RPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_IB_OFFSET + sdma_offsets[i], 0);
+
+		/* set the wb address whether it's enabled or not */
+		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i],
+		       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
+		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i],
+		       lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC);
+
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
+
+		WREG32(mmSDMA0_GFX_RB_BASE + sdma_offsets[i], ring->gpu_addr >> 8);
+		WREG32(mmSDMA0_GFX_RB_BASE_HI + sdma_offsets[i], ring->gpu_addr >> 40);
+
+		ring->wptr = 0;
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], lower_32_bits(ring->wptr) << 2);
+
+		/* enable DMA RB */
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+
+		ib_cntl = RREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i]);
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1);
+#ifdef __BIG_ENDIAN
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1);
+#endif
+		/* enable DMA IBs */
+		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
+
+		ring->ready = true;
+	}
+
+	sdma_v2_4_enable(adev, true);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		r = amdgpu_ring_test_ring(ring);
+		if (r) {
+			ring->ready = false;
+			return r;
+		}
+
+		if (adev->mman.buffer_funcs_ring == ring)
+			amdgpu_ttm_set_buffer_funcs_status(adev, true);
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v2_4_rlc_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the compute DMA queues and enable them (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v2_4_rlc_resume(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+	return 0;
+}
+
+/**
+ * sdma_v2_4_load_microcode - load the sDMA ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the sDMA0/1 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int sdma_v2_4_load_microcode(struct amdgpu_device *adev)
+{
+	const struct sdma_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+	int i, j;
+
+	/* halt the MEs */
+	sdma_v2_4_enable(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (!adev->sdma.instance[i].fw)
+			return -EINVAL;
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		amdgpu_ucode_print_sdma_hdr(&hdr->header);
+		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+		fw_data = (const __le32 *)
+			(adev->sdma.instance[i].fw->data +
+			 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], 0);
+		for (j = 0; j < fw_size; j++)
+			WREG32(mmSDMA0_UCODE_DATA + sdma_offsets[i], le32_to_cpup(fw_data++));
+		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], adev->sdma.instance[i].fw_version);
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v2_4_start - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the DMA engines and enable them (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v2_4_start(struct amdgpu_device *adev)
+{
+	int r;
+
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		r = sdma_v2_4_load_microcode(adev);
+		if (r)
+			return r;
+	}
+
+	/* halt the engine before programing */
+	sdma_v2_4_enable(adev, false);
+
+	/* start the gfx rings and rlc compute queues */
+	r = sdma_v2_4_gfx_resume(adev);
+	if (r)
+		return r;
+	r = sdma_v2_4_rlc_resume(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * sdma_v2_4_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory. (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v2_4_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_alloc(ring, 5);
+	if (r) {
+		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_device_wb_free(adev, index);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(1));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_device_wb_free(adev, index);
+
+	return r;
+}
+
+/**
+ * sdma_v2_4_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test a simple IB in the DMA ring (VI).
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned index;
+	u32 tmp = 0;
+	u64 gpu_addr;
+	long r;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, 256, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err0;
+	}
+
+	ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr);
+	ib.ptr[3] = SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(1);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.ptr[5] = SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[6] = SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[7] = SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+	ib.length_dw = 8;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err1;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
+		goto err1;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto err1;
+	}
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+	if (tmp == 0xDEADBEEF) {
+		DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
+	} else {
+		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+		r = -EINVAL;
+	}
+
+err1:
+	amdgpu_ib_free(adev, &ib, NULL);
+	dma_fence_put(f);
+err0:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+/**
+ * sdma_v2_4_vm_copy_pte - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using sDMA (CIK).
+ */
+static void sdma_v2_4_vm_copy_pte(struct amdgpu_ib *ib,
+				  uint64_t pe, uint64_t src,
+				  unsigned count)
+{
+	unsigned bytes = count * 8;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+	ib->ptr[ib->length_dw++] = bytes;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+}
+
+/**
+ * sdma_v2_4_vm_write_pte - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @value: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ *
+ * Update PTEs by writing them manually using sDMA (CIK).
+ */
+static void sdma_v2_4_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
+				   uint64_t value, unsigned count,
+				   uint32_t incr)
+{
+	unsigned ndw = count * 2;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib->ptr[ib->length_dw++] = pe;
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = ndw;
+	for (; ndw > 0; ndw -= 2) {
+		ib->ptr[ib->length_dw++] = lower_32_bits(value);
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		value += incr;
+	}
+}
+
+/**
+ * sdma_v2_4_vm_set_pte_pde - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA (CIK).
+ */
+static void sdma_v2_4_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe,
+				     uint64_t addr, unsigned count,
+				     uint32_t incr, uint64_t flags)
+{
+	/* for physically contiguous pages (vram) */
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_GEN_PTEPDE);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = lower_32_bits(flags); /* mask */
+	ib->ptr[ib->length_dw++] = upper_32_bits(flags);
+	ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = incr; /* increment size */
+	ib->ptr[ib->length_dw++] = 0;
+	ib->ptr[ib->length_dw++] = count; /* number of entries */
+}
+
+/**
+ * sdma_v2_4_ring_pad_ib - pad the IB to the required number of dw
+ *
+ * @ib: indirect buffer to fill with padding
+ *
+ */
+static void sdma_v2_4_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	u32 pad_count;
+	int i;
+
+	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	for (i = 0; i < pad_count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP) |
+				SDMA_PKT_NOP_HEADER_COUNT(pad_count - 1);
+		else
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+}
+
+/**
+ * sdma_v2_4_ring_emit_pipeline_sync - sync the pipeline
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Make sure all previous operations are completed (CIK).
+ */
+static void sdma_v2_4_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	/* wait for idle */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3) | /* equal */
+			  SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(1));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+	amdgpu_ring_write(ring, seq); /* reference */
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(4)); /* retry count, poll interval */
+}
+
+/**
+ * sdma_v2_4_ring_emit_vm_flush - cik vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vm: amdgpu_vm pointer
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA (VI).
+ */
+static void sdma_v2_4_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for flush */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(0)); /* always */
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST << 2);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0); /* reference */
+	amdgpu_ring_write(ring, 0); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
+}
+
+static void sdma_v2_4_ring_emit_wreg(struct amdgpu_ring *ring,
+				     uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, val);
+}
+
+static int sdma_v2_4_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->sdma.num_instances = SDMA_MAX_INSTANCE;
+
+	sdma_v2_4_set_ring_funcs(adev);
+	sdma_v2_4_set_buffer_funcs(adev);
+	sdma_v2_4_set_vm_pte_funcs(adev);
+	sdma_v2_4_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int sdma_v2_4_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* SDMA trap event */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_TRAP,
+			      &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	/* SDMA Privileged inst */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 241,
+			      &adev->sdma.illegal_inst_irq);
+	if (r)
+		return r;
+
+	/* SDMA Privileged inst */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_SRBM_WRITE,
+			      &adev->sdma.illegal_inst_irq);
+	if (r)
+		return r;
+
+	r = sdma_v2_4_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load sdma firmware!\n");
+		return r;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		ring->use_doorbell = false;
+		snprintf(ring->name, sizeof(ring->name), "sdma%d", i);
+		r = amdgpu_ring_init(adev, ring, 1024,
+				     &adev->sdma.trap_irq,
+				     (i == 0) ?
+				     AMDGPU_SDMA_IRQ_TRAP0 :
+				     AMDGPU_SDMA_IRQ_TRAP1);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+static int sdma_v2_4_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+
+	sdma_v2_4_free_microcode(adev);
+	return 0;
+}
+
+static int sdma_v2_4_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	sdma_v2_4_init_golden_registers(adev);
+
+	r = sdma_v2_4_start(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int sdma_v2_4_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	sdma_v2_4_enable(adev, false);
+
+	return 0;
+}
+
+static int sdma_v2_4_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return sdma_v2_4_hw_fini(adev);
+}
+
+static int sdma_v2_4_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return sdma_v2_4_hw_init(adev);
+}
+
+static bool sdma_v2_4_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS2);
+
+	if (tmp & (SRBM_STATUS2__SDMA_BUSY_MASK |
+		   SRBM_STATUS2__SDMA1_BUSY_MASK))
+	    return false;
+
+	return true;
+}
+
+static int sdma_v2_4_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmSRBM_STATUS2) & (SRBM_STATUS2__SDMA_BUSY_MASK |
+				SRBM_STATUS2__SDMA1_BUSY_MASK);
+
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int sdma_v2_4_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS2);
+
+	if (tmp & SRBM_STATUS2__SDMA_BUSY_MASK) {
+		/* sdma0 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET);
+		tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 0);
+		WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp);
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK;
+	}
+	if (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK) {
+		/* sdma1 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET);
+		tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 0);
+		WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp);
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK;
+	}
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int sdma_v2_4_set_trap_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *src,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	switch (type) {
+	case AMDGPU_SDMA_IRQ_TRAP0:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 0);
+			WREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 1);
+			WREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	case AMDGPU_SDMA_IRQ_TRAP1:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 0);
+			WREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 1);
+			WREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v2_4_process_trap_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	u8 instance_id, queue_id;
+
+	instance_id = (entry->ring_id & 0x3) >> 0;
+	queue_id = (entry->ring_id & 0xc) >> 2;
+	DRM_DEBUG("IH: SDMA trap\n");
+	switch (instance_id) {
+	case 0:
+		switch (queue_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[0].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		}
+		break;
+	case 1:
+		switch (queue_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[1].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		}
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v2_4_process_illegal_inst_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in SDMA command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int sdma_v2_4_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	/* XXX handled via the smc on VI */
+	return 0;
+}
+
+static int sdma_v2_4_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs sdma_v2_4_ip_funcs = {
+	.name = "sdma_v2_4",
+	.early_init = sdma_v2_4_early_init,
+	.late_init = NULL,
+	.sw_init = sdma_v2_4_sw_init,
+	.sw_fini = sdma_v2_4_sw_fini,
+	.hw_init = sdma_v2_4_hw_init,
+	.hw_fini = sdma_v2_4_hw_fini,
+	.suspend = sdma_v2_4_suspend,
+	.resume = sdma_v2_4_resume,
+	.is_idle = sdma_v2_4_is_idle,
+	.wait_for_idle = sdma_v2_4_wait_for_idle,
+	.soft_reset = sdma_v2_4_soft_reset,
+	.set_clockgating_state = sdma_v2_4_set_clockgating_state,
+	.set_powergating_state = sdma_v2_4_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs sdma_v2_4_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xf,
+	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
+	.support_64bit_ptrs = false,
+	.get_rptr = sdma_v2_4_ring_get_rptr,
+	.get_wptr = sdma_v2_4_ring_get_wptr,
+	.set_wptr = sdma_v2_4_ring_set_wptr,
+	.emit_frame_size =
+		6 + /* sdma_v2_4_ring_emit_hdp_flush */
+		3 + /* hdp invalidate */
+		6 + /* sdma_v2_4_ring_emit_pipeline_sync */
+		VI_FLUSH_GPU_TLB_NUM_WREG * 3 + 6 + /* sdma_v2_4_ring_emit_vm_flush */
+		10 + 10 + 10, /* sdma_v2_4_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 7 + 6, /* sdma_v2_4_ring_emit_ib */
+	.emit_ib = sdma_v2_4_ring_emit_ib,
+	.emit_fence = sdma_v2_4_ring_emit_fence,
+	.emit_pipeline_sync = sdma_v2_4_ring_emit_pipeline_sync,
+	.emit_vm_flush = sdma_v2_4_ring_emit_vm_flush,
+	.emit_hdp_flush = sdma_v2_4_ring_emit_hdp_flush,
+	.test_ring = sdma_v2_4_ring_test_ring,
+	.test_ib = sdma_v2_4_ring_test_ib,
+	.insert_nop = sdma_v2_4_ring_insert_nop,
+	.pad_ib = sdma_v2_4_ring_pad_ib,
+	.emit_wreg = sdma_v2_4_ring_emit_wreg,
+};
+
+static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->sdma.instance[i].ring.funcs = &sdma_v2_4_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs sdma_v2_4_trap_irq_funcs = {
+	.set = sdma_v2_4_set_trap_irq_state,
+	.process = sdma_v2_4_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v2_4_illegal_inst_irq_funcs = {
+	.process = sdma_v2_4_process_illegal_inst_irq,
+};
+
+static void sdma_v2_4_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_LAST;
+	adev->sdma.trap_irq.funcs = &sdma_v2_4_trap_irq_funcs;
+	adev->sdma.illegal_inst_irq.funcs = &sdma_v2_4_illegal_inst_irq_funcs;
+}
+
+/**
+ * sdma_v2_4_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Copy GPU buffers using the DMA engine (VI).
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void sdma_v2_4_emit_copy_buffer(struct amdgpu_ib *ib,
+				       uint64_t src_offset,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+	ib->ptr[ib->length_dw++] = byte_count;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+}
+
+/**
+ * sdma_v2_4_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine (VI).
+ */
+static void sdma_v2_4_emit_fill_buffer(struct amdgpu_ib *ib,
+				       uint32_t src_data,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_CONST_FILL);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = byte_count;
+}
+
+static const struct amdgpu_buffer_funcs sdma_v2_4_buffer_funcs = {
+	.copy_max_bytes = 0x1fffff,
+	.copy_num_dw = 7,
+	.emit_copy_buffer = sdma_v2_4_emit_copy_buffer,
+
+	.fill_max_bytes = 0x1fffff,
+	.fill_num_dw = 7,
+	.emit_fill_buffer = sdma_v2_4_emit_fill_buffer,
+};
+
+static void sdma_v2_4_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mman.buffer_funcs == NULL) {
+		adev->mman.buffer_funcs = &sdma_v2_4_buffer_funcs;
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+	}
+}
+
+static const struct amdgpu_vm_pte_funcs sdma_v2_4_vm_pte_funcs = {
+	.copy_pte_num_dw = 7,
+	.copy_pte = sdma_v2_4_vm_copy_pte,
+
+	.write_pte = sdma_v2_4_vm_write_pte,
+	.set_pte_pde = sdma_v2_4_vm_set_pte_pde,
+};
+
+static void sdma_v2_4_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	if (adev->vm_manager.vm_pte_funcs == NULL) {
+		adev->vm_manager.vm_pte_funcs = &sdma_v2_4_vm_pte_funcs;
+		for (i = 0; i < adev->sdma.num_instances; i++)
+			adev->vm_manager.vm_pte_rings[i] =
+				&adev->sdma.instance[i].ring;
+
+		adev->vm_manager.vm_pte_num_rings = adev->sdma.num_instances;
+	}
+}
+
+const struct amdgpu_ip_block_version sdma_v2_4_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_SDMA,
+	.major = 2,
+	.minor = 4,
+	.rev = 0,
+	.funcs = &sdma_v2_4_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/sdma_v2_4.h b/sys/dev/pci/drm/amd/amdgpu/sdma_v2_4.h
new file mode 100644
index 00000000000..28b43372921
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/sdma_v2_4.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SDMA_V2_4_H__
+#define __SDMA_V2_4_H__
+
+extern const struct amdgpu_ip_block_version sdma_v2_4_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/sdma_v3_0.c b/sys/dev/pci/drm/amd/amdgpu/sdma_v3_0.c
new file mode 100644
index 00000000000..0189627cb69
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/sdma_v3_0.c
@@ -0,0 +1,1783 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_trace.h"
+#include "vi.h"
+#include "vid.h"
+
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+
+#include "gmc/gmc_8_1_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_enum.h"
+#include "gca/gfx_8_0_sh_mask.h"
+
+#include "bif/bif_5_0_d.h"
+#include "bif/bif_5_0_sh_mask.h"
+
+#include "tonga_sdma_pkt_open.h"
+
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev);
+static void sdma_v3_0_set_buffer_funcs(struct amdgpu_device *adev);
+static void sdma_v3_0_set_vm_pte_funcs(struct amdgpu_device *adev);
+static void sdma_v3_0_set_irq_funcs(struct amdgpu_device *adev);
+
+MODULE_FIRMWARE("amdgpu/tonga_sdma.bin");
+MODULE_FIRMWARE("amdgpu/tonga_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_sdma.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/fiji_sdma.bin");
+MODULE_FIRMWARE("amdgpu/fiji_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/stoney_sdma.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_sdma.bin");
+MODULE_FIRMWARE("amdgpu/polaris10_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_sdma.bin");
+MODULE_FIRMWARE("amdgpu/polaris11_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_sdma.bin");
+MODULE_FIRMWARE("amdgpu/polaris12_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/vegam_sdma.bin");
+MODULE_FIRMWARE("amdgpu/vegam_sdma1.bin");
+
+
+static const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
+{
+	SDMA0_REGISTER_OFFSET,
+	SDMA1_REGISTER_OFFSET
+};
+
+static const u32 golden_settings_tonga_a11[] =
+{
+	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA0_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA1_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+};
+
+static const u32 tonga_mgcg_cgcg_init[] =
+{
+	mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
+	mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100
+};
+
+static const u32 golden_settings_fiji_a10[] =
+{
+	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA0_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA1_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+};
+
+static const u32 fiji_mgcg_cgcg_init[] =
+{
+	mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
+	mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100
+};
+
+static const u32 golden_settings_polaris11_a11[] =
+{
+	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA0_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA1_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+};
+
+static const u32 golden_settings_polaris10_a11[] =
+{
+	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA0_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA1_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+};
+
+static const u32 cz_golden_settings_a11[] =
+{
+	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA0_GFX_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA0_POWER_CNTL, 0x00000800, 0x0003c800,
+	mmSDMA0_RLC0_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA0_RLC1_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA1_GFX_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA1_POWER_CNTL, 0x00000800, 0x0003c800,
+	mmSDMA1_RLC0_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA1_RLC1_IB_CNTL, 0x00000100, 0x00000100,
+};
+
+static const u32 cz_mgcg_cgcg_init[] =
+{
+	mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
+	mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100
+};
+
+static const u32 stoney_golden_settings_a11[] =
+{
+	mmSDMA0_GFX_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA0_POWER_CNTL, 0x00000800, 0x0003c800,
+	mmSDMA0_RLC0_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA0_RLC1_IB_CNTL, 0x00000100, 0x00000100,
+};
+
+static const u32 stoney_mgcg_cgcg_init[] =
+{
+	mmSDMA0_CLK_CTRL, 0xffffffff, 0x00000100,
+};
+
+/*
+ * sDMA - System DMA
+ * Starting with CIK, the GPU has new asynchronous
+ * DMA engines.  These engines are used for compute
+ * and gfx.  There are two DMA engines (SDMA0, SDMA1)
+ * and each one supports 1 ring buffer used for gfx
+ * and 2 queues used for compute.
+ *
+ * The programming model is very similar to the CP
+ * (ring buffer, IBs, etc.), but sDMA has it's own
+ * packet format that is different from the PM4 format
+ * used by the CP. sDMA supports copying data, writing
+ * embedded data, solid fills, and a number of other
+ * things.  It also has support for tiling/detiling of
+ * buffers.
+ */
+
+static void sdma_v3_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+		amdgpu_device_program_register_sequence(adev,
+							fiji_mgcg_cgcg_init,
+							ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_fiji_a10,
+							ARRAY_SIZE(golden_settings_fiji_a10));
+		break;
+	case CHIP_TONGA:
+		amdgpu_device_program_register_sequence(adev,
+							tonga_mgcg_cgcg_init,
+							ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_tonga_a11,
+							ARRAY_SIZE(golden_settings_tonga_a11));
+		break;
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_polaris11_a11,
+							ARRAY_SIZE(golden_settings_polaris11_a11));
+		break;
+	case CHIP_POLARIS10:
+		amdgpu_device_program_register_sequence(adev,
+							golden_settings_polaris10_a11,
+							ARRAY_SIZE(golden_settings_polaris10_a11));
+		break;
+	case CHIP_CARRIZO:
+		amdgpu_device_program_register_sequence(adev,
+							cz_mgcg_cgcg_init,
+							ARRAY_SIZE(cz_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							cz_golden_settings_a11,
+							ARRAY_SIZE(cz_golden_settings_a11));
+		break;
+	case CHIP_STONEY:
+		amdgpu_device_program_register_sequence(adev,
+							stoney_mgcg_cgcg_init,
+							ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		amdgpu_device_program_register_sequence(adev,
+							stoney_golden_settings_a11,
+							ARRAY_SIZE(stoney_golden_settings_a11));
+		break;
+	default:
+		break;
+	}
+}
+
+static void sdma_v3_0_free_microcode(struct amdgpu_device *adev)
+{
+	int i;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		release_firmware(adev->sdma.instance[i].fw);
+		adev->sdma.instance[i].fw = NULL;
+	}
+}
+
+/**
+ * sdma_v3_0_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v3_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err = 0, i;
+	struct amdgpu_firmware_info *info = NULL;
+	const struct common_firmware_header *header = NULL;
+	const struct sdma_firmware_header_v1_0 *hdr;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_TONGA:
+		chip_name = "tonga";
+		break;
+	case CHIP_FIJI:
+		chip_name = "fiji";
+		break;
+	case CHIP_POLARIS10:
+		chip_name = "polaris10";
+		break;
+	case CHIP_POLARIS11:
+		chip_name = "polaris11";
+		break;
+	case CHIP_POLARIS12:
+		chip_name = "polaris12";
+		break;
+	case CHIP_VEGAM:
+		chip_name = "vegam";
+		break;
+	case CHIP_CARRIZO:
+		chip_name = "carrizo";
+		break;
+	case CHIP_STONEY:
+		chip_name = "stoney";
+		break;
+	default: BUG();
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (i == 0)
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma.bin", chip_name);
+		else
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma1.bin", chip_name);
+		err = request_firmware(&adev->sdma.instance[i].fw, fw_name, adev->dev);
+		if (err)
+			goto out;
+		err = amdgpu_ucode_validate(adev->sdma.instance[i].fw);
+		if (err)
+			goto out;
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		adev->sdma.instance[i].fw_version = le32_to_cpu(hdr->header.ucode_version);
+		adev->sdma.instance[i].feature_version = le32_to_cpu(hdr->ucode_feature_version);
+		if (adev->sdma.instance[i].feature_version >= 20)
+			adev->sdma.instance[i].burst_nop = true;
+
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_SMU) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA0 + i];
+			info->ucode_id = AMDGPU_UCODE_ID_SDMA0 + i;
+			info->fw = adev->sdma.instance[i].fw;
+			header = (const struct common_firmware_header *)info->fw->data;
+			adev->firmware.fw_size +=
+				roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+		}
+	}
+out:
+	if (err) {
+		pr_err("sdma_v3_0: Failed to load firmware \"%s\"\n", fw_name);
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			release_firmware(adev->sdma.instance[i].fw);
+			adev->sdma.instance[i].fw = NULL;
+		}
+	}
+	return err;
+}
+
+/**
+ * sdma_v3_0_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware (VI+).
+ */
+static uint64_t sdma_v3_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	/* XXX check if swapping is necessary on BE */
+	return ring->adev->wb.wb[ring->rptr_offs] >> 2;
+}
+
+/**
+ * sdma_v3_0_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware (VI+).
+ */
+static uint64_t sdma_v3_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 wptr;
+
+	if (ring->use_doorbell || ring->use_pollmem) {
+		/* XXX check if swapping is necessary on BE */
+		wptr = ring->adev->wb.wb[ring->wptr_offs] >> 2;
+	} else {
+		wptr = RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >> 2;
+	}
+
+	return wptr;
+}
+
+/**
+ * sdma_v3_0_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware (VI+).
+ */
+static void sdma_v3_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		u32 *wb = (u32 *)&adev->wb.wb[ring->wptr_offs];
+		/* XXX check if swapping is necessary on BE */
+		WRITE_ONCE(*wb, (lower_32_bits(ring->wptr) << 2));
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr) << 2);
+	} else if (ring->use_pollmem) {
+		u32 *wb = (u32 *)&adev->wb.wb[ring->wptr_offs];
+
+		WRITE_ONCE(*wb, (lower_32_bits(ring->wptr) << 2));
+	} else {
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me], lower_32_bits(ring->wptr) << 2);
+	}
+}
+
+static void sdma_v3_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			amdgpu_ring_write(ring, ring->funcs->nop |
+				SDMA_PKT_NOP_HEADER_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->funcs->nop);
+}
+
+/**
+ * sdma_v3_0_ring_emit_ib - Schedule an IB on the DMA engine
+ *
+ * @ring: amdgpu ring pointer
+ * @ib: IB object to schedule
+ *
+ * Schedule an IB in the DMA ring (VI).
+ */
+static void sdma_v3_0_ring_emit_ib(struct amdgpu_ring *ring,
+				   struct amdgpu_ib *ib,
+				   unsigned vmid, bool ctx_switch)
+{
+	/* IB packet must end on a 8 DW boundary */
+	sdma_v3_0_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8);
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
+	/* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0);
+
+}
+
+/**
+ * sdma_v3_0_ring_emit_hdp_flush - emit an hdp flush on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Emit an hdp flush packet on the requested DMA ring.
+ */
+static void sdma_v3_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	u32 ref_and_mask = 0;
+
+	if (ring->me == 0)
+		ref_and_mask = REG_SET_FIELD(ref_and_mask, GPU_HDP_FLUSH_DONE, SDMA0, 1);
+	else
+		ref_and_mask = REG_SET_FIELD(ref_and_mask, GPU_HDP_FLUSH_DONE, SDMA1, 1);
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(1) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* == */
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_DONE << 2);
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_REQ << 2);
+	amdgpu_ring_write(ring, ref_and_mask); /* reference */
+	amdgpu_ring_write(ring, ref_and_mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
+}
+
+/**
+ * sdma_v3_0_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @fence: amdgpu fence object
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed (VI).
+ */
+static void sdma_v3_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				      unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE));
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+
+	/* generate an interrupt */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_TRAP));
+	amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0));
+}
+
+/**
+ * sdma_v3_0_gfx_stop - stop the gfx async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the gfx async dma ring buffers (VI).
+ */
+static void sdma_v3_0_gfx_stop(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *sdma0 = &adev->sdma.instance[0].ring;
+	struct amdgpu_ring *sdma1 = &adev->sdma.instance[1].ring;
+	u32 rb_cntl, ib_cntl;
+	int i;
+
+	if ((adev->mman.buffer_funcs_ring == sdma0) ||
+	    (adev->mman.buffer_funcs_ring == sdma1))
+		amdgpu_ttm_set_buffer_funcs_status(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 0);
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+		ib_cntl = RREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i]);
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0);
+		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
+	}
+	sdma0->ready = false;
+	sdma1->ready = false;
+}
+
+/**
+ * sdma_v3_0_rlc_stop - stop the compute async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute async dma queues (VI).
+ */
+static void sdma_v3_0_rlc_stop(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+}
+
+/**
+ * sdma_v3_0_ctx_switch_enable - stop the async dma engines context switch
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs context switch.
+ *
+ * Halt or unhalt the async dma engines context switch (VI).
+ */
+static void sdma_v3_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl, phase_quantum = 0;
+	int i;
+
+	if (amdgpu_sdma_phase_quantum) {
+		unsigned value = amdgpu_sdma_phase_quantum;
+		unsigned unit = 0;
+
+		while (value > (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
+				SDMA0_PHASE0_QUANTUM__VALUE__SHIFT)) {
+			value = (value + 1) >> 1;
+			unit++;
+		}
+		if (unit > (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
+			    SDMA0_PHASE0_QUANTUM__UNIT__SHIFT)) {
+			value = (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
+				 SDMA0_PHASE0_QUANTUM__VALUE__SHIFT);
+			unit = (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
+				SDMA0_PHASE0_QUANTUM__UNIT__SHIFT);
+			WARN_ONCE(1,
+			"clamping sdma_phase_quantum to %uK clock cycles\n",
+				  value << unit);
+		}
+		phase_quantum =
+			value << SDMA0_PHASE0_QUANTUM__VALUE__SHIFT |
+			unit  << SDMA0_PHASE0_QUANTUM__UNIT__SHIFT;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		f32_cntl = RREG32(mmSDMA0_CNTL + sdma_offsets[i]);
+		if (enable) {
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+					AUTO_CTXSW_ENABLE, 1);
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+					ATC_L1_ENABLE, 1);
+			if (amdgpu_sdma_phase_quantum) {
+				WREG32(mmSDMA0_PHASE0_QUANTUM + sdma_offsets[i],
+				       phase_quantum);
+				WREG32(mmSDMA0_PHASE1_QUANTUM + sdma_offsets[i],
+				       phase_quantum);
+			}
+		} else {
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+					AUTO_CTXSW_ENABLE, 0);
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+					ATC_L1_ENABLE, 1);
+		}
+
+		WREG32(mmSDMA0_CNTL + sdma_offsets[i], f32_cntl);
+	}
+}
+
+/**
+ * sdma_v3_0_enable - stop the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs.
+ *
+ * Halt or unhalt the async dma engines (VI).
+ */
+static void sdma_v3_0_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl;
+	int i;
+
+	if (!enable) {
+		sdma_v3_0_gfx_stop(adev);
+		sdma_v3_0_rlc_stop(adev);
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		f32_cntl = RREG32(mmSDMA0_F32_CNTL + sdma_offsets[i]);
+		if (enable)
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, 0);
+		else
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, 1);
+		WREG32(mmSDMA0_F32_CNTL + sdma_offsets[i], f32_cntl);
+	}
+}
+
+/**
+ * sdma_v3_0_gfx_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the gfx DMA ring buffers and enable them (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v3_0_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl, ib_cntl, wptr_poll_cntl;
+	u32 rb_bufsz;
+	u32 wb_offset;
+	u32 doorbell;
+	u64 wptr_gpu_addr;
+	int i, j, r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		amdgpu_ring_clear_ring(ring);
+		wb_offset = (ring->rptr_offs * 4);
+
+		mutex_lock(&adev->srbm_mutex);
+		for (j = 0; j < 16; j++) {
+			vi_srbm_select(adev, 0, 0, 0, j);
+			/* SDMA GFX */
+			WREG32(mmSDMA0_GFX_VIRTUAL_ADDR + sdma_offsets[i], 0);
+			WREG32(mmSDMA0_GFX_APE1_CNTL + sdma_offsets[i], 0);
+		}
+		vi_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		WREG32(mmSDMA0_TILING_CONFIG + sdma_offsets[i],
+		       adev->gfx.config.gb_addr_config & 0x70);
+
+		WREG32(mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i], 0);
+
+		/* Set ring buffer size in dwords */
+		rb_bufsz = order_base_2(ring->ring_size / 4);
+		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz);
+#ifdef __BIG_ENDIAN
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL,
+					RPTR_WRITEBACK_SWAP_ENABLE, 1);
+#endif
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+
+		/* Initialize the ring buffer's read and write pointers */
+		ring->wptr = 0;
+		WREG32(mmSDMA0_GFX_RB_RPTR + sdma_offsets[i], 0);
+		sdma_v3_0_ring_set_wptr(ring);
+		WREG32(mmSDMA0_GFX_IB_RPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_IB_OFFSET + sdma_offsets[i], 0);
+
+		/* set the wb address whether it's enabled or not */
+		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i],
+		       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
+		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i],
+		       lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC);
+
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
+
+		WREG32(mmSDMA0_GFX_RB_BASE + sdma_offsets[i], ring->gpu_addr >> 8);
+		WREG32(mmSDMA0_GFX_RB_BASE_HI + sdma_offsets[i], ring->gpu_addr >> 40);
+
+		doorbell = RREG32(mmSDMA0_GFX_DOORBELL + sdma_offsets[i]);
+
+		if (ring->use_doorbell) {
+			doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL,
+						 OFFSET, ring->doorbell_index);
+			doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 1);
+		} else {
+			doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 0);
+		}
+		WREG32(mmSDMA0_GFX_DOORBELL + sdma_offsets[i], doorbell);
+
+		/* setup the wptr shadow polling */
+		wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+
+		WREG32(mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO + sdma_offsets[i],
+		       lower_32_bits(wptr_gpu_addr));
+		WREG32(mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI + sdma_offsets[i],
+		       upper_32_bits(wptr_gpu_addr));
+		wptr_poll_cntl = RREG32(mmSDMA0_GFX_RB_WPTR_POLL_CNTL + sdma_offsets[i]);
+		if (ring->use_pollmem) {
+			/*wptr polling is not enogh fast, directly clean the wptr register */
+			WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], 0);
+			wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl,
+						       SDMA0_GFX_RB_WPTR_POLL_CNTL,
+						       ENABLE, 1);
+		} else {
+			wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl,
+						       SDMA0_GFX_RB_WPTR_POLL_CNTL,
+						       ENABLE, 0);
+		}
+		WREG32(mmSDMA0_GFX_RB_WPTR_POLL_CNTL + sdma_offsets[i], wptr_poll_cntl);
+
+		/* enable DMA RB */
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+
+		ib_cntl = RREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i]);
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1);
+#ifdef __BIG_ENDIAN
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1);
+#endif
+		/* enable DMA IBs */
+		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
+
+		ring->ready = true;
+	}
+
+	/* unhalt the MEs */
+	sdma_v3_0_enable(adev, true);
+	/* enable sdma ring preemption */
+	sdma_v3_0_ctx_switch_enable(adev, true);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		r = amdgpu_ring_test_ring(ring);
+		if (r) {
+			ring->ready = false;
+			return r;
+		}
+
+		if (adev->mman.buffer_funcs_ring == ring)
+			amdgpu_ttm_set_buffer_funcs_status(adev, true);
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v3_0_rlc_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the compute DMA queues and enable them (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v3_0_rlc_resume(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+	return 0;
+}
+
+/**
+ * sdma_v3_0_load_microcode - load the sDMA ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the sDMA0/1 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int sdma_v3_0_load_microcode(struct amdgpu_device *adev)
+{
+	const struct sdma_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+	int i, j;
+
+	/* halt the MEs */
+	sdma_v3_0_enable(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (!adev->sdma.instance[i].fw)
+			return -EINVAL;
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		amdgpu_ucode_print_sdma_hdr(&hdr->header);
+		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+		fw_data = (const __le32 *)
+			(adev->sdma.instance[i].fw->data +
+				le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], 0);
+		for (j = 0; j < fw_size; j++)
+			WREG32(mmSDMA0_UCODE_DATA + sdma_offsets[i], le32_to_cpup(fw_data++));
+		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], adev->sdma.instance[i].fw_version);
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v3_0_start - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the DMA engines and enable them (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v3_0_start(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
+		r = sdma_v3_0_load_microcode(adev);
+		if (r)
+			return r;
+	}
+
+	/* disable sdma engine before programing it */
+	sdma_v3_0_ctx_switch_enable(adev, false);
+	sdma_v3_0_enable(adev, false);
+
+	/* start the gfx rings and rlc compute queues */
+	r = sdma_v3_0_gfx_resume(adev);
+	if (r)
+		return r;
+	r = sdma_v3_0_rlc_resume(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * sdma_v3_0_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory. (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v3_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_alloc(ring, 5);
+	if (r) {
+		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_device_wb_free(adev, index);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(1));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_device_wb_free(adev, index);
+
+	return r;
+}
+
+/**
+ * sdma_v3_0_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test a simple IB in the DMA ring (VI).
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned index;
+	u32 tmp = 0;
+	u64 gpu_addr;
+	long r;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, 256, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err0;
+	}
+
+	ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr);
+	ib.ptr[3] = SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(1);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.ptr[5] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[6] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[7] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.length_dw = 8;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err1;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
+		goto err1;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto err1;
+	}
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+	if (tmp == 0xDEADBEEF) {
+		DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
+	} else {
+		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+		r = -EINVAL;
+	}
+err1:
+	amdgpu_ib_free(adev, &ib, NULL);
+	dma_fence_put(f);
+err0:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+/**
+ * sdma_v3_0_vm_copy_pte - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using sDMA (CIK).
+ */
+static void sdma_v3_0_vm_copy_pte(struct amdgpu_ib *ib,
+				  uint64_t pe, uint64_t src,
+				  unsigned count)
+{
+	unsigned bytes = count * 8;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+	ib->ptr[ib->length_dw++] = bytes;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+}
+
+/**
+ * sdma_v3_0_vm_write_pte - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @value: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ *
+ * Update PTEs by writing them manually using sDMA (CIK).
+ */
+static void sdma_v3_0_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
+				   uint64_t value, unsigned count,
+				   uint32_t incr)
+{
+	unsigned ndw = count * 2;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = ndw;
+	for (; ndw > 0; ndw -= 2) {
+		ib->ptr[ib->length_dw++] = lower_32_bits(value);
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		value += incr;
+	}
+}
+
+/**
+ * sdma_v3_0_vm_set_pte_pde - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA (CIK).
+ */
+static void sdma_v3_0_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe,
+				     uint64_t addr, unsigned count,
+				     uint32_t incr, uint64_t flags)
+{
+	/* for physically contiguous pages (vram) */
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_GEN_PTEPDE);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = lower_32_bits(flags); /* mask */
+	ib->ptr[ib->length_dw++] = upper_32_bits(flags);
+	ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = incr; /* increment size */
+	ib->ptr[ib->length_dw++] = 0;
+	ib->ptr[ib->length_dw++] = count; /* number of entries */
+}
+
+/**
+ * sdma_v3_0_ring_pad_ib - pad the IB to the required number of dw
+ *
+ * @ib: indirect buffer to fill with padding
+ *
+ */
+static void sdma_v3_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	u32 pad_count;
+	int i;
+
+	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	for (i = 0; i < pad_count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP) |
+				SDMA_PKT_NOP_HEADER_COUNT(pad_count - 1);
+		else
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+}
+
+/**
+ * sdma_v3_0_ring_emit_pipeline_sync - sync the pipeline
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Make sure all previous operations are completed (CIK).
+ */
+static void sdma_v3_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	/* wait for idle */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3) | /* equal */
+			  SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(1));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+	amdgpu_ring_write(ring, seq); /* reference */
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(4)); /* retry count, poll interval */
+}
+
+/**
+ * sdma_v3_0_ring_emit_vm_flush - cik vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vm: amdgpu_vm pointer
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA (VI).
+ */
+static void sdma_v3_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for flush */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(0)); /* always */
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST << 2);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0); /* reference */
+	amdgpu_ring_write(ring, 0); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
+}
+
+static void sdma_v3_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				     uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, val);
+}
+
+static int sdma_v3_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	switch (adev->asic_type) {
+	case CHIP_STONEY:
+		adev->sdma.num_instances = 1;
+		break;
+	default:
+		adev->sdma.num_instances = SDMA_MAX_INSTANCE;
+		break;
+	}
+
+	sdma_v3_0_set_ring_funcs(adev);
+	sdma_v3_0_set_buffer_funcs(adev);
+	sdma_v3_0_set_vm_pte_funcs(adev);
+	sdma_v3_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int sdma_v3_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* SDMA trap event */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_TRAP,
+			      &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	/* SDMA Privileged inst */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 241,
+			      &adev->sdma.illegal_inst_irq);
+	if (r)
+		return r;
+
+	/* SDMA Privileged inst */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_SDMA_SRBM_WRITE,
+			      &adev->sdma.illegal_inst_irq);
+	if (r)
+		return r;
+
+	r = sdma_v3_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load sdma firmware!\n");
+		return r;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		if (!amdgpu_sriov_vf(adev)) {
+			ring->use_doorbell = true;
+			ring->doorbell_index = (i == 0) ?
+				AMDGPU_DOORBELL_sDMA_ENGINE0 : AMDGPU_DOORBELL_sDMA_ENGINE1;
+		} else {
+			ring->use_pollmem = true;
+		}
+
+		snprintf(ring->name, sizeof(ring->name), "sdma%d", i);
+		r = amdgpu_ring_init(adev, ring, 1024,
+				     &adev->sdma.trap_irq,
+				     (i == 0) ?
+				     AMDGPU_SDMA_IRQ_TRAP0 :
+				     AMDGPU_SDMA_IRQ_TRAP1);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+static int sdma_v3_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+
+	sdma_v3_0_free_microcode(adev);
+	return 0;
+}
+
+static int sdma_v3_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	sdma_v3_0_init_golden_registers(adev);
+
+	r = sdma_v3_0_start(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int sdma_v3_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	sdma_v3_0_ctx_switch_enable(adev, false);
+	sdma_v3_0_enable(adev, false);
+
+	return 0;
+}
+
+static int sdma_v3_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return sdma_v3_0_hw_fini(adev);
+}
+
+static int sdma_v3_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return sdma_v3_0_hw_init(adev);
+}
+
+static bool sdma_v3_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS2);
+
+	if (tmp & (SRBM_STATUS2__SDMA_BUSY_MASK |
+		   SRBM_STATUS2__SDMA1_BUSY_MASK))
+	    return false;
+
+	return true;
+}
+
+static int sdma_v3_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmSRBM_STATUS2) & (SRBM_STATUS2__SDMA_BUSY_MASK |
+				SRBM_STATUS2__SDMA1_BUSY_MASK);
+
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static bool sdma_v3_0_check_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset = 0;
+	u32 tmp = RREG32(mmSRBM_STATUS2);
+
+	if ((tmp & SRBM_STATUS2__SDMA_BUSY_MASK) ||
+	    (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK)) {
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK;
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK;
+	}
+
+	if (srbm_soft_reset) {
+		adev->sdma.srbm_soft_reset = srbm_soft_reset;
+		return true;
+	} else {
+		adev->sdma.srbm_soft_reset = 0;
+		return false;
+	}
+}
+
+static int sdma_v3_0_pre_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset = 0;
+
+	if (!adev->sdma.srbm_soft_reset)
+		return 0;
+
+	srbm_soft_reset = adev->sdma.srbm_soft_reset;
+
+	if (REG_GET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA) ||
+	    REG_GET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA1)) {
+		sdma_v3_0_ctx_switch_enable(adev, false);
+		sdma_v3_0_enable(adev, false);
+	}
+
+	return 0;
+}
+
+static int sdma_v3_0_post_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset = 0;
+
+	if (!adev->sdma.srbm_soft_reset)
+		return 0;
+
+	srbm_soft_reset = adev->sdma.srbm_soft_reset;
+
+	if (REG_GET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA) ||
+	    REG_GET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA1)) {
+		sdma_v3_0_gfx_resume(adev);
+		sdma_v3_0_rlc_resume(adev);
+	}
+
+	return 0;
+}
+
+static int sdma_v3_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset = 0;
+	u32 tmp;
+
+	if (!adev->sdma.srbm_soft_reset)
+		return 0;
+
+	srbm_soft_reset = adev->sdma.srbm_soft_reset;
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int sdma_v3_0_set_trap_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	switch (type) {
+	case AMDGPU_SDMA_IRQ_TRAP0:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 0);
+			WREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 1);
+			WREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	case AMDGPU_SDMA_IRQ_TRAP1:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 0);
+			WREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 1);
+			WREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v3_0_process_trap_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	u8 instance_id, queue_id;
+
+	instance_id = (entry->ring_id & 0x3) >> 0;
+	queue_id = (entry->ring_id & 0xc) >> 2;
+	DRM_DEBUG("IH: SDMA trap\n");
+	switch (instance_id) {
+	case 0:
+		switch (queue_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[0].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		}
+		break;
+	case 1:
+		switch (queue_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[1].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		}
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v3_0_process_illegal_inst_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in SDMA command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static void sdma_v3_0_update_sdma_medium_grain_clock_gating(
+		struct amdgpu_device *adev,
+		bool enable)
+{
+	uint32_t temp, data;
+	int i;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_MGCG)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			temp = data = RREG32(mmSDMA0_CLK_CTRL + sdma_offsets[i]);
+			data &= ~(SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE6_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE5_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+				  SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK);
+			if (data != temp)
+				WREG32(mmSDMA0_CLK_CTRL + sdma_offsets[i], data);
+		}
+	} else {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			temp = data = RREG32(mmSDMA0_CLK_CTRL + sdma_offsets[i]);
+			data |= SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK |
+				SDMA0_CLK_CTRL__SOFT_OVERRIDE6_MASK |
+				SDMA0_CLK_CTRL__SOFT_OVERRIDE5_MASK |
+				SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+				SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+				SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+				SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+				SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK;
+
+			if (data != temp)
+				WREG32(mmSDMA0_CLK_CTRL + sdma_offsets[i], data);
+		}
+	}
+}
+
+static void sdma_v3_0_update_sdma_medium_grain_light_sleep(
+		struct amdgpu_device *adev,
+		bool enable)
+{
+	uint32_t temp, data;
+	int i;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_LS)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			temp = data = RREG32(mmSDMA0_POWER_CNTL + sdma_offsets[i]);
+			data |= SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+
+			if (temp != data)
+				WREG32(mmSDMA0_POWER_CNTL + sdma_offsets[i], data);
+		}
+	} else {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			temp = data = RREG32(mmSDMA0_POWER_CNTL + sdma_offsets[i]);
+			data &= ~SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+
+			if (temp != data)
+				WREG32(mmSDMA0_POWER_CNTL + sdma_offsets[i], data);
+		}
+	}
+}
+
+static int sdma_v3_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		sdma_v3_0_update_sdma_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		sdma_v3_0_update_sdma_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v3_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static void sdma_v3_0_get_clockgating_state(void *handle, u32 *flags)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_SDMA_MGCG */
+	data = RREG32(mmSDMA0_CLK_CTRL + sdma_offsets[0]);
+	if (!(data & SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK))
+		*flags |= AMD_CG_SUPPORT_SDMA_MGCG;
+
+	/* AMD_CG_SUPPORT_SDMA_LS */
+	data = RREG32(mmSDMA0_POWER_CNTL + sdma_offsets[0]);
+	if (data & SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK)
+		*flags |= AMD_CG_SUPPORT_SDMA_LS;
+}
+
+static const struct amd_ip_funcs sdma_v3_0_ip_funcs = {
+	.name = "sdma_v3_0",
+	.early_init = sdma_v3_0_early_init,
+	.late_init = NULL,
+	.sw_init = sdma_v3_0_sw_init,
+	.sw_fini = sdma_v3_0_sw_fini,
+	.hw_init = sdma_v3_0_hw_init,
+	.hw_fini = sdma_v3_0_hw_fini,
+	.suspend = sdma_v3_0_suspend,
+	.resume = sdma_v3_0_resume,
+	.is_idle = sdma_v3_0_is_idle,
+	.wait_for_idle = sdma_v3_0_wait_for_idle,
+	.check_soft_reset = sdma_v3_0_check_soft_reset,
+	.pre_soft_reset = sdma_v3_0_pre_soft_reset,
+	.post_soft_reset = sdma_v3_0_post_soft_reset,
+	.soft_reset = sdma_v3_0_soft_reset,
+	.set_clockgating_state = sdma_v3_0_set_clockgating_state,
+	.set_powergating_state = sdma_v3_0_set_powergating_state,
+	.get_clockgating_state = sdma_v3_0_get_clockgating_state,
+};
+
+static const struct amdgpu_ring_funcs sdma_v3_0_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xf,
+	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
+	.support_64bit_ptrs = false,
+	.get_rptr = sdma_v3_0_ring_get_rptr,
+	.get_wptr = sdma_v3_0_ring_get_wptr,
+	.set_wptr = sdma_v3_0_ring_set_wptr,
+	.emit_frame_size =
+		6 + /* sdma_v3_0_ring_emit_hdp_flush */
+		3 + /* hdp invalidate */
+		6 + /* sdma_v3_0_ring_emit_pipeline_sync */
+		VI_FLUSH_GPU_TLB_NUM_WREG * 3 + 6 + /* sdma_v3_0_ring_emit_vm_flush */
+		10 + 10 + 10, /* sdma_v3_0_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 7 + 6, /* sdma_v3_0_ring_emit_ib */
+	.emit_ib = sdma_v3_0_ring_emit_ib,
+	.emit_fence = sdma_v3_0_ring_emit_fence,
+	.emit_pipeline_sync = sdma_v3_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = sdma_v3_0_ring_emit_vm_flush,
+	.emit_hdp_flush = sdma_v3_0_ring_emit_hdp_flush,
+	.test_ring = sdma_v3_0_ring_test_ring,
+	.test_ib = sdma_v3_0_ring_test_ib,
+	.insert_nop = sdma_v3_0_ring_insert_nop,
+	.pad_ib = sdma_v3_0_ring_pad_ib,
+	.emit_wreg = sdma_v3_0_ring_emit_wreg,
+};
+
+static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->sdma.instance[i].ring.funcs = &sdma_v3_0_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs sdma_v3_0_trap_irq_funcs = {
+	.set = sdma_v3_0_set_trap_irq_state,
+	.process = sdma_v3_0_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v3_0_illegal_inst_irq_funcs = {
+	.process = sdma_v3_0_process_illegal_inst_irq,
+};
+
+static void sdma_v3_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_LAST;
+	adev->sdma.trap_irq.funcs = &sdma_v3_0_trap_irq_funcs;
+	adev->sdma.illegal_inst_irq.funcs = &sdma_v3_0_illegal_inst_irq_funcs;
+}
+
+/**
+ * sdma_v3_0_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Copy GPU buffers using the DMA engine (VI).
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void sdma_v3_0_emit_copy_buffer(struct amdgpu_ib *ib,
+				       uint64_t src_offset,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+	ib->ptr[ib->length_dw++] = byte_count;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+}
+
+/**
+ * sdma_v3_0_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine (VI).
+ */
+static void sdma_v3_0_emit_fill_buffer(struct amdgpu_ib *ib,
+				       uint32_t src_data,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_CONST_FILL);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = byte_count;
+}
+
+static const struct amdgpu_buffer_funcs sdma_v3_0_buffer_funcs = {
+	.copy_max_bytes = 0x3fffe0, /* not 0x3fffff due to HW limitation */
+	.copy_num_dw = 7,
+	.emit_copy_buffer = sdma_v3_0_emit_copy_buffer,
+
+	.fill_max_bytes = 0x3fffe0, /* not 0x3fffff due to HW limitation */
+	.fill_num_dw = 5,
+	.emit_fill_buffer = sdma_v3_0_emit_fill_buffer,
+};
+
+static void sdma_v3_0_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mman.buffer_funcs == NULL) {
+		adev->mman.buffer_funcs = &sdma_v3_0_buffer_funcs;
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+	}
+}
+
+static const struct amdgpu_vm_pte_funcs sdma_v3_0_vm_pte_funcs = {
+	.copy_pte_num_dw = 7,
+	.copy_pte = sdma_v3_0_vm_copy_pte,
+
+	.write_pte = sdma_v3_0_vm_write_pte,
+	.set_pte_pde = sdma_v3_0_vm_set_pte_pde,
+};
+
+static void sdma_v3_0_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	if (adev->vm_manager.vm_pte_funcs == NULL) {
+		adev->vm_manager.vm_pte_funcs = &sdma_v3_0_vm_pte_funcs;
+		for (i = 0; i < adev->sdma.num_instances; i++)
+			adev->vm_manager.vm_pte_rings[i] =
+				&adev->sdma.instance[i].ring;
+
+		adev->vm_manager.vm_pte_num_rings = adev->sdma.num_instances;
+	}
+}
+
+const struct amdgpu_ip_block_version sdma_v3_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_SDMA,
+	.major = 3,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &sdma_v3_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version sdma_v3_1_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_SDMA,
+	.major = 3,
+	.minor = 1,
+	.rev = 0,
+	.funcs = &sdma_v3_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/sdma_v3_0.h b/sys/dev/pci/drm/amd/amdgpu/sdma_v3_0.h
new file mode 100644
index 00000000000..7aa223d35f1
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/sdma_v3_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SDMA_V3_0_H__
+#define __SDMA_V3_0_H__
+
+extern const struct amdgpu_ip_block_version sdma_v3_0_ip_block;
+extern const struct amdgpu_ip_block_version sdma_v3_1_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/sdma_v4_0.c b/sys/dev/pci/drm/amd/amdgpu/sdma_v4_0.c
new file mode 100644
index 00000000000..49e96a6cb3a
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/sdma_v4_0.c
@@ -0,0 +1,1775 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_trace.h"
+
+#include "sdma0/sdma0_4_0_offset.h"
+#include "sdma0/sdma0_4_0_sh_mask.h"
+#include "sdma1/sdma1_4_0_offset.h"
+#include "sdma1/sdma1_4_0_sh_mask.h"
+#include "hdp/hdp_4_0_offset.h"
+#include "sdma0/sdma0_4_1_default.h"
+
+#include "soc15_common.h"
+#include "soc15.h"
+#include "vega10_sdma_pkt_open.h"
+
+#include "ivsrcid/sdma0/irqsrcs_sdma0_4_0.h"
+#include "ivsrcid/sdma1/irqsrcs_sdma1_4_0.h"
+
+MODULE_FIRMWARE("amdgpu/vega10_sdma.bin");
+MODULE_FIRMWARE("amdgpu/vega10_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/vega12_sdma.bin");
+MODULE_FIRMWARE("amdgpu/vega12_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/vega20_sdma.bin");
+MODULE_FIRMWARE("amdgpu/vega20_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/raven_sdma.bin");
+
+#define SDMA0_POWER_CNTL__ON_OFF_CONDITION_HOLD_TIME_MASK  0x000000F8L
+#define SDMA0_POWER_CNTL__ON_OFF_STATUS_DURATION_TIME_MASK 0xFC000000L
+
+static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev);
+static void sdma_v4_0_set_buffer_funcs(struct amdgpu_device *adev);
+static void sdma_v4_0_set_vm_pte_funcs(struct amdgpu_device *adev);
+static void sdma_v4_0_set_irq_funcs(struct amdgpu_device *adev);
+
+static const struct soc15_reg_golden golden_settings_sdma_4[] = {
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CHICKEN_BITS, 0xfe931f07, 0x02831d07),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CLK_CTRL, 0xff000ff0, 0x3f000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_PAGE_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_POWER_CNTL, 0x003ff006, 0x0003c000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_PAGE, 0x000003ff, 0x000003c0),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_WATERMK, 0xfc000000, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CLK_CTRL, 0xffffffff, 0x3f000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GFX_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GFX_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_PAGE_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_PAGE_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_POWER_CNTL, 0x003ff000, 0x0003c000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC0_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC0_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC1_IB_CNTL, 0x800f0100, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC1_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_PAGE, 0x000003ff, 0x000003c0),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_WATERMK, 0xfc000000, 0x00000000)
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_vg10[] = {
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831d07),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0018773f, 0x00104002),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002)
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_vg12[] = {
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00104001),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831d07),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0018773f, 0x00104001),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104001)
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_4_1[] =
+{
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CHICKEN_BITS, 0xfe931f07, 0x02831d07),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CLK_CTRL, 0xffffffff, 0x3f000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_IB_CNTL, 0x800f0111, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_POWER_CNTL, 0xfc3fffff, 0x40000051),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_IB_CNTL, 0x800f0111, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_IB_CNTL, 0x800f0111, 0x00000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_PAGE, 0x000003ff, 0x000003c0),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_WATERMK, 0xfc000000, 0x00000000)
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_4_2[] =
+{
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CHICKEN_BITS, 0xfe931f07, 0x02831d07),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_CLK_CTRL, 0xffffffff, 0x3f000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff0, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_PAGE, 0x000003ff, 0x000003c0),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831d07),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CLK_CTRL, 0xffffffff, 0x3f000100),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0000773f, 0x00004002),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GFX_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_PAGE_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC0_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_RLC1_RB_WPTR_POLL_CNTL, 0xfffffff7, 0x00403000),
+	SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_UTCL1_PAGE, 0x000003ff, 0x000003c0)
+};
+
+static const struct soc15_reg_golden golden_settings_sdma_rv1[] =
+{
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00000002),
+	SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00000002)
+};
+
+static u32 sdma_v4_0_get_reg_offset(struct amdgpu_device *adev,
+		u32 instance, u32 offset)
+{
+	return ( 0 == instance ? (adev->reg_offset[SDMA0_HWIP][0][0] + offset) :
+			(adev->reg_offset[SDMA1_HWIP][0][0] + offset));
+}
+
+static void sdma_v4_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+		soc15_program_register_sequence(adev,
+						 golden_settings_sdma_4,
+						 ARRAY_SIZE(golden_settings_sdma_4));
+		soc15_program_register_sequence(adev,
+						 golden_settings_sdma_vg10,
+						 ARRAY_SIZE(golden_settings_sdma_vg10));
+		break;
+	case CHIP_VEGA12:
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_4,
+						ARRAY_SIZE(golden_settings_sdma_4));
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_vg12,
+						ARRAY_SIZE(golden_settings_sdma_vg12));
+		break;
+	case CHIP_VEGA20:
+		soc15_program_register_sequence(adev,
+						golden_settings_sdma_4_2,
+						ARRAY_SIZE(golden_settings_sdma_4_2));
+		break;
+	case CHIP_RAVEN:
+		soc15_program_register_sequence(adev,
+						 golden_settings_sdma_4_1,
+						 ARRAY_SIZE(golden_settings_sdma_4_1));
+		soc15_program_register_sequence(adev,
+						 golden_settings_sdma_rv1,
+						 ARRAY_SIZE(golden_settings_sdma_rv1));
+		break;
+	default:
+		break;
+	}
+}
+
+/**
+ * sdma_v4_0_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+
+// emulation only, won't work on real chip
+// vega10 real chip need to use PSP to load firmware
+static int sdma_v4_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err = 0, i;
+	struct amdgpu_firmware_info *info = NULL;
+	const struct common_firmware_header *header = NULL;
+	const struct sdma_firmware_header_v1_0 *hdr;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+		chip_name = "vega10";
+		break;
+	case CHIP_VEGA12:
+		chip_name = "vega12";
+		break;
+	case CHIP_VEGA20:
+		chip_name = "vega20";
+		break;
+	case CHIP_RAVEN:
+		chip_name = "raven";
+		break;
+	default:
+		BUG();
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (i == 0)
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma.bin", chip_name);
+		else
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma1.bin", chip_name);
+		err = request_firmware(&adev->sdma.instance[i].fw, fw_name, adev->dev);
+		if (err)
+			goto out;
+		err = amdgpu_ucode_validate(adev->sdma.instance[i].fw);
+		if (err)
+			goto out;
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		adev->sdma.instance[i].fw_version = le32_to_cpu(hdr->header.ucode_version);
+		adev->sdma.instance[i].feature_version = le32_to_cpu(hdr->ucode_feature_version);
+		if (adev->sdma.instance[i].feature_version >= 20)
+			adev->sdma.instance[i].burst_nop = true;
+		DRM_DEBUG("psp_load == '%s'\n",
+				adev->firmware.load_type == AMDGPU_FW_LOAD_PSP ? "true" : "false");
+
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA0 + i];
+			info->ucode_id = AMDGPU_UCODE_ID_SDMA0 + i;
+			info->fw = adev->sdma.instance[i].fw;
+			header = (const struct common_firmware_header *)info->fw->data;
+			adev->firmware.fw_size +=
+				roundup2(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+		}
+	}
+out:
+	if (err) {
+		DRM_ERROR("sdma_v4_0: Failed to load firmware \"%s\"\n", fw_name);
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			release_firmware(adev->sdma.instance[i].fw);
+			adev->sdma.instance[i].fw = NULL;
+		}
+	}
+	return err;
+}
+
+/**
+ * sdma_v4_0_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware (VEGA10+).
+ */
+static uint64_t sdma_v4_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	u64 *rptr;
+
+	/* XXX check if swapping is necessary on BE */
+	rptr = ((u64 *)&ring->adev->wb.wb[ring->rptr_offs]);
+
+	DRM_DEBUG("rptr before shift == 0x%016llx\n", *rptr);
+	return ((*rptr) >> 2);
+}
+
+/**
+ * sdma_v4_0_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware (VEGA10+).
+ */
+static uint64_t sdma_v4_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u64 wptr;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		wptr = READ_ONCE(*((u64 *)&adev->wb.wb[ring->wptr_offs]));
+		DRM_DEBUG("wptr/doorbell before shift == 0x%016llx\n", wptr);
+	} else {
+		u32 lowbit, highbit;
+
+		lowbit = RREG32(sdma_v4_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR)) >> 2;
+		highbit = RREG32(sdma_v4_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI)) >> 2;
+
+		DRM_DEBUG("wptr [%i]high== 0x%08x low==0x%08x\n",
+				ring->me, highbit, lowbit);
+		wptr = highbit;
+		wptr = wptr << 32;
+		wptr |= lowbit;
+	}
+
+	return wptr >> 2;
+}
+
+/**
+ * sdma_v4_0_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware (VEGA10+).
+ */
+static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	DRM_DEBUG("Setting write pointer\n");
+	if (ring->use_doorbell) {
+		u64 *wb = (u64 *)&adev->wb.wb[ring->wptr_offs];
+
+		DRM_DEBUG("Using doorbell -- "
+				"wptr_offs == 0x%08x "
+				"lower_32_bits(ring->wptr) << 2 == 0x%08x "
+				"upper_32_bits(ring->wptr) << 2 == 0x%08x\n",
+				ring->wptr_offs,
+				lower_32_bits(ring->wptr << 2),
+				upper_32_bits(ring->wptr << 2));
+		/* XXX check if swapping is necessary on BE */
+		WRITE_ONCE(*wb, (ring->wptr << 2));
+		DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n",
+				ring->doorbell_index, ring->wptr << 2);
+		WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
+	} else {
+		DRM_DEBUG("Not using doorbell -- "
+				"mmSDMA%i_GFX_RB_WPTR == 0x%08x "
+				"mmSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n",
+				ring->me,
+				lower_32_bits(ring->wptr << 2),
+				ring->me,
+				upper_32_bits(ring->wptr << 2));
+		WREG32(sdma_v4_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr << 2));
+		WREG32(sdma_v4_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr << 2));
+	}
+}
+
+static void sdma_v4_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			amdgpu_ring_write(ring, ring->funcs->nop |
+				SDMA_PKT_NOP_HEADER_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->funcs->nop);
+}
+
+/**
+ * sdma_v4_0_ring_emit_ib - Schedule an IB on the DMA engine
+ *
+ * @ring: amdgpu ring pointer
+ * @ib: IB object to schedule
+ *
+ * Schedule an IB in the DMA ring (VEGA10).
+ */
+static void sdma_v4_0_ring_emit_ib(struct amdgpu_ring *ring,
+					struct amdgpu_ib *ib,
+					unsigned vmid, bool ctx_switch)
+{
+	/* IB packet must end on a 8 DW boundary */
+	sdma_v4_0_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8);
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid & 0xf));
+	/* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0);
+
+}
+
+static void sdma_v4_0_wait_reg_mem(struct amdgpu_ring *ring,
+				   int mem_space, int hdp,
+				   uint32_t addr0, uint32_t addr1,
+				   uint32_t ref, uint32_t mask,
+				   uint32_t inv)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(hdp) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(mem_space) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* == */
+	if (mem_space) {
+		/* memory */
+		amdgpu_ring_write(ring, addr0);
+		amdgpu_ring_write(ring, addr1);
+	} else {
+		/* registers */
+		amdgpu_ring_write(ring, addr0 << 2);
+		amdgpu_ring_write(ring, addr1 << 2);
+	}
+	amdgpu_ring_write(ring, ref); /* reference */
+	amdgpu_ring_write(ring, mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(inv)); /* retry count, poll interval */
+}
+
+/**
+ * sdma_v4_0_ring_emit_hdp_flush - emit an hdp flush on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Emit an hdp flush packet on the requested DMA ring.
+ */
+static void sdma_v4_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 ref_and_mask = 0;
+	const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio_funcs->hdp_flush_reg;
+
+	if (ring->me == 0)
+		ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0;
+	else
+		ref_and_mask = nbio_hf_reg->ref_and_mask_sdma1;
+
+	sdma_v4_0_wait_reg_mem(ring, 0, 1,
+			       adev->nbio_funcs->get_hdp_flush_done_offset(adev),
+			       adev->nbio_funcs->get_hdp_flush_req_offset(adev),
+			       ref_and_mask, ref_and_mask, 10);
+}
+
+/**
+ * sdma_v4_0_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @fence: amdgpu fence object
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed (VEGA10).
+ */
+static void sdma_v4_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				      unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE));
+	/* zero in first two bits */
+	BUG_ON(addr & 0x3);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE));
+		/* zero in first two bits */
+		BUG_ON(addr & 0x3);
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+
+	/* generate an interrupt */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_TRAP));
+	amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0));
+}
+
+
+/**
+ * sdma_v4_0_gfx_stop - stop the gfx async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the gfx async dma ring buffers (VEGA10).
+ */
+static void sdma_v4_0_gfx_stop(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *sdma0 = &adev->sdma.instance[0].ring;
+	struct amdgpu_ring *sdma1 = &adev->sdma.instance[1].ring;
+	u32 rb_cntl, ib_cntl;
+	int i;
+
+	if ((adev->mman.buffer_funcs_ring == sdma0) ||
+	    (adev->mman.buffer_funcs_ring == sdma1))
+			amdgpu_ttm_set_buffer_funcs_status(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		rb_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL));
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 0);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+		ib_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL));
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
+	}
+
+	sdma0->ready = false;
+	sdma1->ready = false;
+}
+
+/**
+ * sdma_v4_0_rlc_stop - stop the compute async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute async dma queues (VEGA10).
+ */
+static void sdma_v4_0_rlc_stop(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+}
+
+/**
+ * sdma_v_0_ctx_switch_enable - stop the async dma engines context switch
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs context switch.
+ *
+ * Halt or unhalt the async dma engines context switch (VEGA10).
+ */
+static void sdma_v4_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl, phase_quantum = 0;
+	int i;
+
+	if (amdgpu_sdma_phase_quantum) {
+		unsigned value = amdgpu_sdma_phase_quantum;
+		unsigned unit = 0;
+
+		while (value > (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
+				SDMA0_PHASE0_QUANTUM__VALUE__SHIFT)) {
+			value = (value + 1) >> 1;
+			unit++;
+		}
+		if (unit > (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
+			    SDMA0_PHASE0_QUANTUM__UNIT__SHIFT)) {
+			value = (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
+				 SDMA0_PHASE0_QUANTUM__VALUE__SHIFT);
+			unit = (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
+				SDMA0_PHASE0_QUANTUM__UNIT__SHIFT);
+			WARN_ONCE(1,
+			"clamping sdma_phase_quantum to %uK clock cycles\n",
+				  value << unit);
+		}
+		phase_quantum =
+			value << SDMA0_PHASE0_QUANTUM__VALUE__SHIFT |
+			unit  << SDMA0_PHASE0_QUANTUM__UNIT__SHIFT;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		f32_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_CNTL));
+		f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+				AUTO_CTXSW_ENABLE, enable ? 1 : 0);
+		if (enable && amdgpu_sdma_phase_quantum) {
+			WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_PHASE0_QUANTUM),
+			       phase_quantum);
+			WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_PHASE1_QUANTUM),
+			       phase_quantum);
+			WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_PHASE2_QUANTUM),
+			       phase_quantum);
+		}
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_CNTL), f32_cntl);
+	}
+
+}
+
+/**
+ * sdma_v4_0_enable - stop the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs.
+ *
+ * Halt or unhalt the async dma engines (VEGA10).
+ */
+static void sdma_v4_0_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl;
+	int i;
+
+	if (enable == false) {
+		sdma_v4_0_gfx_stop(adev);
+		sdma_v4_0_rlc_stop(adev);
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		f32_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL));
+		f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, enable ? 0 : 1);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL), f32_cntl);
+	}
+}
+
+/**
+ * sdma_v4_0_gfx_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the gfx DMA ring buffers and enable them (VEGA10).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v4_0_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl, ib_cntl, wptr_poll_cntl;
+	u32 rb_bufsz;
+	u32 wb_offset;
+	u32 doorbell;
+	u32 doorbell_offset;
+	u32 temp;
+	u64 wptr_gpu_addr;
+	int i, r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		wb_offset = (ring->rptr_offs * 4);
+
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL), 0);
+
+		/* Set ring buffer size in dwords */
+		rb_bufsz = order_base_2(ring->ring_size / 4);
+		rb_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL));
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz);
+#ifdef __BIG_ENDIAN
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL,
+					RPTR_WRITEBACK_SWAP_ENABLE, 1);
+#endif
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+
+		/* Initialize the ring buffer's read and write pointers */
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR), 0);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_HI), 0);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), 0);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), 0);
+
+		/* set the wb address whether it's enabled or not */
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_HI),
+		       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_LO),
+		       lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC);
+
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
+
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE), ring->gpu_addr >> 8);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE_HI), ring->gpu_addr >> 40);
+
+		ring->wptr = 0;
+
+		/* before programing wptr to a less value, need set minor_ptr_update first */
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 1);
+
+		if (!amdgpu_sriov_vf(adev)) { /* only bare-metal use register write for wptr */
+			WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr) << 2);
+			WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr) << 2);
+		}
+
+		doorbell = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL));
+		doorbell_offset = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET));
+
+		if (ring->use_doorbell) {
+			doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 1);
+			doorbell_offset = REG_SET_FIELD(doorbell_offset, SDMA0_GFX_DOORBELL_OFFSET,
+					OFFSET, ring->doorbell_index);
+		} else {
+			doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 0);
+		}
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL), doorbell);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET), doorbell_offset);
+		adev->nbio_funcs->sdma_doorbell_range(adev, i, ring->use_doorbell,
+						      ring->doorbell_index);
+
+		if (amdgpu_sriov_vf(adev))
+			sdma_v4_0_ring_set_wptr(ring);
+
+		/* set minor_ptr_update to 0 after wptr programed */
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 0);
+
+		/* set utc l1 enable flag always to 1 */
+		temp = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_CNTL));
+		temp = REG_SET_FIELD(temp, SDMA0_CNTL, UTC_L1_ENABLE, 1);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_CNTL), temp);
+
+		if (!amdgpu_sriov_vf(adev)) {
+			/* unhalt engine */
+			temp = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL));
+			temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, HALT, 0);
+			WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL), temp);
+		}
+
+		/* setup the wptr shadow polling */
+		wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO),
+		       lower_32_bits(wptr_gpu_addr));
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI),
+		       upper_32_bits(wptr_gpu_addr));
+		wptr_poll_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL));
+		if (amdgpu_sriov_vf(adev))
+			wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, SDMA0_GFX_RB_WPTR_POLL_CNTL, F32_POLL_ENABLE, 1);
+		else
+			wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, SDMA0_GFX_RB_WPTR_POLL_CNTL, F32_POLL_ENABLE, 0);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL), wptr_poll_cntl);
+
+		/* enable DMA RB */
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+
+		ib_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL));
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1);
+#ifdef __BIG_ENDIAN
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1);
+#endif
+		/* enable DMA IBs */
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
+
+		ring->ready = true;
+
+		if (amdgpu_sriov_vf(adev)) { /* bare-metal sequence doesn't need below to lines */
+			sdma_v4_0_ctx_switch_enable(adev, true);
+			sdma_v4_0_enable(adev, true);
+		}
+
+		r = amdgpu_ring_test_ring(ring);
+		if (r) {
+			ring->ready = false;
+			return r;
+		}
+
+		if (adev->mman.buffer_funcs_ring == ring)
+			amdgpu_ttm_set_buffer_funcs_status(adev, true);
+
+	}
+
+	return 0;
+}
+
+static void
+sdma_v4_1_update_power_gating(struct amdgpu_device *adev, bool enable)
+{
+	uint32_t def, data;
+
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_SDMA)) {
+		/* disable idle interrupt */
+		def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CNTL));
+		data |= SDMA0_CNTL__CTXEMPTY_INT_ENABLE_MASK;
+
+		if (data != def)
+			WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CNTL), data);
+	} else {
+		/* disable idle interrupt */
+		def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CNTL));
+		data &= ~SDMA0_CNTL__CTXEMPTY_INT_ENABLE_MASK;
+		if (data != def)
+			WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CNTL), data);
+	}
+}
+
+static void sdma_v4_1_init_power_gating(struct amdgpu_device *adev)
+{
+	uint32_t def, data;
+
+	/* Enable HW based PG. */
+	def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL));
+	data |= SDMA0_POWER_CNTL__PG_CNTL_ENABLE_MASK;
+	if (data != def)
+		WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL), data);
+
+	/* enable interrupt */
+	def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CNTL));
+	data |= SDMA0_CNTL__CTXEMPTY_INT_ENABLE_MASK;
+	if (data != def)
+		WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CNTL), data);
+
+	/* Configure hold time to filter in-valid power on/off request. Use default right now */
+	def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL));
+	data &= ~SDMA0_POWER_CNTL__ON_OFF_CONDITION_HOLD_TIME_MASK;
+	data |= (mmSDMA0_POWER_CNTL_DEFAULT & SDMA0_POWER_CNTL__ON_OFF_CONDITION_HOLD_TIME_MASK);
+	/* Configure switch time for hysteresis purpose. Use default right now */
+	data &= ~SDMA0_POWER_CNTL__ON_OFF_STATUS_DURATION_TIME_MASK;
+	data |= (mmSDMA0_POWER_CNTL_DEFAULT & SDMA0_POWER_CNTL__ON_OFF_STATUS_DURATION_TIME_MASK);
+	if(data != def)
+		WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL), data);
+}
+
+static void sdma_v4_0_init_pg(struct amdgpu_device *adev)
+{
+	if (!(adev->pg_flags & AMD_PG_SUPPORT_SDMA))
+		return;
+
+	switch (adev->asic_type) {
+	case CHIP_RAVEN:
+		sdma_v4_1_init_power_gating(adev);
+		sdma_v4_1_update_power_gating(adev, true);
+		break;
+	default:
+		break;
+	}
+}
+
+/**
+ * sdma_v4_0_rlc_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the compute DMA queues and enable them (VEGA10).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v4_0_rlc_resume(struct amdgpu_device *adev)
+{
+	sdma_v4_0_init_pg(adev);
+
+	return 0;
+}
+
+/**
+ * sdma_v4_0_load_microcode - load the sDMA ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the sDMA0/1 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int sdma_v4_0_load_microcode(struct amdgpu_device *adev)
+{
+	const struct sdma_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+	int i, j;
+
+	/* halt the MEs */
+	sdma_v4_0_enable(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (!adev->sdma.instance[i].fw)
+			return -EINVAL;
+
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		amdgpu_ucode_print_sdma_hdr(&hdr->header);
+		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+
+		fw_data = (const __le32 *)
+			(adev->sdma.instance[i].fw->data +
+				le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_UCODE_ADDR), 0);
+
+		for (j = 0; j < fw_size; j++)
+			WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_UCODE_DATA), le32_to_cpup(fw_data++));
+
+		WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_UCODE_ADDR), adev->sdma.instance[i].fw_version);
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v4_0_start - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the DMA engines and enable them (VEGA10).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v4_0_start(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (amdgpu_sriov_vf(adev)) {
+		sdma_v4_0_ctx_switch_enable(adev, false);
+		sdma_v4_0_enable(adev, false);
+
+		/* set RB registers */
+		r = sdma_v4_0_gfx_resume(adev);
+		return r;
+	}
+
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
+		r = sdma_v4_0_load_microcode(adev);
+		if (r)
+			return r;
+	}
+
+	/* unhalt the MEs */
+	sdma_v4_0_enable(adev, true);
+	/* enable sdma ring preemption */
+	sdma_v4_0_ctx_switch_enable(adev, true);
+
+	/* start the gfx rings and rlc compute queues */
+	r = sdma_v4_0_gfx_resume(adev);
+	if (r)
+		return r;
+	r = sdma_v4_0_rlc_resume(adev);
+
+	return r;
+}
+
+/**
+ * sdma_v4_0_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory. (VEGA10).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v4_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_alloc(ring, 5);
+	if (r) {
+		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_device_wb_free(adev, index);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_device_wb_free(adev, index);
+
+	return r;
+}
+
+/**
+ * sdma_v4_0_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test a simple IB in the DMA ring (VEGA10).
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v4_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct dma_fence *f = NULL;
+	unsigned index;
+	long r;
+	u32 tmp = 0;
+	u64 gpu_addr;
+
+	r = amdgpu_device_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(adev, NULL, 256, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
+		goto err0;
+	}
+
+	ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr);
+	ib.ptr[3] = SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(0);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.ptr[5] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[6] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[7] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.length_dw = 8;
+
+	r = amdgpu_ib_schedule(ring, 1, &ib, NULL, &f);
+	if (r)
+		goto err1;
+
+	r = dma_fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
+		goto err1;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+		goto err1;
+	}
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+	if (tmp == 0xDEADBEEF) {
+		DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
+	} else {
+		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+		r = -EINVAL;
+	}
+err1:
+	amdgpu_ib_free(adev, &ib, NULL);
+	dma_fence_put(f);
+err0:
+	amdgpu_device_wb_free(adev, index);
+	return r;
+}
+
+
+/**
+ * sdma_v4_0_vm_copy_pte - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using sDMA (VEGA10).
+ */
+static void sdma_v4_0_vm_copy_pte(struct amdgpu_ib *ib,
+				  uint64_t pe, uint64_t src,
+				  unsigned count)
+{
+	unsigned bytes = count * 8;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+	ib->ptr[ib->length_dw++] = bytes - 1;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+
+}
+
+/**
+ * sdma_v4_0_vm_write_pte - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update PTEs by writing them manually using sDMA (VEGA10).
+ */
+static void sdma_v4_0_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
+				   uint64_t value, unsigned count,
+				   uint32_t incr)
+{
+	unsigned ndw = count * 2;
+
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = ndw - 1;
+	for (; ndw > 0; ndw -= 2) {
+		ib->ptr[ib->length_dw++] = lower_32_bits(value);
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		value += incr;
+	}
+}
+
+/**
+ * sdma_v4_0_vm_set_pte_pde - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA (VEGA10).
+ */
+static void sdma_v4_0_vm_set_pte_pde(struct amdgpu_ib *ib,
+				     uint64_t pe,
+				     uint64_t addr, unsigned count,
+				     uint32_t incr, uint64_t flags)
+{
+	/* for physically contiguous pages (vram) */
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_PTEPDE);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = lower_32_bits(flags); /* mask */
+	ib->ptr[ib->length_dw++] = upper_32_bits(flags);
+	ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */
+	ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+	ib->ptr[ib->length_dw++] = incr; /* increment size */
+	ib->ptr[ib->length_dw++] = 0;
+	ib->ptr[ib->length_dw++] = count - 1; /* number of entries */
+}
+
+/**
+ * sdma_v4_0_ring_pad_ib - pad the IB to the required number of dw
+ *
+ * @ib: indirect buffer to fill with padding
+ *
+ */
+static void sdma_v4_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	u32 pad_count;
+	int i;
+
+	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	for (i = 0; i < pad_count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP) |
+				SDMA_PKT_NOP_HEADER_COUNT(pad_count - 1);
+		else
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+}
+
+
+/**
+ * sdma_v4_0_ring_emit_pipeline_sync - sync the pipeline
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Make sure all previous operations are completed (CIK).
+ */
+static void sdma_v4_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	/* wait for idle */
+	sdma_v4_0_wait_reg_mem(ring, 1, 0,
+			       addr & 0xfffffffc,
+			       upper_32_bits(addr) & 0xffffffff,
+			       seq, 0xffffffff, 4);
+}
+
+
+/**
+ * sdma_v4_0_ring_emit_vm_flush - vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vm: amdgpu_vm pointer
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA (VEGA10).
+ */
+static void sdma_v4_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+}
+
+static void sdma_v4_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				     uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	amdgpu_ring_write(ring, reg);
+	amdgpu_ring_write(ring, val);
+}
+
+static void sdma_v4_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					 uint32_t val, uint32_t mask)
+{
+	sdma_v4_0_wait_reg_mem(ring, 0, 0, reg, 0, val, mask, 10);
+}
+
+static int sdma_v4_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->asic_type == CHIP_RAVEN)
+		adev->sdma.num_instances = 1;
+	else
+		adev->sdma.num_instances = 2;
+
+	sdma_v4_0_set_ring_funcs(adev);
+	sdma_v4_0_set_buffer_funcs(adev);
+	sdma_v4_0_set_vm_pte_funcs(adev);
+	sdma_v4_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+
+static int sdma_v4_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* SDMA trap event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_SDMA0, SDMA0_4_0__SRCID__SDMA_TRAP,
+			      &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	/* SDMA trap event */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_SDMA1, SDMA1_4_0__SRCID__SDMA_TRAP,
+			      &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	r = sdma_v4_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load sdma firmware!\n");
+		return r;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		ring->use_doorbell = true;
+
+		DRM_INFO("use_doorbell being set to: [%s]\n",
+				ring->use_doorbell?"true":"false");
+
+		ring->doorbell_index = (i == 0) ?
+			(AMDGPU_DOORBELL64_sDMA_ENGINE0 << 1) //get DWORD offset
+			: (AMDGPU_DOORBELL64_sDMA_ENGINE1 << 1); // get DWORD offset
+
+		snprintf(ring->name, sizeof(ring->name), "sdma%d", i);
+		r = amdgpu_ring_init(adev, ring, 1024,
+				     &adev->sdma.trap_irq,
+				     (i == 0) ?
+				     AMDGPU_SDMA_IRQ_TRAP0 :
+				     AMDGPU_SDMA_IRQ_TRAP1);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+static int sdma_v4_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		release_firmware(adev->sdma.instance[i].fw);
+		adev->sdma.instance[i].fw = NULL;
+	}
+
+	return 0;
+}
+
+static int sdma_v4_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	sdma_v4_0_init_golden_registers(adev);
+
+	r = sdma_v4_0_start(adev);
+
+	return r;
+}
+
+static int sdma_v4_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	sdma_v4_0_ctx_switch_enable(adev, false);
+	sdma_v4_0_enable(adev, false);
+
+	return 0;
+}
+
+static int sdma_v4_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return sdma_v4_0_hw_fini(adev);
+}
+
+static int sdma_v4_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return sdma_v4_0_hw_init(adev);
+}
+
+static bool sdma_v4_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		u32 tmp = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_STATUS_REG));
+
+		if (!(tmp & SDMA0_STATUS_REG__IDLE_MASK))
+			return false;
+	}
+
+	return true;
+}
+
+static int sdma_v4_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 sdma0, sdma1;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		sdma0 = RREG32(sdma_v4_0_get_reg_offset(adev, 0, mmSDMA0_STATUS_REG));
+		sdma1 = RREG32(sdma_v4_0_get_reg_offset(adev, 1, mmSDMA0_STATUS_REG));
+
+		if (sdma0 & sdma1 & SDMA0_STATUS_REG__IDLE_MASK)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int sdma_v4_0_soft_reset(void *handle)
+{
+	/* todo */
+
+	return 0;
+}
+
+static int sdma_v4_0_set_trap_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	u32 reg_offset = (type == AMDGPU_SDMA_IRQ_TRAP0) ?
+		sdma_v4_0_get_reg_offset(adev, 0, mmSDMA0_CNTL) :
+		sdma_v4_0_get_reg_offset(adev, 1, mmSDMA0_CNTL);
+
+	sdma_cntl = RREG32(reg_offset);
+	sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE,
+		       state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0);
+	WREG32(reg_offset, sdma_cntl);
+
+	return 0;
+}
+
+static int sdma_v4_0_process_trap_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: SDMA trap\n");
+	switch (entry->client_id) {
+	case SOC15_IH_CLIENTID_SDMA0:
+		switch (entry->ring_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[0].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		case 3:
+			/* XXX page queue*/
+			break;
+		}
+		break;
+	case SOC15_IH_CLIENTID_SDMA1:
+		switch (entry->ring_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[1].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		case 3:
+			/* XXX page queue*/
+			break;
+		}
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v4_0_process_illegal_inst_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in SDMA command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+
+static void sdma_v4_0_update_medium_grain_clock_gating(
+		struct amdgpu_device *adev,
+		bool enable)
+{
+	uint32_t data, def;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_MGCG)) {
+		/* enable sdma0 clock gating */
+		def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CLK_CTRL));
+		data &= ~(SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK |
+			  SDMA0_CLK_CTRL__SOFT_OVERRIDE6_MASK |
+			  SDMA0_CLK_CTRL__SOFT_OVERRIDE5_MASK |
+			  SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+			  SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+			  SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+			  SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+			  SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK);
+		if (def != data)
+			WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CLK_CTRL), data);
+
+		if (adev->sdma.num_instances > 1) {
+			def = data = RREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_CLK_CTRL));
+			data &= ~(SDMA1_CLK_CTRL__SOFT_OVERRIDE7_MASK |
+				  SDMA1_CLK_CTRL__SOFT_OVERRIDE6_MASK |
+				  SDMA1_CLK_CTRL__SOFT_OVERRIDE5_MASK |
+				  SDMA1_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+				  SDMA1_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+				  SDMA1_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+				  SDMA1_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+				  SDMA1_CLK_CTRL__SOFT_OVERRIDE0_MASK);
+			if (def != data)
+				WREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_CLK_CTRL), data);
+		}
+	} else {
+		/* disable sdma0 clock gating */
+		def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CLK_CTRL));
+		data |= (SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK |
+			 SDMA0_CLK_CTRL__SOFT_OVERRIDE6_MASK |
+			 SDMA0_CLK_CTRL__SOFT_OVERRIDE5_MASK |
+			 SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+			 SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+			 SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+			 SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+			 SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK);
+
+		if (def != data)
+			WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CLK_CTRL), data);
+
+		if (adev->sdma.num_instances > 1) {
+			def = data = RREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_CLK_CTRL));
+			data |= (SDMA1_CLK_CTRL__SOFT_OVERRIDE7_MASK |
+				 SDMA1_CLK_CTRL__SOFT_OVERRIDE6_MASK |
+				 SDMA1_CLK_CTRL__SOFT_OVERRIDE5_MASK |
+				 SDMA1_CLK_CTRL__SOFT_OVERRIDE4_MASK |
+				 SDMA1_CLK_CTRL__SOFT_OVERRIDE3_MASK |
+				 SDMA1_CLK_CTRL__SOFT_OVERRIDE2_MASK |
+				 SDMA1_CLK_CTRL__SOFT_OVERRIDE1_MASK |
+				 SDMA1_CLK_CTRL__SOFT_OVERRIDE0_MASK);
+			if (def != data)
+				WREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_CLK_CTRL), data);
+		}
+	}
+}
+
+
+static void sdma_v4_0_update_medium_grain_light_sleep(
+		struct amdgpu_device *adev,
+		bool enable)
+{
+	uint32_t data, def;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_LS)) {
+		/* 1-not override: enable sdma0 mem light sleep */
+		def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL));
+		data |= SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+		if (def != data)
+			WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL), data);
+
+		/* 1-not override: enable sdma1 mem light sleep */
+		if (adev->sdma.num_instances > 1) {
+			def = data = RREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_POWER_CNTL));
+			data |= SDMA1_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+			if (def != data)
+				WREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_POWER_CNTL), data);
+		}
+	} else {
+		/* 0-override:disable sdma0 mem light sleep */
+		def = data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL));
+		data &= ~SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+		if (def != data)
+			WREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL), data);
+
+		/* 0-override:disable sdma1 mem light sleep */
+		if (adev->sdma.num_instances > 1) {
+			def = data = RREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_POWER_CNTL));
+			data &= ~SDMA1_POWER_CNTL__MEM_POWER_OVERRIDE_MASK;
+			if (def != data)
+				WREG32(SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_POWER_CNTL), data);
+		}
+	}
+}
+
+static int sdma_v4_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+	case CHIP_RAVEN:
+		sdma_v4_0_update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		sdma_v4_0_update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v4_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	switch (adev->asic_type) {
+	case CHIP_RAVEN:
+		sdma_v4_1_update_power_gating(adev,
+				state == AMD_PG_STATE_GATE ? true : false);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void sdma_v4_0_get_clockgating_state(void *handle, u32 *flags)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_SDMA_MGCG */
+	data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_CLK_CTRL));
+	if (!(data & SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK))
+		*flags |= AMD_CG_SUPPORT_SDMA_MGCG;
+
+	/* AMD_CG_SUPPORT_SDMA_LS */
+	data = RREG32(SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_POWER_CNTL));
+	if (data & SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK)
+		*flags |= AMD_CG_SUPPORT_SDMA_LS;
+}
+
+const struct amd_ip_funcs sdma_v4_0_ip_funcs = {
+	.name = "sdma_v4_0",
+	.early_init = sdma_v4_0_early_init,
+	.late_init = NULL,
+	.sw_init = sdma_v4_0_sw_init,
+	.sw_fini = sdma_v4_0_sw_fini,
+	.hw_init = sdma_v4_0_hw_init,
+	.hw_fini = sdma_v4_0_hw_fini,
+	.suspend = sdma_v4_0_suspend,
+	.resume = sdma_v4_0_resume,
+	.is_idle = sdma_v4_0_is_idle,
+	.wait_for_idle = sdma_v4_0_wait_for_idle,
+	.soft_reset = sdma_v4_0_soft_reset,
+	.set_clockgating_state = sdma_v4_0_set_clockgating_state,
+	.set_powergating_state = sdma_v4_0_set_powergating_state,
+	.get_clockgating_state = sdma_v4_0_get_clockgating_state,
+};
+
+static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xf,
+	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
+	.support_64bit_ptrs = true,
+	.vmhub = AMDGPU_MMHUB,
+	.get_rptr = sdma_v4_0_ring_get_rptr,
+	.get_wptr = sdma_v4_0_ring_get_wptr,
+	.set_wptr = sdma_v4_0_ring_set_wptr,
+	.emit_frame_size =
+		6 + /* sdma_v4_0_ring_emit_hdp_flush */
+		3 + /* hdp invalidate */
+		6 + /* sdma_v4_0_ring_emit_pipeline_sync */
+		/* sdma_v4_0_ring_emit_vm_flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 +
+		10 + 10 + 10, /* sdma_v4_0_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 7 + 6, /* sdma_v4_0_ring_emit_ib */
+	.emit_ib = sdma_v4_0_ring_emit_ib,
+	.emit_fence = sdma_v4_0_ring_emit_fence,
+	.emit_pipeline_sync = sdma_v4_0_ring_emit_pipeline_sync,
+	.emit_vm_flush = sdma_v4_0_ring_emit_vm_flush,
+	.emit_hdp_flush = sdma_v4_0_ring_emit_hdp_flush,
+	.test_ring = sdma_v4_0_ring_test_ring,
+	.test_ib = sdma_v4_0_ring_test_ib,
+	.insert_nop = sdma_v4_0_ring_insert_nop,
+	.pad_ib = sdma_v4_0_ring_pad_ib,
+	.emit_wreg = sdma_v4_0_ring_emit_wreg,
+	.emit_reg_wait = sdma_v4_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+};
+
+static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		adev->sdma.instance[i].ring.funcs = &sdma_v4_0_ring_funcs;
+		adev->sdma.instance[i].ring.me = i;
+	}
+}
+
+static const struct amdgpu_irq_src_funcs sdma_v4_0_trap_irq_funcs = {
+	.set = sdma_v4_0_set_trap_irq_state,
+	.process = sdma_v4_0_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v4_0_illegal_inst_irq_funcs = {
+	.process = sdma_v4_0_process_illegal_inst_irq,
+};
+
+static void sdma_v4_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_LAST;
+	adev->sdma.trap_irq.funcs = &sdma_v4_0_trap_irq_funcs;
+	adev->sdma.illegal_inst_irq.funcs = &sdma_v4_0_illegal_inst_irq_funcs;
+}
+
+/**
+ * sdma_v4_0_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Copy GPU buffers using the DMA engine (VEGA10/12).
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void sdma_v4_0_emit_copy_buffer(struct amdgpu_ib *ib,
+				       uint64_t src_offset,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+	ib->ptr[ib->length_dw++] = byte_count - 1;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+}
+
+/**
+ * sdma_v4_0_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine (VEGA10/12).
+ */
+static void sdma_v4_0_emit_fill_buffer(struct amdgpu_ib *ib,
+				       uint32_t src_data,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_CONST_FILL);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = byte_count - 1;
+}
+
+static const struct amdgpu_buffer_funcs sdma_v4_0_buffer_funcs = {
+	.copy_max_bytes = 0x400000,
+	.copy_num_dw = 7,
+	.emit_copy_buffer = sdma_v4_0_emit_copy_buffer,
+
+	.fill_max_bytes = 0x400000,
+	.fill_num_dw = 5,
+	.emit_fill_buffer = sdma_v4_0_emit_fill_buffer,
+};
+
+static void sdma_v4_0_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mman.buffer_funcs == NULL) {
+		adev->mman.buffer_funcs = &sdma_v4_0_buffer_funcs;
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+	}
+}
+
+static const struct amdgpu_vm_pte_funcs sdma_v4_0_vm_pte_funcs = {
+	.copy_pte_num_dw = 7,
+	.copy_pte = sdma_v4_0_vm_copy_pte,
+
+	.write_pte = sdma_v4_0_vm_write_pte,
+	.set_pte_pde = sdma_v4_0_vm_set_pte_pde,
+};
+
+static void sdma_v4_0_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	if (adev->vm_manager.vm_pte_funcs == NULL) {
+		adev->vm_manager.vm_pte_funcs = &sdma_v4_0_vm_pte_funcs;
+		for (i = 0; i < adev->sdma.num_instances; i++)
+			adev->vm_manager.vm_pte_rings[i] =
+				&adev->sdma.instance[i].ring;
+
+		adev->vm_manager.vm_pte_num_rings = adev->sdma.num_instances;
+	}
+}
+
+const struct amdgpu_ip_block_version sdma_v4_0_ip_block = {
+	.type = AMD_IP_BLOCK_TYPE_SDMA,
+	.major = 4,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &sdma_v4_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/sdma_v4_0.h b/sys/dev/pci/drm/amd/amdgpu/sdma_v4_0.h
new file mode 100644
index 00000000000..5c5a7479a06
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/sdma_v4_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SDMA_V4_0_H__
+#define __SDMA_V4_0_H__
+
+extern const struct amd_ip_funcs sdma_v4_0_ip_funcs;
+extern const struct amdgpu_ip_block_version sdma_v4_0_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/si.h b/sys/dev/pci/drm/amd/amdgpu/si.h
new file mode 100644
index 00000000000..06ed7212a0d
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/si.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SI_H__
+#define __SI_H__
+
+#define SI_FLUSH_GPU_TLB_NUM_WREG	2
+
+void si_srbm_select(struct amdgpu_device *adev,
+		     u32 me, u32 pipe, u32 queue, u32 vmid);
+int si_set_ip_blocks(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/si_dma.h b/sys/dev/pci/drm/amd/amdgpu/si_dma.h
new file mode 100644
index 00000000000..5ac1b8452fb
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/si_dma.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SI_DMA_H__
+#define __SI_DMA_H__
+
+extern const struct amdgpu_ip_block_version si_dma_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/si_dpm.h b/sys/dev/pci/drm/amd/amdgpu/si_dpm.h
new file mode 100644
index 00000000000..6b7d292b919
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/si_dpm.h
@@ -0,0 +1,1015 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __SI_DPM_H__
+#define __SI_DPM_H__
+
+#include "amdgpu_atombios.h"
+#include "sislands_smc.h"
+
+#define MC_CG_CONFIG                                    0x96f
+#define MC_ARB_CG                                       0x9fa
+#define		CG_ARB_REQ(x)				((x) << 0)
+#define		CG_ARB_REQ_MASK				(0xff << 0)
+
+#define	MC_ARB_DRAM_TIMING_1				0x9fc
+#define	MC_ARB_DRAM_TIMING_2				0x9fd
+#define	MC_ARB_DRAM_TIMING_3				0x9fe
+#define	MC_ARB_DRAM_TIMING2_1				0x9ff
+#define	MC_ARB_DRAM_TIMING2_2				0xa00
+#define	MC_ARB_DRAM_TIMING2_3				0xa01
+
+#define MAX_NO_OF_MVDD_VALUES 2
+#define MAX_NO_VREG_STEPS 32
+#define NISLANDS_MAX_SMC_PERFORMANCE_LEVELS_PER_SWSTATE 16
+#define SMC_NISLANDS_MC_REGISTER_ARRAY_SIZE 32
+#define SMC_NISLANDS_MC_REGISTER_ARRAY_SET_COUNT 20
+#define RV770_ASI_DFLT                                1000
+#define CYPRESS_HASI_DFLT                               400000
+#define PCIE_PERF_REQ_PECI_GEN1         2
+#define PCIE_PERF_REQ_PECI_GEN2         3
+#define PCIE_PERF_REQ_PECI_GEN3         4
+#define RV770_DEFAULT_VCLK_FREQ  53300 /* 10 khz */
+#define RV770_DEFAULT_DCLK_FREQ  40000 /* 10 khz */
+
+#define SMC_EVERGREEN_MC_REGISTER_ARRAY_SIZE 16
+
+#define RV770_SMC_TABLE_ADDRESS 0xB000
+#define RV770_SMC_PERFORMANCE_LEVELS_PER_SWSTATE    3
+
+#define SMC_STROBE_RATIO    0x0F
+#define SMC_STROBE_ENABLE   0x10
+
+#define SMC_MC_EDC_RD_FLAG  0x01
+#define SMC_MC_EDC_WR_FLAG  0x02
+#define SMC_MC_RTT_ENABLE   0x04
+#define SMC_MC_STUTTER_EN   0x08
+
+#define RV770_SMC_VOLTAGEMASK_VDDC 0
+#define RV770_SMC_VOLTAGEMASK_MVDD 1
+#define RV770_SMC_VOLTAGEMASK_VDDCI 2
+#define RV770_SMC_VOLTAGEMASK_MAX  4
+
+#define NISLANDS_MAX_SMC_PERFORMANCE_LEVELS_PER_SWSTATE 16
+#define NISLANDS_SMC_STROBE_RATIO    0x0F
+#define NISLANDS_SMC_STROBE_ENABLE   0x10
+
+#define NISLANDS_SMC_MC_EDC_RD_FLAG  0x01
+#define NISLANDS_SMC_MC_EDC_WR_FLAG  0x02
+#define NISLANDS_SMC_MC_RTT_ENABLE   0x04
+#define NISLANDS_SMC_MC_STUTTER_EN   0x08
+
+#define MAX_NO_VREG_STEPS 32
+
+#define NISLANDS_SMC_VOLTAGEMASK_VDDC  0
+#define NISLANDS_SMC_VOLTAGEMASK_MVDD  1
+#define NISLANDS_SMC_VOLTAGEMASK_VDDCI 2
+#define NISLANDS_SMC_VOLTAGEMASK_MAX   4
+
+#define SISLANDS_MCREGISTERTABLE_INITIAL_SLOT               0
+#define SISLANDS_MCREGISTERTABLE_ACPI_SLOT                  1
+#define SISLANDS_MCREGISTERTABLE_ULV_SLOT                   2
+#define SISLANDS_MCREGISTERTABLE_FIRST_DRIVERSTATE_SLOT     3
+
+#define SISLANDS_LEAKAGE_INDEX0     0xff01
+#define SISLANDS_MAX_LEAKAGE_COUNT  4
+
+#define SISLANDS_MAX_HARDWARE_POWERLEVELS 5
+#define SISLANDS_INITIAL_STATE_ARB_INDEX    0
+#define SISLANDS_ACPI_STATE_ARB_INDEX       1
+#define SISLANDS_ULV_STATE_ARB_INDEX        2
+#define SISLANDS_DRIVER_STATE_ARB_INDEX     3
+
+#define SISLANDS_DPM2_MAX_PULSE_SKIP        256
+
+#define SISLANDS_DPM2_NEAR_TDP_DEC          10
+#define SISLANDS_DPM2_ABOVE_SAFE_INC        5
+#define SISLANDS_DPM2_BELOW_SAFE_INC        20
+
+#define SISLANDS_DPM2_TDP_SAFE_LIMIT_PERCENT            80
+
+#define SISLANDS_DPM2_MAXPS_PERCENT_H                   99
+#define SISLANDS_DPM2_MAXPS_PERCENT_M                   99
+
+#define SISLANDS_DPM2_SQ_RAMP_MAX_POWER                 0x3FFF
+#define SISLANDS_DPM2_SQ_RAMP_MIN_POWER                 0x12
+#define SISLANDS_DPM2_SQ_RAMP_MAX_POWER_DELTA           0x15
+#define SISLANDS_DPM2_SQ_RAMP_STI_SIZE                  0x1E
+#define SISLANDS_DPM2_SQ_RAMP_LTI_RATIO                 0xF
+
+#define SISLANDS_DPM2_PWREFFICIENCYRATIO_MARGIN         10
+
+#define SISLANDS_VRC_DFLT                               0xC000B3
+#define SISLANDS_ULVVOLTAGECHANGEDELAY_DFLT             1687
+#define SISLANDS_CGULVPARAMETER_DFLT                    0x00040035
+#define SISLANDS_CGULVCONTROL_DFLT                      0x1f007550
+
+#define SI_ASI_DFLT                                10000
+#define SI_BSP_DFLT                                0x41EB
+#define SI_BSU_DFLT                                0x2
+#define SI_AH_DFLT                                 5
+#define SI_RLP_DFLT                                25
+#define SI_RMP_DFLT                                65
+#define SI_LHP_DFLT                                40
+#define SI_LMP_DFLT                                15
+#define SI_TD_DFLT                                 0
+#define SI_UTC_DFLT_00                             0x24
+#define SI_UTC_DFLT_01                             0x22
+#define SI_UTC_DFLT_02                             0x22
+#define SI_UTC_DFLT_03                             0x22
+#define SI_UTC_DFLT_04                             0x22
+#define SI_UTC_DFLT_05                             0x22
+#define SI_UTC_DFLT_06                             0x22
+#define SI_UTC_DFLT_07                             0x22
+#define SI_UTC_DFLT_08                             0x22
+#define SI_UTC_DFLT_09                             0x22
+#define SI_UTC_DFLT_10                             0x22
+#define SI_UTC_DFLT_11                             0x22
+#define SI_UTC_DFLT_12                             0x22
+#define SI_UTC_DFLT_13                             0x22
+#define SI_UTC_DFLT_14                             0x22
+#define SI_DTC_DFLT_00                             0x24
+#define SI_DTC_DFLT_01                             0x22
+#define SI_DTC_DFLT_02                             0x22
+#define SI_DTC_DFLT_03                             0x22
+#define SI_DTC_DFLT_04                             0x22
+#define SI_DTC_DFLT_05                             0x22
+#define SI_DTC_DFLT_06                             0x22
+#define SI_DTC_DFLT_07                             0x22
+#define SI_DTC_DFLT_08                             0x22
+#define SI_DTC_DFLT_09                             0x22
+#define SI_DTC_DFLT_10                             0x22
+#define SI_DTC_DFLT_11                             0x22
+#define SI_DTC_DFLT_12                             0x22
+#define SI_DTC_DFLT_13                             0x22
+#define SI_DTC_DFLT_14                             0x22
+#define SI_VRC_DFLT                                0x0000C003
+#define SI_VOLTAGERESPONSETIME_DFLT                1000
+#define SI_BACKBIASRESPONSETIME_DFLT               1000
+#define SI_VRU_DFLT                                0x3
+#define SI_SPLLSTEPTIME_DFLT                       0x1000
+#define SI_SPLLSTEPUNIT_DFLT                       0x3
+#define SI_TPU_DFLT                                0
+#define SI_TPC_DFLT                                0x200
+#define SI_SSTU_DFLT                               0
+#define SI_SST_DFLT                                0x00C8
+#define SI_GICST_DFLT                              0x200
+#define SI_FCT_DFLT                                0x0400
+#define SI_FCTU_DFLT                               0
+#define SI_CTXCGTT3DRPHC_DFLT                      0x20
+#define SI_CTXCGTT3DRSDC_DFLT                      0x40
+#define SI_VDDC3DOORPHC_DFLT                       0x100
+#define SI_VDDC3DOORSDC_DFLT                       0x7
+#define SI_VDDC3DOORSU_DFLT                        0
+#define SI_MPLLLOCKTIME_DFLT                       100
+#define SI_MPLLRESETTIME_DFLT                      150
+#define SI_VCOSTEPPCT_DFLT                          20
+#define SI_ENDINGVCOSTEPPCT_DFLT                    5
+#define SI_REFERENCEDIVIDER_DFLT                    4
+
+#define SI_PM_NUMBER_OF_TC 15
+#define SI_PM_NUMBER_OF_SCLKS 20
+#define SI_PM_NUMBER_OF_MCLKS 4
+#define SI_PM_NUMBER_OF_VOLTAGE_LEVELS 4
+#define SI_PM_NUMBER_OF_ACTIVITY_LEVELS 3
+
+/* XXX are these ok? */
+#define SI_TEMP_RANGE_MIN (90 * 1000)
+#define SI_TEMP_RANGE_MAX (120 * 1000)
+
+#define FDO_PWM_MODE_STATIC  1
+#define FDO_PWM_MODE_STATIC_RPM 5
+
+enum ni_dc_cac_level
+{
+	NISLANDS_DCCAC_LEVEL_0 = 0,
+	NISLANDS_DCCAC_LEVEL_1,
+	NISLANDS_DCCAC_LEVEL_2,
+	NISLANDS_DCCAC_LEVEL_3,
+	NISLANDS_DCCAC_LEVEL_4,
+	NISLANDS_DCCAC_LEVEL_5,
+	NISLANDS_DCCAC_LEVEL_6,
+	NISLANDS_DCCAC_LEVEL_7,
+	NISLANDS_DCCAC_MAX_LEVELS
+};
+
+enum si_cac_config_reg_type
+{
+	SISLANDS_CACCONFIG_MMR = 0,
+	SISLANDS_CACCONFIG_CGIND,
+	SISLANDS_CACCONFIG_MAX
+};
+
+enum si_power_level {
+	SI_POWER_LEVEL_LOW = 0,
+	SI_POWER_LEVEL_MEDIUM = 1,
+	SI_POWER_LEVEL_HIGH = 2,
+	SI_POWER_LEVEL_CTXSW = 3,
+};
+
+enum si_td {
+	SI_TD_AUTO,
+	SI_TD_UP,
+	SI_TD_DOWN,
+};
+
+enum si_display_watermark {
+	SI_DISPLAY_WATERMARK_LOW = 0,
+	SI_DISPLAY_WATERMARK_HIGH = 1,
+};
+
+enum si_display_gap
+{
+    SI_PM_DISPLAY_GAP_VBLANK_OR_WM = 0,
+    SI_PM_DISPLAY_GAP_VBLANK       = 1,
+    SI_PM_DISPLAY_GAP_WATERMARK    = 2,
+    SI_PM_DISPLAY_GAP_IGNORE       = 3,
+};
+
+extern const struct amdgpu_ip_block_version si_smu_ip_block;
+
+struct ni_leakage_coeffients
+{
+	u32 at;
+	u32 bt;
+	u32 av;
+	u32 bv;
+	s32 t_slope;
+	s32 t_intercept;
+	u32 t_ref;
+};
+
+struct SMC_Evergreen_MCRegisterAddress
+{
+    uint16_t s0;
+    uint16_t s1;
+};
+
+typedef struct SMC_Evergreen_MCRegisterAddress SMC_Evergreen_MCRegisterAddress;
+
+struct evergreen_mc_reg_entry {
+	u32 mclk_max;
+	u32 mc_data[SMC_EVERGREEN_MC_REGISTER_ARRAY_SIZE];
+};
+
+struct evergreen_mc_reg_table {
+	u8 last;
+	u8 num_entries;
+	u16 valid_flag;
+	struct evergreen_mc_reg_entry mc_reg_table_entry[MAX_AC_TIMING_ENTRIES];
+	SMC_Evergreen_MCRegisterAddress mc_reg_address[SMC_EVERGREEN_MC_REGISTER_ARRAY_SIZE];
+};
+
+struct SMC_Evergreen_MCRegisterSet
+{
+    uint32_t value[SMC_EVERGREEN_MC_REGISTER_ARRAY_SIZE];
+};
+
+typedef struct SMC_Evergreen_MCRegisterSet SMC_Evergreen_MCRegisterSet;
+
+struct SMC_Evergreen_MCRegisters
+{
+    uint8_t                             last;
+    uint8_t                             reserved[3];
+    SMC_Evergreen_MCRegisterAddress     address[SMC_EVERGREEN_MC_REGISTER_ARRAY_SIZE];
+    SMC_Evergreen_MCRegisterSet         data[5];
+};
+
+typedef struct SMC_Evergreen_MCRegisters SMC_Evergreen_MCRegisters;
+
+struct SMC_NIslands_MCRegisterSet
+{
+    uint32_t value[SMC_NISLANDS_MC_REGISTER_ARRAY_SIZE];
+};
+
+typedef struct SMC_NIslands_MCRegisterSet SMC_NIslands_MCRegisterSet;
+
+struct ni_mc_reg_entry {
+	u32 mclk_max;
+	u32 mc_data[SMC_NISLANDS_MC_REGISTER_ARRAY_SIZE];
+};
+
+struct SMC_NIslands_MCRegisterAddress
+{
+    uint16_t s0;
+    uint16_t s1;
+};
+
+typedef struct SMC_NIslands_MCRegisterAddress SMC_NIslands_MCRegisterAddress;
+
+struct SMC_NIslands_MCRegisters
+{
+    uint8_t                             last;
+    uint8_t                             reserved[3];
+    SMC_NIslands_MCRegisterAddress      address[SMC_NISLANDS_MC_REGISTER_ARRAY_SIZE];
+    SMC_NIslands_MCRegisterSet          data[SMC_NISLANDS_MC_REGISTER_ARRAY_SET_COUNT];
+};
+
+typedef struct SMC_NIslands_MCRegisters SMC_NIslands_MCRegisters;
+
+struct evergreen_ulv_param {
+	bool supported;
+	struct rv7xx_pl *pl;
+};
+
+struct evergreen_arb_registers {
+	u32 mc_arb_dram_timing;
+	u32 mc_arb_dram_timing2;
+	u32 mc_arb_rfsh_rate;
+	u32 mc_arb_burst_time;
+};
+
+struct at {
+	u32 rlp;
+	u32 rmp;
+	u32 lhp;
+	u32 lmp;
+};
+
+struct ni_clock_registers {
+	u32 cg_spll_func_cntl;
+	u32 cg_spll_func_cntl_2;
+	u32 cg_spll_func_cntl_3;
+	u32 cg_spll_func_cntl_4;
+	u32 cg_spll_spread_spectrum;
+	u32 cg_spll_spread_spectrum_2;
+	u32 mclk_pwrmgt_cntl;
+	u32 dll_cntl;
+	u32 mpll_ad_func_cntl;
+	u32 mpll_ad_func_cntl_2;
+	u32 mpll_dq_func_cntl;
+	u32 mpll_dq_func_cntl_2;
+	u32 mpll_ss1;
+	u32 mpll_ss2;
+};
+
+struct RV770_SMC_SCLK_VALUE
+{
+    uint32_t        vCG_SPLL_FUNC_CNTL;
+    uint32_t        vCG_SPLL_FUNC_CNTL_2;
+    uint32_t        vCG_SPLL_FUNC_CNTL_3;
+    uint32_t        vCG_SPLL_SPREAD_SPECTRUM;
+    uint32_t        vCG_SPLL_SPREAD_SPECTRUM_2;
+    uint32_t        sclk_value;
+};
+
+typedef struct RV770_SMC_SCLK_VALUE RV770_SMC_SCLK_VALUE;
+
+struct RV770_SMC_MCLK_VALUE
+{
+    uint32_t        vMPLL_AD_FUNC_CNTL;
+    uint32_t        vMPLL_AD_FUNC_CNTL_2;
+    uint32_t        vMPLL_DQ_FUNC_CNTL;
+    uint32_t        vMPLL_DQ_FUNC_CNTL_2;
+    uint32_t        vMCLK_PWRMGT_CNTL;
+    uint32_t        vDLL_CNTL;
+    uint32_t        vMPLL_SS;
+    uint32_t        vMPLL_SS2;
+    uint32_t        mclk_value;
+};
+
+typedef struct RV770_SMC_MCLK_VALUE RV770_SMC_MCLK_VALUE;
+
+
+struct RV730_SMC_MCLK_VALUE
+{
+    uint32_t        vMCLK_PWRMGT_CNTL;
+    uint32_t        vDLL_CNTL;
+    uint32_t        vMPLL_FUNC_CNTL;
+    uint32_t        vMPLL_FUNC_CNTL2;
+    uint32_t        vMPLL_FUNC_CNTL3;
+    uint32_t        vMPLL_SS;
+    uint32_t        vMPLL_SS2;
+    uint32_t        mclk_value;
+};
+
+typedef struct RV730_SMC_MCLK_VALUE RV730_SMC_MCLK_VALUE;
+
+struct RV770_SMC_VOLTAGE_VALUE
+{
+    uint16_t             value;
+    uint8_t              index;
+    uint8_t              padding;
+};
+
+typedef struct RV770_SMC_VOLTAGE_VALUE RV770_SMC_VOLTAGE_VALUE;
+
+union RV7XX_SMC_MCLK_VALUE
+{
+    RV770_SMC_MCLK_VALUE    mclk770;
+    RV730_SMC_MCLK_VALUE    mclk730;
+};
+
+typedef union RV7XX_SMC_MCLK_VALUE RV7XX_SMC_MCLK_VALUE, *LPRV7XX_SMC_MCLK_VALUE;
+
+struct RV770_SMC_HW_PERFORMANCE_LEVEL
+{
+    uint8_t                 arbValue;
+    union{
+        uint8_t             seqValue;
+        uint8_t             ACIndex;
+    };
+    uint8_t                 displayWatermark;
+    uint8_t                 gen2PCIE;
+    uint8_t                 gen2XSP;
+    uint8_t                 backbias;
+    uint8_t                 strobeMode;
+    uint8_t                 mcFlags;
+    uint32_t                aT;
+    uint32_t                bSP;
+    RV770_SMC_SCLK_VALUE    sclk;
+    RV7XX_SMC_MCLK_VALUE    mclk;
+    RV770_SMC_VOLTAGE_VALUE vddc;
+    RV770_SMC_VOLTAGE_VALUE mvdd;
+    RV770_SMC_VOLTAGE_VALUE vddci;
+    uint8_t                 reserved1;
+    uint8_t                 reserved2;
+    uint8_t                 stateFlags;
+    uint8_t                 padding;
+};
+
+typedef struct RV770_SMC_HW_PERFORMANCE_LEVEL RV770_SMC_HW_PERFORMANCE_LEVEL;
+
+struct RV770_SMC_SWSTATE
+{
+    uint8_t           flags;
+    uint8_t           padding1;
+    uint8_t           padding2;
+    uint8_t           padding3;
+    RV770_SMC_HW_PERFORMANCE_LEVEL levels[RV770_SMC_PERFORMANCE_LEVELS_PER_SWSTATE];
+};
+
+typedef struct RV770_SMC_SWSTATE RV770_SMC_SWSTATE;
+
+struct RV770_SMC_VOLTAGEMASKTABLE
+{
+    uint8_t  highMask[RV770_SMC_VOLTAGEMASK_MAX];
+    uint32_t lowMask[RV770_SMC_VOLTAGEMASK_MAX];
+};
+
+typedef struct RV770_SMC_VOLTAGEMASKTABLE RV770_SMC_VOLTAGEMASKTABLE;
+
+struct RV770_SMC_STATETABLE
+{
+    uint8_t             thermalProtectType;
+    uint8_t             systemFlags;
+    uint8_t             maxVDDCIndexInPPTable;
+    uint8_t             extraFlags;
+    uint8_t             highSMIO[MAX_NO_VREG_STEPS];
+    uint32_t            lowSMIO[MAX_NO_VREG_STEPS];
+    RV770_SMC_VOLTAGEMASKTABLE voltageMaskTable;
+    RV770_SMC_SWSTATE   initialState;
+    RV770_SMC_SWSTATE   ACPIState;
+    RV770_SMC_SWSTATE   driverState;
+    RV770_SMC_SWSTATE   ULVState;
+};
+
+typedef struct RV770_SMC_STATETABLE RV770_SMC_STATETABLE;
+
+struct vddc_table_entry {
+	u16 vddc;
+	u8 vddc_index;
+	u8 high_smio;
+	u32 low_smio;
+};
+
+struct rv770_clock_registers {
+	u32 cg_spll_func_cntl;
+	u32 cg_spll_func_cntl_2;
+	u32 cg_spll_func_cntl_3;
+	u32 cg_spll_spread_spectrum;
+	u32 cg_spll_spread_spectrum_2;
+	u32 mpll_ad_func_cntl;
+	u32 mpll_ad_func_cntl_2;
+	u32 mpll_dq_func_cntl;
+	u32 mpll_dq_func_cntl_2;
+	u32 mclk_pwrmgt_cntl;
+	u32 dll_cntl;
+	u32 mpll_ss1;
+	u32 mpll_ss2;
+};
+
+struct rv730_clock_registers {
+	u32 cg_spll_func_cntl;
+	u32 cg_spll_func_cntl_2;
+	u32 cg_spll_func_cntl_3;
+	u32 cg_spll_spread_spectrum;
+	u32 cg_spll_spread_spectrum_2;
+	u32 mclk_pwrmgt_cntl;
+	u32 dll_cntl;
+	u32 mpll_func_cntl;
+	u32 mpll_func_cntl2;
+	u32 mpll_func_cntl3;
+	u32 mpll_ss;
+	u32 mpll_ss2;
+};
+
+union r7xx_clock_registers {
+	struct rv770_clock_registers rv770;
+	struct rv730_clock_registers rv730;
+};
+
+struct rv7xx_power_info {
+	/* flags */
+	bool mem_gddr5;
+	bool pcie_gen2;
+	bool dynamic_pcie_gen2;
+	bool acpi_pcie_gen2;
+	bool boot_in_gen2;
+	bool voltage_control; /* vddc */
+	bool mvdd_control;
+	bool sclk_ss;
+	bool mclk_ss;
+	bool dynamic_ss;
+	bool gfx_clock_gating;
+	bool mg_clock_gating;
+	bool mgcgtssm;
+	bool power_gating;
+	bool thermal_protection;
+	bool display_gap;
+	bool dcodt;
+	bool ulps;
+	/* registers */
+	union r7xx_clock_registers clk_regs;
+	u32 s0_vid_lower_smio_cntl;
+	/* voltage */
+	u32 vddc_mask_low;
+	u32 mvdd_mask_low;
+	u32 mvdd_split_frequency;
+	u32 mvdd_low_smio[MAX_NO_OF_MVDD_VALUES];
+	u16 max_vddc;
+	u16 max_vddc_in_table;
+	u16 min_vddc_in_table;
+	struct vddc_table_entry vddc_table[MAX_NO_VREG_STEPS];
+	u8 valid_vddc_entries;
+	/* dc odt */
+	u32 mclk_odt_threshold;
+	u8 odt_value_0[2];
+	u8 odt_value_1[2];
+	/* stored values */
+	u32 boot_sclk;
+	u16 acpi_vddc;
+	u32 ref_div;
+	u32 active_auto_throttle_sources;
+	u32 mclk_stutter_mode_threshold;
+	u32 mclk_strobe_mode_threshold;
+	u32 mclk_edc_enable_threshold;
+	u32 bsp;
+	u32 bsu;
+	u32 pbsp;
+	u32 pbsu;
+	u32 dsp;
+	u32 psp;
+	u32 asi;
+	u32 pasi;
+	u32 vrc;
+	u32 restricted_levels;
+	u32 rlp;
+	u32 rmp;
+	u32 lhp;
+	u32 lmp;
+	/* smc offsets */
+	u16 state_table_start;
+	u16 soft_regs_start;
+	u16 sram_end;
+	/* scratch structs */
+	RV770_SMC_STATETABLE smc_statetable;
+};
+
+struct rv7xx_pl {
+	u32 sclk;
+	u32 mclk;
+	u16 vddc;
+	u16 vddci; /* eg+ only */
+	u32 flags;
+	enum amdgpu_pcie_gen pcie_gen; /* si+ only */
+};
+
+struct rv7xx_ps {
+	struct rv7xx_pl high;
+	struct rv7xx_pl medium;
+	struct rv7xx_pl low;
+	bool dc_compatible;
+};
+
+struct si_ps {
+	u16 performance_level_count;
+	bool dc_compatible;
+	struct rv7xx_pl performance_levels[NISLANDS_MAX_SMC_PERFORMANCE_LEVELS_PER_SWSTATE];
+};
+
+struct ni_mc_reg_table {
+	u8 last;
+	u8 num_entries;
+	u16 valid_flag;
+	struct ni_mc_reg_entry mc_reg_table_entry[MAX_AC_TIMING_ENTRIES];
+	SMC_NIslands_MCRegisterAddress mc_reg_address[SMC_NISLANDS_MC_REGISTER_ARRAY_SIZE];
+};
+
+struct ni_cac_data
+{
+	struct ni_leakage_coeffients leakage_coefficients;
+	u32 i_leakage;
+	s32 leakage_minimum_temperature;
+	u32 pwr_const;
+	u32 dc_cac_value;
+	u32 bif_cac_value;
+	u32 lkge_pwr;
+	u8 mc_wr_weight;
+	u8 mc_rd_weight;
+	u8 allow_ovrflw;
+	u8 num_win_tdp;
+	u8 l2num_win_tdp;
+	u8 lts_truncate_n;
+};
+
+struct evergreen_power_info {
+	/* must be first! */
+	struct rv7xx_power_info rv7xx;
+	/* flags */
+	bool vddci_control;
+	bool dynamic_ac_timing;
+	bool abm;
+	bool mcls;
+	bool light_sleep;
+	bool memory_transition;
+	bool pcie_performance_request;
+	bool pcie_performance_request_registered;
+	bool sclk_deep_sleep;
+	bool dll_default_on;
+	bool ls_clock_gating;
+	bool smu_uvd_hs;
+	bool uvd_enabled;
+	/* stored values */
+	u16 acpi_vddci;
+	u8 mvdd_high_index;
+	u8 mvdd_low_index;
+	u32 mclk_edc_wr_enable_threshold;
+	struct evergreen_mc_reg_table mc_reg_table;
+	struct atom_voltage_table vddc_voltage_table;
+	struct atom_voltage_table vddci_voltage_table;
+	struct evergreen_arb_registers bootup_arb_registers;
+	struct evergreen_ulv_param ulv;
+	struct at ats[2];
+	/* smc offsets */
+	u16 mc_reg_table_start;
+	struct amdgpu_ps current_rps;
+	struct rv7xx_ps current_ps;
+	struct amdgpu_ps requested_rps;
+	struct rv7xx_ps requested_ps;
+};
+
+struct PP_NIslands_Dpm2PerfLevel
+{
+    uint8_t     MaxPS;
+    uint8_t     TgtAct;
+    uint8_t     MaxPS_StepInc;
+    uint8_t     MaxPS_StepDec;
+    uint8_t     PSST;
+    uint8_t     NearTDPDec;
+    uint8_t     AboveSafeInc;
+    uint8_t     BelowSafeInc;
+    uint8_t     PSDeltaLimit;
+    uint8_t     PSDeltaWin;
+    uint8_t     Reserved[6];
+};
+
+typedef struct PP_NIslands_Dpm2PerfLevel PP_NIslands_Dpm2PerfLevel;
+
+struct PP_NIslands_DPM2Parameters
+{
+    uint32_t    TDPLimit;
+    uint32_t    NearTDPLimit;
+    uint32_t    SafePowerLimit;
+    uint32_t    PowerBoostLimit;
+};
+typedef struct PP_NIslands_DPM2Parameters PP_NIslands_DPM2Parameters;
+
+struct NISLANDS_SMC_SCLK_VALUE
+{
+    uint32_t        vCG_SPLL_FUNC_CNTL;
+    uint32_t        vCG_SPLL_FUNC_CNTL_2;
+    uint32_t        vCG_SPLL_FUNC_CNTL_3;
+    uint32_t        vCG_SPLL_FUNC_CNTL_4;
+    uint32_t        vCG_SPLL_SPREAD_SPECTRUM;
+    uint32_t        vCG_SPLL_SPREAD_SPECTRUM_2;
+    uint32_t        sclk_value;
+};
+
+typedef struct NISLANDS_SMC_SCLK_VALUE NISLANDS_SMC_SCLK_VALUE;
+
+struct NISLANDS_SMC_MCLK_VALUE
+{
+    uint32_t        vMPLL_FUNC_CNTL;
+    uint32_t        vMPLL_FUNC_CNTL_1;
+    uint32_t        vMPLL_FUNC_CNTL_2;
+    uint32_t        vMPLL_AD_FUNC_CNTL;
+    uint32_t        vMPLL_AD_FUNC_CNTL_2;
+    uint32_t        vMPLL_DQ_FUNC_CNTL;
+    uint32_t        vMPLL_DQ_FUNC_CNTL_2;
+    uint32_t        vMCLK_PWRMGT_CNTL;
+    uint32_t        vDLL_CNTL;
+    uint32_t        vMPLL_SS;
+    uint32_t        vMPLL_SS2;
+    uint32_t        mclk_value;
+};
+
+typedef struct NISLANDS_SMC_MCLK_VALUE NISLANDS_SMC_MCLK_VALUE;
+
+struct NISLANDS_SMC_VOLTAGE_VALUE
+{
+    uint16_t             value;
+    uint8_t              index;
+    uint8_t              padding;
+};
+
+typedef struct NISLANDS_SMC_VOLTAGE_VALUE NISLANDS_SMC_VOLTAGE_VALUE;
+
+struct NISLANDS_SMC_HW_PERFORMANCE_LEVEL
+{
+    uint8_t                     arbValue;
+    uint8_t                     ACIndex;
+    uint8_t                     displayWatermark;
+    uint8_t                     gen2PCIE;
+    uint8_t                     reserved1;
+    uint8_t                     reserved2;
+    uint8_t                     strobeMode;
+    uint8_t                     mcFlags;
+    uint32_t                    aT;
+    uint32_t                    bSP;
+    NISLANDS_SMC_SCLK_VALUE     sclk;
+    NISLANDS_SMC_MCLK_VALUE     mclk;
+    NISLANDS_SMC_VOLTAGE_VALUE  vddc;
+    NISLANDS_SMC_VOLTAGE_VALUE  mvdd;
+    NISLANDS_SMC_VOLTAGE_VALUE  vddci;
+    NISLANDS_SMC_VOLTAGE_VALUE  std_vddc;
+    uint32_t                    powergate_en;
+    uint8_t                     hUp;
+    uint8_t                     hDown;
+    uint8_t                     stateFlags;
+    uint8_t                     arbRefreshState;
+    uint32_t                    SQPowerThrottle;
+    uint32_t                    SQPowerThrottle_2;
+    uint32_t                    reserved[2];
+    PP_NIslands_Dpm2PerfLevel   dpm2;
+};
+
+typedef struct NISLANDS_SMC_HW_PERFORMANCE_LEVEL NISLANDS_SMC_HW_PERFORMANCE_LEVEL;
+
+struct NISLANDS_SMC_SWSTATE
+{
+    uint8_t                             flags;
+    uint8_t                             levelCount;
+    uint8_t                             padding2;
+    uint8_t                             padding3;
+    NISLANDS_SMC_HW_PERFORMANCE_LEVEL   levels[1];
+};
+
+typedef struct NISLANDS_SMC_SWSTATE NISLANDS_SMC_SWSTATE;
+
+struct NISLANDS_SMC_VOLTAGEMASKTABLE
+{
+    uint8_t  highMask[NISLANDS_SMC_VOLTAGEMASK_MAX];
+    uint32_t lowMask[NISLANDS_SMC_VOLTAGEMASK_MAX];
+};
+
+typedef struct NISLANDS_SMC_VOLTAGEMASKTABLE NISLANDS_SMC_VOLTAGEMASKTABLE;
+
+#define NISLANDS_MAX_NO_VREG_STEPS 32
+
+struct NISLANDS_SMC_STATETABLE
+{
+    uint8_t                             thermalProtectType;
+    uint8_t                             systemFlags;
+    uint8_t                             maxVDDCIndexInPPTable;
+    uint8_t                             extraFlags;
+    uint8_t                             highSMIO[NISLANDS_MAX_NO_VREG_STEPS];
+    uint32_t                            lowSMIO[NISLANDS_MAX_NO_VREG_STEPS];
+    NISLANDS_SMC_VOLTAGEMASKTABLE       voltageMaskTable;
+    PP_NIslands_DPM2Parameters          dpm2Params;
+    NISLANDS_SMC_SWSTATE                initialState;
+    NISLANDS_SMC_SWSTATE                ACPIState;
+    NISLANDS_SMC_SWSTATE                ULVState;
+    NISLANDS_SMC_SWSTATE                driverState;
+    NISLANDS_SMC_HW_PERFORMANCE_LEVEL   dpmLevels[NISLANDS_MAX_SMC_PERFORMANCE_LEVELS_PER_SWSTATE - 1];
+};
+
+typedef struct NISLANDS_SMC_STATETABLE NISLANDS_SMC_STATETABLE;
+
+struct ni_power_info {
+	/* must be first! */
+	struct evergreen_power_info eg;
+	struct ni_clock_registers clock_registers;
+	struct ni_mc_reg_table mc_reg_table;
+	u32 mclk_rtt_mode_threshold;
+	/* flags */
+	bool use_power_boost_limit;
+	bool support_cac_long_term_average;
+	bool cac_enabled;
+	bool cac_configuration_required;
+	bool driver_calculate_cac_leakage;
+	bool pc_enabled;
+	bool enable_power_containment;
+	bool enable_cac;
+	bool enable_sq_ramping;
+	/* smc offsets */
+	u16 arb_table_start;
+	u16 fan_table_start;
+	u16 cac_table_start;
+	u16 spll_table_start;
+	/* CAC stuff */
+	struct ni_cac_data cac_data;
+	u32 dc_cac_table[NISLANDS_DCCAC_MAX_LEVELS];
+	const struct ni_cac_weights *cac_weights;
+	u8 lta_window_size;
+	u8 lts_truncate;
+	struct si_ps current_ps;
+	struct si_ps requested_ps;
+	/* scratch structs */
+	SMC_NIslands_MCRegisters smc_mc_reg_table;
+	NISLANDS_SMC_STATETABLE smc_statetable;
+};
+
+struct si_cac_config_reg
+{
+	u32 offset;
+	u32 mask;
+	u32 shift;
+	u32 value;
+	enum si_cac_config_reg_type type;
+};
+
+struct si_powertune_data
+{
+	u32 cac_window;
+	u32 l2_lta_window_size_default;
+	u8 lts_truncate_default;
+	u8 shift_n_default;
+	u8 operating_temp;
+	struct ni_leakage_coeffients leakage_coefficients;
+	u32 fixed_kt;
+	u32 lkge_lut_v0_percent;
+	u8 dc_cac[NISLANDS_DCCAC_MAX_LEVELS];
+	bool enable_powertune_by_default;
+};
+
+struct si_dyn_powertune_data
+{
+	u32 cac_leakage;
+	s32 leakage_minimum_temperature;
+	u32 wintime;
+	u32 l2_lta_window_size;
+	u8 lts_truncate;
+	u8 shift_n;
+	u8 dc_pwr_value;
+	bool disable_uvd_powertune;
+};
+
+struct si_dte_data
+{
+	u32 tau[SMC_SISLANDS_DTE_MAX_FILTER_STAGES];
+	u32 r[SMC_SISLANDS_DTE_MAX_FILTER_STAGES];
+	u32 k;
+	u32 t0;
+	u32 max_t;
+	u8 window_size;
+	u8 temp_select;
+	u8 dte_mode;
+	u8 tdep_count;
+	u8 t_limits[SMC_SISLANDS_DTE_MAX_TEMPERATURE_DEPENDENT_ARRAY_SIZE];
+	u32 tdep_tau[SMC_SISLANDS_DTE_MAX_TEMPERATURE_DEPENDENT_ARRAY_SIZE];
+	u32 tdep_r[SMC_SISLANDS_DTE_MAX_TEMPERATURE_DEPENDENT_ARRAY_SIZE];
+	u32 t_threshold;
+	bool enable_dte_by_default;
+};
+
+struct si_clock_registers {
+	u32 cg_spll_func_cntl;
+	u32 cg_spll_func_cntl_2;
+	u32 cg_spll_func_cntl_3;
+	u32 cg_spll_func_cntl_4;
+	u32 cg_spll_spread_spectrum;
+	u32 cg_spll_spread_spectrum_2;
+	u32 dll_cntl;
+	u32 mclk_pwrmgt_cntl;
+	u32 mpll_ad_func_cntl;
+	u32 mpll_dq_func_cntl;
+	u32 mpll_func_cntl;
+	u32 mpll_func_cntl_1;
+	u32 mpll_func_cntl_2;
+	u32 mpll_ss1;
+	u32 mpll_ss2;
+};
+
+struct si_mc_reg_entry {
+	u32 mclk_max;
+	u32 mc_data[SMC_SISLANDS_MC_REGISTER_ARRAY_SIZE];
+};
+
+struct si_mc_reg_table {
+	u8 last;
+	u8 num_entries;
+	u16 valid_flag;
+	struct si_mc_reg_entry mc_reg_table_entry[MAX_AC_TIMING_ENTRIES];
+	SMC_NIslands_MCRegisterAddress mc_reg_address[SMC_SISLANDS_MC_REGISTER_ARRAY_SIZE];
+};
+
+struct si_leakage_voltage_entry
+{
+	u16 voltage;
+	u16 leakage_index;
+};
+
+struct si_leakage_voltage
+{
+	u16 count;
+	struct si_leakage_voltage_entry entries[SISLANDS_MAX_LEAKAGE_COUNT];
+};
+
+
+struct si_ulv_param {
+	bool supported;
+	u32 cg_ulv_control;
+	u32 cg_ulv_parameter;
+	u32 volt_change_delay;
+	struct rv7xx_pl pl;
+	bool one_pcie_lane_in_ulv;
+};
+
+struct si_power_info {
+	/* must be first! */
+	struct ni_power_info ni;
+	struct si_clock_registers clock_registers;
+	struct si_mc_reg_table mc_reg_table;
+	struct atom_voltage_table mvdd_voltage_table;
+	struct atom_voltage_table vddc_phase_shed_table;
+	struct si_leakage_voltage leakage_voltage;
+	u16 mvdd_bootup_value;
+	struct si_ulv_param ulv;
+	u32 max_cu;
+	/* pcie gen */
+	enum amdgpu_pcie_gen force_pcie_gen;
+	enum amdgpu_pcie_gen boot_pcie_gen;
+	enum amdgpu_pcie_gen acpi_pcie_gen;
+	u32 sys_pcie_mask;
+	/* flags */
+	bool enable_dte;
+	bool enable_ppm;
+	bool vddc_phase_shed_control;
+	bool pspp_notify_required;
+	bool sclk_deep_sleep_above_low;
+	bool voltage_control_svi2;
+	bool vddci_control_svi2;
+	/* smc offsets */
+	u32 sram_end;
+	u32 state_table_start;
+	u32 soft_regs_start;
+	u32 mc_reg_table_start;
+	u32 arb_table_start;
+	u32 cac_table_start;
+	u32 dte_table_start;
+	u32 spll_table_start;
+	u32 papm_cfg_table_start;
+	u32 fan_table_start;
+	/* CAC stuff */
+	const struct si_cac_config_reg *cac_weights;
+	const struct si_cac_config_reg *lcac_config;
+	const struct si_cac_config_reg *cac_override;
+	const struct si_powertune_data *powertune_data;
+	struct si_dyn_powertune_data dyn_powertune_data;
+	/* DTE stuff */
+	struct si_dte_data dte_data;
+	/* scratch structs */
+	SMC_SIslands_MCRegisters smc_mc_reg_table;
+	SISLANDS_SMC_STATETABLE smc_statetable;
+	PP_SIslands_PAPMParameters papm_parm;
+	/* SVI2 */
+	u8 svd_gpio_id;
+	u8 svc_gpio_id;
+	/* fan control */
+	bool fan_ctrl_is_in_default_mode;
+	u32 t_min;
+	u32 fan_ctrl_default_mode;
+	bool fan_is_controlled_by_smc;
+};
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/si_enums.h b/sys/dev/pci/drm/amd/amdgpu/si_enums.h
new file mode 100644
index 00000000000..dc9e0e6b455
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/si_enums.h
@@ -0,0 +1,272 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef SI_ENUMS_H
+#define SI_ENUMS_H
+
+#define VBLANK_INT_MASK                (1 << 0)
+#define DC_HPDx_INT_EN                 (1 << 16)
+#define VBLANK_ACK                     (1 << 4)
+#define VLINE_ACK                      (1 << 4)
+
+#define CURSOR_WIDTH 64
+#define CURSOR_HEIGHT 64
+
+#define VGA_VSTATUS_CNTL               0xFFFCFFFF
+#define PRIORITY_MARK_MASK             0x7fff
+#define PRIORITY_OFF                   (1 << 16)
+#define PRIORITY_ALWAYS_ON             (1 << 20)
+#define INTERLEAVE_EN                  (1 << 0)
+
+#define LATENCY_WATERMARK_MASK(x)      ((x) << 16)
+#define DC_LB_MEMORY_CONFIG(x)         ((x) << 20)
+#define ICON_DEGAMMA_MODE(x)           (((x) & 0x3) << 8)
+
+#define GRPH_ENDIAN_SWAP(x)            (((x) & 0x3) << 0)
+#define GRPH_ENDIAN_NONE               0
+#define GRPH_ENDIAN_8IN16              1
+#define GRPH_ENDIAN_8IN32              2
+#define GRPH_ENDIAN_8IN64              3
+
+#define GRPH_DEPTH(x)                  (((x) & 0x3) << 0)
+#define GRPH_DEPTH_8BPP                0
+#define GRPH_DEPTH_16BPP               1
+#define GRPH_DEPTH_32BPP               2
+
+#define GRPH_FORMAT(x)                 (((x) & 0x7) << 8)
+#define GRPH_FORMAT_INDEXED            0
+#define GRPH_FORMAT_ARGB1555           0
+#define GRPH_FORMAT_ARGB565            1
+#define GRPH_FORMAT_ARGB4444           2
+#define GRPH_FORMAT_AI88               3
+#define GRPH_FORMAT_MONO16             4
+#define GRPH_FORMAT_BGRA5551           5
+#define GRPH_FORMAT_ARGB8888           0
+#define GRPH_FORMAT_ARGB2101010        1
+#define GRPH_FORMAT_32BPP_DIG          2
+#define GRPH_FORMAT_8B_ARGB2101010     3
+#define GRPH_FORMAT_BGRA1010102        4
+#define GRPH_FORMAT_8B_BGRA1010102     5
+#define GRPH_FORMAT_RGB111110          6
+#define GRPH_FORMAT_BGR101111          7
+
+#define GRPH_NUM_BANKS(x)              (((x) & 0x3) << 2)
+#define GRPH_ARRAY_MODE(x)             (((x) & 0x7) << 20)
+#define GRPH_ARRAY_LINEAR_GENERAL      0
+#define GRPH_ARRAY_LINEAR_ALIGNED      1
+#define GRPH_ARRAY_1D_TILED_THIN1      2
+#define GRPH_ARRAY_2D_TILED_THIN1      4
+#define GRPH_TILE_SPLIT(x)             (((x) & 0x7) << 13)
+#define GRPH_BANK_WIDTH(x)             (((x) & 0x3) << 6)
+#define GRPH_BANK_HEIGHT(x)            (((x) & 0x3) << 11)
+#define GRPH_MACRO_TILE_ASPECT(x)      (((x) & 0x3) << 18)
+#define GRPH_ARRAY_MODE(x)             (((x) & 0x7) << 20)
+#define GRPH_PIPE_CONFIG(x)                   (((x) & 0x1f) << 24)
+
+#define CURSOR_EN                      (1 << 0)
+#define CURSOR_MODE(x)                 (((x) & 0x3) << 8)
+#define CURSOR_MONO                    0
+#define CURSOR_24_1                    1
+#define CURSOR_24_8_PRE_MULT           2
+#define CURSOR_24_8_UNPRE_MULT         3
+#define CURSOR_2X_MAGNIFY              (1 << 16)
+#define CURSOR_FORCE_MC_ON             (1 << 20)
+#define CURSOR_URGENT_CONTROL(x)       (((x) & 0x7) << 24)
+#define CURSOR_URGENT_ALWAYS           0
+#define CURSOR_URGENT_1_8              1
+#define CURSOR_URGENT_1_4              2
+#define CURSOR_URGENT_3_8              3
+#define CURSOR_URGENT_1_2              4
+#define CURSOR_UPDATE_PENDING          (1 << 0)
+#define CURSOR_UPDATE_TAKEN            (1 << 1)
+#define CURSOR_UPDATE_LOCK             (1 << 16)
+#define CURSOR_DISABLE_MULTIPLE_UPDATE (1 << 24)
+
+#define AMDGPU_NUM_OF_VMIDS                     8
+#define SI_CRTC0_REGISTER_OFFSET                0
+#define SI_CRTC1_REGISTER_OFFSET                0x300
+#define SI_CRTC2_REGISTER_OFFSET                0x2600
+#define SI_CRTC3_REGISTER_OFFSET                0x2900
+#define SI_CRTC4_REGISTER_OFFSET                0x2c00
+#define SI_CRTC5_REGISTER_OFFSET                0x2f00
+
+#define DMA0_REGISTER_OFFSET 0x000
+#define DMA1_REGISTER_OFFSET 0x200
+#define ES_AND_GS_AUTO       3
+#define RADEON_PACKET_TYPE3  3
+#define CE_PARTITION_BASE    3
+#define BUF_SWAP_32BIT       (2 << 16)
+
+#define GFX_POWER_STATUS                           (1 << 1)
+#define GFX_CLOCK_STATUS                           (1 << 2)
+#define GFX_LS_STATUS                              (1 << 3)
+#define RLC_BUSY_STATUS                            (1 << 0)
+
+#define RLC_PUD(x)                               ((x) << 0)
+#define RLC_PUD_MASK                             (0xff << 0)
+#define RLC_PDD(x)                               ((x) << 8)
+#define RLC_PDD_MASK                             (0xff << 8)
+#define RLC_TTPD(x)                              ((x) << 16)
+#define RLC_TTPD_MASK                            (0xff << 16)
+#define RLC_MSD(x)                               ((x) << 24)
+#define RLC_MSD_MASK                             (0xff << 24)
+#define WRITE_DATA_ENGINE_SEL(x) ((x) << 30)
+#define WRITE_DATA_DST_SEL(x) ((x) << 8)
+#define EVENT_TYPE(x) ((x) << 0)
+#define EVENT_INDEX(x) ((x) << 8)
+#define WAIT_REG_MEM_MEM_SPACE(x)               ((x) << 4)
+#define WAIT_REG_MEM_FUNCTION(x)                ((x) << 0)
+#define WAIT_REG_MEM_ENGINE(x)                  ((x) << 8)
+
+#define GFX6_NUM_GFX_RINGS     1
+#define GFX6_NUM_COMPUTE_RINGS 2
+#define RLC_SAVE_AND_RESTORE_STARTING_OFFSET 0x90
+#define RLC_CLEAR_STATE_DESCRIPTOR_OFFSET    0x3D
+
+#define TAHITI_GB_ADDR_CONFIG_GOLDEN        0x12011003
+#define VERDE_GB_ADDR_CONFIG_GOLDEN         0x02010002
+#define HAINAN_GB_ADDR_CONFIG_GOLDEN        0x02011003
+
+#define PACKET3(op, n)  ((RADEON_PACKET_TYPE3 << 30) |                  \
+                         (((op) & 0xFF) << 8) |                         \
+                         ((n) & 0x3FFF) << 16)
+#define PACKET3_COMPUTE(op, n) (PACKET3(op, n) | 1 << 1)
+#define	PACKET3_NOP					0x10
+#define	PACKET3_SET_BASE				0x11
+#define		PACKET3_BASE_INDEX(x)                  ((x) << 0)
+#define	PACKET3_CLEAR_STATE				0x12
+#define	PACKET3_INDEX_BUFFER_SIZE			0x13
+#define	PACKET3_DISPATCH_DIRECT				0x15
+#define	PACKET3_DISPATCH_INDIRECT			0x16
+#define	PACKET3_ALLOC_GDS				0x1B
+#define	PACKET3_WRITE_GDS_RAM				0x1C
+#define	PACKET3_ATOMIC_GDS				0x1D
+#define	PACKET3_ATOMIC					0x1E
+#define	PACKET3_OCCLUSION_QUERY				0x1F
+#define	PACKET3_SET_PREDICATION				0x20
+#define	PACKET3_REG_RMW					0x21
+#define	PACKET3_COND_EXEC				0x22
+#define	PACKET3_PRED_EXEC				0x23
+#define	PACKET3_DRAW_INDIRECT				0x24
+#define	PACKET3_DRAW_INDEX_INDIRECT			0x25
+#define	PACKET3_INDEX_BASE				0x26
+#define	PACKET3_DRAW_INDEX_2				0x27
+#define	PACKET3_CONTEXT_CONTROL				0x28
+#define	PACKET3_INDEX_TYPE				0x2A
+#define	PACKET3_DRAW_INDIRECT_MULTI			0x2C
+#define	PACKET3_DRAW_INDEX_AUTO				0x2D
+#define	PACKET3_DRAW_INDEX_IMMD				0x2E
+#define	PACKET3_NUM_INSTANCES				0x2F
+#define	PACKET3_DRAW_INDEX_MULTI_AUTO			0x30
+#define	PACKET3_INDIRECT_BUFFER_CONST			0x31
+#define	PACKET3_INDIRECT_BUFFER				0x3F
+#define	PACKET3_STRMOUT_BUFFER_UPDATE			0x34
+#define	PACKET3_DRAW_INDEX_OFFSET_2			0x35
+#define	PACKET3_DRAW_INDEX_MULTI_ELEMENT		0x36
+#define	PACKET3_WRITE_DATA				0x37
+#define	PACKET3_DRAW_INDEX_INDIRECT_MULTI		0x38
+#define	PACKET3_MEM_SEMAPHORE				0x39
+#define	PACKET3_MPEG_INDEX				0x3A
+#define	PACKET3_COPY_DW					0x3B
+#define	PACKET3_WAIT_REG_MEM				0x3C
+#define	PACKET3_MEM_WRITE				0x3D
+#define	PACKET3_COPY_DATA				0x40
+#define	PACKET3_CP_DMA					0x41
+#              define PACKET3_CP_DMA_DST_SEL(x)    ((x) << 20)
+#              define PACKET3_CP_DMA_ENGINE(x)     ((x) << 27)
+#              define PACKET3_CP_DMA_SRC_SEL(x)    ((x) << 29)
+#              define PACKET3_CP_DMA_CP_SYNC       (1 << 31)
+#              define PACKET3_CP_DMA_DIS_WC        (1 << 21)
+#              define PACKET3_CP_DMA_CMD_SRC_SWAP(x) ((x) << 22)
+#              define PACKET3_CP_DMA_CMD_DST_SWAP(x) ((x) << 24)
+#              define PACKET3_CP_DMA_CMD_SAS       (1 << 26)
+#              define PACKET3_CP_DMA_CMD_DAS       (1 << 27)
+#              define PACKET3_CP_DMA_CMD_SAIC      (1 << 28)
+#              define PACKET3_CP_DMA_CMD_DAIC      (1 << 29)
+#              define PACKET3_CP_DMA_CMD_RAW_WAIT  (1 << 30)
+#define	PACKET3_PFP_SYNC_ME				0x42
+#define	PACKET3_SURFACE_SYNC				0x43
+#              define PACKET3_DEST_BASE_0_ENA      (1 << 0)
+#              define PACKET3_DEST_BASE_1_ENA      (1 << 1)
+#              define PACKET3_CB0_DEST_BASE_ENA    (1 << 6)
+#              define PACKET3_CB1_DEST_BASE_ENA    (1 << 7)
+#              define PACKET3_CB2_DEST_BASE_ENA    (1 << 8)
+#              define PACKET3_CB3_DEST_BASE_ENA    (1 << 9)
+#              define PACKET3_CB4_DEST_BASE_ENA    (1 << 10)
+#              define PACKET3_CB5_DEST_BASE_ENA    (1 << 11)
+#              define PACKET3_CB6_DEST_BASE_ENA    (1 << 12)
+#              define PACKET3_CB7_DEST_BASE_ENA    (1 << 13)
+#              define PACKET3_DB_DEST_BASE_ENA     (1 << 14)
+#              define PACKET3_DEST_BASE_2_ENA      (1 << 19)
+#              define PACKET3_DEST_BASE_3_ENA      (1 << 21)
+#              define PACKET3_TCL1_ACTION_ENA      (1 << 22)
+#              define PACKET3_TC_ACTION_ENA        (1 << 23)
+#              define PACKET3_CB_ACTION_ENA        (1 << 25)
+#              define PACKET3_DB_ACTION_ENA        (1 << 26)
+#              define PACKET3_SH_KCACHE_ACTION_ENA (1 << 27)
+#              define PACKET3_SH_ICACHE_ACTION_ENA (1 << 29)
+#define	PACKET3_ME_INITIALIZE				0x44
+#define		PACKET3_ME_INITIALIZE_DEVICE_ID(x) ((x) << 16)
+#define	PACKET3_COND_WRITE				0x45
+#define	PACKET3_EVENT_WRITE				0x46
+#define	PACKET3_EVENT_WRITE_EOP				0x47
+#define	PACKET3_EVENT_WRITE_EOS				0x48
+#define	PACKET3_PREAMBLE_CNTL				0x4A
+#              define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE     (2 << 28)
+#              define PACKET3_PREAMBLE_END_CLEAR_STATE       (3 << 28)
+#define	PACKET3_ONE_REG_WRITE				0x57
+#define	PACKET3_LOAD_CONFIG_REG				0x5F
+#define	PACKET3_LOAD_CONTEXT_REG			0x60
+#define	PACKET3_LOAD_SH_REG				0x61
+#define	PACKET3_SET_CONFIG_REG				0x68
+#define		PACKET3_SET_CONFIG_REG_START			0x00002000
+#define		PACKET3_SET_CONFIG_REG_END			0x00002c00
+#define	PACKET3_SET_CONTEXT_REG				0x69
+#define		PACKET3_SET_CONTEXT_REG_START			0x000a000
+#define		PACKET3_SET_CONTEXT_REG_END			0x000a400
+#define	PACKET3_SET_CONTEXT_REG_INDIRECT		0x73
+#define	PACKET3_SET_RESOURCE_INDIRECT			0x74
+#define	PACKET3_SET_SH_REG				0x76
+#define		PACKET3_SET_SH_REG_START			0x00002c00
+#define		PACKET3_SET_SH_REG_END				0x00003000
+#define	PACKET3_SET_SH_REG_OFFSET			0x77
+#define	PACKET3_ME_WRITE				0x7A
+#define	PACKET3_SCRATCH_RAM_WRITE			0x7D
+#define	PACKET3_SCRATCH_RAM_READ			0x7E
+#define	PACKET3_CE_WRITE				0x7F
+#define	PACKET3_LOAD_CONST_RAM				0x80
+#define	PACKET3_WRITE_CONST_RAM				0x81
+#define	PACKET3_WRITE_CONST_RAM_OFFSET			0x82
+#define	PACKET3_DUMP_CONST_RAM				0x83
+#define	PACKET3_INCREMENT_CE_COUNTER			0x84
+#define	PACKET3_INCREMENT_DE_COUNTER			0x85
+#define	PACKET3_WAIT_ON_CE_COUNTER			0x86
+#define	PACKET3_WAIT_ON_DE_COUNTER			0x87
+#define	PACKET3_WAIT_ON_DE_COUNTER_DIFF			0x88
+#define	PACKET3_SET_CE_DE_COUNTERS			0x89
+#define	PACKET3_WAIT_ON_AVAIL_BUFFER			0x8A
+#define	PACKET3_SWITCH_BUFFER				0x8B
+#define PACKET3_SEM_WAIT_ON_SIGNAL    (0x1 << 12)
+#define PACKET3_SEM_SEL_SIGNAL	    (0x6 << 29)
+#define PACKET3_SEM_SEL_WAIT	    (0x7 << 29)
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/si_ih.c b/sys/dev/pci/drm/amd/amdgpu/si_ih.c
new file mode 100644
index 00000000000..60dad63098a
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/si_ih.c
@@ -0,0 +1,322 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "sid.h"
+#include "si_ih.h"
+
+static void si_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+static void si_ih_enable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_cntl = RREG32(IH_CNTL);
+	u32 ih_rb_cntl = RREG32(IH_RB_CNTL);
+
+	ih_cntl |= ENABLE_INTR;
+	ih_rb_cntl |= IH_RB_ENABLE;
+	WREG32(IH_CNTL, ih_cntl);
+	WREG32(IH_RB_CNTL, ih_rb_cntl);
+	adev->irq.ih.enabled = true;
+}
+  
+static void si_ih_disable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32(IH_RB_CNTL);
+	u32 ih_cntl = RREG32(IH_CNTL);
+
+	ih_rb_cntl &= ~IH_RB_ENABLE;
+	ih_cntl &= ~ENABLE_INTR;
+	WREG32(IH_RB_CNTL, ih_rb_cntl);
+	WREG32(IH_CNTL, ih_cntl);
+	WREG32(IH_RB_RPTR, 0);
+	WREG32(IH_RB_WPTR, 0);
+	adev->irq.ih.enabled = false;
+	adev->irq.ih.rptr = 0;
+}
+
+static int si_ih_irq_init(struct amdgpu_device *adev)
+{
+	int rb_bufsz;
+	u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
+	u64 wptr_off;
+
+	si_ih_disable_interrupts(adev);
+	WREG32(INTERRUPT_CNTL2, adev->irq.ih.gpu_addr >> 8);
+	interrupt_cntl = RREG32(INTERRUPT_CNTL);
+	interrupt_cntl &= ~IH_DUMMY_RD_OVERRIDE;
+	interrupt_cntl &= ~IH_REQ_NONSNOOP_EN;
+	WREG32(INTERRUPT_CNTL, interrupt_cntl);
+
+	WREG32(IH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
+	rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
+
+	ih_rb_cntl = IH_WPTR_OVERFLOW_ENABLE |
+		     IH_WPTR_OVERFLOW_CLEAR |
+		     (rb_bufsz << 1) |
+		     IH_WPTR_WRITEBACK_ENABLE;
+
+	wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
+	WREG32(IH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
+	WREG32(IH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+	WREG32(IH_RB_CNTL, ih_rb_cntl);
+	WREG32(IH_RB_RPTR, 0);
+	WREG32(IH_RB_WPTR, 0);
+
+	ih_cntl = MC_WRREQ_CREDIT(0x10) | MC_WR_CLEAN_CNT(0x10) | MC_VMID(0);
+	if (adev->irq.msi_enabled)
+		ih_cntl |= RPTR_REARM;
+	WREG32(IH_CNTL, ih_cntl);
+
+	pci_set_master(adev->pdev);
+	si_ih_enable_interrupts(adev);
+
+	return 0;
+}
+
+static void si_ih_irq_disable(struct amdgpu_device *adev)
+{
+	si_ih_disable_interrupts(adev);
+	mdelay(1);
+}
+
+static u32 si_ih_get_wptr(struct amdgpu_device *adev)
+{
+	u32 wptr, tmp;
+
+	wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
+
+	if (wptr & IH_RB_WPTR__RB_OVERFLOW_MASK) {
+		wptr &= ~IH_RB_WPTR__RB_OVERFLOW_MASK;
+		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
+		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
+		tmp = RREG32(IH_RB_CNTL);
+		tmp |= IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK;
+		WREG32(IH_RB_CNTL, tmp);
+	}
+	return (wptr & adev->irq.ih.ptr_mask);
+}
+
+/**
+ * si_ih_prescreen_iv - prescreen an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns true if the interrupt vector should be further processed.
+ */
+static bool si_ih_prescreen_iv(struct amdgpu_device *adev)
+{
+	/* Process all interrupts */
+	return true;
+}
+
+static void si_ih_decode_iv(struct amdgpu_device *adev,
+			     struct amdgpu_iv_entry *entry)
+{
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	uint32_t dw[4];
+
+	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+
+	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->src_id = dw[0] & 0xff;
+	entry->src_data[0] = dw[1] & 0xfffffff;
+	entry->ring_id = dw[2] & 0xff;
+	entry->vmid = (dw[2] >> 8) & 0xff;
+
+	adev->irq.ih.rptr += 16;
+}
+
+static void si_ih_set_rptr(struct amdgpu_device *adev)
+{
+	WREG32(IH_RB_RPTR, adev->irq.ih.rptr);
+}
+
+static int si_ih_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	si_ih_set_interrupt_funcs(adev);
+
+	return 0;
+}
+
+static int si_ih_sw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	if (r)
+		return r;
+
+	return amdgpu_irq_init(adev);
+}
+
+static int si_ih_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_fini(adev);
+	amdgpu_ih_ring_fini(adev);
+
+	return 0;
+}
+
+static int si_ih_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return si_ih_irq_init(adev);
+}
+
+static int si_ih_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	si_ih_irq_disable(adev);
+
+	return 0;
+}
+
+static int si_ih_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return si_ih_hw_fini(adev);
+}
+
+static int si_ih_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return si_ih_hw_init(adev);
+}
+
+static bool si_ih_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(SRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		return false;
+
+	return true;
+}
+
+static int si_ih_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (si_ih_is_idle(handle))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static int si_ih_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	u32 srbm_soft_reset = 0;
+	u32 tmp = RREG32(SRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_IH_MASK;
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(SRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(SRBM_SOFT_RESET, tmp);
+		tmp = RREG32(SRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(SRBM_SOFT_RESET, tmp);
+		tmp = RREG32(SRBM_SOFT_RESET);
+
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int si_ih_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int si_ih_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs si_ih_ip_funcs = {
+	.name = "si_ih",
+	.early_init = si_ih_early_init,
+	.late_init = NULL,
+	.sw_init = si_ih_sw_init,
+	.sw_fini = si_ih_sw_fini,
+	.hw_init = si_ih_hw_init,
+	.hw_fini = si_ih_hw_fini,
+	.suspend = si_ih_suspend,
+	.resume = si_ih_resume,
+	.is_idle = si_ih_is_idle,
+	.wait_for_idle = si_ih_wait_for_idle,
+	.soft_reset = si_ih_soft_reset,
+	.set_clockgating_state = si_ih_set_clockgating_state,
+	.set_powergating_state = si_ih_set_powergating_state,
+};
+
+static const struct amdgpu_ih_funcs si_ih_funcs = {
+	.get_wptr = si_ih_get_wptr,
+	.prescreen_iv = si_ih_prescreen_iv,
+	.decode_iv = si_ih_decode_iv,
+	.set_rptr = si_ih_set_rptr
+};
+
+static void si_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih_funcs == NULL)
+		adev->irq.ih_funcs = &si_ih_funcs;
+}
+
+const struct amdgpu_ip_block_version si_ih_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_IH,
+	.major = 1,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &si_ih_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/si_ih.h b/sys/dev/pci/drm/amd/amdgpu/si_ih.h
new file mode 100644
index 00000000000..42e64a53e24
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/si_ih.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SI_IH_H__
+#define __SI_IH_H__
+
+extern const struct amdgpu_ip_block_version si_ih_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/sid.h b/sys/dev/pci/drm/amd/amdgpu/sid.h
new file mode 100644
index 00000000000..c57eff15937
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/sid.h
@@ -0,0 +1,2461 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+#ifndef SI_H
+#define SI_H
+
+#define TAHITI_RB_BITMAP_WIDTH_PER_SH  2
+
+#define TAHITI_GB_ADDR_CONFIG_GOLDEN        0x12011003
+#define VERDE_GB_ADDR_CONFIG_GOLDEN         0x12010002
+#define HAINAN_GB_ADDR_CONFIG_GOLDEN        0x02010001
+
+#define SI_MAX_SH_GPRS		 	256
+#define SI_MAX_TEMP_GPRS         	16
+#define SI_MAX_SH_THREADS        	256
+#define SI_MAX_SH_STACK_ENTRIES  	4096
+#define SI_MAX_FRC_EOV_CNT       	16384
+#define SI_MAX_BACKENDS          	8
+#define SI_MAX_BACKENDS_MASK     	0xFF
+#define SI_MAX_BACKENDS_PER_SE_MASK     0x0F
+#define SI_MAX_SIMDS             	12
+#define SI_MAX_SIMDS_MASK        	0x0FFF
+#define SI_MAX_SIMDS_PER_SE_MASK        0x00FF
+#define SI_MAX_PIPES            	8
+#define SI_MAX_PIPES_MASK        	0xFF
+#define SI_MAX_PIPES_PER_SIMD_MASK      0x3F
+#define SI_MAX_LDS_NUM           	0xFFFF
+#define SI_MAX_TCC               	16
+#define SI_MAX_TCC_MASK          	0xFFFF
+
+#define AMDGPU_NUM_OF_VMIDS 		8
+
+/* SMC IND accessor regs */
+#define SMC_IND_INDEX_0                              0x80
+#define SMC_IND_DATA_0                               0x81
+
+#define SMC_IND_ACCESS_CNTL                          0x8A
+#       define AUTO_INCREMENT_IND_0                  (1 << 0)
+#define SMC_MESSAGE_0                                0x8B
+#define SMC_RESP_0                                   0x8C
+
+/* CG IND registers are accessed via SMC indirect space + SMC_CG_IND_START */
+#define SMC_CG_IND_START                    0xc0030000
+#define SMC_CG_IND_END                      0xc0040000
+
+#define	CG_CGTT_LOCAL_0				0x400
+#define	CG_CGTT_LOCAL_1				0x401
+
+/* SMC IND registers */
+#define	SMC_SYSCON_RESET_CNTL				0x80000000
+#       define RST_REG                                  (1 << 0)
+#define	SMC_SYSCON_CLOCK_CNTL_0				0x80000004
+#       define CK_DISABLE                               (1 << 0)
+#       define CKEN                                     (1 << 24)
+
+#define VGA_HDP_CONTROL  				0xCA
+#define		VGA_MEMORY_DISABLE				(1 << 4)
+
+#define DCCG_DISP_SLOW_SELECT_REG                       0x13F
+#define		DCCG_DISP1_SLOW_SELECT(x)		((x) << 0)
+#define		DCCG_DISP1_SLOW_SELECT_MASK		(7 << 0)
+#define		DCCG_DISP1_SLOW_SELECT_SHIFT		0
+#define		DCCG_DISP2_SLOW_SELECT(x)		((x) << 4)
+#define		DCCG_DISP2_SLOW_SELECT_MASK		(7 << 4)
+#define		DCCG_DISP2_SLOW_SELECT_SHIFT		4
+
+#define	CG_SPLL_FUNC_CNTL				0x180
+#define		SPLL_RESET				(1 << 0)
+#define		SPLL_SLEEP				(1 << 1)
+#define		SPLL_BYPASS_EN				(1 << 3)
+#define		SPLL_REF_DIV(x)				((x) << 4)
+#define		SPLL_REF_DIV_MASK			(0x3f << 4)
+#define		SPLL_PDIV_A(x)				((x) << 20)
+#define		SPLL_PDIV_A_MASK			(0x7f << 20)
+#define		SPLL_PDIV_A_SHIFT			20
+#define	CG_SPLL_FUNC_CNTL_2				0x181
+#define		SCLK_MUX_SEL(x)				((x) << 0)
+#define		SCLK_MUX_SEL_MASK			(0x1ff << 0)
+#define		SPLL_CTLREQ_CHG				(1 << 23)
+#define		SCLK_MUX_UPDATE				(1 << 26)
+#define	CG_SPLL_FUNC_CNTL_3				0x182
+#define		SPLL_FB_DIV(x)				((x) << 0)
+#define		SPLL_FB_DIV_MASK			(0x3ffffff << 0)
+#define		SPLL_FB_DIV_SHIFT			0
+#define		SPLL_DITHEN				(1 << 28)
+#define	CG_SPLL_FUNC_CNTL_4				0x183
+
+#define	SPLL_STATUS					0x185
+#define		SPLL_CHG_STATUS				(1 << 1)
+#define	SPLL_CNTL_MODE					0x186
+#define		SPLL_SW_DIR_CONTROL			(1 << 0)
+#	define SPLL_REFCLK_SEL(x)			((x) << 26)
+#	define SPLL_REFCLK_SEL_MASK			(3 << 26)
+
+#define	CG_SPLL_SPREAD_SPECTRUM				0x188
+#define		SSEN					(1 << 0)
+#define		CLK_S(x)				((x) << 4)
+#define		CLK_S_MASK				(0xfff << 4)
+#define		CLK_S_SHIFT				4
+#define	CG_SPLL_SPREAD_SPECTRUM_2			0x189
+#define		CLK_V(x)				((x) << 0)
+#define		CLK_V_MASK				(0x3ffffff << 0)
+#define		CLK_V_SHIFT				0
+
+#define	CG_SPLL_AUTOSCALE_CNTL				0x18b
+#       define AUTOSCALE_ON_SS_CLEAR                    (1 << 9)
+
+/* discrete uvd clocks */
+#define	CG_UPLL_FUNC_CNTL				0x18d
+#	define UPLL_RESET_MASK				0x00000001
+#	define UPLL_SLEEP_MASK				0x00000002
+#	define UPLL_BYPASS_EN_MASK			0x00000004
+#	define UPLL_CTLREQ_MASK				0x00000008
+#	define UPLL_VCO_MODE_MASK			0x00000600
+#	define UPLL_REF_DIV_MASK			0x003F0000
+#	define UPLL_CTLACK_MASK				0x40000000
+#	define UPLL_CTLACK2_MASK			0x80000000
+#define	CG_UPLL_FUNC_CNTL_2				0x18e
+#	define UPLL_PDIV_A(x)				((x) << 0)
+#	define UPLL_PDIV_A_MASK				0x0000007F
+#	define UPLL_PDIV_B(x)				((x) << 8)
+#	define UPLL_PDIV_B_MASK				0x00007F00
+#	define VCLK_SRC_SEL(x)				((x) << 20)
+#	define VCLK_SRC_SEL_MASK			0x01F00000
+#	define DCLK_SRC_SEL(x)				((x) << 25)
+#	define DCLK_SRC_SEL_MASK			0x3E000000
+#define	CG_UPLL_FUNC_CNTL_3				0x18f
+#	define UPLL_FB_DIV(x)				((x) << 0)
+#	define UPLL_FB_DIV_MASK				0x01FFFFFF
+#define	CG_UPLL_FUNC_CNTL_4                             0x191
+#	define UPLL_SPARE_ISPARE9			0x00020000
+#define	CG_UPLL_FUNC_CNTL_5				0x192
+#	define RESET_ANTI_MUX_MASK			0x00000200
+#define	CG_UPLL_SPREAD_SPECTRUM				0x194
+#	define SSEN_MASK				0x00000001
+
+#define	MPLL_BYPASSCLK_SEL				0x197
+#	define MPLL_CLKOUT_SEL(x)			((x) << 8)
+#	define MPLL_CLKOUT_SEL_MASK			0xFF00
+
+#define CG_CLKPIN_CNTL                                    0x198
+#       define XTALIN_DIVIDE                              (1 << 1)
+#       define BCLK_AS_XCLK                               (1 << 2)
+#define CG_CLKPIN_CNTL_2                                  0x199
+#       define FORCE_BIF_REFCLK_EN                        (1 << 3)
+#       define MUX_TCLK_TO_XCLK                           (1 << 8)
+
+#define	THM_CLK_CNTL					0x19b
+#	define CMON_CLK_SEL(x)				((x) << 0)
+#	define CMON_CLK_SEL_MASK			0xFF
+#	define TMON_CLK_SEL(x)				((x) << 8)
+#	define TMON_CLK_SEL_MASK			0xFF00
+#define	MISC_CLK_CNTL					0x19c
+#	define DEEP_SLEEP_CLK_SEL(x)			((x) << 0)
+#	define DEEP_SLEEP_CLK_SEL_MASK			0xFF
+#	define ZCLK_SEL(x)				((x) << 8)
+#	define ZCLK_SEL_MASK				0xFF00
+
+#define	CG_THERMAL_CTRL					0x1c0
+#define 	DPM_EVENT_SRC(x)			((x) << 0)
+#define 	DPM_EVENT_SRC_MASK			(7 << 0)
+#define		DIG_THERM_DPM(x)			((x) << 14)
+#define		DIG_THERM_DPM_MASK			0x003FC000
+#define		DIG_THERM_DPM_SHIFT			14
+#define	CG_THERMAL_STATUS				0x1c1
+#define		FDO_PWM_DUTY(x)				((x) << 9)
+#define		FDO_PWM_DUTY_MASK			(0xff << 9)
+#define		FDO_PWM_DUTY_SHIFT			9
+#define	CG_THERMAL_INT					0x1c2
+#define		DIG_THERM_INTH(x)			((x) << 8)
+#define		DIG_THERM_INTH_MASK			0x0000FF00
+#define		DIG_THERM_INTH_SHIFT			8
+#define		DIG_THERM_INTL(x)			((x) << 16)
+#define		DIG_THERM_INTL_MASK			0x00FF0000
+#define		DIG_THERM_INTL_SHIFT			16
+#define 	THERM_INT_MASK_HIGH			(1 << 24)
+#define 	THERM_INT_MASK_LOW			(1 << 25)
+
+#define	CG_MULT_THERMAL_CTRL					0x1c4
+#define		TEMP_SEL(x)					((x) << 20)
+#define		TEMP_SEL_MASK					(0xff << 20)
+#define		TEMP_SEL_SHIFT					20
+#define	CG_MULT_THERMAL_STATUS					0x1c5
+#define		ASIC_MAX_TEMP(x)				((x) << 0)
+#define		ASIC_MAX_TEMP_MASK				0x000001ff
+#define		ASIC_MAX_TEMP_SHIFT				0
+#define		CTF_TEMP(x)					((x) << 9)
+#define		CTF_TEMP_MASK					0x0003fe00
+#define		CTF_TEMP_SHIFT					9
+
+#define	CG_FDO_CTRL0					0x1d5
+#define		FDO_STATIC_DUTY(x)			((x) << 0)
+#define		FDO_STATIC_DUTY_MASK			0x000000FF
+#define		FDO_STATIC_DUTY_SHIFT			0
+#define	CG_FDO_CTRL1					0x1d6
+#define		FMAX_DUTY100(x)				((x) << 0)
+#define		FMAX_DUTY100_MASK			0x000000FF
+#define		FMAX_DUTY100_SHIFT			0
+#define	CG_FDO_CTRL2					0x1d7
+#define		TMIN(x)					((x) << 0)
+#define		TMIN_MASK				0x000000FF
+#define		TMIN_SHIFT				0
+#define		FDO_PWM_MODE(x)				((x) << 11)
+#define		FDO_PWM_MODE_MASK			(7 << 11)
+#define		FDO_PWM_MODE_SHIFT			11
+#define		TACH_PWM_RESP_RATE(x)			((x) << 25)
+#define		TACH_PWM_RESP_RATE_MASK			(0x7f << 25)
+#define		TACH_PWM_RESP_RATE_SHIFT		25
+
+#define CG_TACH_CTRL                                    0x1dc
+#       define EDGE_PER_REV(x)                          ((x) << 0)
+#       define EDGE_PER_REV_MASK                        (0x7 << 0)
+#       define EDGE_PER_REV_SHIFT                       0
+#       define TARGET_PERIOD(x)                         ((x) << 3)
+#       define TARGET_PERIOD_MASK                       0xfffffff8
+#       define TARGET_PERIOD_SHIFT                      3
+#define CG_TACH_STATUS                                  0x1dd
+#       define TACH_PERIOD(x)                           ((x) << 0)
+#       define TACH_PERIOD_MASK                         0xffffffff
+#       define TACH_PERIOD_SHIFT                        0
+
+#define GENERAL_PWRMGT                                  0x1e0
+#       define GLOBAL_PWRMGT_EN                         (1 << 0)
+#       define STATIC_PM_EN                             (1 << 1)
+#       define THERMAL_PROTECTION_DIS                   (1 << 2)
+#       define THERMAL_PROTECTION_TYPE                  (1 << 3)
+#       define SW_SMIO_INDEX(x)                         ((x) << 6)
+#       define SW_SMIO_INDEX_MASK                       (1 << 6)
+#       define SW_SMIO_INDEX_SHIFT                      6
+#       define VOLT_PWRMGT_EN                           (1 << 10)
+#       define DYN_SPREAD_SPECTRUM_EN                   (1 << 23)
+#define CG_TPC                                            0x1e1
+#define SCLK_PWRMGT_CNTL                                  0x1e2
+#       define SCLK_PWRMGT_OFF                            (1 << 0)
+#       define SCLK_LOW_D1                                (1 << 1)
+#       define FIR_RESET                                  (1 << 4)
+#       define FIR_FORCE_TREND_SEL                        (1 << 5)
+#       define FIR_TREND_MODE                             (1 << 6)
+#       define DYN_GFX_CLK_OFF_EN                         (1 << 7)
+#       define GFX_CLK_FORCE_ON                           (1 << 8)
+#       define GFX_CLK_REQUEST_OFF                        (1 << 9)
+#       define GFX_CLK_FORCE_OFF                          (1 << 10)
+#       define GFX_CLK_OFF_ACPI_D1                        (1 << 11)
+#       define GFX_CLK_OFF_ACPI_D2                        (1 << 12)
+#       define GFX_CLK_OFF_ACPI_D3                        (1 << 13)
+#       define DYN_LIGHT_SLEEP_EN                         (1 << 14)
+
+#define TARGET_AND_CURRENT_PROFILE_INDEX                  0x1e6
+#       define CURRENT_STATE_INDEX_MASK                   (0xf << 4)
+#       define CURRENT_STATE_INDEX_SHIFT                  4
+
+#define CG_FTV                                            0x1ef
+
+#define CG_FFCT_0                                         0x1f0
+#       define UTC_0(x)                                   ((x) << 0)
+#       define UTC_0_MASK                                 (0x3ff << 0)
+#       define DTC_0(x)                                   ((x) << 10)
+#       define DTC_0_MASK                                 (0x3ff << 10)
+
+#define CG_BSP                                          0x1ff
+#       define BSP(x)					((x) << 0)
+#       define BSP_MASK					(0xffff << 0)
+#       define BSU(x)					((x) << 16)
+#       define BSU_MASK					(0xf << 16)
+#define CG_AT                                           0x200
+#       define CG_R(x)					((x) << 0)
+#       define CG_R_MASK				(0xffff << 0)
+#       define CG_L(x)					((x) << 16)
+#       define CG_L_MASK				(0xffff << 16)
+
+#define CG_GIT                                          0x201
+#       define CG_GICST(x)                              ((x) << 0)
+#       define CG_GICST_MASK                            (0xffff << 0)
+#       define CG_GIPOT(x)                              ((x) << 16)
+#       define CG_GIPOT_MASK                            (0xffff << 16)
+
+#define CG_SSP                                            0x203
+#       define SST(x)                                     ((x) << 0)
+#       define SST_MASK                                   (0xffff << 0)
+#       define SSTU(x)                                    ((x) << 16)
+#       define SSTU_MASK                                  (0xf << 16)
+
+#define CG_DISPLAY_GAP_CNTL                               0x20a
+#       define DISP1_GAP(x)                               ((x) << 0)
+#       define DISP1_GAP_MASK                             (3 << 0)
+#       define DISP2_GAP(x)                               ((x) << 2)
+#       define DISP2_GAP_MASK                             (3 << 2)
+#       define VBI_TIMER_COUNT(x)                         ((x) << 4)
+#       define VBI_TIMER_COUNT_MASK                       (0x3fff << 4)
+#       define VBI_TIMER_UNIT(x)                          ((x) << 20)
+#       define VBI_TIMER_UNIT_MASK                        (7 << 20)
+#       define DISP1_GAP_MCHG(x)                          ((x) << 24)
+#       define DISP1_GAP_MCHG_MASK                        (3 << 24)
+#       define DISP2_GAP_MCHG(x)                          ((x) << 26)
+#       define DISP2_GAP_MCHG_MASK                        (3 << 26)
+
+#define	CG_ULV_CONTROL					0x21e
+#define	CG_ULV_PARAMETER				0x21f
+
+#define	SMC_SCRATCH0					0x221
+
+#define	CG_CAC_CTRL					0x22e
+#	define CAC_WINDOW(x)				((x) << 0)
+#	define CAC_WINDOW_MASK				0x00ffffff
+
+#define DMIF_ADDR_CONFIG  				0x2F5
+
+#define DMIF_ADDR_CALC  				0x300
+
+#define	PIPE0_DMIF_BUFFER_CONTROL			  0x0328
+#       define DMIF_BUFFERS_ALLOCATED(x)                  ((x) << 0)
+#       define DMIF_BUFFERS_ALLOCATED_COMPLETED           (1 << 4)
+
+#define	SRBM_STATUS				        0x394
+#define		GRBM_RQ_PENDING 			(1 << 5)
+#define		VMC_BUSY 				(1 << 8)
+#define		MCB_BUSY 				(1 << 9)
+#define		MCB_NON_DISPLAY_BUSY 			(1 << 10)
+#define		MCC_BUSY 				(1 << 11)
+#define		MCD_BUSY 				(1 << 12)
+#define		SEM_BUSY 				(1 << 14)
+#define		IH_BUSY 				(1 << 17)
+
+#define	SRBM_SOFT_RESET				        0x398
+#define		SOFT_RESET_BIF				(1 << 1)
+#define		SOFT_RESET_DC				(1 << 5)
+#define		SOFT_RESET_DMA1				(1 << 6)
+#define		SOFT_RESET_GRBM				(1 << 8)
+#define		SOFT_RESET_HDP				(1 << 9)
+#define		SOFT_RESET_IH				(1 << 10)
+#define		SOFT_RESET_MC				(1 << 11)
+#define		SOFT_RESET_ROM				(1 << 14)
+#define		SOFT_RESET_SEM				(1 << 15)
+#define		SOFT_RESET_VMC				(1 << 17)
+#define		SOFT_RESET_DMA				(1 << 20)
+#define		SOFT_RESET_TST				(1 << 21)
+#define		SOFT_RESET_REGBB			(1 << 22)
+#define		SOFT_RESET_ORB				(1 << 23)
+
+#define	CC_SYS_RB_BACKEND_DISABLE			0x3A0
+#define	GC_USER_SYS_RB_BACKEND_DISABLE			0x3A1
+
+#define SRBM_READ_ERROR					0x3A6
+#define SRBM_INT_CNTL					0x3A8
+#define SRBM_INT_ACK					0x3AA
+
+#define	SRBM_STATUS2				        0x3B1
+#define		DMA_BUSY 				(1 << 5)
+#define		DMA1_BUSY 				(1 << 6)
+
+#define VM_L2_CNTL					0x500
+#define		ENABLE_L2_CACHE					(1 << 0)
+#define		ENABLE_L2_FRAGMENT_PROCESSING			(1 << 1)
+#define		L2_CACHE_PTE_ENDIAN_SWAP_MODE(x)		((x) << 2)
+#define		L2_CACHE_PDE_ENDIAN_SWAP_MODE(x)		((x) << 4)
+#define		ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE		(1 << 9)
+#define		ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE	(1 << 10)
+#define		EFFECTIVE_L2_QUEUE_SIZE(x)			(((x) & 7) << 15)
+#define		CONTEXT1_IDENTITY_ACCESS_MODE(x)		(((x) & 3) << 19)
+#define VM_L2_CNTL2					0x501
+#define		INVALIDATE_ALL_L1_TLBS				(1 << 0)
+#define		INVALIDATE_L2_CACHE				(1 << 1)
+#define		INVALIDATE_CACHE_MODE(x)			((x) << 26)
+#define			INVALIDATE_PTE_AND_PDE_CACHES		0
+#define			INVALIDATE_ONLY_PTE_CACHES		1
+#define			INVALIDATE_ONLY_PDE_CACHES		2
+#define VM_L2_CNTL3					0x502
+#define		BANK_SELECT(x)					((x) << 0)
+#define		L2_CACHE_UPDATE_MODE(x)				((x) << 6)
+#define		L2_CACHE_BIGK_FRAGMENT_SIZE(x)			((x) << 15)
+#define		L2_CACHE_BIGK_ASSOCIATIVITY			(1 << 20)
+#define	VM_L2_STATUS					0x503
+#define		L2_BUSY						(1 << 0)
+#define VM_CONTEXT0_CNTL				0x504
+#define		ENABLE_CONTEXT					(1 << 0)
+#define		PAGE_TABLE_DEPTH(x)				(((x) & 3) << 1)
+#define		RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT		(1 << 3)
+#define		RANGE_PROTECTION_FAULT_ENABLE_DEFAULT		(1 << 4)
+#define		DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT	(1 << 6)
+#define		DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT	(1 << 7)
+#define		PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT		(1 << 9)
+#define		PDE0_PROTECTION_FAULT_ENABLE_DEFAULT		(1 << 10)
+#define		VALID_PROTECTION_FAULT_ENABLE_INTERRUPT		(1 << 12)
+#define		VALID_PROTECTION_FAULT_ENABLE_DEFAULT		(1 << 13)
+#define		READ_PROTECTION_FAULT_ENABLE_INTERRUPT		(1 << 15)
+#define		READ_PROTECTION_FAULT_ENABLE_DEFAULT		(1 << 16)
+#define		WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT		(1 << 18)
+#define		WRITE_PROTECTION_FAULT_ENABLE_DEFAULT		(1 << 19)
+#define		PAGE_TABLE_BLOCK_SIZE(x)			(((x) & 0xF) << 24)
+#define VM_CONTEXT1_CNTL				0x505
+#define VM_CONTEXT0_CNTL2				0x50C
+#define VM_CONTEXT1_CNTL2				0x50D
+#define	VM_CONTEXT8_PAGE_TABLE_BASE_ADDR		0x50E
+#define	VM_CONTEXT9_PAGE_TABLE_BASE_ADDR		0x50F
+#define	VM_CONTEXT10_PAGE_TABLE_BASE_ADDR		0x510
+#define	VM_CONTEXT11_PAGE_TABLE_BASE_ADDR		0x511
+#define	VM_CONTEXT12_PAGE_TABLE_BASE_ADDR		0x512
+#define	VM_CONTEXT13_PAGE_TABLE_BASE_ADDR		0x513
+#define	VM_CONTEXT14_PAGE_TABLE_BASE_ADDR		0x514
+#define	VM_CONTEXT15_PAGE_TABLE_BASE_ADDR		0x515
+
+#define	VM_CONTEXT1_PROTECTION_FAULT_ADDR		0x53f
+#define	VM_CONTEXT1_PROTECTION_FAULT_STATUS		0x537
+#define		PROTECTIONS_MASK			(0xf << 0)
+#define		PROTECTIONS_SHIFT			0
+		/* bit 0: range
+		 * bit 1: pde0
+		 * bit 2: valid
+		 * bit 3: read
+		 * bit 4: write
+		 */
+#define		MEMORY_CLIENT_ID_MASK			(0xff << 12)
+#define		MEMORY_CLIENT_ID_SHIFT			12
+#define		MEMORY_CLIENT_RW_MASK			(1 << 24)
+#define		MEMORY_CLIENT_RW_SHIFT			24
+#define		FAULT_VMID_MASK				(0xf << 25)
+#define		FAULT_VMID_SHIFT			25
+
+#define VM_INVALIDATE_REQUEST				0x51E
+#define VM_INVALIDATE_RESPONSE				0x51F
+
+#define VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR	0x546
+#define VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR	0x547
+
+#define	VM_CONTEXT0_PAGE_TABLE_BASE_ADDR		0x54F
+#define	VM_CONTEXT1_PAGE_TABLE_BASE_ADDR		0x550
+#define	VM_CONTEXT2_PAGE_TABLE_BASE_ADDR		0x551
+#define	VM_CONTEXT3_PAGE_TABLE_BASE_ADDR		0x552
+#define	VM_CONTEXT4_PAGE_TABLE_BASE_ADDR		0x553
+#define	VM_CONTEXT5_PAGE_TABLE_BASE_ADDR		0x554
+#define	VM_CONTEXT6_PAGE_TABLE_BASE_ADDR		0x555
+#define	VM_CONTEXT7_PAGE_TABLE_BASE_ADDR		0x556
+#define	VM_CONTEXT0_PAGE_TABLE_START_ADDR		0x557
+#define	VM_CONTEXT1_PAGE_TABLE_START_ADDR		0x558
+
+#define	VM_CONTEXT0_PAGE_TABLE_END_ADDR			0x55F
+#define	VM_CONTEXT1_PAGE_TABLE_END_ADDR			0x560
+
+#define VM_L2_CG           				0x570
+#define		MC_CG_ENABLE				(1 << 18)
+#define		MC_LS_ENABLE				(1 << 19)
+
+#define MC_SHARED_CHMAP						0x801
+#define		NOOFCHAN_SHIFT					12
+#define		NOOFCHAN_MASK					0x0000f000
+#define MC_SHARED_CHREMAP					0x802
+
+#define	MC_VM_FB_LOCATION				0x809
+#define	MC_VM_AGP_TOP					0x80A
+#define	MC_VM_AGP_BOT					0x80B
+#define	MC_VM_AGP_BASE					0x80C
+#define	MC_VM_SYSTEM_APERTURE_LOW_ADDR			0x80D
+#define	MC_VM_SYSTEM_APERTURE_HIGH_ADDR			0x80E
+#define	MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR		0x80F
+
+#define	MC_VM_MX_L1_TLB_CNTL				0x819
+#define		ENABLE_L1_TLB					(1 << 0)
+#define		ENABLE_L1_FRAGMENT_PROCESSING			(1 << 1)
+#define		SYSTEM_ACCESS_MODE_PA_ONLY			(0 << 3)
+#define		SYSTEM_ACCESS_MODE_USE_SYS_MAP			(1 << 3)
+#define		SYSTEM_ACCESS_MODE_IN_SYS			(2 << 3)
+#define		SYSTEM_ACCESS_MODE_NOT_IN_SYS			(3 << 3)
+#define		SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU	(0 << 5)
+#define		ENABLE_ADVANCED_DRIVER_MODEL			(1 << 6)
+
+#define MC_SHARED_BLACKOUT_CNTL           		0x82B
+
+#define MC_HUB_MISC_HUB_CG           			0x82E
+#define MC_HUB_MISC_VM_CG           			0x82F
+
+#define MC_HUB_MISC_SIP_CG           			0x830
+
+#define MC_XPB_CLK_GAT           			0x91E
+
+#define MC_CITF_MISC_RD_CG           			0x992
+#define MC_CITF_MISC_WR_CG           			0x993
+#define MC_CITF_MISC_VM_CG           			0x994
+
+#define	MC_ARB_RAMCFG					0x9D8
+#define		NOOFBANK_SHIFT					0
+#define		NOOFBANK_MASK					0x00000003
+#define		NOOFRANK_SHIFT					2
+#define		NOOFRANK_MASK					0x00000004
+#define		NOOFROWS_SHIFT					3
+#define		NOOFROWS_MASK					0x00000038
+#define		NOOFCOLS_SHIFT					6
+#define		NOOFCOLS_MASK					0x000000C0
+#define		CHANSIZE_SHIFT					8
+#define		CHANSIZE_MASK					0x00000100
+#define		CHANSIZE_OVERRIDE				(1 << 11)
+#define		NOOFGROUPS_SHIFT				12
+#define		NOOFGROUPS_MASK					0x00001000
+
+#define	MC_ARB_DRAM_TIMING				0x9DD
+#define	MC_ARB_DRAM_TIMING2				0x9DE
+
+#define MC_ARB_BURST_TIME                               0xA02
+#define		STATE0(x)				((x) << 0)
+#define		STATE0_MASK				(0x1f << 0)
+#define		STATE0_SHIFT				0
+#define		STATE1(x)				((x) << 5)
+#define		STATE1_MASK				(0x1f << 5)
+#define		STATE1_SHIFT				5
+#define		STATE2(x)				((x) << 10)
+#define		STATE2_MASK				(0x1f << 10)
+#define		STATE2_SHIFT				10
+#define		STATE3(x)				((x) << 15)
+#define		STATE3_MASK				(0x1f << 15)
+#define		STATE3_SHIFT				15
+
+#define	MC_SEQ_TRAIN_WAKEUP_CNTL			0xA3A
+#define		TRAIN_DONE_D0      			(1 << 30)
+#define		TRAIN_DONE_D1      			(1 << 31)
+
+#define MC_SEQ_SUP_CNTL           			0xA32
+#define		RUN_MASK      				(1 << 0)
+#define MC_SEQ_SUP_PGM           			0xA33
+#define MC_PMG_AUTO_CMD           			0xA34
+
+#define MC_IO_PAD_CNTL_D0           			0xA74
+#define		MEM_FALL_OUT_CMD      			(1 << 8)
+
+#define MC_SEQ_RAS_TIMING                               0xA28
+#define MC_SEQ_CAS_TIMING                               0xA29
+#define MC_SEQ_MISC_TIMING                              0xA2A
+#define MC_SEQ_MISC_TIMING2                             0xA2B
+#define MC_SEQ_PMG_TIMING                               0xA2C
+#define MC_SEQ_RD_CTL_D0                                0xA2D
+#define MC_SEQ_RD_CTL_D1                                0xA2E
+#define MC_SEQ_WR_CTL_D0                                0xA2F
+#define MC_SEQ_WR_CTL_D1                                0xA30
+
+#define MC_SEQ_MISC0           				0xA80
+#define 	MC_SEQ_MISC0_VEN_ID_SHIFT               8
+#define 	MC_SEQ_MISC0_VEN_ID_MASK                0x00000f00
+#define 	MC_SEQ_MISC0_VEN_ID_VALUE               3
+#define 	MC_SEQ_MISC0_REV_ID_SHIFT               12
+#define 	MC_SEQ_MISC0_REV_ID_MASK                0x0000f000
+#define 	MC_SEQ_MISC0_REV_ID_VALUE               1
+#define 	MC_SEQ_MISC0_GDDR5_SHIFT                28
+#define 	MC_SEQ_MISC0_GDDR5_MASK                 0xf0000000
+#define 	MC_SEQ_MISC0_GDDR5_VALUE                5
+#define MC_SEQ_MISC1                                    0xA81
+#define MC_SEQ_RESERVE_M                                0xA82
+#define MC_PMG_CMD_EMRS                                 0xA83
+
+#define MC_SEQ_IO_DEBUG_INDEX           		0xA91
+#define MC_SEQ_IO_DEBUG_DATA           			0xA92
+
+#define MC_SEQ_MISC5                                    0xA95
+#define MC_SEQ_MISC6                                    0xA96
+
+#define MC_SEQ_MISC7                                    0xA99
+
+#define MC_SEQ_RAS_TIMING_LP                            0xA9B
+#define MC_SEQ_CAS_TIMING_LP                            0xA9C
+#define MC_SEQ_MISC_TIMING_LP                           0xA9D
+#define MC_SEQ_MISC_TIMING2_LP                          0xA9E
+#define MC_SEQ_WR_CTL_D0_LP                             0xA9F
+#define MC_SEQ_WR_CTL_D1_LP                             0xAA0
+#define MC_SEQ_PMG_CMD_EMRS_LP                          0xAA1
+#define MC_SEQ_PMG_CMD_MRS_LP                           0xAA2
+
+#define MC_PMG_CMD_MRS                                  0xAAB
+
+#define MC_SEQ_RD_CTL_D0_LP                             0xAC7
+#define MC_SEQ_RD_CTL_D1_LP                             0xAC8
+
+#define MC_PMG_CMD_MRS1                                 0xAD1
+#define MC_SEQ_PMG_CMD_MRS1_LP                          0xAD2
+#define MC_SEQ_PMG_TIMING_LP                            0xAD3
+
+#define MC_SEQ_WR_CTL_2                                 0xAD5
+#define MC_SEQ_WR_CTL_2_LP                              0xAD6
+#define MC_PMG_CMD_MRS2                                 0xAD7
+#define MC_SEQ_PMG_CMD_MRS2_LP                          0xAD8
+
+#define	MCLK_PWRMGT_CNTL				0xAE8
+#       define DLL_SPEED(x)				((x) << 0)
+#       define DLL_SPEED_MASK				(0x1f << 0)
+#       define DLL_READY                                (1 << 6)
+#       define MC_INT_CNTL                              (1 << 7)
+#       define MRDCK0_PDNB                              (1 << 8)
+#       define MRDCK1_PDNB                              (1 << 9)
+#       define MRDCK0_RESET                             (1 << 16)
+#       define MRDCK1_RESET                             (1 << 17)
+#       define DLL_READY_READ                           (1 << 24)
+#define	DLL_CNTL					0xAE9
+#       define MRDCK0_BYPASS                            (1 << 24)
+#       define MRDCK1_BYPASS                            (1 << 25)
+
+#define	MPLL_CNTL_MODE					0xAEC
+#       define MPLL_MCLK_SEL                            (1 << 11)
+#define	MPLL_FUNC_CNTL					0xAED
+#define		BWCTRL(x)				((x) << 20)
+#define		BWCTRL_MASK				(0xff << 20)
+#define	MPLL_FUNC_CNTL_1				0xAEE
+#define		VCO_MODE(x)				((x) << 0)
+#define		VCO_MODE_MASK				(3 << 0)
+#define		CLKFRAC(x)				((x) << 4)
+#define		CLKFRAC_MASK				(0xfff << 4)
+#define		CLKF(x)					((x) << 16)
+#define		CLKF_MASK				(0xfff << 16)
+#define	MPLL_FUNC_CNTL_2				0xAEF
+#define	MPLL_AD_FUNC_CNTL				0xAF0
+#define		YCLK_POST_DIV(x)			((x) << 0)
+#define		YCLK_POST_DIV_MASK			(7 << 0)
+#define	MPLL_DQ_FUNC_CNTL				0xAF1
+#define		YCLK_SEL(x)				((x) << 4)
+#define		YCLK_SEL_MASK				(1 << 4)
+
+#define	MPLL_SS1					0xAF3
+#define		CLKV(x)					((x) << 0)
+#define		CLKV_MASK				(0x3ffffff << 0)
+#define	MPLL_SS2					0xAF4
+#define		CLKS(x)					((x) << 0)
+#define		CLKS_MASK				(0xfff << 0)
+
+#define	HDP_HOST_PATH_CNTL				0xB00
+#define 	CLOCK_GATING_DIS			(1 << 23)
+#define	HDP_NONSURFACE_BASE				0xB01
+#define	HDP_NONSURFACE_INFO				0xB02
+#define	HDP_NONSURFACE_SIZE				0xB03
+
+#define HDP_DEBUG0  					0xBCC
+
+#define HDP_ADDR_CONFIG  				0xBD2
+#define HDP_MISC_CNTL					0xBD3
+#define 	HDP_FLUSH_INVALIDATE_CACHE			(1 << 0)
+#define HDP_MEM_POWER_LS				0xBD4
+#define 	HDP_LS_ENABLE				(1 << 0)
+
+#define ATC_MISC_CG           				0xCD4
+
+#define IH_RB_CNTL                                        0xF80
+#       define IH_RB_ENABLE                               (1 << 0)
+#       define IH_IB_SIZE(x)                              ((x) << 1) /* log2 */
+#       define IH_RB_FULL_DRAIN_ENABLE                    (1 << 6)
+#       define IH_WPTR_WRITEBACK_ENABLE                   (1 << 8)
+#       define IH_WPTR_WRITEBACK_TIMER(x)                 ((x) << 9) /* log2 */
+#       define IH_WPTR_OVERFLOW_ENABLE                    (1 << 16)
+#       define IH_WPTR_OVERFLOW_CLEAR                     (1 << 31)
+#define IH_RB_BASE                                        0xF81
+#define IH_RB_RPTR                                        0xF82
+#define IH_RB_WPTR                                        0xF83
+#       define RB_OVERFLOW                                (1 << 0)
+#       define WPTR_OFFSET_MASK                           0x3fffc
+#define IH_RB_WPTR_ADDR_HI                                0xF84
+#define IH_RB_WPTR_ADDR_LO                                0xF85
+#define IH_CNTL                                           0xF86
+#       define ENABLE_INTR                                (1 << 0)
+#       define IH_MC_SWAP(x)                              ((x) << 1)
+#       define IH_MC_SWAP_NONE                            0
+#       define IH_MC_SWAP_16BIT                           1
+#       define IH_MC_SWAP_32BIT                           2
+#       define IH_MC_SWAP_64BIT                           3
+#       define RPTR_REARM                                 (1 << 4)
+#       define MC_WRREQ_CREDIT(x)                         ((x) << 15)
+#       define MC_WR_CLEAN_CNT(x)                         ((x) << 20)
+#       define MC_VMID(x)                                 ((x) << 25)
+
+#define	CONFIG_MEMSIZE					0x150A
+
+#define INTERRUPT_CNTL                                    0x151A
+#       define IH_DUMMY_RD_OVERRIDE                       (1 << 0)
+#       define IH_DUMMY_RD_EN                             (1 << 1)
+#       define IH_REQ_NONSNOOP_EN                         (1 << 3)
+#       define GEN_IH_INT_EN                              (1 << 8)
+#define INTERRUPT_CNTL2                                   0x151B
+
+#define HDP_MEM_COHERENCY_FLUSH_CNTL			0x1520
+
+#define	BIF_FB_EN						0x1524
+#define		FB_READ_EN					(1 << 0)
+#define		FB_WRITE_EN					(1 << 1)
+
+#define HDP_REG_COHERENCY_FLUSH_CNTL			0x1528
+
+/* DCE6 ELD audio interface */
+#define AZ_F0_CODEC_ENDPOINT_INDEX                       0x1780
+#       define AZ_ENDPOINT_REG_INDEX(x)                  (((x) & 0xff) << 0)
+#       define AZ_ENDPOINT_REG_WRITE_EN                  (1 << 8)
+#define AZ_F0_CODEC_ENDPOINT_DATA                        0x1781
+
+#define AZ_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER          0x25
+#define		SPEAKER_ALLOCATION(x)			(((x) & 0x7f) << 0)
+#define		SPEAKER_ALLOCATION_MASK			(0x7f << 0)
+#define		SPEAKER_ALLOCATION_SHIFT		0
+#define		HDMI_CONNECTION				(1 << 16)
+#define		DP_CONNECTION				(1 << 17)
+
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0        0x28 /* LPCM */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR1        0x29 /* AC3 */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR2        0x2A /* MPEG1 */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR3        0x2B /* MP3 */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR4        0x2C /* MPEG2 */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR5        0x2D /* AAC */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR6        0x2E /* DTS */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR7        0x2F /* ATRAC */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR8        0x30 /* one bit audio - leave at 0 (default) */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR9        0x31 /* Dolby Digital */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR10       0x32 /* DTS-HD */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR11       0x33 /* MAT-MLP */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR12       0x34 /* DTS */
+#define AZ_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13       0x35 /* WMA Pro */
+#       define MAX_CHANNELS(x)                            (((x) & 0x7) << 0)
+/* max channels minus one.  7 = 8 channels */
+#       define SUPPORTED_FREQUENCIES(x)                   (((x) & 0xff) << 8)
+#       define DESCRIPTOR_BYTE_2(x)                       (((x) & 0xff) << 16)
+#       define SUPPORTED_FREQUENCIES_STEREO(x)            (((x) & 0xff) << 24) /* LPCM only */
+/* SUPPORTED_FREQUENCIES, SUPPORTED_FREQUENCIES_STEREO
+ * bit0 = 32 kHz
+ * bit1 = 44.1 kHz
+ * bit2 = 48 kHz
+ * bit3 = 88.2 kHz
+ * bit4 = 96 kHz
+ * bit5 = 176.4 kHz
+ * bit6 = 192 kHz
+ */
+
+#define AZ_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC         0x37
+#       define VIDEO_LIPSYNC(x)                           (((x) & 0xff) << 0)
+#       define AUDIO_LIPSYNC(x)                           (((x) & 0xff) << 8)
+/* VIDEO_LIPSYNC, AUDIO_LIPSYNC
+ * 0   = invalid
+ * x   = legal delay value
+ * 255 = sync not supported
+ */
+#define AZ_F0_CODEC_PIN_CONTROL_RESPONSE_HBR             0x38
+#       define HBR_CAPABLE                                (1 << 0) /* enabled by default */
+
+#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO0               0x3a
+#       define MANUFACTURER_ID(x)                        (((x) & 0xffff) << 0)
+#       define PRODUCT_ID(x)                             (((x) & 0xffff) << 16)
+#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO1               0x3b
+#       define SINK_DESCRIPTION_LEN(x)                   (((x) & 0xff) << 0)
+#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO2               0x3c
+#       define PORT_ID0(x)                               (((x) & 0xffffffff) << 0)
+#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO3               0x3d
+#       define PORT_ID1(x)                               (((x) & 0xffffffff) << 0)
+#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO4               0x3e
+#       define DESCRIPTION0(x)                           (((x) & 0xff) << 0)
+#       define DESCRIPTION1(x)                           (((x) & 0xff) << 8)
+#       define DESCRIPTION2(x)                           (((x) & 0xff) << 16)
+#       define DESCRIPTION3(x)                           (((x) & 0xff) << 24)
+#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO5               0x3f
+#       define DESCRIPTION4(x)                           (((x) & 0xff) << 0)
+#       define DESCRIPTION5(x)                           (((x) & 0xff) << 8)
+#       define DESCRIPTION6(x)                           (((x) & 0xff) << 16)
+#       define DESCRIPTION7(x)                           (((x) & 0xff) << 24)
+#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO6               0x40
+#       define DESCRIPTION8(x)                           (((x) & 0xff) << 0)
+#       define DESCRIPTION9(x)                           (((x) & 0xff) << 8)
+#       define DESCRIPTION10(x)                          (((x) & 0xff) << 16)
+#       define DESCRIPTION11(x)                          (((x) & 0xff) << 24)
+#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO7               0x41
+#       define DESCRIPTION12(x)                          (((x) & 0xff) << 0)
+#       define DESCRIPTION13(x)                          (((x) & 0xff) << 8)
+#       define DESCRIPTION14(x)                          (((x) & 0xff) << 16)
+#       define DESCRIPTION15(x)                          (((x) & 0xff) << 24)
+#define AZ_F0_CODEC_PIN_CONTROL_SINK_INFO8               0x42
+#       define DESCRIPTION16(x)                          (((x) & 0xff) << 0)
+#       define DESCRIPTION17(x)                          (((x) & 0xff) << 8)
+
+#define AZ_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL         0x54
+#       define AUDIO_ENABLED                             (1 << 31)
+
+#define AZ_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT  0x56
+#define		PORT_CONNECTIVITY_MASK				(3 << 30)
+#define		PORT_CONNECTIVITY_SHIFT				30
+
+#define	DC_LB_MEMORY_SPLIT					0x1AC3
+#define		DC_LB_MEMORY_CONFIG(x)				((x) << 20)
+
+#define	PRIORITY_A_CNT						0x1AC6
+#define		PRIORITY_MARK_MASK				0x7fff
+#define		PRIORITY_OFF					(1 << 16)
+#define		PRIORITY_ALWAYS_ON				(1 << 20)
+#define	PRIORITY_B_CNT						0x1AC7
+
+#define	DPG_PIPE_ARBITRATION_CONTROL3				0x1B32
+#       define LATENCY_WATERMARK_MASK(x)			((x) << 16)
+#define	DPG_PIPE_LATENCY_CONTROL				0x1B33
+#       define LATENCY_LOW_WATERMARK(x)				((x) << 0)
+#       define LATENCY_HIGH_WATERMARK(x)			((x) << 16)
+
+/* 0x6bb8, 0x77b8, 0x103b8, 0x10fb8, 0x11bb8, 0x127b8 */
+#define VLINE_STATUS                                    0x1AEE
+#       define VLINE_OCCURRED                           (1 << 0)
+#       define VLINE_ACK                                (1 << 4)
+#       define VLINE_STAT                               (1 << 12)
+#       define VLINE_INTERRUPT                          (1 << 16)
+#       define VLINE_INTERRUPT_TYPE                     (1 << 17)
+/* 0x6bbc, 0x77bc, 0x103bc, 0x10fbc, 0x11bbc, 0x127bc */
+#define VBLANK_STATUS                                   0x1AEF
+#       define VBLANK_OCCURRED                          (1 << 0)
+#       define VBLANK_ACK                               (1 << 4)
+#       define VBLANK_STAT                              (1 << 12)
+#       define VBLANK_INTERRUPT                         (1 << 16)
+#       define VBLANK_INTERRUPT_TYPE                    (1 << 17)
+
+/* 0x6b40, 0x7740, 0x10340, 0x10f40, 0x11b40, 0x12740 */
+#define INT_MASK                                        0x1AD0
+#       define VBLANK_INT_MASK                          (1 << 0)
+#       define VLINE_INT_MASK                           (1 << 4)
+
+#define DISP_INTERRUPT_STATUS                           0x183D
+#       define LB_D1_VLINE_INTERRUPT                    (1 << 2)
+#       define LB_D1_VBLANK_INTERRUPT                   (1 << 3)
+#       define DC_HPD1_INTERRUPT                        (1 << 17)
+#       define DC_HPD1_RX_INTERRUPT                     (1 << 18)
+#       define DACA_AUTODETECT_INTERRUPT                (1 << 22)
+#       define DACB_AUTODETECT_INTERRUPT                (1 << 23)
+#       define DC_I2C_SW_DONE_INTERRUPT                 (1 << 24)
+#       define DC_I2C_HW_DONE_INTERRUPT                 (1 << 25)
+#define DISP_INTERRUPT_STATUS_CONTINUE                  0x183E
+#       define LB_D2_VLINE_INTERRUPT                    (1 << 2)
+#       define LB_D2_VBLANK_INTERRUPT                   (1 << 3)
+#       define DC_HPD2_INTERRUPT                        (1 << 17)
+#       define DC_HPD2_RX_INTERRUPT                     (1 << 18)
+#       define DISP_TIMER_INTERRUPT                     (1 << 24)
+#define DISP_INTERRUPT_STATUS_CONTINUE2                 0x183F
+#       define LB_D3_VLINE_INTERRUPT                    (1 << 2)
+#       define LB_D3_VBLANK_INTERRUPT                   (1 << 3)
+#       define DC_HPD3_INTERRUPT                        (1 << 17)
+#       define DC_HPD3_RX_INTERRUPT                     (1 << 18)
+#define DISP_INTERRUPT_STATUS_CONTINUE3                 0x1840
+#       define LB_D4_VLINE_INTERRUPT                    (1 << 2)
+#       define LB_D4_VBLANK_INTERRUPT                   (1 << 3)
+#       define DC_HPD4_INTERRUPT                        (1 << 17)
+#       define DC_HPD4_RX_INTERRUPT                     (1 << 18)
+#define DISP_INTERRUPT_STATUS_CONTINUE4                 0x1853
+#       define LB_D5_VLINE_INTERRUPT                    (1 << 2)
+#       define LB_D5_VBLANK_INTERRUPT                   (1 << 3)
+#       define DC_HPD5_INTERRUPT                        (1 << 17)
+#       define DC_HPD5_RX_INTERRUPT                     (1 << 18)
+#define DISP_INTERRUPT_STATUS_CONTINUE5                 0x1854
+#       define LB_D6_VLINE_INTERRUPT                    (1 << 2)
+#       define LB_D6_VBLANK_INTERRUPT                   (1 << 3)
+#       define DC_HPD6_INTERRUPT                        (1 << 17)
+#       define DC_HPD6_RX_INTERRUPT                     (1 << 18)
+
+/* 0x6858, 0x7458, 0x10058, 0x10c58, 0x11858, 0x12458 */
+#define GRPH_INT_STATUS                                 0x1A16
+#       define GRPH_PFLIP_INT_OCCURRED                  (1 << 0)
+#       define GRPH_PFLIP_INT_CLEAR                     (1 << 8)
+/* 0x685c, 0x745c, 0x1005c, 0x10c5c, 0x1185c, 0x1245c */
+#define	GRPH_INT_CONTROL			        0x1A17
+#       define GRPH_PFLIP_INT_MASK                      (1 << 0)
+#       define GRPH_PFLIP_INT_TYPE                      (1 << 8)
+
+#define	DAC_AUTODETECT_INT_CONTROL			0x19F2
+
+#define DC_HPD1_INT_STATUS                              0x1807
+#define DC_HPD2_INT_STATUS                              0x180A
+#define DC_HPD3_INT_STATUS                              0x180D
+#define DC_HPD4_INT_STATUS                              0x1810
+#define DC_HPD5_INT_STATUS                              0x1813
+#define DC_HPD6_INT_STATUS                              0x1816
+#       define DC_HPDx_INT_STATUS                       (1 << 0)
+#       define DC_HPDx_SENSE                            (1 << 1)
+#       define DC_HPDx_RX_INT_STATUS                    (1 << 8)
+
+#define DC_HPD1_INT_CONTROL                             0x1808
+#define DC_HPD2_INT_CONTROL                             0x180B
+#define DC_HPD3_INT_CONTROL                             0x180E
+#define DC_HPD4_INT_CONTROL                             0x1811
+#define DC_HPD5_INT_CONTROL                             0x1814
+#define DC_HPD6_INT_CONTROL                             0x1817
+#       define DC_HPDx_INT_ACK                          (1 << 0)
+#       define DC_HPDx_INT_POLARITY                     (1 << 8)
+#       define DC_HPDx_INT_EN                           (1 << 16)
+#       define DC_HPDx_RX_INT_ACK                       (1 << 20)
+#       define DC_HPDx_RX_INT_EN                        (1 << 24)
+
+#define DC_HPD1_CONTROL                                   0x1809
+#define DC_HPD2_CONTROL                                   0x180C
+#define DC_HPD3_CONTROL                                   0x180F
+#define DC_HPD4_CONTROL                                   0x1812
+#define DC_HPD5_CONTROL                                   0x1815
+#define DC_HPD6_CONTROL                                   0x1818
+#       define DC_HPDx_CONNECTION_TIMER(x)                ((x) << 0)
+#       define DC_HPDx_RX_INT_TIMER(x)                    ((x) << 16)
+#       define DC_HPDx_EN                                 (1 << 28)
+
+#define DPG_PIPE_STUTTER_CONTROL                          0x1B35
+#       define STUTTER_ENABLE                             (1 << 0)
+
+/* 0x6e98, 0x7a98, 0x10698, 0x11298, 0x11e98, 0x12a98 */
+#define CRTC_STATUS_FRAME_COUNT                         0x1BA6
+
+/* Audio clocks */
+#define DCCG_AUDIO_DTO_SOURCE                           0x05ac
+#       define DCCG_AUDIO_DTO0_SOURCE_SEL(x) ((x) << 0) /* crtc0 - crtc5 */
+#       define DCCG_AUDIO_DTO_SEL            (1 << 4)   /* 0=dto0 1=dto1 */
+
+#define DCCG_AUDIO_DTO0_PHASE                           0x05b0
+#define DCCG_AUDIO_DTO0_MODULE                          0x05b4
+#define DCCG_AUDIO_DTO1_PHASE                           0x05c0
+#define DCCG_AUDIO_DTO1_MODULE                          0x05c4
+
+#define AFMT_AUDIO_SRC_CONTROL                          0x1c4f
+#define		AFMT_AUDIO_SRC_SELECT(x)		(((x) & 7) << 0)
+/* AFMT_AUDIO_SRC_SELECT
+ * 0 = stream0
+ * 1 = stream1
+ * 2 = stream2
+ * 3 = stream3
+ * 4 = stream4
+ * 5 = stream5
+ */
+
+#define	GRBM_CNTL					0x2000
+#define		GRBM_READ_TIMEOUT(x)				((x) << 0)
+
+#define	GRBM_STATUS2					0x2002
+#define		RLC_RQ_PENDING 					(1 << 0)
+#define		RLC_BUSY 					(1 << 8)
+#define		TC_BUSY 					(1 << 9)
+
+#define	GRBM_STATUS					0x2004
+#define		CMDFIFO_AVAIL_MASK				0x0000000F
+#define		RING2_RQ_PENDING				(1 << 4)
+#define		SRBM_RQ_PENDING					(1 << 5)
+#define		RING1_RQ_PENDING				(1 << 6)
+#define		CF_RQ_PENDING					(1 << 7)
+#define		PF_RQ_PENDING					(1 << 8)
+#define		GDS_DMA_RQ_PENDING				(1 << 9)
+#define		GRBM_EE_BUSY					(1 << 10)
+#define		DB_CLEAN					(1 << 12)
+#define		CB_CLEAN					(1 << 13)
+#define		TA_BUSY 					(1 << 14)
+#define		GDS_BUSY 					(1 << 15)
+#define		VGT_BUSY					(1 << 17)
+#define		IA_BUSY_NO_DMA					(1 << 18)
+#define		IA_BUSY						(1 << 19)
+#define		SX_BUSY 					(1 << 20)
+#define		SPI_BUSY					(1 << 22)
+#define		BCI_BUSY					(1 << 23)
+#define		SC_BUSY 					(1 << 24)
+#define		PA_BUSY 					(1 << 25)
+#define		DB_BUSY 					(1 << 26)
+#define		CP_COHERENCY_BUSY      				(1 << 28)
+#define		CP_BUSY 					(1 << 29)
+#define		CB_BUSY 					(1 << 30)
+#define		GUI_ACTIVE					(1 << 31)
+#define	GRBM_STATUS_SE0					0x2005
+#define	GRBM_STATUS_SE1					0x2006
+#define		SE_DB_CLEAN					(1 << 1)
+#define		SE_CB_CLEAN					(1 << 2)
+#define		SE_BCI_BUSY					(1 << 22)
+#define		SE_VGT_BUSY					(1 << 23)
+#define		SE_PA_BUSY					(1 << 24)
+#define		SE_TA_BUSY					(1 << 25)
+#define		SE_SX_BUSY					(1 << 26)
+#define		SE_SPI_BUSY					(1 << 27)
+#define		SE_SC_BUSY					(1 << 29)
+#define		SE_DB_BUSY					(1 << 30)
+#define		SE_CB_BUSY					(1 << 31)
+
+#define	GRBM_SOFT_RESET					0x2008
+#define		SOFT_RESET_CP					(1 << 0)
+#define		SOFT_RESET_CB					(1 << 1)
+#define		SOFT_RESET_RLC					(1 << 2)
+#define		SOFT_RESET_DB					(1 << 3)
+#define		SOFT_RESET_GDS					(1 << 4)
+#define		SOFT_RESET_PA					(1 << 5)
+#define		SOFT_RESET_SC					(1 << 6)
+#define		SOFT_RESET_BCI					(1 << 7)
+#define		SOFT_RESET_SPI					(1 << 8)
+#define		SOFT_RESET_SX					(1 << 10)
+#define		SOFT_RESET_TC					(1 << 11)
+#define		SOFT_RESET_TA					(1 << 12)
+#define		SOFT_RESET_VGT					(1 << 14)
+#define		SOFT_RESET_IA					(1 << 15)
+
+#define GRBM_GFX_INDEX          			0x200B
+#define		INSTANCE_INDEX(x)			((x) << 0)
+#define		SH_INDEX(x)     			((x) << 8)
+#define		SE_INDEX(x)     			((x) << 16)
+#define		SH_BROADCAST_WRITES      		(1 << 29)
+#define		INSTANCE_BROADCAST_WRITES      		(1 << 30)
+#define		SE_BROADCAST_WRITES      		(1 << 31)
+
+#define GRBM_INT_CNTL                                   0x2018
+#       define RDERR_INT_ENABLE                         (1 << 0)
+#       define GUI_IDLE_INT_ENABLE                      (1 << 19)
+
+#define	CP_STRMOUT_CNTL					0x213F
+#define	SCRATCH_REG0					0x2140
+#define	SCRATCH_REG1					0x2141
+#define	SCRATCH_REG2					0x2142
+#define	SCRATCH_REG3					0x2143
+#define	SCRATCH_REG4					0x2144
+#define	SCRATCH_REG5					0x2145
+#define	SCRATCH_REG6					0x2146
+#define	SCRATCH_REG7					0x2147
+
+#define	SCRATCH_UMSK					0x2150
+#define	SCRATCH_ADDR					0x2151
+
+#define	CP_SEM_WAIT_TIMER				0x216F
+
+#define	CP_SEM_INCOMPLETE_TIMER_CNTL			0x2172
+
+#define CP_ME_CNTL					0x21B6
+#define		CP_CE_HALT					(1 << 24)
+#define		CP_PFP_HALT					(1 << 26)
+#define		CP_ME_HALT					(1 << 28)
+
+#define	CP_COHER_CNTL2					0x217A
+
+#define	CP_RB2_RPTR					0x21BE
+#define	CP_RB1_RPTR					0x21BF
+#define	CP_RB0_RPTR					0x21C0
+#define	CP_RB_WPTR_DELAY				0x21C1
+
+#define	CP_QUEUE_THRESHOLDS				0x21D8
+#define		ROQ_IB1_START(x)				((x) << 0)
+#define		ROQ_IB2_START(x)				((x) << 8)
+#define CP_MEQ_THRESHOLDS				0x21D9
+#define		MEQ1_START(x)				((x) << 0)
+#define		MEQ2_START(x)				((x) << 8)
+
+#define	CP_PERFMON_CNTL					0x21FF
+
+#define	VGT_VTX_VECT_EJECT_REG				0x222C
+
+#define	VGT_CACHE_INVALIDATION				0x2231
+#define		CACHE_INVALIDATION(x)				((x) << 0)
+#define			VC_ONLY						0
+#define			TC_ONLY						1
+#define			VC_AND_TC					2
+#define		AUTO_INVLD_EN(x)				((x) << 6)
+#define			NO_AUTO						0
+#define			ES_AUTO						1
+#define			GS_AUTO						2
+#define			ES_AND_GS_AUTO					3
+#define	VGT_ESGS_RING_SIZE				0x2232
+#define	VGT_GSVS_RING_SIZE				0x2233
+
+#define	VGT_GS_VERTEX_REUSE				0x2235
+
+#define	VGT_PRIMITIVE_TYPE				0x2256
+#define	VGT_INDEX_TYPE					0x2257
+
+#define	VGT_NUM_INDICES					0x225C
+#define	VGT_NUM_INSTANCES				0x225D
+
+#define	VGT_TF_RING_SIZE				0x2262
+
+#define	VGT_HS_OFFCHIP_PARAM				0x226C
+
+#define	VGT_TF_MEMORY_BASE				0x226E
+
+#define CC_GC_SHADER_ARRAY_CONFIG			0x226F
+#define		INACTIVE_CUS_MASK			0xFFFF0000
+#define		INACTIVE_CUS_SHIFT			16
+#define GC_USER_SHADER_ARRAY_CONFIG			0x2270
+
+#define	PA_CL_ENHANCE					0x2285
+#define		CLIP_VTX_REORDER_ENA				(1 << 0)
+#define		NUM_CLIP_SEQ(x)					((x) << 1)
+
+#define	PA_SU_LINE_STIPPLE_VALUE			0x2298
+
+#define	PA_SC_LINE_STIPPLE_STATE			0x22C4
+
+#define	PA_SC_FORCE_EOV_MAX_CNTS			0x22C9
+#define		FORCE_EOV_MAX_CLK_CNT(x)			((x) << 0)
+#define		FORCE_EOV_MAX_REZ_CNT(x)			((x) << 16)
+
+#define	PA_SC_FIFO_SIZE					0x22F3
+#define		SC_FRONTEND_PRIM_FIFO_SIZE(x)			((x) << 0)
+#define		SC_BACKEND_PRIM_FIFO_SIZE(x)			((x) << 6)
+#define		SC_HIZ_TILE_FIFO_SIZE(x)			((x) << 15)
+#define		SC_EARLYZ_TILE_FIFO_SIZE(x)			((x) << 23)
+
+#define	PA_SC_ENHANCE					0x22FC
+
+#define	SQ_CONFIG					0x2300
+
+#define	SQC_CACHES					0x2302
+
+#define SQ_POWER_THROTTLE                               0x2396
+#define		MIN_POWER(x)				((x) << 0)
+#define		MIN_POWER_MASK				(0x3fff << 0)
+#define		MIN_POWER_SHIFT				0
+#define		MAX_POWER(x)				((x) << 16)
+#define		MAX_POWER_MASK				(0x3fff << 16)
+#define		MAX_POWER_SHIFT				0
+#define SQ_POWER_THROTTLE2                              0x2397
+#define		MAX_POWER_DELTA(x)			((x) << 0)
+#define		MAX_POWER_DELTA_MASK			(0x3fff << 0)
+#define		MAX_POWER_DELTA_SHIFT			0
+#define		STI_SIZE(x)				((x) << 16)
+#define		STI_SIZE_MASK				(0x3ff << 16)
+#define		STI_SIZE_SHIFT				16
+#define		LTI_RATIO(x)				((x) << 27)
+#define		LTI_RATIO_MASK				(0xf << 27)
+#define		LTI_RATIO_SHIFT				27
+
+#define	SX_DEBUG_1					0x2418
+
+#define	SPI_STATIC_THREAD_MGMT_1			0x2438
+#define	SPI_STATIC_THREAD_MGMT_2			0x2439
+#define	SPI_STATIC_THREAD_MGMT_3			0x243A
+#define	SPI_PS_MAX_WAVE_ID				0x243B
+
+#define	SPI_CONFIG_CNTL					0x2440
+
+#define	SPI_CONFIG_CNTL_1				0x244F
+#define		VTX_DONE_DELAY(x)				((x) << 0)
+#define		INTERP_ONE_PRIM_PER_ROW				(1 << 4)
+
+#define	CGTS_TCC_DISABLE				0x2452
+#define	CGTS_USER_TCC_DISABLE				0x2453
+#define		TCC_DISABLE_MASK				0xFFFF0000
+#define		TCC_DISABLE_SHIFT				16
+#define	CGTS_SM_CTRL_REG				0x2454
+#define		OVERRIDE				(1 << 21)
+#define		LS_OVERRIDE				(1 << 22)
+
+#define	SPI_LB_CU_MASK					0x24D5
+
+#define	TA_CNTL_AUX					0x2542
+
+#define CC_RB_BACKEND_DISABLE				0x263D
+#define		BACKEND_DISABLE(x)     			((x) << 16)
+#define GB_ADDR_CONFIG  				0x263E
+#define		NUM_PIPES(x)				((x) << 0)
+#define		NUM_PIPES_MASK				0x00000007
+#define		NUM_PIPES_SHIFT				0
+#define		PIPE_INTERLEAVE_SIZE(x)			((x) << 4)
+#define		PIPE_INTERLEAVE_SIZE_MASK		0x00000070
+#define		PIPE_INTERLEAVE_SIZE_SHIFT		4
+#define		NUM_SHADER_ENGINES(x)			((x) << 12)
+#define		NUM_SHADER_ENGINES_MASK			0x00003000
+#define		NUM_SHADER_ENGINES_SHIFT		12
+#define		SHADER_ENGINE_TILE_SIZE(x)     		((x) << 16)
+#define		SHADER_ENGINE_TILE_SIZE_MASK		0x00070000
+#define		SHADER_ENGINE_TILE_SIZE_SHIFT		16
+#define		NUM_GPUS(x)     			((x) << 20)
+#define		NUM_GPUS_MASK				0x00700000
+#define		NUM_GPUS_SHIFT				20
+#define		MULTI_GPU_TILE_SIZE(x)     		((x) << 24)
+#define		MULTI_GPU_TILE_SIZE_MASK		0x03000000
+#define		MULTI_GPU_TILE_SIZE_SHIFT		24
+#define		ROW_SIZE(x)             		((x) << 28)
+#define		ROW_SIZE_MASK				0x30000000
+#define		ROW_SIZE_SHIFT				28
+
+#define	GB_TILE_MODE0					0x2644
+#       define MICRO_TILE_MODE(x)				((x) << 0)
+#              define	ADDR_SURF_DISPLAY_MICRO_TILING		0
+#              define	ADDR_SURF_THIN_MICRO_TILING		1
+#              define	ADDR_SURF_DEPTH_MICRO_TILING		2
+#       define ARRAY_MODE(x)					((x) << 2)
+#              define	ARRAY_LINEAR_GENERAL			0
+#              define	ARRAY_LINEAR_ALIGNED			1
+#              define	ARRAY_1D_TILED_THIN1			2
+#              define	ARRAY_2D_TILED_THIN1			4
+#       define PIPE_CONFIG(x)					((x) << 6)
+#              define	ADDR_SURF_P2				0
+#              define	ADDR_SURF_P4_8x16			4
+#              define	ADDR_SURF_P4_16x16			5
+#              define	ADDR_SURF_P4_16x32			6
+#              define	ADDR_SURF_P4_32x32			7
+#              define	ADDR_SURF_P8_16x16_8x16			8
+#              define	ADDR_SURF_P8_16x32_8x16			9
+#              define	ADDR_SURF_P8_32x32_8x16			10
+#              define	ADDR_SURF_P8_16x32_16x16		11
+#              define	ADDR_SURF_P8_32x32_16x16		12
+#              define	ADDR_SURF_P8_32x32_16x32		13
+#              define	ADDR_SURF_P8_32x64_32x32		14
+#       define TILE_SPLIT(x)					((x) << 11)
+#              define	ADDR_SURF_TILE_SPLIT_64B		0
+#              define	ADDR_SURF_TILE_SPLIT_128B		1
+#              define	ADDR_SURF_TILE_SPLIT_256B		2
+#              define	ADDR_SURF_TILE_SPLIT_512B		3
+#              define	ADDR_SURF_TILE_SPLIT_1KB		4
+#              define	ADDR_SURF_TILE_SPLIT_2KB		5
+#              define	ADDR_SURF_TILE_SPLIT_4KB		6
+#       define BANK_WIDTH(x)					((x) << 14)
+#              define	ADDR_SURF_BANK_WIDTH_1			0
+#              define	ADDR_SURF_BANK_WIDTH_2			1
+#              define	ADDR_SURF_BANK_WIDTH_4			2
+#              define	ADDR_SURF_BANK_WIDTH_8			3
+#       define BANK_HEIGHT(x)					((x) << 16)
+#              define	ADDR_SURF_BANK_HEIGHT_1			0
+#              define	ADDR_SURF_BANK_HEIGHT_2			1
+#              define	ADDR_SURF_BANK_HEIGHT_4			2
+#              define	ADDR_SURF_BANK_HEIGHT_8			3
+#       define MACRO_TILE_ASPECT(x)				((x) << 18)
+#              define	ADDR_SURF_MACRO_ASPECT_1		0
+#              define	ADDR_SURF_MACRO_ASPECT_2		1
+#              define	ADDR_SURF_MACRO_ASPECT_4		2
+#              define	ADDR_SURF_MACRO_ASPECT_8		3
+#       define NUM_BANKS(x)					((x) << 20)
+#              define	ADDR_SURF_2_BANK			0
+#              define	ADDR_SURF_4_BANK			1
+#              define	ADDR_SURF_8_BANK			2
+#              define	ADDR_SURF_16_BANK			3
+#define	GB_TILE_MODE1					0x2645
+#define	GB_TILE_MODE2					0x2646
+#define	GB_TILE_MODE3					0x2647
+#define	GB_TILE_MODE4					0x2648
+#define	GB_TILE_MODE5					0x2649
+#define	GB_TILE_MODE6					0x264a
+#define	GB_TILE_MODE7					0x264b
+#define	GB_TILE_MODE8					0x264c
+#define	GB_TILE_MODE9					0x264d
+#define	GB_TILE_MODE10					0x264e
+#define	GB_TILE_MODE11					0x264f
+#define	GB_TILE_MODE12					0x2650
+#define	GB_TILE_MODE13					0x2651
+#define	GB_TILE_MODE14					0x2652
+#define	GB_TILE_MODE15					0x2653
+#define	GB_TILE_MODE16					0x2654
+#define	GB_TILE_MODE17					0x2655
+#define	GB_TILE_MODE18					0x2656
+#define	GB_TILE_MODE19					0x2657
+#define	GB_TILE_MODE20					0x2658
+#define	GB_TILE_MODE21					0x2659
+#define	GB_TILE_MODE22					0x265a
+#define	GB_TILE_MODE23					0x265b
+#define	GB_TILE_MODE24					0x265c
+#define	GB_TILE_MODE25					0x265d
+#define	GB_TILE_MODE26					0x265e
+#define	GB_TILE_MODE27					0x265f
+#define	GB_TILE_MODE28					0x2660
+#define	GB_TILE_MODE29					0x2661
+#define	GB_TILE_MODE30					0x2662
+#define	GB_TILE_MODE31					0x2663
+
+#define	CB_PERFCOUNTER0_SELECT0				0x2688
+#define	CB_PERFCOUNTER0_SELECT1				0x2689
+#define	CB_PERFCOUNTER1_SELECT0				0x268A
+#define	CB_PERFCOUNTER1_SELECT1				0x268B
+#define	CB_PERFCOUNTER2_SELECT0				0x268C
+#define	CB_PERFCOUNTER2_SELECT1				0x268D
+#define	CB_PERFCOUNTER3_SELECT0				0x268E
+#define	CB_PERFCOUNTER3_SELECT1				0x268F
+
+#define	CB_CGTT_SCLK_CTRL				0x2698
+
+#define	GC_USER_RB_BACKEND_DISABLE			0x26DF
+#define		BACKEND_DISABLE_MASK			0x00FF0000
+#define		BACKEND_DISABLE_SHIFT			16
+
+#define	TCP_CHAN_STEER_LO				0x2B03
+#define	TCP_CHAN_STEER_HI				0x2B94
+
+#define	CP_RB0_BASE					0x3040
+#define	CP_RB0_CNTL					0x3041
+#define		RB_BUFSZ(x)					((x) << 0)
+#define		RB_BLKSZ(x)					((x) << 8)
+#define		BUF_SWAP_32BIT					(2 << 16)
+#define		RB_NO_UPDATE					(1 << 27)
+#define		RB_RPTR_WR_ENA					(1 << 31)
+
+#define	CP_RB0_RPTR_ADDR				0x3043
+#define	CP_RB0_RPTR_ADDR_HI				0x3044
+#define	CP_RB0_WPTR					0x3045
+
+#define	CP_PFP_UCODE_ADDR				0x3054
+#define	CP_PFP_UCODE_DATA				0x3055
+#define	CP_ME_RAM_RADDR					0x3056
+#define	CP_ME_RAM_WADDR					0x3057
+#define	CP_ME_RAM_DATA					0x3058
+
+#define	CP_CE_UCODE_ADDR				0x305A
+#define	CP_CE_UCODE_DATA				0x305B
+
+#define	CP_RB1_BASE					0x3060
+#define	CP_RB1_CNTL					0x3061
+#define	CP_RB1_RPTR_ADDR				0x3062
+#define	CP_RB1_RPTR_ADDR_HI				0x3063
+#define	CP_RB1_WPTR					0x3064
+#define	CP_RB2_BASE					0x3065
+#define	CP_RB2_CNTL					0x3066
+#define	CP_RB2_RPTR_ADDR				0x3067
+#define	CP_RB2_RPTR_ADDR_HI				0x3068
+#define	CP_RB2_WPTR					0x3069
+#define CP_INT_CNTL_RING0                               0x306A
+#define CP_INT_CNTL_RING1                               0x306B
+#define CP_INT_CNTL_RING2                               0x306C
+#       define CNTX_BUSY_INT_ENABLE                     (1 << 19)
+#       define CNTX_EMPTY_INT_ENABLE                    (1 << 20)
+#       define WAIT_MEM_SEM_INT_ENABLE                  (1 << 21)
+#       define TIME_STAMP_INT_ENABLE                    (1 << 26)
+#       define CP_RINGID2_INT_ENABLE                    (1 << 29)
+#       define CP_RINGID1_INT_ENABLE                    (1 << 30)
+#       define CP_RINGID0_INT_ENABLE                    (1 << 31)
+#define CP_INT_STATUS_RING0                             0x306D
+#define CP_INT_STATUS_RING1                             0x306E
+#define CP_INT_STATUS_RING2                             0x306F
+#       define WAIT_MEM_SEM_INT_STAT                    (1 << 21)
+#       define TIME_STAMP_INT_STAT                      (1 << 26)
+#       define CP_RINGID2_INT_STAT                      (1 << 29)
+#       define CP_RINGID1_INT_STAT                      (1 << 30)
+#       define CP_RINGID0_INT_STAT                      (1 << 31)
+
+#define	CP_MEM_SLP_CNTL					0x3079
+#       define CP_MEM_LS_EN                             (1 << 0)
+
+#define	CP_DEBUG					0x307F
+
+#define RLC_CNTL                                          0x30C0
+#       define RLC_ENABLE                                 (1 << 0)
+#define RLC_RL_BASE                                       0x30C1
+#define RLC_RL_SIZE                                       0x30C2
+#define RLC_LB_CNTL                                       0x30C3
+#       define LOAD_BALANCE_ENABLE                        (1 << 0)
+#define RLC_SAVE_AND_RESTORE_BASE                         0x30C4
+#define RLC_LB_CNTR_MAX                                   0x30C5
+#define RLC_LB_CNTR_INIT                                  0x30C6
+
+#define RLC_CLEAR_STATE_RESTORE_BASE                      0x30C8
+
+#define RLC_UCODE_ADDR                                    0x30CB
+#define RLC_UCODE_DATA                                    0x30CC
+
+#define RLC_GPU_CLOCK_COUNT_LSB                           0x30CE
+#define RLC_GPU_CLOCK_COUNT_MSB                           0x30CF
+#define RLC_CAPTURE_GPU_CLOCK_COUNT                       0x30D0
+#define RLC_MC_CNTL                                       0x30D1
+#define RLC_UCODE_CNTL                                    0x30D2
+#define RLC_STAT                                          0x30D3
+#       define RLC_BUSY_STATUS                            (1 << 0)
+#       define GFX_POWER_STATUS                           (1 << 1)
+#       define GFX_CLOCK_STATUS                           (1 << 2)
+#       define GFX_LS_STATUS                              (1 << 3)
+
+#define	RLC_PG_CNTL					0x30D7
+#	define GFX_PG_ENABLE				(1 << 0)
+#	define GFX_PG_SRC				(1 << 1)
+
+#define	RLC_CGTT_MGCG_OVERRIDE				0x3100
+#define	RLC_CGCG_CGLS_CTRL				0x3101
+#	define CGCG_EN					(1 << 0)
+#	define CGLS_EN					(1 << 1)
+
+#define	RLC_TTOP_D					0x3105
+#	define RLC_PUD(x)				((x) << 0)
+#	define RLC_PUD_MASK				(0xff << 0)
+#	define RLC_PDD(x)				((x) << 8)
+#	define RLC_PDD_MASK				(0xff << 8)
+#	define RLC_TTPD(x)				((x) << 16)
+#	define RLC_TTPD_MASK				(0xff << 16)
+#	define RLC_MSD(x)				((x) << 24)
+#	define RLC_MSD_MASK				(0xff << 24)
+
+#define RLC_LB_INIT_CU_MASK                               0x3107
+
+#define	RLC_PG_AO_CU_MASK				0x310B
+#define	RLC_MAX_PG_CU					0x310C
+#	define MAX_PU_CU(x)				((x) << 0)
+#	define MAX_PU_CU_MASK				(0xff << 0)
+#define	RLC_AUTO_PG_CTRL				0x310C
+#	define AUTO_PG_EN				(1 << 0)
+#	define GRBM_REG_SGIT(x)				((x) << 3)
+#	define GRBM_REG_SGIT_MASK			(0xffff << 3)
+#	define PG_AFTER_GRBM_REG_ST(x)			((x) << 19)
+#	define PG_AFTER_GRBM_REG_ST_MASK		(0x1fff << 19)
+
+#define RLC_SERDES_WR_MASTER_MASK_0                       0x3115
+#define RLC_SERDES_WR_MASTER_MASK_1                       0x3116
+#define RLC_SERDES_WR_CTRL                                0x3117
+
+#define RLC_SERDES_MASTER_BUSY_0                          0x3119
+#define RLC_SERDES_MASTER_BUSY_1                          0x311A
+
+#define RLC_GCPM_GENERAL_3                                0x311E
+
+#define	DB_RENDER_CONTROL				0xA000
+
+#define DB_DEPTH_INFO                                   0xA00F
+
+#define PA_SC_RASTER_CONFIG                             0xA0D4
+#	define RB_MAP_PKR0(x)				((x) << 0)
+#	define RB_MAP_PKR0_MASK				(0x3 << 0)
+#	define RB_MAP_PKR1(x)				((x) << 2)
+#	define RB_MAP_PKR1_MASK				(0x3 << 2)
+#       define RASTER_CONFIG_RB_MAP_0                   0
+#       define RASTER_CONFIG_RB_MAP_1                   1
+#       define RASTER_CONFIG_RB_MAP_2                   2
+#       define RASTER_CONFIG_RB_MAP_3                   3
+#	define RB_XSEL2(x)				((x) << 4)
+#	define RB_XSEL2_MASK				(0x3 << 4)
+#	define RB_XSEL					(1 << 6)
+#	define RB_YSEL					(1 << 7)
+#	define PKR_MAP(x)				((x) << 8)
+#	define PKR_MAP_MASK				(0x3 << 8)
+#       define RASTER_CONFIG_PKR_MAP_0			0
+#       define RASTER_CONFIG_PKR_MAP_1			1
+#       define RASTER_CONFIG_PKR_MAP_2			2
+#       define RASTER_CONFIG_PKR_MAP_3			3
+#	define PKR_XSEL(x)				((x) << 10)
+#	define PKR_XSEL_MASK				(0x3 << 10)
+#	define PKR_YSEL(x)				((x) << 12)
+#	define PKR_YSEL_MASK				(0x3 << 12)
+#	define SC_MAP(x)				((x) << 16)
+#	define SC_MAP_MASK				(0x3 << 16)
+#	define SC_XSEL(x)				((x) << 18)
+#	define SC_XSEL_MASK				(0x3 << 18)
+#	define SC_YSEL(x)				((x) << 20)
+#	define SC_YSEL_MASK				(0x3 << 20)
+#	define SE_MAP(x)				((x) << 24)
+#	define SE_MAP_MASK				(0x3 << 24)
+#       define RASTER_CONFIG_SE_MAP_0			0
+#       define RASTER_CONFIG_SE_MAP_1			1
+#       define RASTER_CONFIG_SE_MAP_2			2
+#       define RASTER_CONFIG_SE_MAP_3			3
+#	define SE_XSEL(x)				((x) << 26)
+#	define SE_XSEL_MASK				(0x3 << 26)
+#	define SE_YSEL(x)				((x) << 28)
+#	define SE_YSEL_MASK				(0x3 << 28)
+
+
+#define VGT_EVENT_INITIATOR                             0xA2A4
+#       define SAMPLE_STREAMOUTSTATS1                   (1 << 0)
+#       define SAMPLE_STREAMOUTSTATS2                   (2 << 0)
+#       define SAMPLE_STREAMOUTSTATS3                   (3 << 0)
+#       define CACHE_FLUSH_TS                           (4 << 0)
+#       define CACHE_FLUSH                              (6 << 0)
+#       define CS_PARTIAL_FLUSH                         (7 << 0)
+#       define VGT_STREAMOUT_RESET                      (10 << 0)
+#       define END_OF_PIPE_INCR_DE                      (11 << 0)
+#       define END_OF_PIPE_IB_END                       (12 << 0)
+#       define RST_PIX_CNT                              (13 << 0)
+#       define VS_PARTIAL_FLUSH                         (15 << 0)
+#       define PS_PARTIAL_FLUSH                         (16 << 0)
+#       define CACHE_FLUSH_AND_INV_TS_EVENT             (20 << 0)
+#       define ZPASS_DONE                               (21 << 0)
+#       define CACHE_FLUSH_AND_INV_EVENT                (22 << 0)
+#       define PERFCOUNTER_START                        (23 << 0)
+#       define PERFCOUNTER_STOP                         (24 << 0)
+#       define PIPELINESTAT_START                       (25 << 0)
+#       define PIPELINESTAT_STOP                        (26 << 0)
+#       define PERFCOUNTER_SAMPLE                       (27 << 0)
+#       define SAMPLE_PIPELINESTAT                      (30 << 0)
+#       define SAMPLE_STREAMOUTSTATS                    (32 << 0)
+#       define RESET_VTX_CNT                            (33 << 0)
+#       define VGT_FLUSH                                (36 << 0)
+#       define BOTTOM_OF_PIPE_TS                        (40 << 0)
+#       define DB_CACHE_FLUSH_AND_INV                   (42 << 0)
+#       define FLUSH_AND_INV_DB_DATA_TS                 (43 << 0)
+#       define FLUSH_AND_INV_DB_META                    (44 << 0)
+#       define FLUSH_AND_INV_CB_DATA_TS                 (45 << 0)
+#       define FLUSH_AND_INV_CB_META                    (46 << 0)
+#       define CS_DONE                                  (47 << 0)
+#       define PS_DONE                                  (48 << 0)
+#       define FLUSH_AND_INV_CB_PIXEL_DATA              (49 << 0)
+#       define THREAD_TRACE_START                       (51 << 0)
+#       define THREAD_TRACE_STOP                        (52 << 0)
+#       define THREAD_TRACE_FLUSH                       (54 << 0)
+#       define THREAD_TRACE_FINISH                      (55 << 0)
+
+/* PIF PHY0 registers idx/data 0x8/0xc */
+#define PB0_PIF_CNTL                                      0x10
+#       define LS2_EXIT_TIME(x)                           ((x) << 17)
+#       define LS2_EXIT_TIME_MASK                         (0x7 << 17)
+#       define LS2_EXIT_TIME_SHIFT                        17
+#define PB0_PIF_PAIRING                                   0x11
+#       define MULTI_PIF                                  (1 << 25)
+#define PB0_PIF_PWRDOWN_0                                 0x12
+#       define PLL_POWER_STATE_IN_TXS2_0(x)               ((x) << 7)
+#       define PLL_POWER_STATE_IN_TXS2_0_MASK             (0x7 << 7)
+#       define PLL_POWER_STATE_IN_TXS2_0_SHIFT            7
+#       define PLL_POWER_STATE_IN_OFF_0(x)                ((x) << 10)
+#       define PLL_POWER_STATE_IN_OFF_0_MASK              (0x7 << 10)
+#       define PLL_POWER_STATE_IN_OFF_0_SHIFT             10
+#       define PLL_RAMP_UP_TIME_0(x)                      ((x) << 24)
+#       define PLL_RAMP_UP_TIME_0_MASK                    (0x7 << 24)
+#       define PLL_RAMP_UP_TIME_0_SHIFT                   24
+#define PB0_PIF_PWRDOWN_1                                 0x13
+#       define PLL_POWER_STATE_IN_TXS2_1(x)               ((x) << 7)
+#       define PLL_POWER_STATE_IN_TXS2_1_MASK             (0x7 << 7)
+#       define PLL_POWER_STATE_IN_TXS2_1_SHIFT            7
+#       define PLL_POWER_STATE_IN_OFF_1(x)                ((x) << 10)
+#       define PLL_POWER_STATE_IN_OFF_1_MASK              (0x7 << 10)
+#       define PLL_POWER_STATE_IN_OFF_1_SHIFT             10
+#       define PLL_RAMP_UP_TIME_1(x)                      ((x) << 24)
+#       define PLL_RAMP_UP_TIME_1_MASK                    (0x7 << 24)
+#       define PLL_RAMP_UP_TIME_1_SHIFT                   24
+
+#define PB0_PIF_PWRDOWN_2                                 0x17
+#       define PLL_POWER_STATE_IN_TXS2_2(x)               ((x) << 7)
+#       define PLL_POWER_STATE_IN_TXS2_2_MASK             (0x7 << 7)
+#       define PLL_POWER_STATE_IN_TXS2_2_SHIFT            7
+#       define PLL_POWER_STATE_IN_OFF_2(x)                ((x) << 10)
+#       define PLL_POWER_STATE_IN_OFF_2_MASK              (0x7 << 10)
+#       define PLL_POWER_STATE_IN_OFF_2_SHIFT             10
+#       define PLL_RAMP_UP_TIME_2(x)                      ((x) << 24)
+#       define PLL_RAMP_UP_TIME_2_MASK                    (0x7 << 24)
+#       define PLL_RAMP_UP_TIME_2_SHIFT                   24
+#define PB0_PIF_PWRDOWN_3                                 0x18
+#       define PLL_POWER_STATE_IN_TXS2_3(x)               ((x) << 7)
+#       define PLL_POWER_STATE_IN_TXS2_3_MASK             (0x7 << 7)
+#       define PLL_POWER_STATE_IN_TXS2_3_SHIFT            7
+#       define PLL_POWER_STATE_IN_OFF_3(x)                ((x) << 10)
+#       define PLL_POWER_STATE_IN_OFF_3_MASK              (0x7 << 10)
+#       define PLL_POWER_STATE_IN_OFF_3_SHIFT             10
+#       define PLL_RAMP_UP_TIME_3(x)                      ((x) << 24)
+#       define PLL_RAMP_UP_TIME_3_MASK                    (0x7 << 24)
+#       define PLL_RAMP_UP_TIME_3_SHIFT                   24
+/* PIF PHY1 registers idx/data 0x10/0x14 */
+#define PB1_PIF_CNTL                                      0x10
+#define PB1_PIF_PAIRING                                   0x11
+#define PB1_PIF_PWRDOWN_0                                 0x12
+#define PB1_PIF_PWRDOWN_1                                 0x13
+
+#define PB1_PIF_PWRDOWN_2                                 0x17
+#define PB1_PIF_PWRDOWN_3                                 0x18
+/* PCIE registers idx/data 0x30/0x34 */
+#define PCIE_CNTL2                                        0x1c /* PCIE */
+#       define SLV_MEM_LS_EN                              (1 << 16)
+#       define SLV_MEM_AGGRESSIVE_LS_EN                   (1 << 17)
+#       define MST_MEM_LS_EN                              (1 << 18)
+#       define REPLAY_MEM_LS_EN                           (1 << 19)
+#define PCIE_LC_STATUS1                                   0x28 /* PCIE */
+#       define LC_REVERSE_RCVR                            (1 << 0)
+#       define LC_REVERSE_XMIT                            (1 << 1)
+#       define LC_OPERATING_LINK_WIDTH_MASK               (0x7 << 2)
+#       define LC_OPERATING_LINK_WIDTH_SHIFT              2
+#       define LC_DETECTED_LINK_WIDTH_MASK                (0x7 << 5)
+#       define LC_DETECTED_LINK_WIDTH_SHIFT               5
+
+#define PCIE_P_CNTL                                       0x40 /* PCIE */
+#       define P_IGNORE_EDB_ERR                           (1 << 6)
+
+/* PCIE PORT registers idx/data 0x38/0x3c */
+#define PCIE_LC_CNTL                                      0xa0
+#       define LC_L0S_INACTIVITY(x)                       ((x) << 8)
+#       define LC_L0S_INACTIVITY_MASK                     (0xf << 8)
+#       define LC_L0S_INACTIVITY_SHIFT                    8
+#       define LC_L1_INACTIVITY(x)                        ((x) << 12)
+#       define LC_L1_INACTIVITY_MASK                      (0xf << 12)
+#       define LC_L1_INACTIVITY_SHIFT                     12
+#       define LC_PMI_TO_L1_DIS                           (1 << 16)
+#       define LC_ASPM_TO_L1_DIS                          (1 << 24)
+#define PCIE_LC_LINK_WIDTH_CNTL                           0xa2 /* PCIE_P */
+#       define LC_LINK_WIDTH_SHIFT                        0
+#       define LC_LINK_WIDTH_MASK                         0x7
+#       define LC_LINK_WIDTH_X0                           0
+#       define LC_LINK_WIDTH_X1                           1
+#       define LC_LINK_WIDTH_X2                           2
+#       define LC_LINK_WIDTH_X4                           3
+#       define LC_LINK_WIDTH_X8                           4
+#       define LC_LINK_WIDTH_X16                          6
+#       define LC_LINK_WIDTH_RD_SHIFT                     4
+#       define LC_LINK_WIDTH_RD_MASK                      0x70
+#       define LC_RECONFIG_ARC_MISSING_ESCAPE             (1 << 7)
+#       define LC_RECONFIG_NOW                            (1 << 8)
+#       define LC_RENEGOTIATION_SUPPORT                   (1 << 9)
+#       define LC_RENEGOTIATE_EN                          (1 << 10)
+#       define LC_SHORT_RECONFIG_EN                       (1 << 11)
+#       define LC_UPCONFIGURE_SUPPORT                     (1 << 12)
+#       define LC_UPCONFIGURE_DIS                         (1 << 13)
+#       define LC_DYN_LANES_PWR_STATE(x)                  ((x) << 21)
+#       define LC_DYN_LANES_PWR_STATE_MASK                (0x3 << 21)
+#       define LC_DYN_LANES_PWR_STATE_SHIFT               21
+#define PCIE_LC_N_FTS_CNTL                                0xa3 /* PCIE_P */
+#       define LC_XMIT_N_FTS(x)                           ((x) << 0)
+#       define LC_XMIT_N_FTS_MASK                         (0xff << 0)
+#       define LC_XMIT_N_FTS_SHIFT                        0
+#       define LC_XMIT_N_FTS_OVERRIDE_EN                  (1 << 8)
+#       define LC_N_FTS_MASK                              (0xff << 24)
+#define PCIE_LC_SPEED_CNTL                                0xa4 /* PCIE_P */
+#       define LC_GEN2_EN_STRAP                           (1 << 0)
+#       define LC_GEN3_EN_STRAP                           (1 << 1)
+#       define LC_TARGET_LINK_SPEED_OVERRIDE_EN           (1 << 2)
+#       define LC_TARGET_LINK_SPEED_OVERRIDE_MASK         (0x3 << 3)
+#       define LC_TARGET_LINK_SPEED_OVERRIDE_SHIFT        3
+#       define LC_FORCE_EN_SW_SPEED_CHANGE                (1 << 5)
+#       define LC_FORCE_DIS_SW_SPEED_CHANGE               (1 << 6)
+#       define LC_FORCE_EN_HW_SPEED_CHANGE                (1 << 7)
+#       define LC_FORCE_DIS_HW_SPEED_CHANGE               (1 << 8)
+#       define LC_INITIATE_LINK_SPEED_CHANGE              (1 << 9)
+#       define LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_MASK      (0x3 << 10)
+#       define LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_SHIFT     10
+#       define LC_CURRENT_DATA_RATE_MASK                  (0x3 << 13) /* 0/1/2 = gen1/2/3 */
+#       define LC_CURRENT_DATA_RATE_SHIFT                 13
+#       define LC_CLR_FAILED_SPD_CHANGE_CNT               (1 << 16)
+#       define LC_OTHER_SIDE_EVER_SENT_GEN2               (1 << 18)
+#       define LC_OTHER_SIDE_SUPPORTS_GEN2                (1 << 19)
+#       define LC_OTHER_SIDE_EVER_SENT_GEN3               (1 << 20)
+#       define LC_OTHER_SIDE_SUPPORTS_GEN3                (1 << 21)
+
+#define PCIE_LC_CNTL2                                     0xb1
+#       define LC_ALLOW_PDWN_IN_L1                        (1 << 17)
+#       define LC_ALLOW_PDWN_IN_L23                       (1 << 18)
+
+#define PCIE_LC_CNTL3                                     0xb5 /* PCIE_P */
+#       define LC_GO_TO_RECOVERY                          (1 << 30)
+#define PCIE_LC_CNTL4                                     0xb6 /* PCIE_P */
+#       define LC_REDO_EQ                                 (1 << 5)
+#       define LC_SET_QUIESCE                             (1 << 13)
+
+/*
+ * UVD
+ */
+#define UVD_UDEC_ADDR_CONFIG				0x3bd3
+#define UVD_UDEC_DB_ADDR_CONFIG				0x3bd4
+#define UVD_UDEC_DBW_ADDR_CONFIG			0x3bd5
+#define UVD_RBC_RB_RPTR					0x3da4
+#define UVD_RBC_RB_WPTR					0x3da5
+#define UVD_STATUS					0x3daf
+
+#define	UVD_CGC_CTRL					0x3dc2
+#	define DCM					(1 << 0)
+#	define CG_DT(x)					((x) << 2)
+#	define CG_DT_MASK				(0xf << 2)
+#	define CLK_OD(x)				((x) << 6)
+#	define CLK_OD_MASK				(0x1f << 6)
+
+ /* UVD CTX indirect */
+#define	UVD_CGC_MEM_CTRL				0xC0
+#define	UVD_CGC_CTRL2					0xC1
+#	define DYN_OR_EN				(1 << 0)
+#	define DYN_RR_EN				(1 << 1)
+#	define G_DIV_ID(x)				((x) << 2)
+#	define G_DIV_ID_MASK				(0x7 << 2)
+
+/*
+ * PM4
+ */
+#define PACKET0(reg, n)	((RADEON_PACKET_TYPE0 << 30) |			\
+			 (((reg) >> 2) & 0xFFFF) |			\
+			 ((n) & 0x3FFF) << 16)
+#define CP_PACKET2			0x80000000
+#define		PACKET2_PAD_SHIFT		0
+#define		PACKET2_PAD_MASK		(0x3fffffff << 0)
+
+#define PACKET2(v)	(CP_PACKET2 | REG_SET(PACKET2_PAD, (v)))
+#define RADEON_PACKET_TYPE3 3
+#define PACKET3(op, n)	((RADEON_PACKET_TYPE3 << 30) |			\
+			 (((op) & 0xFF) << 8) |				\
+			 ((n) & 0x3FFF) << 16)
+
+#define PACKET3_COMPUTE(op, n) (PACKET3(op, n) | 1 << 1)
+
+/* Packet 3 types */
+#define	PACKET3_NOP					0x10
+#define	PACKET3_SET_BASE				0x11
+#define		PACKET3_BASE_INDEX(x)                  ((x) << 0)
+#define			GDS_PARTITION_BASE		2
+#define			CE_PARTITION_BASE		3
+#define	PACKET3_CLEAR_STATE				0x12
+#define	PACKET3_INDEX_BUFFER_SIZE			0x13
+#define	PACKET3_DISPATCH_DIRECT				0x15
+#define	PACKET3_DISPATCH_INDIRECT			0x16
+#define	PACKET3_ALLOC_GDS				0x1B
+#define	PACKET3_WRITE_GDS_RAM				0x1C
+#define	PACKET3_ATOMIC_GDS				0x1D
+#define	PACKET3_ATOMIC					0x1E
+#define	PACKET3_OCCLUSION_QUERY				0x1F
+#define	PACKET3_SET_PREDICATION				0x20
+#define	PACKET3_REG_RMW					0x21
+#define	PACKET3_COND_EXEC				0x22
+#define	PACKET3_PRED_EXEC				0x23
+#define	PACKET3_DRAW_INDIRECT				0x24
+#define	PACKET3_DRAW_INDEX_INDIRECT			0x25
+#define	PACKET3_INDEX_BASE				0x26
+#define	PACKET3_DRAW_INDEX_2				0x27
+#define	PACKET3_CONTEXT_CONTROL				0x28
+#define	PACKET3_INDEX_TYPE				0x2A
+#define	PACKET3_DRAW_INDIRECT_MULTI			0x2C
+#define	PACKET3_DRAW_INDEX_AUTO				0x2D
+#define	PACKET3_DRAW_INDEX_IMMD				0x2E
+#define	PACKET3_NUM_INSTANCES				0x2F
+#define	PACKET3_DRAW_INDEX_MULTI_AUTO			0x30
+#define	PACKET3_INDIRECT_BUFFER_CONST			0x31
+#define	PACKET3_INDIRECT_BUFFER				0x3F
+#define	PACKET3_STRMOUT_BUFFER_UPDATE			0x34
+#define	PACKET3_DRAW_INDEX_OFFSET_2			0x35
+#define	PACKET3_DRAW_INDEX_MULTI_ELEMENT		0x36
+#define	PACKET3_WRITE_DATA				0x37
+#define		WRITE_DATA_DST_SEL(x)                   ((x) << 8)
+                /* 0 - register
+		 * 1 - memory (sync - via GRBM)
+		 * 2 - tc/l2
+		 * 3 - gds
+		 * 4 - reserved
+		 * 5 - memory (async - direct)
+		 */
+#define		WR_ONE_ADDR                             (1 << 16)
+#define		WR_CONFIRM                              (1 << 20)
+#define		WRITE_DATA_ENGINE_SEL(x)                ((x) << 30)
+                /* 0 - me
+		 * 1 - pfp
+		 * 2 - ce
+		 */
+#define	PACKET3_DRAW_INDEX_INDIRECT_MULTI		0x38
+#define	PACKET3_MEM_SEMAPHORE				0x39
+#define	PACKET3_MPEG_INDEX				0x3A
+#define	PACKET3_COPY_DW					0x3B
+#define	PACKET3_WAIT_REG_MEM				0x3C
+#define		WAIT_REG_MEM_FUNCTION(x)                ((x) << 0)
+                /* 0 - always
+		 * 1 - <
+		 * 2 - <=
+		 * 3 - ==
+		 * 4 - !=
+		 * 5 - >=
+		 * 6 - >
+		 */
+#define		WAIT_REG_MEM_MEM_SPACE(x)               ((x) << 4)
+                /* 0 - reg
+		 * 1 - mem
+		 */
+#define		WAIT_REG_MEM_ENGINE(x)                  ((x) << 8)
+                /* 0 - me
+		 * 1 - pfp
+		 */
+#define	PACKET3_MEM_WRITE				0x3D
+#define	PACKET3_COPY_DATA				0x40
+#define	PACKET3_CP_DMA					0x41
+/* 1. header
+ * 2. SRC_ADDR_LO or DATA [31:0]
+ * 3. CP_SYNC [31] | SRC_SEL [30:29] | ENGINE [27] | DST_SEL [21:20] |
+ *    SRC_ADDR_HI [7:0]
+ * 4. DST_ADDR_LO [31:0]
+ * 5. DST_ADDR_HI [7:0]
+ * 6. COMMAND [30:21] | BYTE_COUNT [20:0]
+ */
+#              define PACKET3_CP_DMA_DST_SEL(x)    ((x) << 20)
+                /* 0 - DST_ADDR
+		 * 1 - GDS
+		 */
+#              define PACKET3_CP_DMA_ENGINE(x)     ((x) << 27)
+                /* 0 - ME
+		 * 1 - PFP
+		 */
+#              define PACKET3_CP_DMA_SRC_SEL(x)    ((x) << 29)
+                /* 0 - SRC_ADDR
+		 * 1 - GDS
+		 * 2 - DATA
+		 */
+#              define PACKET3_CP_DMA_CP_SYNC       (1 << 31)
+/* COMMAND */
+#              define PACKET3_CP_DMA_DIS_WC        (1 << 21)
+#              define PACKET3_CP_DMA_CMD_SRC_SWAP(x) ((x) << 22)
+                /* 0 - none
+		 * 1 - 8 in 16
+		 * 2 - 8 in 32
+		 * 3 - 8 in 64
+		 */
+#              define PACKET3_CP_DMA_CMD_DST_SWAP(x) ((x) << 24)
+                /* 0 - none
+		 * 1 - 8 in 16
+		 * 2 - 8 in 32
+		 * 3 - 8 in 64
+		 */
+#              define PACKET3_CP_DMA_CMD_SAS       (1 << 26)
+                /* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_CP_DMA_CMD_DAS       (1 << 27)
+                /* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_CP_DMA_CMD_SAIC      (1 << 28)
+#              define PACKET3_CP_DMA_CMD_DAIC      (1 << 29)
+#              define PACKET3_CP_DMA_CMD_RAW_WAIT  (1 << 30)
+#define	PACKET3_PFP_SYNC_ME				0x42
+#define	PACKET3_SURFACE_SYNC				0x43
+#              define PACKET3_DEST_BASE_0_ENA      (1 << 0)
+#              define PACKET3_DEST_BASE_1_ENA      (1 << 1)
+#              define PACKET3_CB0_DEST_BASE_ENA    (1 << 6)
+#              define PACKET3_CB1_DEST_BASE_ENA    (1 << 7)
+#              define PACKET3_CB2_DEST_BASE_ENA    (1 << 8)
+#              define PACKET3_CB3_DEST_BASE_ENA    (1 << 9)
+#              define PACKET3_CB4_DEST_BASE_ENA    (1 << 10)
+#              define PACKET3_CB5_DEST_BASE_ENA    (1 << 11)
+#              define PACKET3_CB6_DEST_BASE_ENA    (1 << 12)
+#              define PACKET3_CB7_DEST_BASE_ENA    (1 << 13)
+#              define PACKET3_DB_DEST_BASE_ENA     (1 << 14)
+#              define PACKET3_DEST_BASE_2_ENA      (1 << 19)
+#              define PACKET3_DEST_BASE_3_ENA      (1 << 21)
+#              define PACKET3_TCL1_ACTION_ENA      (1 << 22)
+#              define PACKET3_TC_ACTION_ENA        (1 << 23)
+#              define PACKET3_CB_ACTION_ENA        (1 << 25)
+#              define PACKET3_DB_ACTION_ENA        (1 << 26)
+#              define PACKET3_SH_KCACHE_ACTION_ENA (1 << 27)
+#              define PACKET3_SH_ICACHE_ACTION_ENA (1 << 29)
+#define	PACKET3_ME_INITIALIZE				0x44
+#define		PACKET3_ME_INITIALIZE_DEVICE_ID(x) ((x) << 16)
+#define	PACKET3_COND_WRITE				0x45
+#define	PACKET3_EVENT_WRITE				0x46
+#define		EVENT_TYPE(x)                           ((x) << 0)
+#define		EVENT_INDEX(x)                          ((x) << 8)
+                /* 0 - any non-TS event
+		 * 1 - ZPASS_DONE
+		 * 2 - SAMPLE_PIPELINESTAT
+		 * 3 - SAMPLE_STREAMOUTSTAT*
+		 * 4 - *S_PARTIAL_FLUSH
+		 * 5 - EOP events
+		 * 6 - EOS events
+		 * 7 - CACHE_FLUSH, CACHE_FLUSH_AND_INV_EVENT
+		 */
+#define		INV_L2                                  (1 << 20)
+                /* INV TC L2 cache when EVENT_INDEX = 7 */
+#define	PACKET3_EVENT_WRITE_EOP				0x47
+#define		DATA_SEL(x)                             ((x) << 29)
+                /* 0 - discard
+		 * 1 - send low 32bit data
+		 * 2 - send 64bit data
+		 * 3 - send 64bit counter value
+		 */
+#define		INT_SEL(x)                              ((x) << 24)
+                /* 0 - none
+		 * 1 - interrupt only (DATA_SEL = 0)
+		 * 2 - interrupt when data write is confirmed
+		 */
+#define	PACKET3_EVENT_WRITE_EOS				0x48
+#define	PACKET3_PREAMBLE_CNTL				0x4A
+#              define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE     (2 << 28)
+#              define PACKET3_PREAMBLE_END_CLEAR_STATE       (3 << 28)
+#define	PACKET3_ONE_REG_WRITE				0x57
+#define	PACKET3_LOAD_CONFIG_REG				0x5F
+#define	PACKET3_LOAD_CONTEXT_REG			0x60
+#define	PACKET3_LOAD_SH_REG				0x61
+#define	PACKET3_SET_CONFIG_REG				0x68
+#define		PACKET3_SET_CONFIG_REG_START			0x00002000
+#define		PACKET3_SET_CONFIG_REG_END			0x00002c00
+#define	PACKET3_SET_CONTEXT_REG				0x69
+#define		PACKET3_SET_CONTEXT_REG_START			0x000a000
+#define		PACKET3_SET_CONTEXT_REG_END			0x000a400
+#define	PACKET3_SET_CONTEXT_REG_INDIRECT		0x73
+#define	PACKET3_SET_RESOURCE_INDIRECT			0x74
+#define	PACKET3_SET_SH_REG				0x76
+#define		PACKET3_SET_SH_REG_START			0x00002c00
+#define		PACKET3_SET_SH_REG_END				0x00003000
+#define	PACKET3_SET_SH_REG_OFFSET			0x77
+#define	PACKET3_ME_WRITE				0x7A
+#define	PACKET3_SCRATCH_RAM_WRITE			0x7D
+#define	PACKET3_SCRATCH_RAM_READ			0x7E
+#define	PACKET3_CE_WRITE				0x7F
+#define	PACKET3_LOAD_CONST_RAM				0x80
+#define	PACKET3_WRITE_CONST_RAM				0x81
+#define	PACKET3_WRITE_CONST_RAM_OFFSET			0x82
+#define	PACKET3_DUMP_CONST_RAM				0x83
+#define	PACKET3_INCREMENT_CE_COUNTER			0x84
+#define	PACKET3_INCREMENT_DE_COUNTER			0x85
+#define	PACKET3_WAIT_ON_CE_COUNTER			0x86
+#define	PACKET3_WAIT_ON_DE_COUNTER			0x87
+#define	PACKET3_WAIT_ON_DE_COUNTER_DIFF			0x88
+#define	PACKET3_SET_CE_DE_COUNTERS			0x89
+#define	PACKET3_WAIT_ON_AVAIL_BUFFER			0x8A
+#define	PACKET3_SWITCH_BUFFER				0x8B
+
+/* ASYNC DMA - first instance at 0xd000, second at 0xd800 */
+#define DMA0_REGISTER_OFFSET                              0x0 /* not a register */
+#define DMA1_REGISTER_OFFSET                              0x200 /* not a register */
+
+#define DMA_RB_CNTL                                       0x3400
+#       define DMA_RB_ENABLE                              (1 << 0)
+#       define DMA_RB_SIZE(x)                             ((x) << 1) /* log2 */
+#       define DMA_RB_SWAP_ENABLE                         (1 << 9) /* 8IN32 */
+#       define DMA_RPTR_WRITEBACK_ENABLE                  (1 << 12)
+#       define DMA_RPTR_WRITEBACK_SWAP_ENABLE             (1 << 13)  /* 8IN32 */
+#       define DMA_RPTR_WRITEBACK_TIMER(x)                ((x) << 16) /* log2 */
+#define DMA_RB_BASE                                       0x3401
+#define DMA_RB_RPTR                                       0x3402
+#define DMA_RB_WPTR                                       0x3403
+
+#define DMA_RB_RPTR_ADDR_HI                               0x3407
+#define DMA_RB_RPTR_ADDR_LO                               0x3408
+
+#define DMA_IB_CNTL                                       0x3409
+#       define DMA_IB_ENABLE                              (1 << 0)
+#       define DMA_IB_SWAP_ENABLE                         (1 << 4)
+#       define CMD_VMID_FORCE                             (1 << 31)
+#define DMA_IB_RPTR                                       0x340a
+#define DMA_CNTL                                          0x340b
+#       define TRAP_ENABLE                                (1 << 0)
+#       define SEM_INCOMPLETE_INT_ENABLE                  (1 << 1)
+#       define SEM_WAIT_INT_ENABLE                        (1 << 2)
+#       define DATA_SWAP_ENABLE                           (1 << 3)
+#       define FENCE_SWAP_ENABLE                          (1 << 4)
+#       define CTXEMPTY_INT_ENABLE                        (1 << 28)
+#define DMA_STATUS_REG                                    0x340d
+#       define DMA_IDLE                                   (1 << 0)
+#define DMA_TILING_CONFIG  				  0x342e
+
+#define	DMA_POWER_CNTL					0x342f
+#       define MEM_POWER_OVERRIDE                       (1 << 8)
+#define	DMA_CLK_CTRL					0x3430
+
+#define	DMA_PG						0x3435
+#	define PG_CNTL_ENABLE				(1 << 0)
+#define	DMA_PGFSM_CONFIG				0x3436
+#define	DMA_PGFSM_WRITE					0x3437
+
+#define DMA_PACKET(cmd, b, t, s, n)	((((cmd) & 0xF) << 28) |	\
+					 (((b) & 0x1) << 26) |		\
+					 (((t) & 0x1) << 23) |		\
+					 (((s) & 0x1) << 22) |		\
+					 (((n) & 0xFFFFF) << 0))
+
+#define DMA_IB_PACKET(cmd, vmid, n)	((((cmd) & 0xF) << 28) |	\
+					 (((vmid) & 0xF) << 20) |	\
+					 (((n) & 0xFFFFF) << 0))
+
+#define DMA_PTE_PDE_PACKET(n)		((2 << 28) |			\
+					 (1 << 26) |			\
+					 (1 << 21) |			\
+					 (((n) & 0xFFFFF) << 0))
+
+/* async DMA Packet types */
+#define	DMA_PACKET_WRITE				  0x2
+#define	DMA_PACKET_COPY					  0x3
+#define	DMA_PACKET_INDIRECT_BUFFER			  0x4
+#define	DMA_PACKET_SEMAPHORE				  0x5
+#define	DMA_PACKET_FENCE				  0x6
+#define	DMA_PACKET_TRAP					  0x7
+#define	DMA_PACKET_SRBM_WRITE				  0x9
+#define	DMA_PACKET_CONSTANT_FILL			  0xd
+#define	DMA_PACKET_POLL_REG_MEM				  0xe
+#define	DMA_PACKET_NOP					  0xf
+
+#define VCE_STATUS					0x20004
+#define VCE_VCPU_CNTL					0x20014
+#define		VCE_CLK_EN				(1 << 0)
+#define VCE_VCPU_CACHE_OFFSET0				0x20024
+#define VCE_VCPU_CACHE_SIZE0				0x20028
+#define VCE_VCPU_CACHE_OFFSET1				0x2002c
+#define VCE_VCPU_CACHE_SIZE1				0x20030
+#define VCE_VCPU_CACHE_OFFSET2				0x20034
+#define VCE_VCPU_CACHE_SIZE2				0x20038
+#define VCE_SOFT_RESET					0x20120
+#define 	VCE_ECPU_SOFT_RESET			(1 << 0)
+#define 	VCE_FME_SOFT_RESET			(1 << 2)
+#define VCE_RB_BASE_LO2					0x2016c
+#define VCE_RB_BASE_HI2					0x20170
+#define VCE_RB_SIZE2					0x20174
+#define VCE_RB_RPTR2					0x20178
+#define VCE_RB_WPTR2					0x2017c
+#define VCE_RB_BASE_LO					0x20180
+#define VCE_RB_BASE_HI					0x20184
+#define VCE_RB_SIZE					0x20188
+#define VCE_RB_RPTR					0x2018c
+#define VCE_RB_WPTR					0x20190
+#define VCE_CLOCK_GATING_A				0x202f8
+#define VCE_CLOCK_GATING_B				0x202fc
+#define VCE_UENC_CLOCK_GATING				0x205bc
+#define VCE_UENC_REG_CLOCK_GATING			0x205c0
+#define VCE_FW_REG_STATUS				0x20e10
+#	define VCE_FW_REG_STATUS_BUSY			(1 << 0)
+#	define VCE_FW_REG_STATUS_PASS			(1 << 3)
+#	define VCE_FW_REG_STATUS_DONE			(1 << 11)
+#define VCE_LMI_FW_START_KEYSEL				0x20e18
+#define VCE_LMI_FW_PERIODIC_CTRL			0x20e20
+#define VCE_LMI_CTRL2					0x20e74
+#define VCE_LMI_CTRL					0x20e98
+#define VCE_LMI_VM_CTRL					0x20ea0
+#define VCE_LMI_SWAP_CNTL				0x20eb4
+#define VCE_LMI_SWAP_CNTL1				0x20eb8
+#define VCE_LMI_CACHE_CTRL				0x20ef4
+
+#define VCE_CMD_NO_OP					0x00000000
+#define VCE_CMD_END					0x00000001
+#define VCE_CMD_IB					0x00000002
+#define VCE_CMD_FENCE					0x00000003
+#define VCE_CMD_TRAP					0x00000004
+#define VCE_CMD_IB_AUTO					0x00000005
+#define VCE_CMD_SEMAPHORE				0x00000006
+
+
+//#dce stupp
+/* display controller offsets used for crtc/cur/lut/grph/viewport/etc. */
+#define SI_CRTC0_REGISTER_OFFSET                0 //(0x6df0 - 0x6df0)/4
+#define SI_CRTC1_REGISTER_OFFSET                0x300 //(0x79f0 - 0x6df0)/4
+#define SI_CRTC2_REGISTER_OFFSET                0x2600 //(0x105f0 - 0x6df0)/4
+#define SI_CRTC3_REGISTER_OFFSET                0x2900 //(0x111f0 - 0x6df0)/4
+#define SI_CRTC4_REGISTER_OFFSET                0x2c00 //(0x11df0 - 0x6df0)/4
+#define SI_CRTC5_REGISTER_OFFSET                0x2f00 //(0x129f0 - 0x6df0)/4
+
+#define CURSOR_WIDTH 64
+#define CURSOR_HEIGHT 64
+#define AMDGPU_MM_INDEX		        0x0000
+#define AMDGPU_MM_DATA		        0x0001
+
+#define VERDE_NUM_CRTC 6
+#define	BLACKOUT_MODE_MASK			0x00000007
+#define	VGA_RENDER_CONTROL			0xC0
+#define R_000300_VGA_RENDER_CONTROL             0xC0
+#define C_000300_VGA_VSTATUS_CNTL               0xFFFCFFFF
+#define EVERGREEN_CRTC_STATUS                   0x1BA3
+#define EVERGREEN_CRTC_V_BLANK                  (1 << 0)
+#define EVERGREEN_CRTC_STATUS_POSITION          0x1BA4
+/* CRTC blocks at 0x6df0, 0x79f0, 0x105f0, 0x111f0, 0x11df0, 0x129f0 */
+#define EVERGREEN_CRTC_V_BLANK_START_END                0x1b8d
+#define EVERGREEN_CRTC_CONTROL                          0x1b9c
+#define EVERGREEN_CRTC_MASTER_EN                 (1 << 0)
+#define EVERGREEN_CRTC_DISP_READ_REQUEST_DISABLE (1 << 24)
+#define EVERGREEN_CRTC_BLANK_CONTROL                    0x1b9d
+#define EVERGREEN_CRTC_BLANK_DATA_EN             (1 << 8)
+#define EVERGREEN_CRTC_V_BLANK                   (1 << 0)
+#define EVERGREEN_CRTC_STATUS_HV_COUNT                  0x1ba8
+#define EVERGREEN_CRTC_UPDATE_LOCK                      0x1bb5
+#define EVERGREEN_MASTER_UPDATE_LOCK                    0x1bbd
+#define EVERGREEN_MASTER_UPDATE_MODE                    0x1bbe
+#define EVERGREEN_GRPH_UPDATE_LOCK               (1 << 16)
+#define EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH     0x1a07
+#define EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH   0x1a08
+#define EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS          0x1a04
+#define EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS        0x1a05
+#define EVERGREEN_GRPH_UPDATE                           0x1a11
+#define EVERGREEN_VGA_MEMORY_BASE_ADDRESS               0xc4
+#define EVERGREEN_VGA_MEMORY_BASE_ADDRESS_HIGH          0xc9
+#define EVERGREEN_GRPH_SURFACE_UPDATE_PENDING    (1 << 2)
+
+#define EVERGREEN_DATA_FORMAT                           0x1ac0
+#       define EVERGREEN_INTERLEAVE_EN                  (1 << 0)
+
+#define MC_SHARED_CHMAP__NOOFCHAN_MASK 0xf000
+#define MC_SHARED_CHMAP__NOOFCHAN__SHIFT 0xc
+
+#define R600_D1GRPH_ARRAY_MODE_LINEAR_GENERAL            (0 << 20)
+#define R600_D1GRPH_ARRAY_MODE_LINEAR_ALIGNED            (1 << 20)
+#define R600_D1GRPH_ARRAY_MODE_1D_TILED_THIN1            (2 << 20)
+#define R600_D1GRPH_ARRAY_MODE_2D_TILED_THIN1            (4 << 20)
+
+#define R700_D1GRPH_PRIMARY_SURFACE_ADDRESS_HIGH                0x1a45
+#define R700_D2GRPH_PRIMARY_SURFACE_ADDRESS_HIGH                0x1845
+
+#define R700_D2GRPH_SECONDARY_SURFACE_ADDRESS_HIGH              0x1847
+#define R700_D1GRPH_SECONDARY_SURFACE_ADDRESS_HIGH              0x1a47
+
+#define DISP_INTERRUPT_STATUS__LB_D1_VBLANK_INTERRUPT_MASK 0x8
+#define DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VBLANK_INTERRUPT_MASK 0x8
+#define DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VBLANK_INTERRUPT_MASK 0x8
+#define DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VBLANK_INTERRUPT_MASK 0x8
+#define DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VBLANK_INTERRUPT_MASK 0x8
+#define DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VBLANK_INTERRUPT_MASK 0x8
+
+#define DISP_INTERRUPT_STATUS__LB_D1_VLINE_INTERRUPT_MASK 0x4
+#define DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VLINE_INTERRUPT_MASK 0x4
+#define DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VLINE_INTERRUPT_MASK 0x4
+#define DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VLINE_INTERRUPT_MASK 0x4
+#define DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VLINE_INTERRUPT_MASK 0x4
+#define DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VLINE_INTERRUPT_MASK 0x4
+
+#define DISP_INTERRUPT_STATUS__DC_HPD1_INTERRUPT_MASK 0x20000
+#define DISP_INTERRUPT_STATUS_CONTINUE__DC_HPD2_INTERRUPT_MASK 0x20000
+#define DISP_INTERRUPT_STATUS_CONTINUE2__DC_HPD3_INTERRUPT_MASK 0x20000
+#define DISP_INTERRUPT_STATUS_CONTINUE3__DC_HPD4_INTERRUPT_MASK 0x20000
+#define DISP_INTERRUPT_STATUS_CONTINUE4__DC_HPD5_INTERRUPT_MASK 0x20000
+#define DISP_INTERRUPT_STATUS_CONTINUE5__DC_HPD6_INTERRUPT_MASK 0x20000
+
+#define GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_OCCURRED_MASK 0x1
+#define GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_CLEAR_MASK 0x100
+
+#define DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK 0x1
+
+#define R600_D1GRPH_SWAP_CONTROL                               0x1843
+#define R600_D1GRPH_SWAP_ENDIAN_NONE                    (0 << 0)
+#define R600_D1GRPH_SWAP_ENDIAN_16BIT                   (1 << 0)
+#define R600_D1GRPH_SWAP_ENDIAN_32BIT                   (2 << 0)
+#define R600_D1GRPH_SWAP_ENDIAN_64BIT                   (3 << 0)
+
+#define AVIVO_D1VGA_CONTROL					0x00cc
+#       define AVIVO_DVGA_CONTROL_MODE_ENABLE            (1 << 0)
+#       define AVIVO_DVGA_CONTROL_TIMING_SELECT          (1 << 8)
+#       define AVIVO_DVGA_CONTROL_SYNC_POLARITY_SELECT   (1 << 9)
+#       define AVIVO_DVGA_CONTROL_OVERSCAN_TIMING_SELECT (1 << 10)
+#       define AVIVO_DVGA_CONTROL_OVERSCAN_COLOR_EN      (1 << 16)
+#       define AVIVO_DVGA_CONTROL_ROTATE                 (1 << 24)
+#define AVIVO_D2VGA_CONTROL					0x00ce
+
+#define R600_BUS_CNTL                                           0x1508
+#       define R600_BIOS_ROM_DIS                                (1 << 1)
+
+#define R600_ROM_CNTL                              0x580
+#       define R600_SCK_OVERWRITE                  (1 << 1)
+#       define R600_SCK_PRESCALE_CRYSTAL_CLK_SHIFT 28
+#       define R600_SCK_PRESCALE_CRYSTAL_CLK_MASK  (0xf << 28)
+
+#define GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK 0x1
+
+#define FMT_BIT_DEPTH_CONTROL                0x1bf2
+#define FMT_TRUNCATE_EN               (1 << 0)
+#define FMT_TRUNCATE_DEPTH            (1 << 4)
+#define FMT_SPATIAL_DITHER_EN         (1 << 8)
+#define FMT_SPATIAL_DITHER_MODE(x)    ((x) << 9)
+#define FMT_SPATIAL_DITHER_DEPTH      (1 << 12)
+#define FMT_FRAME_RANDOM_ENABLE       (1 << 13)
+#define FMT_RGB_RANDOM_ENABLE         (1 << 14)
+#define FMT_HIGHPASS_RANDOM_ENABLE    (1 << 15)
+#define FMT_TEMPORAL_DITHER_EN        (1 << 16)
+#define FMT_TEMPORAL_DITHER_DEPTH     (1 << 20)
+#define FMT_TEMPORAL_DITHER_OFFSET(x) ((x) << 21)
+#define FMT_TEMPORAL_LEVEL            (1 << 24)
+#define FMT_TEMPORAL_DITHER_RESET     (1 << 25)
+#define FMT_25FRC_SEL(x)              ((x) << 26)
+#define FMT_50FRC_SEL(x)              ((x) << 28)
+#define FMT_75FRC_SEL(x)              ((x) << 30)
+
+#define EVERGREEN_DC_LUT_CONTROL                        0x1a80
+#define EVERGREEN_DC_LUT_BLACK_OFFSET_BLUE              0x1a81
+#define EVERGREEN_DC_LUT_BLACK_OFFSET_GREEN             0x1a82
+#define EVERGREEN_DC_LUT_BLACK_OFFSET_RED               0x1a83
+#define EVERGREEN_DC_LUT_WHITE_OFFSET_BLUE              0x1a84
+#define EVERGREEN_DC_LUT_WHITE_OFFSET_GREEN             0x1a85
+#define EVERGREEN_DC_LUT_WHITE_OFFSET_RED               0x1a86
+#define EVERGREEN_DC_LUT_30_COLOR                       0x1a7c
+#define EVERGREEN_DC_LUT_RW_INDEX                       0x1a79
+#define EVERGREEN_DC_LUT_WRITE_EN_MASK                  0x1a7e
+#define EVERGREEN_DC_LUT_RW_MODE                        0x1a78
+
+#define EVERGREEN_GRPH_ENABLE                           0x1a00
+#define EVERGREEN_GRPH_CONTROL                          0x1a01
+#define EVERGREEN_GRPH_DEPTH(x)                  (((x) & 0x3) << 0)
+#define EVERGREEN_GRPH_DEPTH_8BPP                0
+#define EVERGREEN_GRPH_DEPTH_16BPP               1
+#define EVERGREEN_GRPH_DEPTH_32BPP               2
+#define EVERGREEN_GRPH_NUM_BANKS(x)              (((x) & 0x3) << 2)
+#define EVERGREEN_ADDR_SURF_2_BANK               0
+#define EVERGREEN_ADDR_SURF_4_BANK               1
+#define EVERGREEN_ADDR_SURF_8_BANK               2
+#define EVERGREEN_ADDR_SURF_16_BANK              3
+#define EVERGREEN_GRPH_Z(x)                      (((x) & 0x3) << 4)
+#define EVERGREEN_GRPH_BANK_WIDTH(x)             (((x) & 0x3) << 6)
+#define EVERGREEN_ADDR_SURF_BANK_WIDTH_1         0
+#define EVERGREEN_ADDR_SURF_BANK_WIDTH_2         1
+#define EVERGREEN_ADDR_SURF_BANK_WIDTH_4         2
+#define EVERGREEN_ADDR_SURF_BANK_WIDTH_8         3
+#define EVERGREEN_GRPH_FORMAT(x)                 (((x) & 0x7) << 8)
+
+#define EVERGREEN_GRPH_FORMAT_INDEXED            0
+#define EVERGREEN_GRPH_FORMAT_ARGB1555           0
+#define EVERGREEN_GRPH_FORMAT_ARGB565            1
+#define EVERGREEN_GRPH_FORMAT_ARGB4444           2
+#define EVERGREEN_GRPH_FORMAT_AI88               3
+#define EVERGREEN_GRPH_FORMAT_MONO16             4
+#define EVERGREEN_GRPH_FORMAT_BGRA5551           5
+
+/* 32 BPP */
+#define EVERGREEN_GRPH_FORMAT_ARGB8888           0
+#define EVERGREEN_GRPH_FORMAT_ARGB2101010        1
+#define EVERGREEN_GRPH_FORMAT_32BPP_DIG          2
+#define EVERGREEN_GRPH_FORMAT_8B_ARGB2101010     3
+#define EVERGREEN_GRPH_FORMAT_BGRA1010102        4
+#define EVERGREEN_GRPH_FORMAT_8B_BGRA1010102     5
+#define EVERGREEN_GRPH_FORMAT_RGB111110          6
+#define EVERGREEN_GRPH_FORMAT_BGR101111          7
+#define EVERGREEN_GRPH_BANK_HEIGHT(x)            (((x) & 0x3) << 11)
+#define EVERGREEN_ADDR_SURF_BANK_HEIGHT_1        0
+#define EVERGREEN_ADDR_SURF_BANK_HEIGHT_2        1
+#define EVERGREEN_ADDR_SURF_BANK_HEIGHT_4        2
+#define EVERGREEN_ADDR_SURF_BANK_HEIGHT_8        3
+#define EVERGREEN_GRPH_TILE_SPLIT(x)             (((x) & 0x7) << 13)
+#define EVERGREEN_ADDR_SURF_TILE_SPLIT_64B       0
+#define EVERGREEN_ADDR_SURF_TILE_SPLIT_128B      1
+#define EVERGREEN_ADDR_SURF_TILE_SPLIT_256B      2
+#define EVERGREEN_ADDR_SURF_TILE_SPLIT_512B      3
+#define EVERGREEN_ADDR_SURF_TILE_SPLIT_1KB       4
+#define EVERGREEN_ADDR_SURF_TILE_SPLIT_2KB       5
+#define EVERGREEN_ADDR_SURF_TILE_SPLIT_4KB       6
+#define EVERGREEN_GRPH_MACRO_TILE_ASPECT(x)      (((x) & 0x3) << 18)
+#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_1  0
+#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_2  1
+#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_4  2
+#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_8  3
+#define EVERGREEN_GRPH_ARRAY_MODE(x)             (((x) & 0x7) << 20)
+#define EVERGREEN_GRPH_ARRAY_LINEAR_GENERAL      0
+#define EVERGREEN_GRPH_ARRAY_LINEAR_ALIGNED      1
+#define EVERGREEN_GRPH_ARRAY_1D_TILED_THIN1      2
+#define EVERGREEN_GRPH_ARRAY_2D_TILED_THIN1      4
+#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_1  0
+#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_2  1
+#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_4  2
+#define EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_8  3
+
+#define EVERGREEN_GRPH_SWAP_CONTROL                     0x1a03
+#define EVERGREEN_GRPH_ENDIAN_SWAP(x)            (((x) & 0x3) << 0)
+#       define EVERGREEN_GRPH_ENDIAN_NONE               0
+#       define EVERGREEN_GRPH_ENDIAN_8IN16              1
+#       define EVERGREEN_GRPH_ENDIAN_8IN32              2
+#       define EVERGREEN_GRPH_ENDIAN_8IN64              3
+
+#define EVERGREEN_D3VGA_CONTROL                         0xf8
+#define EVERGREEN_D4VGA_CONTROL                         0xf9
+#define EVERGREEN_D5VGA_CONTROL                         0xfa
+#define EVERGREEN_D6VGA_CONTROL                         0xfb
+
+#define EVERGREEN_GRPH_SURFACE_ADDRESS_MASK      0xffffff00
+
+#define EVERGREEN_GRPH_LUT_10BIT_BYPASS_CONTROL         0x1a02
+#define EVERGREEN_LUT_10BIT_BYPASS_EN            (1 << 8)
+
+#define EVERGREEN_GRPH_PITCH                            0x1a06
+#define EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH     0x1a07
+#define EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH   0x1a08
+#define EVERGREEN_GRPH_SURFACE_OFFSET_X                 0x1a09
+#define EVERGREEN_GRPH_SURFACE_OFFSET_Y                 0x1a0a
+#define EVERGREEN_GRPH_X_START                          0x1a0b
+#define EVERGREEN_GRPH_Y_START                          0x1a0c
+#define EVERGREEN_GRPH_X_END                            0x1a0d
+#define EVERGREEN_GRPH_Y_END                            0x1a0e
+#define EVERGREEN_GRPH_UPDATE                           0x1a11
+#define EVERGREEN_GRPH_SURFACE_UPDATE_PENDING    (1 << 2)
+#define EVERGREEN_GRPH_UPDATE_LOCK               (1 << 16)
+#define EVERGREEN_GRPH_FLIP_CONTROL                     0x1a12
+#define EVERGREEN_GRPH_SURFACE_UPDATE_H_RETRACE_EN (1 << 0)
+
+#define EVERGREEN_VIEWPORT_START                        0x1b5c
+#define EVERGREEN_VIEWPORT_SIZE                         0x1b5d
+#define EVERGREEN_DESKTOP_HEIGHT                        0x1ac1
+
+/* CUR blocks at 0x6998, 0x7598, 0x10198, 0x10d98, 0x11998, 0x12598 */
+#define EVERGREEN_CUR_CONTROL                           0x1a66
+#       define EVERGREEN_CURSOR_EN                      (1 << 0)
+#       define EVERGREEN_CURSOR_MODE(x)                 (((x) & 0x3) << 8)
+#       define EVERGREEN_CURSOR_MONO                    0
+#       define EVERGREEN_CURSOR_24_1                    1
+#       define EVERGREEN_CURSOR_24_8_PRE_MULT           2
+#       define EVERGREEN_CURSOR_24_8_UNPRE_MULT         3
+#       define EVERGREEN_CURSOR_2X_MAGNIFY              (1 << 16)
+#       define EVERGREEN_CURSOR_FORCE_MC_ON             (1 << 20)
+#       define EVERGREEN_CURSOR_URGENT_CONTROL(x)       (((x) & 0x7) << 24)
+#       define EVERGREEN_CURSOR_URGENT_ALWAYS           0
+#       define EVERGREEN_CURSOR_URGENT_1_8              1
+#       define EVERGREEN_CURSOR_URGENT_1_4              2
+#       define EVERGREEN_CURSOR_URGENT_3_8              3
+#       define EVERGREEN_CURSOR_URGENT_1_2              4
+#define EVERGREEN_CUR_SURFACE_ADDRESS                   0x1a67
+#       define EVERGREEN_CUR_SURFACE_ADDRESS_MASK       0xfffff000
+#define EVERGREEN_CUR_SIZE                              0x1a68
+#define EVERGREEN_CUR_SURFACE_ADDRESS_HIGH              0x1a69
+#define EVERGREEN_CUR_POSITION                          0x1a6a
+#define EVERGREEN_CUR_HOT_SPOT                          0x1a6b
+#define EVERGREEN_CUR_COLOR1                            0x1a6c
+#define EVERGREEN_CUR_COLOR2                            0x1a6d
+#define EVERGREEN_CUR_UPDATE                            0x1a6e
+#       define EVERGREEN_CURSOR_UPDATE_PENDING          (1 << 0)
+#       define EVERGREEN_CURSOR_UPDATE_TAKEN            (1 << 1)
+#       define EVERGREEN_CURSOR_UPDATE_LOCK             (1 << 16)
+#       define EVERGREEN_CURSOR_DISABLE_MULTIPLE_UPDATE (1 << 24)
+
+
+#define NI_INPUT_CSC_CONTROL                           0x1a35
+#       define NI_INPUT_CSC_GRPH_MODE(x)               (((x) & 0x3) << 0)
+#       define NI_INPUT_CSC_BYPASS                     0
+#       define NI_INPUT_CSC_PROG_COEFF                 1
+#       define NI_INPUT_CSC_PROG_SHARED_MATRIXA        2
+#       define NI_INPUT_CSC_OVL_MODE(x)                (((x) & 0x3) << 4)
+
+#define NI_OUTPUT_CSC_CONTROL                          0x1a3c
+#       define NI_OUTPUT_CSC_GRPH_MODE(x)              (((x) & 0x7) << 0)
+#       define NI_OUTPUT_CSC_BYPASS                    0
+#       define NI_OUTPUT_CSC_TV_RGB                    1
+#       define NI_OUTPUT_CSC_YCBCR_601                 2
+#       define NI_OUTPUT_CSC_YCBCR_709                 3
+#       define NI_OUTPUT_CSC_PROG_COEFF                4
+#       define NI_OUTPUT_CSC_PROG_SHARED_MATRIXB       5
+#       define NI_OUTPUT_CSC_OVL_MODE(x)               (((x) & 0x7) << 4)
+
+#define NI_DEGAMMA_CONTROL                             0x1a58
+#       define NI_GRPH_DEGAMMA_MODE(x)                 (((x) & 0x3) << 0)
+#       define NI_DEGAMMA_BYPASS                       0
+#       define NI_DEGAMMA_SRGB_24                      1
+#       define NI_DEGAMMA_XVYCC_222                    2
+#       define NI_OVL_DEGAMMA_MODE(x)                  (((x) & 0x3) << 4)
+#       define NI_ICON_DEGAMMA_MODE(x)                 (((x) & 0x3) << 8)
+#       define NI_CURSOR_DEGAMMA_MODE(x)               (((x) & 0x3) << 12)
+
+#define NI_GAMUT_REMAP_CONTROL                         0x1a59
+#       define NI_GRPH_GAMUT_REMAP_MODE(x)             (((x) & 0x3) << 0)
+#       define NI_GAMUT_REMAP_BYPASS                   0
+#       define NI_GAMUT_REMAP_PROG_COEFF               1
+#       define NI_GAMUT_REMAP_PROG_SHARED_MATRIXA      2
+#       define NI_GAMUT_REMAP_PROG_SHARED_MATRIXB      3
+#       define NI_OVL_GAMUT_REMAP_MODE(x)              (((x) & 0x3) << 4)
+
+#define NI_REGAMMA_CONTROL                             0x1aa0
+#       define NI_GRPH_REGAMMA_MODE(x)                 (((x) & 0x7) << 0)
+#       define NI_REGAMMA_BYPASS                       0
+#       define NI_REGAMMA_SRGB_24                      1
+#       define NI_REGAMMA_XVYCC_222                    2
+#       define NI_REGAMMA_PROG_A                       3
+#       define NI_REGAMMA_PROG_B                       4
+#       define NI_OVL_REGAMMA_MODE(x)                  (((x) & 0x7) << 4)
+
+
+#define NI_PRESCALE_GRPH_CONTROL                       0x1a2d
+#       define NI_GRPH_PRESCALE_BYPASS                 (1 << 4)
+
+#define NI_PRESCALE_OVL_CONTROL                        0x1a31
+#       define NI_OVL_PRESCALE_BYPASS                  (1 << 4)
+
+#define NI_INPUT_GAMMA_CONTROL                         0x1a10
+#       define NI_GRPH_INPUT_GAMMA_MODE(x)             (((x) & 0x3) << 0)
+#       define NI_INPUT_GAMMA_USE_LUT                  0
+#       define NI_INPUT_GAMMA_BYPASS                   1
+#       define NI_INPUT_GAMMA_SRGB_24                  2
+#       define NI_INPUT_GAMMA_XVYCC_222                3
+#       define NI_OVL_INPUT_GAMMA_MODE(x)              (((x) & 0x3) << 4)
+
+#define IH_RB_WPTR__RB_OVERFLOW_MASK	0x1
+#define IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK 0x80000000
+#define SRBM_STATUS__IH_BUSY_MASK	0x20000
+#define SRBM_SOFT_RESET__SOFT_RESET_IH_MASK	0x400
+
+#define	BLACKOUT_MODE_MASK			0x00000007
+#define	VGA_RENDER_CONTROL			0xC0
+#define R_000300_VGA_RENDER_CONTROL             0xC0
+#define C_000300_VGA_VSTATUS_CNTL               0xFFFCFFFF
+#define EVERGREEN_CRTC_STATUS                   0x1BA3
+#define EVERGREEN_CRTC_V_BLANK                  (1 << 0)
+#define EVERGREEN_CRTC_STATUS_POSITION          0x1BA4
+/* CRTC blocks at 0x6df0, 0x79f0, 0x105f0, 0x111f0, 0x11df0, 0x129f0 */
+#define EVERGREEN_CRTC_V_BLANK_START_END                0x1b8d
+#define EVERGREEN_CRTC_CONTROL                          0x1b9c
+#       define EVERGREEN_CRTC_MASTER_EN                 (1 << 0)
+#       define EVERGREEN_CRTC_DISP_READ_REQUEST_DISABLE (1 << 24)
+#define EVERGREEN_CRTC_BLANK_CONTROL                    0x1b9d
+#       define EVERGREEN_CRTC_BLANK_DATA_EN             (1 << 8)
+#       define EVERGREEN_CRTC_V_BLANK                   (1 << 0)
+#define EVERGREEN_CRTC_STATUS_HV_COUNT                  0x1ba8
+#define EVERGREEN_CRTC_UPDATE_LOCK                      0x1bb5
+#define EVERGREEN_MASTER_UPDATE_LOCK                    0x1bbd
+#define EVERGREEN_MASTER_UPDATE_MODE                    0x1bbe
+#define EVERGREEN_GRPH_UPDATE_LOCK               (1 << 16)
+#define EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH     0x1a07
+#define EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH   0x1a08
+#define EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS          0x1a04
+#define EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS        0x1a05
+#define EVERGREEN_GRPH_UPDATE                           0x1a11
+#define EVERGREEN_VGA_MEMORY_BASE_ADDRESS               0xc4
+#define EVERGREEN_VGA_MEMORY_BASE_ADDRESS_HIGH          0xc9
+#define EVERGREEN_GRPH_SURFACE_UPDATE_PENDING    (1 << 2)
+
+#define mmVM_CONTEXT1_CNTL__xxRANGE_PROTECTION_FAULT_ENABLE_DEFAULT_MASK 0x10
+#define mmVM_CONTEXT1_CNTL__xxRANGE_PROTECTION_FAULT_ENABLE_DEFAULT__SHIFT 0x4
+#define mmVM_CONTEXT1_CNTL__xxDUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT_MASK 0x80
+#define mmVM_CONTEXT1_CNTL__xxDUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT__SHIFT 0x7
+#define mmVM_CONTEXT1_CNTL__xxPDE0_PROTECTION_FAULT_ENABLE_DEFAULT_MASK 0x400
+#define mmVM_CONTEXT1_CNTL__xxPDE0_PROTECTION_FAULT_ENABLE_DEFAULT__SHIFT 0xa
+#define mmVM_CONTEXT1_CNTL__xxVALID_PROTECTION_FAULT_ENABLE_DEFAULT_MASK 0x2000
+#define mmVM_CONTEXT1_CNTL__xxVALID_PROTECTION_FAULT_ENABLE_DEFAULT__SHIFT 0xd
+#define mmVM_CONTEXT1_CNTL__xxREAD_PROTECTION_FAULT_ENABLE_DEFAULT_MASK 0x10000
+#define mmVM_CONTEXT1_CNTL__xxREAD_PROTECTION_FAULT_ENABLE_DEFAULT__SHIFT 0x10
+#define mmVM_CONTEXT1_CNTL__xxWRITE_PROTECTION_FAULT_ENABLE_DEFAULT_MASK 0x80000
+#define mmVM_CONTEXT1_CNTL__xxWRITE_PROTECTION_FAULT_ENABLE_DEFAULT__SHIFT 0x13
+
+#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxVMID_MASK 0x1e000000
+#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxVMID__SHIFT 0x19
+#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxPROTECTIONS_MASK 0xff
+#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxPROTECTIONS__SHIFT 0x0
+#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxMEMORY_CLIENT_ID_MASK 0xff000
+#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxMEMORY_CLIENT_ID__SHIFT 0xc
+#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxMEMORY_CLIENT_RW_MASK 0x1000000
+#define mmVM_CONTEXT1_PROTECTION_FAULT_STATUS__xxMEMORY_CLIENT_RW__SHIFT 0x18
+
+#define mmMC_SHARED_BLACKOUT_CNTL__xxBLACKOUT_MODE_MASK 0x7
+#define mmMC_SHARED_BLACKOUT_CNTL__xxBLACKOUT_MODE__SHIFT 0x0
+
+#define mmBIF_FB_EN__xxFB_READ_EN_MASK 0x1
+#define mmBIF_FB_EN__xxFB_READ_EN__SHIFT 0x0
+#define mmBIF_FB_EN__xxFB_WRITE_EN_MASK 0x2
+#define mmBIF_FB_EN__xxFB_WRITE_EN__SHIFT 0x1
+
+#define mmSRBM_SOFT_RESET__xxSOFT_RESET_VMC_MASK 0x20000
+#define mmSRBM_SOFT_RESET__xxSOFT_RESET_VMC__SHIFT 0x11
+#define mmSRBM_SOFT_RESET__xxSOFT_RESET_MC_MASK 0x800
+#define mmSRBM_SOFT_RESET__xxSOFT_RESET_MC__SHIFT 0xb
+
+#define VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK 0x8
+#define VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT__SHIFT 0x3
+#define VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK 0x40
+#define VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT__SHIFT 0x6
+#define VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK 0x200
+#define VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT__SHIFT 0x9
+#define VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK 0x1000
+#define VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT__SHIFT 0xc
+#define VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK 0x8000
+#define VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT__SHIFT 0xf
+#define VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK 0x40000
+#define VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT__SHIFT 0x12
+
+#define MC_SEQ_MISC0__MT__MASK	0xf0000000
+#define MC_SEQ_MISC0__MT__GDDR1  0x10000000
+#define MC_SEQ_MISC0__MT__DDR2   0x20000000
+#define MC_SEQ_MISC0__MT__GDDR3  0x30000000
+#define MC_SEQ_MISC0__MT__GDDR4  0x40000000
+#define MC_SEQ_MISC0__MT__GDDR5  0x50000000
+#define MC_SEQ_MISC0__MT__HBM    0x60000000
+#define MC_SEQ_MISC0__MT__DDR3   0xB0000000
+
+#define SRBM_STATUS__MCB_BUSY_MASK 0x200
+#define SRBM_STATUS__MCB_BUSY__SHIFT 0x9
+#define SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK 0x400
+#define SRBM_STATUS__MCB_NON_DISPLAY_BUSY__SHIFT 0xa
+#define SRBM_STATUS__MCC_BUSY_MASK 0x800
+#define SRBM_STATUS__MCC_BUSY__SHIFT 0xb
+#define SRBM_STATUS__MCD_BUSY_MASK 0x1000
+#define SRBM_STATUS__MCD_BUSY__SHIFT 0xc
+#define SRBM_STATUS__VMC_BUSY_MASK 0x100
+#define SRBM_STATUS__VMC_BUSY__SHIFT 0x8
+
+
+#define GRBM_STATUS__GUI_ACTIVE_MASK 0x80000000
+#define CP_INT_CNTL_RING__TIME_STAMP_INT_ENABLE_MASK 0x4000000
+#define CP_INT_CNTL_RING0__PRIV_REG_INT_ENABLE_MASK 0x800000
+#define CP_INT_CNTL_RING0__PRIV_INSTR_INT_ENABLE_MASK 0x400000
+#define PACKET3_SEM_WAIT_ON_SIGNAL    (0x1 << 12)
+#define PACKET3_SEM_SEL_SIGNAL	    (0x6 << 29)
+#define PACKET3_SEM_SEL_WAIT	    (0x7 << 29)
+
+#define CONFIG_CNTL	0x1509
+#define CC_DRM_ID_STRAPS	0X1559
+#define AMDGPU_PCIE_INDEX	0xc
+#define AMDGPU_PCIE_DATA	0xd
+
+#define DMA_SEM_INCOMPLETE_TIMER_CNTL                     0x3411
+#define DMA_SEM_WAIT_FAIL_TIMER_CNTL                      0x3412
+#define DMA_MODE                                          0x342f
+#define DMA_RB_RPTR_ADDR_HI                               0x3407
+#define DMA_RB_RPTR_ADDR_LO                               0x3408
+#define DMA_BUSY_MASK 0x20
+#define DMA1_BUSY_MASK 0X40
+#define SDMA_MAX_INSTANCE 2
+
+#define PCIE_BUS_CLK    10000
+#define TCLK            (PCIE_BUS_CLK / 10)
+#define CC_DRM_ID_STRAPS__ATI_REV_ID_MASK		0xf0000000
+#define CC_DRM_ID_STRAPS__ATI_REV_ID__SHIFT 0x1c
+#define	PCIE_PORT_INDEX					0xe
+#define	PCIE_PORT_DATA					0xf
+#define EVERGREEN_PIF_PHY0_INDEX                        0x8
+#define EVERGREEN_PIF_PHY0_DATA                         0xc
+#define EVERGREEN_PIF_PHY1_INDEX                        0x10
+#define EVERGREEN_PIF_PHY1_DATA				0x14
+
+#define	MC_VM_FB_OFFSET					0x81a
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/sislands_smc.h b/sys/dev/pci/drm/amd/amdgpu/sislands_smc.h
new file mode 100644
index 00000000000..d2930eceaf3
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/sislands_smc.h
@@ -0,0 +1,423 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef PP_SISLANDS_SMC_H
+#define PP_SISLANDS_SMC_H
+
+#include "ppsmc.h"
+
+#pragma pack(push, 1)
+
+#define SISLANDS_MAX_SMC_PERFORMANCE_LEVELS_PER_SWSTATE 16
+
+struct PP_SIslands_Dpm2PerfLevel
+{
+    uint8_t MaxPS;
+    uint8_t TgtAct;
+    uint8_t MaxPS_StepInc;
+    uint8_t MaxPS_StepDec;
+    uint8_t PSSamplingTime;
+    uint8_t NearTDPDec;
+    uint8_t AboveSafeInc;
+    uint8_t BelowSafeInc;
+    uint8_t PSDeltaLimit;
+    uint8_t PSDeltaWin;
+    uint16_t PwrEfficiencyRatio;
+    uint8_t Reserved[4];
+};
+
+typedef struct PP_SIslands_Dpm2PerfLevel PP_SIslands_Dpm2PerfLevel;
+
+struct PP_SIslands_DPM2Status
+{
+    uint32_t    dpm2Flags;
+    uint8_t     CurrPSkip;
+    uint8_t     CurrPSkipPowerShift;
+    uint8_t     CurrPSkipTDP;
+    uint8_t     CurrPSkipOCP;
+    uint8_t     MaxSPLLIndex;
+    uint8_t     MinSPLLIndex;
+    uint8_t     CurrSPLLIndex;
+    uint8_t     InfSweepMode;
+    uint8_t     InfSweepDir;
+    uint8_t     TDPexceeded;
+    uint8_t     reserved;
+    uint8_t     SwitchDownThreshold;
+    uint32_t    SwitchDownCounter;
+    uint32_t    SysScalingFactor;
+};
+
+typedef struct PP_SIslands_DPM2Status PP_SIslands_DPM2Status;
+
+struct PP_SIslands_DPM2Parameters
+{
+    uint32_t    TDPLimit;
+    uint32_t    NearTDPLimit;
+    uint32_t    SafePowerLimit;
+    uint32_t    PowerBoostLimit;
+    uint32_t    MinLimitDelta;
+};
+typedef struct PP_SIslands_DPM2Parameters PP_SIslands_DPM2Parameters;
+
+struct PP_SIslands_PAPMStatus
+{
+    uint32_t    EstimatedDGPU_T;
+    uint32_t    EstimatedDGPU_P;
+    uint32_t    EstimatedAPU_T;
+    uint32_t    EstimatedAPU_P;
+    uint8_t     dGPU_T_Limit_Exceeded;
+    uint8_t     reserved[3];
+};
+typedef struct PP_SIslands_PAPMStatus PP_SIslands_PAPMStatus;
+
+struct PP_SIslands_PAPMParameters
+{
+    uint32_t    NearTDPLimitTherm;
+    uint32_t    NearTDPLimitPAPM;
+    uint32_t    PlatformPowerLimit;
+    uint32_t    dGPU_T_Limit;
+    uint32_t    dGPU_T_Warning;
+    uint32_t    dGPU_T_Hysteresis;
+};
+typedef struct PP_SIslands_PAPMParameters PP_SIslands_PAPMParameters;
+
+struct SISLANDS_SMC_SCLK_VALUE
+{
+    uint32_t    vCG_SPLL_FUNC_CNTL;
+    uint32_t    vCG_SPLL_FUNC_CNTL_2;
+    uint32_t    vCG_SPLL_FUNC_CNTL_3;
+    uint32_t    vCG_SPLL_FUNC_CNTL_4;
+    uint32_t    vCG_SPLL_SPREAD_SPECTRUM;
+    uint32_t    vCG_SPLL_SPREAD_SPECTRUM_2;
+    uint32_t    sclk_value;
+};
+
+typedef struct SISLANDS_SMC_SCLK_VALUE SISLANDS_SMC_SCLK_VALUE;
+
+struct SISLANDS_SMC_MCLK_VALUE
+{
+    uint32_t    vMPLL_FUNC_CNTL;
+    uint32_t    vMPLL_FUNC_CNTL_1;
+    uint32_t    vMPLL_FUNC_CNTL_2;
+    uint32_t    vMPLL_AD_FUNC_CNTL;
+    uint32_t    vMPLL_DQ_FUNC_CNTL;
+    uint32_t    vMCLK_PWRMGT_CNTL;
+    uint32_t    vDLL_CNTL;
+    uint32_t    vMPLL_SS;
+    uint32_t    vMPLL_SS2;
+    uint32_t    mclk_value;
+};
+
+typedef struct SISLANDS_SMC_MCLK_VALUE SISLANDS_SMC_MCLK_VALUE;
+
+struct SISLANDS_SMC_VOLTAGE_VALUE
+{
+    uint16_t    value;
+    uint8_t     index;
+    uint8_t     phase_settings;
+};
+
+typedef struct SISLANDS_SMC_VOLTAGE_VALUE SISLANDS_SMC_VOLTAGE_VALUE;
+
+struct SISLANDS_SMC_HW_PERFORMANCE_LEVEL
+{
+    uint8_t                     ACIndex;
+    uint8_t                     displayWatermark;
+    uint8_t                     gen2PCIE;
+    uint8_t                     UVDWatermark;
+    uint8_t                     VCEWatermark;
+    uint8_t                     strobeMode;
+    uint8_t                     mcFlags;
+    uint8_t                     padding;
+    uint32_t                    aT;
+    uint32_t                    bSP;
+    SISLANDS_SMC_SCLK_VALUE     sclk;
+    SISLANDS_SMC_MCLK_VALUE     mclk;
+    SISLANDS_SMC_VOLTAGE_VALUE  vddc;
+    SISLANDS_SMC_VOLTAGE_VALUE  mvdd;
+    SISLANDS_SMC_VOLTAGE_VALUE  vddci;
+    SISLANDS_SMC_VOLTAGE_VALUE  std_vddc;
+    uint8_t                     hysteresisUp;
+    uint8_t                     hysteresisDown;
+    uint8_t                     stateFlags;
+    uint8_t                     arbRefreshState;
+    uint32_t                    SQPowerThrottle;
+    uint32_t                    SQPowerThrottle_2;
+    uint32_t                    MaxPoweredUpCU;
+    SISLANDS_SMC_VOLTAGE_VALUE  high_temp_vddc;
+    SISLANDS_SMC_VOLTAGE_VALUE  low_temp_vddc;
+    uint32_t                    reserved[2];
+    PP_SIslands_Dpm2PerfLevel   dpm2;
+};
+
+#define SISLANDS_SMC_STROBE_RATIO    0x0F
+#define SISLANDS_SMC_STROBE_ENABLE   0x10
+
+#define SISLANDS_SMC_MC_EDC_RD_FLAG  0x01
+#define SISLANDS_SMC_MC_EDC_WR_FLAG  0x02
+#define SISLANDS_SMC_MC_RTT_ENABLE   0x04
+#define SISLANDS_SMC_MC_STUTTER_EN   0x08
+#define SISLANDS_SMC_MC_PG_EN        0x10
+
+typedef struct SISLANDS_SMC_HW_PERFORMANCE_LEVEL SISLANDS_SMC_HW_PERFORMANCE_LEVEL;
+
+struct SISLANDS_SMC_SWSTATE
+{
+    uint8_t                             flags;
+    uint8_t                             levelCount;
+    uint8_t                             padding2;
+    uint8_t                             padding3;
+    SISLANDS_SMC_HW_PERFORMANCE_LEVEL   levels[1];
+};
+
+typedef struct SISLANDS_SMC_SWSTATE SISLANDS_SMC_SWSTATE;
+
+#define SISLANDS_SMC_VOLTAGEMASK_VDDC  0
+#define SISLANDS_SMC_VOLTAGEMASK_MVDD  1
+#define SISLANDS_SMC_VOLTAGEMASK_VDDCI 2
+#define SISLANDS_SMC_VOLTAGEMASK_VDDC_PHASE_SHEDDING 3
+#define SISLANDS_SMC_VOLTAGEMASK_MAX   4
+
+struct SISLANDS_SMC_VOLTAGEMASKTABLE
+{
+    uint32_t lowMask[SISLANDS_SMC_VOLTAGEMASK_MAX];
+};
+
+typedef struct SISLANDS_SMC_VOLTAGEMASKTABLE SISLANDS_SMC_VOLTAGEMASKTABLE;
+
+#define SISLANDS_MAX_NO_VREG_STEPS 32
+
+struct SISLANDS_SMC_STATETABLE
+{
+    uint8_t                             thermalProtectType;
+    uint8_t                             systemFlags;
+    uint8_t                             maxVDDCIndexInPPTable;
+    uint8_t                             extraFlags;
+    uint32_t                            lowSMIO[SISLANDS_MAX_NO_VREG_STEPS];
+    SISLANDS_SMC_VOLTAGEMASKTABLE       voltageMaskTable;
+    SISLANDS_SMC_VOLTAGEMASKTABLE       phaseMaskTable;
+    PP_SIslands_DPM2Parameters          dpm2Params;
+    SISLANDS_SMC_SWSTATE                initialState;
+    SISLANDS_SMC_SWSTATE                ACPIState;
+    SISLANDS_SMC_SWSTATE                ULVState;
+    SISLANDS_SMC_SWSTATE                driverState;
+    SISLANDS_SMC_HW_PERFORMANCE_LEVEL   dpmLevels[SISLANDS_MAX_SMC_PERFORMANCE_LEVELS_PER_SWSTATE - 1];
+};
+
+typedef struct SISLANDS_SMC_STATETABLE SISLANDS_SMC_STATETABLE;
+
+#define SI_SMC_SOFT_REGISTER_mclk_chg_timeout         0x0
+#define SI_SMC_SOFT_REGISTER_delay_vreg               0xC
+#define SI_SMC_SOFT_REGISTER_delay_acpi               0x28
+#define SI_SMC_SOFT_REGISTER_seq_index                0x5C
+#define SI_SMC_SOFT_REGISTER_mvdd_chg_time            0x60
+#define SI_SMC_SOFT_REGISTER_mclk_switch_lim          0x70
+#define SI_SMC_SOFT_REGISTER_watermark_threshold      0x78
+#define SI_SMC_SOFT_REGISTER_phase_shedding_delay     0x88
+#define SI_SMC_SOFT_REGISTER_ulv_volt_change_delay    0x8C
+#define SI_SMC_SOFT_REGISTER_mc_block_delay           0x98
+#define SI_SMC_SOFT_REGISTER_ticks_per_us             0xA8
+#define SI_SMC_SOFT_REGISTER_crtc_index               0xC4
+#define SI_SMC_SOFT_REGISTER_mclk_change_block_cp_min 0xC8
+#define SI_SMC_SOFT_REGISTER_mclk_change_block_cp_max 0xCC
+#define SI_SMC_SOFT_REGISTER_non_ulv_pcie_link_width  0xF4
+#define SI_SMC_SOFT_REGISTER_tdr_is_about_to_happen   0xFC
+#define SI_SMC_SOFT_REGISTER_vr_hot_gpio              0x100
+#define SI_SMC_SOFT_REGISTER_svi_rework_plat_type     0x118
+#define SI_SMC_SOFT_REGISTER_svi_rework_gpio_id_svd   0x11c
+#define SI_SMC_SOFT_REGISTER_svi_rework_gpio_id_svc   0x120
+
+struct PP_SIslands_FanTable
+{
+	uint8_t  fdo_mode;
+	uint8_t  padding;
+	int16_t  temp_min;
+	int16_t  temp_med;
+	int16_t  temp_max;
+	int16_t  slope1;
+	int16_t  slope2;
+	int16_t  fdo_min;
+	int16_t  hys_up;
+	int16_t  hys_down;
+	int16_t  hys_slope;
+	int16_t  temp_resp_lim;
+	int16_t  temp_curr;
+	int16_t  slope_curr;
+	int16_t  pwm_curr;
+	uint32_t refresh_period;
+	int16_t  fdo_max;
+	uint8_t  temp_src;
+	int8_t  padding2;
+};
+
+typedef struct PP_SIslands_FanTable PP_SIslands_FanTable;
+
+#define SMC_SISLANDS_LKGE_LUT_NUM_OF_TEMP_ENTRIES 16
+#define SMC_SISLANDS_LKGE_LUT_NUM_OF_VOLT_ENTRIES 32
+
+#define SMC_SISLANDS_SCALE_I  7
+#define SMC_SISLANDS_SCALE_R 12
+
+struct PP_SIslands_CacConfig
+{
+    uint16_t   cac_lkge_lut[SMC_SISLANDS_LKGE_LUT_NUM_OF_TEMP_ENTRIES][SMC_SISLANDS_LKGE_LUT_NUM_OF_VOLT_ENTRIES];
+    uint32_t   lkge_lut_V0;
+    uint32_t   lkge_lut_Vstep;
+    uint32_t   WinTime;
+    uint32_t   R_LL;
+    uint32_t   calculation_repeats;
+    uint32_t   l2numWin_TDP;
+    uint32_t   dc_cac;
+    uint8_t    lts_truncate_n;
+    uint8_t    SHIFT_N;
+    uint8_t    log2_PG_LKG_SCALE;
+    uint8_t    cac_temp;
+    uint32_t   lkge_lut_T0;
+    uint32_t   lkge_lut_Tstep;
+};
+
+typedef struct PP_SIslands_CacConfig PP_SIslands_CacConfig;
+
+#define SMC_SISLANDS_MC_REGISTER_ARRAY_SIZE 16
+#define SMC_SISLANDS_MC_REGISTER_ARRAY_SET_COUNT 20
+
+struct SMC_SIslands_MCRegisterAddress
+{
+    uint16_t s0;
+    uint16_t s1;
+};
+
+typedef struct SMC_SIslands_MCRegisterAddress SMC_SIslands_MCRegisterAddress;
+
+struct SMC_SIslands_MCRegisterSet
+{
+    uint32_t value[SMC_SISLANDS_MC_REGISTER_ARRAY_SIZE];
+};
+
+typedef struct SMC_SIslands_MCRegisterSet SMC_SIslands_MCRegisterSet;
+
+struct SMC_SIslands_MCRegisters
+{
+    uint8_t                             last;
+    uint8_t                             reserved[3];
+    SMC_SIslands_MCRegisterAddress      address[SMC_SISLANDS_MC_REGISTER_ARRAY_SIZE];
+    SMC_SIslands_MCRegisterSet          data[SMC_SISLANDS_MC_REGISTER_ARRAY_SET_COUNT];
+};
+
+typedef struct SMC_SIslands_MCRegisters SMC_SIslands_MCRegisters;
+
+struct SMC_SIslands_MCArbDramTimingRegisterSet
+{
+    uint32_t mc_arb_dram_timing;
+    uint32_t mc_arb_dram_timing2;
+    uint8_t  mc_arb_rfsh_rate;
+    uint8_t  mc_arb_burst_time;
+    uint8_t  padding[2];
+};
+
+typedef struct SMC_SIslands_MCArbDramTimingRegisterSet SMC_SIslands_MCArbDramTimingRegisterSet;
+
+struct SMC_SIslands_MCArbDramTimingRegisters
+{
+    uint8_t                                     arb_current;
+    uint8_t                                     reserved[3];
+    SMC_SIslands_MCArbDramTimingRegisterSet     data[16];
+};
+
+typedef struct SMC_SIslands_MCArbDramTimingRegisters SMC_SIslands_MCArbDramTimingRegisters;
+
+struct SMC_SISLANDS_SPLL_DIV_TABLE
+{
+    uint32_t    freq[256];
+    uint32_t    ss[256];
+};
+
+#define SMC_SISLANDS_SPLL_DIV_TABLE_FBDIV_MASK  0x01ffffff
+#define SMC_SISLANDS_SPLL_DIV_TABLE_FBDIV_SHIFT 0
+#define SMC_SISLANDS_SPLL_DIV_TABLE_PDIV_MASK   0xfe000000
+#define SMC_SISLANDS_SPLL_DIV_TABLE_PDIV_SHIFT  25
+#define SMC_SISLANDS_SPLL_DIV_TABLE_CLKV_MASK   0x000fffff
+#define SMC_SISLANDS_SPLL_DIV_TABLE_CLKV_SHIFT  0
+#define SMC_SISLANDS_SPLL_DIV_TABLE_CLKS_MASK   0xfff00000
+#define SMC_SISLANDS_SPLL_DIV_TABLE_CLKS_SHIFT  20
+
+typedef struct SMC_SISLANDS_SPLL_DIV_TABLE SMC_SISLANDS_SPLL_DIV_TABLE;
+
+#define SMC_SISLANDS_DTE_MAX_FILTER_STAGES 5
+
+#define SMC_SISLANDS_DTE_MAX_TEMPERATURE_DEPENDENT_ARRAY_SIZE 16
+
+struct Smc_SIslands_DTE_Configuration
+{
+    uint32_t tau[SMC_SISLANDS_DTE_MAX_FILTER_STAGES];
+    uint32_t R[SMC_SISLANDS_DTE_MAX_FILTER_STAGES];
+    uint32_t K;
+    uint32_t T0;
+    uint32_t MaxT;
+    uint8_t  WindowSize;
+    uint8_t  Tdep_count;
+    uint8_t  temp_select;
+    uint8_t  DTE_mode;
+    uint8_t  T_limits[SMC_SISLANDS_DTE_MAX_TEMPERATURE_DEPENDENT_ARRAY_SIZE];
+    uint32_t Tdep_tau[SMC_SISLANDS_DTE_MAX_TEMPERATURE_DEPENDENT_ARRAY_SIZE];
+    uint32_t Tdep_R[SMC_SISLANDS_DTE_MAX_TEMPERATURE_DEPENDENT_ARRAY_SIZE];
+    uint32_t Tthreshold;
+};
+
+typedef struct Smc_SIslands_DTE_Configuration Smc_SIslands_DTE_Configuration;
+
+#define SMC_SISLANDS_DTE_STATUS_FLAG_DTE_ON 1
+
+#define SISLANDS_SMC_FIRMWARE_HEADER_LOCATION 0x10000
+
+#define SISLANDS_SMC_FIRMWARE_HEADER_version                   0x0
+#define SISLANDS_SMC_FIRMWARE_HEADER_flags                     0x4
+#define SISLANDS_SMC_FIRMWARE_HEADER_softRegisters             0xC
+#define SISLANDS_SMC_FIRMWARE_HEADER_stateTable                0x10
+#define SISLANDS_SMC_FIRMWARE_HEADER_fanTable                  0x14
+#define SISLANDS_SMC_FIRMWARE_HEADER_CacConfigTable            0x18
+#define SISLANDS_SMC_FIRMWARE_HEADER_mcRegisterTable           0x24
+#define SISLANDS_SMC_FIRMWARE_HEADER_mcArbDramAutoRefreshTable 0x30
+#define SISLANDS_SMC_FIRMWARE_HEADER_spllTable                 0x38
+#define SISLANDS_SMC_FIRMWARE_HEADER_DteConfiguration          0x40
+#define SISLANDS_SMC_FIRMWARE_HEADER_PAPMParameters            0x48
+
+#pragma pack(pop)
+
+int amdgpu_si_copy_bytes_to_smc(struct amdgpu_device *adev,
+				u32 smc_start_address,
+				const u8 *src, u32 byte_count, u32 limit);
+void amdgpu_si_start_smc(struct amdgpu_device *adev);
+void amdgpu_si_reset_smc(struct amdgpu_device *adev);
+int amdgpu_si_program_jump_on_start(struct amdgpu_device *adev);
+void amdgpu_si_smc_clock(struct amdgpu_device *adev, bool enable);
+bool amdgpu_si_is_smc_running(struct amdgpu_device *adev);
+PPSMC_Result amdgpu_si_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg);
+PPSMC_Result amdgpu_si_wait_for_smc_inactive(struct amdgpu_device *adev);
+int amdgpu_si_load_smc_ucode(struct amdgpu_device *adev, u32 limit);
+int amdgpu_si_read_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
+				  u32 *value, u32 limit);
+int amdgpu_si_write_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
+				   u32 value, u32 limit);
+
+#endif
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/soc15.c b/sys/dev/pci/drm/amd/amdgpu/soc15.c
new file mode 100644
index 00000000000..be235aa94e0
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/soc15.c
@@ -0,0 +1,1001 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_psp.h"
+#include "atom.h"
+#include "amd_pcie.h"
+
+#include "uvd/uvd_7_0_offset.h"
+#include "gc/gc_9_0_offset.h"
+#include "gc/gc_9_0_sh_mask.h"
+#include "sdma0/sdma0_4_0_offset.h"
+#include "sdma1/sdma1_4_0_offset.h"
+#include "hdp/hdp_4_0_offset.h"
+#include "hdp/hdp_4_0_sh_mask.h"
+#include "smuio/smuio_9_0_offset.h"
+#include "smuio/smuio_9_0_sh_mask.h"
+
+#include "soc15.h"
+#include "soc15_common.h"
+#include "gfx_v9_0.h"
+#include "gmc_v9_0.h"
+#include "gfxhub_v1_0.h"
+#include "mmhub_v1_0.h"
+#include "df_v1_7.h"
+#include "df_v3_6.h"
+#include "vega10_ih.h"
+#include "sdma_v4_0.h"
+#include "uvd_v7_0.h"
+#include "vce_v4_0.h"
+#include "vcn_v1_0.h"
+#include "dce_virtual.h"
+#include "mxgpu_ai.h"
+
+#define mmMP0_MISC_CGTT_CTRL0                                                                   0x01b9
+#define mmMP0_MISC_CGTT_CTRL0_BASE_IDX                                                          0
+#define mmMP0_MISC_LIGHT_SLEEP_CTRL                                                             0x01ba
+#define mmMP0_MISC_LIGHT_SLEEP_CTRL_BASE_IDX                                                    0
+
+/*
+ * Indirect registers accessor
+ */
+static u32 soc15_pcie_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags, address, data;
+	u32 r;
+	address = adev->nbio_funcs->get_pcie_index_offset(adev);
+	data = adev->nbio_funcs->get_pcie_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(address, reg);
+	(void)RREG32(address);
+	r = RREG32(data);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+	return r;
+}
+
+static void soc15_pcie_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags, address, data;
+
+	address = adev->nbio_funcs->get_pcie_index_offset(adev);
+	data = adev->nbio_funcs->get_pcie_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(address, reg);
+	(void)RREG32(address);
+	WREG32(data, v);
+	(void)RREG32(data);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+static u32 soc15_uvd_ctx_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags, address, data;
+	u32 r;
+
+	address = SOC15_REG_OFFSET(UVD, 0, mmUVD_CTX_INDEX);
+	data = SOC15_REG_OFFSET(UVD, 0, mmUVD_CTX_DATA);
+
+	spin_lock_irqsave(&adev->uvd_ctx_idx_lock, flags);
+	WREG32(address, ((reg) & 0x1ff));
+	r = RREG32(data);
+	spin_unlock_irqrestore(&adev->uvd_ctx_idx_lock, flags);
+	return r;
+}
+
+static void soc15_uvd_ctx_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags, address, data;
+
+	address = SOC15_REG_OFFSET(UVD, 0, mmUVD_CTX_INDEX);
+	data = SOC15_REG_OFFSET(UVD, 0, mmUVD_CTX_DATA);
+
+	spin_lock_irqsave(&adev->uvd_ctx_idx_lock, flags);
+	WREG32(address, ((reg) & 0x1ff));
+	WREG32(data, (v));
+	spin_unlock_irqrestore(&adev->uvd_ctx_idx_lock, flags);
+}
+
+static u32 soc15_didt_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags, address, data;
+	u32 r;
+
+	address = SOC15_REG_OFFSET(GC, 0, mmDIDT_IND_INDEX);
+	data = SOC15_REG_OFFSET(GC, 0, mmDIDT_IND_DATA);
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(address, (reg));
+	r = RREG32(data);
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+	return r;
+}
+
+static void soc15_didt_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags, address, data;
+
+	address = SOC15_REG_OFFSET(GC, 0, mmDIDT_IND_INDEX);
+	data = SOC15_REG_OFFSET(GC, 0, mmDIDT_IND_DATA);
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(address, (reg));
+	WREG32(data, (v));
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+}
+
+static u32 soc15_gc_cac_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->gc_cac_idx_lock, flags);
+	WREG32_SOC15(GC, 0, mmGC_CAC_IND_INDEX, (reg));
+	r = RREG32_SOC15(GC, 0, mmGC_CAC_IND_DATA);
+	spin_unlock_irqrestore(&adev->gc_cac_idx_lock, flags);
+	return r;
+}
+
+static void soc15_gc_cac_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->gc_cac_idx_lock, flags);
+	WREG32_SOC15(GC, 0, mmGC_CAC_IND_INDEX, (reg));
+	WREG32_SOC15(GC, 0, mmGC_CAC_IND_DATA, (v));
+	spin_unlock_irqrestore(&adev->gc_cac_idx_lock, flags);
+}
+
+static u32 soc15_se_cac_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->se_cac_idx_lock, flags);
+	WREG32_SOC15(GC, 0, mmSE_CAC_IND_INDEX, (reg));
+	r = RREG32_SOC15(GC, 0, mmSE_CAC_IND_DATA);
+	spin_unlock_irqrestore(&adev->se_cac_idx_lock, flags);
+	return r;
+}
+
+static void soc15_se_cac_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->se_cac_idx_lock, flags);
+	WREG32_SOC15(GC, 0, mmSE_CAC_IND_INDEX, (reg));
+	WREG32_SOC15(GC, 0, mmSE_CAC_IND_DATA, (v));
+	spin_unlock_irqrestore(&adev->se_cac_idx_lock, flags);
+}
+
+static u32 soc15_get_config_memsize(struct amdgpu_device *adev)
+{
+	return adev->nbio_funcs->get_memsize(adev);
+}
+
+static u32 soc15_get_xclk(struct amdgpu_device *adev)
+{
+	return adev->clock.spll.reference_freq;
+}
+
+
+void soc15_grbm_select(struct amdgpu_device *adev,
+		     u32 me, u32 pipe, u32 queue, u32 vmid)
+{
+	u32 grbm_gfx_cntl = 0;
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, PIPEID, pipe);
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, MEID, me);
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, VMID, vmid);
+	grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, QUEUEID, queue);
+
+	WREG32(SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_CNTL), grbm_gfx_cntl);
+}
+
+static void soc15_vga_set_state(struct amdgpu_device *adev, bool state)
+{
+	/* todo */
+}
+
+static bool soc15_read_disabled_bios(struct amdgpu_device *adev)
+{
+	/* todo */
+	return false;
+}
+
+static bool soc15_read_bios_from_rom(struct amdgpu_device *adev,
+				     u8 *bios, u32 length_bytes)
+{
+	u32 *dw_ptr;
+	u32 i, length_dw;
+
+	if (bios == NULL)
+		return false;
+	if (length_bytes == 0)
+		return false;
+	/* APU vbios image is part of sbios image */
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	dw_ptr = (u32 *)bios;
+	length_dw = roundup2(length_bytes, 4) / 4;
+
+	/* set rom index to 0 */
+	WREG32(SOC15_REG_OFFSET(SMUIO, 0, mmROM_INDEX), 0);
+	/* read out the rom data */
+	for (i = 0; i < length_dw; i++)
+		dw_ptr[i] = RREG32(SOC15_REG_OFFSET(SMUIO, 0, mmROM_DATA));
+
+	return true;
+}
+
+struct soc15_allowed_register_entry {
+	uint32_t hwip;
+	uint32_t inst;
+	uint32_t seg;
+	uint32_t reg_offset;
+	bool grbm_indexed;
+};
+
+
+static struct soc15_allowed_register_entry soc15_allowed_read_registers[] = {
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS2)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE0)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE1)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE2)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE3)},
+	{ SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_STATUS_REG)},
+	{ SOC15_REG_ENTRY(SDMA1, 0, mmSDMA1_STATUS_REG)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_STALLED_STAT2)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_STALLED_STAT3)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPF_BUSY_STAT)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPF_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPF_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPC_STALLED_STAT1)},
+	{ SOC15_REG_ENTRY(GC, 0, mmCP_CPC_STATUS)},
+	{ SOC15_REG_ENTRY(GC, 0, mmGB_ADDR_CONFIG)},
+	{ SOC15_REG_ENTRY(GC, 0, mmDB_DEBUG2)},
+};
+
+static uint32_t soc15_read_indexed_register(struct amdgpu_device *adev, u32 se_num,
+					 u32 sh_num, u32 reg_offset)
+{
+	uint32_t val;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	if (se_num != 0xffffffff || sh_num != 0xffffffff)
+		amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+
+	val = RREG32(reg_offset);
+
+	if (se_num != 0xffffffff || sh_num != 0xffffffff)
+		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+	return val;
+}
+
+static uint32_t soc15_get_register_value(struct amdgpu_device *adev,
+					 bool indexed, u32 se_num,
+					 u32 sh_num, u32 reg_offset)
+{
+	if (indexed) {
+		return soc15_read_indexed_register(adev, se_num, sh_num, reg_offset);
+	} else {
+		if (reg_offset == SOC15_REG_OFFSET(GC, 0, mmGB_ADDR_CONFIG))
+			return adev->gfx.config.gb_addr_config;
+		else if (reg_offset == SOC15_REG_OFFSET(GC, 0, mmDB_DEBUG2))
+			return adev->gfx.config.db_debug2;
+		return RREG32(reg_offset);
+	}
+}
+
+static int soc15_read_register(struct amdgpu_device *adev, u32 se_num,
+			    u32 sh_num, u32 reg_offset, u32 *value)
+{
+	uint32_t i;
+	struct soc15_allowed_register_entry  *en;
+
+	*value = 0;
+	for (i = 0; i < ARRAY_SIZE(soc15_allowed_read_registers); i++) {
+		en = &soc15_allowed_read_registers[i];
+		if (reg_offset != (adev->reg_offset[en->hwip][en->inst][en->seg]
+					+ en->reg_offset))
+			continue;
+
+		*value = soc15_get_register_value(adev,
+						  soc15_allowed_read_registers[i].grbm_indexed,
+						  se_num, sh_num, reg_offset);
+		return 0;
+	}
+	return -EINVAL;
+}
+
+
+/**
+ * soc15_program_register_sequence - program an array of registers.
+ *
+ * @adev: amdgpu_device pointer
+ * @regs: pointer to the register array
+ * @array_size: size of the register array
+ *
+ * Programs an array or registers with and and or masks.
+ * This is a helper for setting golden registers.
+ */
+
+void soc15_program_register_sequence(struct amdgpu_device *adev,
+					     const struct soc15_reg_golden *regs,
+					     const u32 array_size)
+{
+	const struct soc15_reg_golden *entry;
+	u32 tmp, reg;
+	int i;
+
+	for (i = 0; i < array_size; ++i) {
+		entry = &regs[i];
+		reg =  adev->reg_offset[entry->hwip][entry->instance][entry->segment] + entry->reg;
+
+		if (entry->and_mask == 0xffffffff) {
+			tmp = entry->or_mask;
+		} else {
+			tmp = RREG32(reg);
+			tmp &= ~(entry->and_mask);
+			tmp |= entry->or_mask;
+		}
+		WREG32(reg, tmp);
+	}
+
+}
+
+
+static int soc15_asic_reset(struct amdgpu_device *adev)
+{
+	u32 i;
+
+	amdgpu_atombios_scratch_regs_engine_hung(adev, true);
+
+	dev_info(adev->dev, "GPU reset\n");
+
+	/* disable BM */
+	pci_clear_master(adev->pdev);
+
+	pci_save_state(adev->pdev);
+
+	psp_gpu_reset(adev);
+
+	pci_restore_state(adev->pdev);
+
+	/* wait for asic to come out of reset */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		u32 memsize = adev->nbio_funcs->get_memsize(adev);
+
+		if (memsize != 0xffffffff)
+			break;
+		udelay(1);
+	}
+
+	amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+
+	return 0;
+}
+
+/*static int soc15_set_uvd_clock(struct amdgpu_device *adev, u32 clock,
+			u32 cntl_reg, u32 status_reg)
+{
+	return 0;
+}*/
+
+static int soc15_set_uvd_clocks(struct amdgpu_device *adev, u32 vclk, u32 dclk)
+{
+	/*int r;
+
+	r = soc15_set_uvd_clock(adev, vclk, ixCG_VCLK_CNTL, ixCG_VCLK_STATUS);
+	if (r)
+		return r;
+
+	r = soc15_set_uvd_clock(adev, dclk, ixCG_DCLK_CNTL, ixCG_DCLK_STATUS);
+	*/
+	return 0;
+}
+
+static int soc15_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk)
+{
+	/* todo */
+
+	return 0;
+}
+
+static void soc15_pcie_gen3_enable(struct amdgpu_device *adev)
+{
+	if (pci_is_root_bus(adev->pdev->bus))
+		return;
+
+	if (amdgpu_pcie_gen2 == 0)
+		return;
+
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	if (!(adev->pm.pcie_gen_mask & (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+					CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)))
+		return;
+
+	/* todo */
+}
+
+static void soc15_program_aspm(struct amdgpu_device *adev)
+{
+
+	if (amdgpu_aspm == 0)
+		return;
+
+	/* todo */
+}
+
+static void soc15_enable_doorbell_aperture(struct amdgpu_device *adev,
+					   bool enable)
+{
+	adev->nbio_funcs->enable_doorbell_aperture(adev, enable);
+	adev->nbio_funcs->enable_doorbell_selfring_aperture(adev, enable);
+}
+
+static const struct amdgpu_ip_block_version vega10_common_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_COMMON,
+	.major = 2,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &soc15_common_ip_funcs,
+};
+
+int soc15_set_ip_blocks(struct amdgpu_device *adev)
+{
+	/* Set IP register base before any HW register access */
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_RAVEN:
+		vega10_reg_base_init(adev);
+		break;
+	case CHIP_VEGA20:
+		vega20_reg_base_init(adev);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if (adev->flags & AMD_IS_APU)
+		adev->nbio_funcs = &nbio_v7_0_funcs;
+	else if (adev->asic_type == CHIP_VEGA20)
+		adev->nbio_funcs = &nbio_v7_0_funcs;
+	else
+		adev->nbio_funcs = &nbio_v6_1_funcs;
+
+	if (adev->asic_type == CHIP_VEGA20)
+		adev->df_funcs = &df_v3_6_funcs;
+	else
+		adev->df_funcs = &df_v1_7_funcs;
+	adev->nbio_funcs->detect_hw_virt(adev);
+
+	if (amdgpu_sriov_vf(adev))
+		adev->virt.ops = &xgpu_ai_virt_ops;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+		amdgpu_device_ip_block_add(adev, &vega10_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v9_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &vega10_ih_ip_block);
+		if (adev->asic_type != CHIP_VEGA20) {
+			amdgpu_device_ip_block_add(adev, &psp_v3_1_ip_block);
+			if (!amdgpu_sriov_vf(adev))
+				amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		}
+		if (adev->enable_virtual_display || amdgpu_sriov_vf(adev))
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#else
+#	warning "Enable CONFIG_DRM_AMD_DC for display support on SOC15."
+#endif
+		amdgpu_device_ip_block_add(adev, &gfx_v9_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v4_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v7_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v4_0_ip_block);
+		break;
+	case CHIP_RAVEN:
+		amdgpu_device_ip_block_add(adev, &vega10_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v9_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &vega10_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &psp_v10_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		if (adev->enable_virtual_display || amdgpu_sriov_vf(adev))
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#else
+#	warning "Enable CONFIG_DRM_AMD_DC for display support on SOC15."
+#endif
+		amdgpu_device_ip_block_add(adev, &gfx_v9_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v4_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &vcn_v1_0_ip_block);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static uint32_t soc15_get_rev_id(struct amdgpu_device *adev)
+{
+	return adev->nbio_funcs->get_rev_id(adev);
+}
+
+static void soc15_flush_hdp(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	adev->nbio_funcs->hdp_flush(adev, ring);
+}
+
+static void soc15_invalidate_hdp(struct amdgpu_device *adev,
+				 struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg)
+		WREG32_SOC15_NO_KIQ(NBIO, 0, mmHDP_READ_CACHE_INVALIDATE, 1);
+	else
+		amdgpu_ring_emit_wreg(ring, SOC15_REG_OFFSET(
+			HDP, 0, mmHDP_READ_CACHE_INVALIDATE), 1);
+}
+
+static bool soc15_need_full_reset(struct amdgpu_device *adev)
+{
+	/* change this when we implement soft reset */
+	return true;
+}
+
+static const struct amdgpu_asic_funcs soc15_asic_funcs =
+{
+	.read_disabled_bios = &soc15_read_disabled_bios,
+	.read_bios_from_rom = &soc15_read_bios_from_rom,
+	.read_register = &soc15_read_register,
+	.reset = &soc15_asic_reset,
+	.set_vga_state = &soc15_vga_set_state,
+	.get_xclk = &soc15_get_xclk,
+	.set_uvd_clocks = &soc15_set_uvd_clocks,
+	.set_vce_clocks = &soc15_set_vce_clocks,
+	.get_config_memsize = &soc15_get_config_memsize,
+	.flush_hdp = &soc15_flush_hdp,
+	.invalidate_hdp = &soc15_invalidate_hdp,
+	.need_full_reset = &soc15_need_full_reset,
+};
+
+static int soc15_common_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->smc_rreg = NULL;
+	adev->smc_wreg = NULL;
+	adev->pcie_rreg = &soc15_pcie_rreg;
+	adev->pcie_wreg = &soc15_pcie_wreg;
+	adev->uvd_ctx_rreg = &soc15_uvd_ctx_rreg;
+	adev->uvd_ctx_wreg = &soc15_uvd_ctx_wreg;
+	adev->didt_rreg = &soc15_didt_rreg;
+	adev->didt_wreg = &soc15_didt_wreg;
+	adev->gc_cac_rreg = &soc15_gc_cac_rreg;
+	adev->gc_cac_wreg = &soc15_gc_cac_wreg;
+	adev->se_cac_rreg = &soc15_se_cac_rreg;
+	adev->se_cac_wreg = &soc15_se_cac_wreg;
+
+	adev->asic_funcs = &soc15_asic_funcs;
+
+	adev->rev_id = soc15_get_rev_id(adev);
+	adev->external_rev_id = 0xFF;
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_DRM_MGCG |
+			AMD_CG_SUPPORT_DRM_LS |
+			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_DF_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS;
+		adev->pg_flags = 0;
+		adev->external_rev_id = 0x1;
+		break;
+	case CHIP_VEGA12:
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_VCE_MGCG |
+			AMD_CG_SUPPORT_UVD_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x14;
+		break;
+	case CHIP_VEGA20:
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_VCE_MGCG |
+			AMD_CG_SUPPORT_UVD_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x28;
+		break;
+	case CHIP_RAVEN:
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_DRM_MGCG |
+			AMD_CG_SUPPORT_DRM_LS |
+			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_VCN_MGCG;
+
+		adev->pg_flags = AMD_PG_SUPPORT_SDMA | AMD_PG_SUPPORT_VCN;
+
+		if (adev->powerplay.pp_feature & PP_GFXOFF_MASK)
+			adev->pg_flags |= AMD_PG_SUPPORT_GFX_PG |
+				AMD_PG_SUPPORT_CP |
+				AMD_PG_SUPPORT_RLC_SMU_HS;
+
+		adev->external_rev_id = 0x1;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		amdgpu_virt_init_setting(adev);
+		xgpu_ai_mailbox_set_irq_funcs(adev);
+	}
+
+	return 0;
+}
+
+static int soc15_common_late_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		xgpu_ai_mailbox_get_irq(adev);
+
+	return 0;
+}
+
+static int soc15_common_sw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		xgpu_ai_mailbox_add_irq_id(adev);
+
+	return 0;
+}
+
+static int soc15_common_sw_fini(void *handle)
+{
+	return 0;
+}
+
+static int soc15_common_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* enable pcie gen2/3 link */
+	soc15_pcie_gen3_enable(adev);
+	/* enable aspm */
+	soc15_program_aspm(adev);
+	/* setup nbio registers */
+	adev->nbio_funcs->init_registers(adev);
+	/* enable the doorbell aperture */
+	soc15_enable_doorbell_aperture(adev, true);
+
+	return 0;
+}
+
+static int soc15_common_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* disable the doorbell aperture */
+	soc15_enable_doorbell_aperture(adev, false);
+	if (amdgpu_sriov_vf(adev))
+		xgpu_ai_mailbox_put_irq(adev);
+
+	return 0;
+}
+
+static int soc15_common_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return soc15_common_hw_fini(adev);
+}
+
+static int soc15_common_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return soc15_common_hw_init(adev);
+}
+
+static bool soc15_common_is_idle(void *handle)
+{
+	return true;
+}
+
+static int soc15_common_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static int soc15_common_soft_reset(void *handle)
+{
+	return 0;
+}
+
+static void soc15_update_hdp_light_sleep(struct amdgpu_device *adev, bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32(SOC15_REG_OFFSET(HDP, 0, mmHDP_MEM_POWER_LS));
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
+		data |= HDP_MEM_POWER_LS__LS_ENABLE_MASK;
+	else
+		data &= ~HDP_MEM_POWER_LS__LS_ENABLE_MASK;
+
+	if (def != data)
+		WREG32(SOC15_REG_OFFSET(HDP, 0, mmHDP_MEM_POWER_LS), data);
+}
+
+static void soc15_update_drm_clock_gating(struct amdgpu_device *adev, bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_CGTT_CTRL0));
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_DRM_MGCG))
+		data &= ~(0x01000000 |
+			  0x02000000 |
+			  0x04000000 |
+			  0x08000000 |
+			  0x10000000 |
+			  0x20000000 |
+			  0x40000000 |
+			  0x80000000);
+	else
+		data |= (0x01000000 |
+			 0x02000000 |
+			 0x04000000 |
+			 0x08000000 |
+			 0x10000000 |
+			 0x20000000 |
+			 0x40000000 |
+			 0x80000000);
+
+	if (def != data)
+		WREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_CGTT_CTRL0), data);
+}
+
+static void soc15_update_drm_light_sleep(struct amdgpu_device *adev, bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_LIGHT_SLEEP_CTRL));
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_DRM_LS))
+		data |= 1;
+	else
+		data &= ~1;
+
+	if (def != data)
+		WREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_LIGHT_SLEEP_CTRL), data);
+}
+
+static void soc15_update_rom_medium_grain_clock_gating(struct amdgpu_device *adev,
+						       bool enable)
+{
+	uint32_t def, data;
+
+	def = data = RREG32(SOC15_REG_OFFSET(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0));
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ROM_MGCG))
+		data &= ~(CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
+			CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK);
+	else
+		data |= CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
+			CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK;
+
+	if (def != data)
+		WREG32(SOC15_REG_OFFSET(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0), data);
+}
+
+static int soc15_common_set_clockgating_state(void *handle,
+					    enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA10:
+	case CHIP_VEGA12:
+	case CHIP_VEGA20:
+		adev->nbio_funcs->update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		adev->nbio_funcs->update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		soc15_update_hdp_light_sleep(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		soc15_update_drm_clock_gating(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		soc15_update_drm_light_sleep(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		soc15_update_rom_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		adev->df_funcs->update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		break;
+	case CHIP_RAVEN:
+		adev->nbio_funcs->update_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		adev->nbio_funcs->update_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		soc15_update_hdp_light_sleep(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		soc15_update_drm_clock_gating(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		soc15_update_drm_light_sleep(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		soc15_update_rom_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE ? true : false);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static void soc15_common_get_clockgating_state(void *handle, u32 *flags)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	adev->nbio_funcs->get_clockgating_state(adev, flags);
+
+	/* AMD_CG_SUPPORT_HDP_LS */
+	data = RREG32(SOC15_REG_OFFSET(HDP, 0, mmHDP_MEM_POWER_LS));
+	if (data & HDP_MEM_POWER_LS__LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_LS;
+
+	/* AMD_CG_SUPPORT_DRM_MGCG */
+	data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_CGTT_CTRL0));
+	if (!(data & 0x01000000))
+		*flags |= AMD_CG_SUPPORT_DRM_MGCG;
+
+	/* AMD_CG_SUPPORT_DRM_LS */
+	data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_LIGHT_SLEEP_CTRL));
+	if (data & 0x1)
+		*flags |= AMD_CG_SUPPORT_DRM_LS;
+
+	/* AMD_CG_SUPPORT_ROM_MGCG */
+	data = RREG32(SOC15_REG_OFFSET(SMUIO, 0, mmCGTT_ROM_CLK_CTRL0));
+	if (!(data & CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK))
+		*flags |= AMD_CG_SUPPORT_ROM_MGCG;
+
+	adev->df_funcs->get_clockgating_state(adev, flags);
+}
+
+static int soc15_common_set_powergating_state(void *handle,
+					    enum amd_powergating_state state)
+{
+	/* todo */
+	return 0;
+}
+
+const struct amd_ip_funcs soc15_common_ip_funcs = {
+	.name = "soc15_common",
+	.early_init = soc15_common_early_init,
+	.late_init = soc15_common_late_init,
+	.sw_init = soc15_common_sw_init,
+	.sw_fini = soc15_common_sw_fini,
+	.hw_init = soc15_common_hw_init,
+	.hw_fini = soc15_common_hw_fini,
+	.suspend = soc15_common_suspend,
+	.resume = soc15_common_resume,
+	.is_idle = soc15_common_is_idle,
+	.wait_for_idle = soc15_common_wait_for_idle,
+	.soft_reset = soc15_common_soft_reset,
+	.set_clockgating_state = soc15_common_set_clockgating_state,
+	.set_powergating_state = soc15_common_set_powergating_state,
+	.get_clockgating_state= soc15_common_get_clockgating_state,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/soc15.h b/sys/dev/pci/drm/amd/amdgpu/soc15.h
new file mode 100644
index 00000000000..1f714b7af52
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/soc15.h
@@ -0,0 +1,60 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SOC15_H__
+#define __SOC15_H__
+
+#include "nbio_v6_1.h"
+#include "nbio_v7_0.h"
+
+#define SOC15_FLUSH_GPU_TLB_NUM_WREG		4
+#define SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT	1
+
+extern const struct amd_ip_funcs soc15_common_ip_funcs;
+
+struct soc15_reg_golden {
+	u32	hwip;
+	u32	instance;
+	u32	segment;
+	u32	reg;
+	u32	and_mask;
+	u32	or_mask;
+};
+
+#define SOC15_REG_ENTRY(ip, inst, reg)	ip##_HWIP, inst, reg##_BASE_IDX, reg
+
+#define SOC15_REG_GOLDEN_VALUE(ip, inst, reg, and_mask, or_mask) \
+	{ ip##_HWIP, inst, reg##_BASE_IDX, reg, and_mask, or_mask }
+
+void soc15_grbm_select(struct amdgpu_device *adev,
+		    u32 me, u32 pipe, u32 queue, u32 vmid);
+int soc15_set_ip_blocks(struct amdgpu_device *adev);
+
+void soc15_program_register_sequence(struct amdgpu_device *adev,
+					     const struct soc15_reg_golden *registers,
+					     const u32 array_size);
+
+int vega10_reg_base_init(struct amdgpu_device *adev);
+int vega20_reg_base_init(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/soc15_common.h b/sys/dev/pci/drm/amd/amdgpu/soc15_common.h
new file mode 100644
index 00000000000..0942f492d2e
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/soc15_common.h
@@ -0,0 +1,67 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SOC15_COMMON_H__
+#define __SOC15_COMMON_H__
+
+/* Register Access Macros */
+#define SOC15_REG_OFFSET(ip, inst, reg)	(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg)
+
+#define WREG32_FIELD15(ip, idx, reg, field, val)	\
+	WREG32(adev->reg_offset[ip##_HWIP][idx][mm##reg##_BASE_IDX] + mm##reg,	\
+	(RREG32(adev->reg_offset[ip##_HWIP][idx][mm##reg##_BASE_IDX] + mm##reg)	\
+	& ~REG_FIELD_MASK(reg, field)) | (val) << REG_FIELD_SHIFT(reg, field))
+
+#define RREG32_SOC15(ip, inst, reg) \
+	RREG32(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg)
+
+#define RREG32_SOC15_OFFSET(ip, inst, reg, offset) \
+	RREG32((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) + offset)
+
+#define WREG32_SOC15(ip, inst, reg, value) \
+	WREG32((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg), value)
+
+#define WREG32_SOC15_NO_KIQ(ip, inst, reg, value) \
+	WREG32_NO_KIQ((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg), value)
+
+#define WREG32_SOC15_OFFSET(ip, inst, reg, offset, value) \
+	WREG32((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg) + offset, value)
+
+#define SOC15_WAIT_ON_RREG(ip, inst, reg, expected_value, mask, ret) \
+	do {							\
+		uint32_t tmp_ = RREG32(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg); \
+		uint32_t loop = adev->usec_timeout;		\
+		while ((tmp_ & (mask)) != (expected_value)) {	\
+			udelay(2);				\
+			tmp_ = RREG32(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg); \
+			loop--;					\
+			if (!loop) {				\
+				ret = -ETIMEDOUT;		\
+				break;				\
+			}					\
+		}						\
+	} while (0)
+
+#endif
+
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/soc15d.h b/sys/dev/pci/drm/amd/amdgpu/soc15d.h
new file mode 100644
index 00000000000..edfe50821cd
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/soc15d.h
@@ -0,0 +1,408 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef SOC15_H
+#define SOC15_H
+
+#define GFX9_NUM_GFX_RINGS     1
+#define GFX9_NUM_COMPUTE_RINGS 8
+
+/*
+ * PM4
+ */
+#define	PACKET_TYPE0	0
+#define	PACKET_TYPE1	1
+#define	PACKET_TYPE2	2
+#define	PACKET_TYPE3	3
+
+#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3)
+#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF)
+#define CP_PACKET0_GET_REG(h) ((h) & 0xFFFF)
+#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF)
+#define PACKET0(reg, n)	((PACKET_TYPE0 << 30) |				\
+			 ((reg) & 0xFFFF) |			\
+			 ((n) & 0x3FFF) << 16)
+#define CP_PACKET2			0x80000000
+#define		PACKET2_PAD_SHIFT		0
+#define		PACKET2_PAD_MASK		(0x3fffffff << 0)
+
+#define PACKET2(v)	(CP_PACKET2 | REG_SET(PACKET2_PAD, (v)))
+
+#define PACKET3(op, n)	((PACKET_TYPE3 << 30) |				\
+			 (((op) & 0xFF) << 8) |				\
+			 ((n) & 0x3FFF) << 16)
+
+#define PACKET3_COMPUTE(op, n) (PACKET3(op, n) | 1 << 1)
+
+#define	PACKETJ_CONDITION_CHECK0	0
+#define	PACKETJ_CONDITION_CHECK1	1
+#define	PACKETJ_CONDITION_CHECK2	2
+#define	PACKETJ_CONDITION_CHECK3	3
+#define	PACKETJ_CONDITION_CHECK4	4
+#define	PACKETJ_CONDITION_CHECK5	5
+#define	PACKETJ_CONDITION_CHECK6	6
+#define	PACKETJ_CONDITION_CHECK7	7
+
+#define	PACKETJ_TYPE0	0
+#define	PACKETJ_TYPE1	1
+#define	PACKETJ_TYPE2	2
+#define	PACKETJ_TYPE3	3
+#define	PACKETJ_TYPE4	4
+#define	PACKETJ_TYPE5	5
+#define	PACKETJ_TYPE6	6
+#define	PACKETJ_TYPE7	7
+
+#define PACKETJ(reg, r, cond, type)	((reg & 0x3FFFF) |			\
+			 ((r & 0x3F) << 18) |			\
+			 ((cond & 0xF) << 24) |				\
+			 ((type & 0xF) << 28))
+
+/* Packet 3 types */
+#define	PACKET3_NOP					0x10
+#define	PACKET3_SET_BASE				0x11
+#define		PACKET3_BASE_INDEX(x)                  ((x) << 0)
+#define			CE_PARTITION_BASE		3
+#define	PACKET3_CLEAR_STATE				0x12
+#define	PACKET3_INDEX_BUFFER_SIZE			0x13
+#define	PACKET3_DISPATCH_DIRECT				0x15
+#define	PACKET3_DISPATCH_INDIRECT			0x16
+#define	PACKET3_ATOMIC_GDS				0x1D
+#define	PACKET3_ATOMIC_MEM				0x1E
+#define	PACKET3_OCCLUSION_QUERY				0x1F
+#define	PACKET3_SET_PREDICATION				0x20
+#define	PACKET3_REG_RMW					0x21
+#define	PACKET3_COND_EXEC				0x22
+#define	PACKET3_PRED_EXEC				0x23
+#define	PACKET3_DRAW_INDIRECT				0x24
+#define	PACKET3_DRAW_INDEX_INDIRECT			0x25
+#define	PACKET3_INDEX_BASE				0x26
+#define	PACKET3_DRAW_INDEX_2				0x27
+#define	PACKET3_CONTEXT_CONTROL				0x28
+#define	PACKET3_INDEX_TYPE				0x2A
+#define	PACKET3_DRAW_INDIRECT_MULTI			0x2C
+#define	PACKET3_DRAW_INDEX_AUTO				0x2D
+#define	PACKET3_NUM_INSTANCES				0x2F
+#define	PACKET3_DRAW_INDEX_MULTI_AUTO			0x30
+#define	PACKET3_INDIRECT_BUFFER_CONST			0x33
+#define	PACKET3_STRMOUT_BUFFER_UPDATE			0x34
+#define	PACKET3_DRAW_INDEX_OFFSET_2			0x35
+#define	PACKET3_DRAW_PREAMBLE				0x36
+#define	PACKET3_WRITE_DATA				0x37
+#define		WRITE_DATA_DST_SEL(x)                   ((x) << 8)
+		/* 0 - register
+		 * 1 - memory (sync - via GRBM)
+		 * 2 - gl2
+		 * 3 - gds
+		 * 4 - reserved
+		 * 5 - memory (async - direct)
+		 */
+#define		WR_ONE_ADDR                             (1 << 16)
+#define		WR_CONFIRM                              (1 << 20)
+#define		WRITE_DATA_CACHE_POLICY(x)              ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 */
+#define		WRITE_DATA_ENGINE_SEL(x)                ((x) << 30)
+		/* 0 - me
+		 * 1 - pfp
+		 * 2 - ce
+		 */
+#define	PACKET3_DRAW_INDEX_INDIRECT_MULTI		0x38
+#define	PACKET3_MEM_SEMAPHORE				0x39
+#              define PACKET3_SEM_USE_MAILBOX       (0x1 << 16)
+#              define PACKET3_SEM_SEL_SIGNAL_TYPE   (0x1 << 20) /* 0 = increment, 1 = write 1 */
+#              define PACKET3_SEM_SEL_SIGNAL	    (0x6 << 29)
+#              define PACKET3_SEM_SEL_WAIT	    (0x7 << 29)
+#define	PACKET3_WAIT_REG_MEM				0x3C
+#define		WAIT_REG_MEM_FUNCTION(x)                ((x) << 0)
+		/* 0 - always
+		 * 1 - <
+		 * 2 - <=
+		 * 3 - ==
+		 * 4 - !=
+		 * 5 - >=
+		 * 6 - >
+		 */
+#define		WAIT_REG_MEM_MEM_SPACE(x)               ((x) << 4)
+		/* 0 - reg
+		 * 1 - mem
+		 */
+#define		WAIT_REG_MEM_OPERATION(x)               ((x) << 6)
+		/* 0 - wait_reg_mem
+		 * 1 - wr_wait_wr_reg
+		 */
+#define		WAIT_REG_MEM_ENGINE(x)                  ((x) << 8)
+		/* 0 - me
+		 * 1 - pfp
+		 */
+#define	PACKET3_INDIRECT_BUFFER				0x3F
+#define		INDIRECT_BUFFER_VALID                   (1 << 23)
+#define		INDIRECT_BUFFER_CACHE_POLICY(x)         ((x) << 28)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#define     INDIRECT_BUFFER_PRE_ENB(x)		 ((x) << 21)
+#define	PACKET3_COPY_DATA				0x40
+#define	PACKET3_PFP_SYNC_ME				0x42
+#define	PACKET3_COND_WRITE				0x45
+#define	PACKET3_EVENT_WRITE				0x46
+#define		EVENT_TYPE(x)                           ((x) << 0)
+#define		EVENT_INDEX(x)                          ((x) << 8)
+		/* 0 - any non-TS event
+		 * 1 - ZPASS_DONE, PIXEL_PIPE_STAT_*
+		 * 2 - SAMPLE_PIPELINESTAT
+		 * 3 - SAMPLE_STREAMOUTSTAT*
+		 * 4 - *S_PARTIAL_FLUSH
+		 */
+#define	PACKET3_RELEASE_MEM				0x49
+#define		EVENT_TYPE(x)                           ((x) << 0)
+#define		EVENT_INDEX(x)                          ((x) << 8)
+#define		EOP_TCL1_VOL_ACTION_EN                  (1 << 12)
+#define		EOP_TC_VOL_ACTION_EN                    (1 << 13) /* L2 */
+#define		EOP_TC_WB_ACTION_EN                     (1 << 15) /* L2 */
+#define		EOP_TCL1_ACTION_EN                      (1 << 16)
+#define		EOP_TC_ACTION_EN                        (1 << 17) /* L2 */
+#define		EOP_TC_NC_ACTION_EN			(1 << 19)
+#define		EOP_TC_MD_ACTION_EN			(1 << 21) /* L2 metadata */
+
+#define		DATA_SEL(x)                             ((x) << 29)
+		/* 0 - discard
+		 * 1 - send low 32bit data
+		 * 2 - send 64bit data
+		 * 3 - send 64bit GPU counter value
+		 * 4 - send 64bit sys counter value
+		 */
+#define		INT_SEL(x)                              ((x) << 24)
+		/* 0 - none
+		 * 1 - interrupt only (DATA_SEL = 0)
+		 * 2 - interrupt when data write is confirmed
+		 */
+#define		DST_SEL(x)                              ((x) << 16)
+		/* 0 - MC
+		 * 1 - TC/L2
+		 */
+
+
+
+#define	PACKET3_PREAMBLE_CNTL				0x4A
+#              define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE     (2 << 28)
+#              define PACKET3_PREAMBLE_END_CLEAR_STATE       (3 << 28)
+#define	PACKET3_DMA_DATA				0x50
+/* 1. header
+ * 2. CONTROL
+ * 3. SRC_ADDR_LO or DATA [31:0]
+ * 4. SRC_ADDR_HI [31:0]
+ * 5. DST_ADDR_LO [31:0]
+ * 6. DST_ADDR_HI [7:0]
+ * 7. COMMAND [30:21] | BYTE_COUNT [20:0]
+ */
+/* CONTROL */
+#              define PACKET3_DMA_DATA_ENGINE(x)     ((x) << 0)
+		/* 0 - ME
+		 * 1 - PFP
+		 */
+#              define PACKET3_DMA_DATA_SRC_CACHE_POLICY(x) ((x) << 13)
+		/* 0 - LRU
+		 * 1 - Stream
+		 */
+#              define PACKET3_DMA_DATA_DST_SEL(x)  ((x) << 20)
+		/* 0 - DST_ADDR using DAS
+		 * 1 - GDS
+		 * 3 - DST_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_DST_CACHE_POLICY(x) ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 */
+#              define PACKET3_DMA_DATA_SRC_SEL(x)  ((x) << 29)
+		/* 0 - SRC_ADDR using SAS
+		 * 1 - GDS
+		 * 2 - DATA
+		 * 3 - SRC_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_CP_SYNC     (1 << 31)
+/* COMMAND */
+#              define PACKET3_DMA_DATA_CMD_SAS     (1 << 26)
+		/* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_DAS     (1 << 27)
+		/* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_SAIC    (1 << 28)
+#              define PACKET3_DMA_DATA_CMD_DAIC    (1 << 29)
+#              define PACKET3_DMA_DATA_CMD_RAW_WAIT  (1 << 30)
+#define	PACKET3_AQUIRE_MEM				0x58
+#define	PACKET3_REWIND					0x59
+#define	PACKET3_LOAD_UCONFIG_REG			0x5E
+#define	PACKET3_LOAD_SH_REG				0x5F
+#define	PACKET3_LOAD_CONFIG_REG				0x60
+#define	PACKET3_LOAD_CONTEXT_REG			0x61
+#define	PACKET3_SET_CONFIG_REG				0x68
+#define		PACKET3_SET_CONFIG_REG_START			0x00002000
+#define		PACKET3_SET_CONFIG_REG_END			0x00002c00
+#define	PACKET3_SET_CONTEXT_REG				0x69
+#define		PACKET3_SET_CONTEXT_REG_START			0x0000a000
+#define		PACKET3_SET_CONTEXT_REG_END			0x0000a400
+#define	PACKET3_SET_CONTEXT_REG_INDIRECT		0x73
+#define	PACKET3_SET_SH_REG				0x76
+#define		PACKET3_SET_SH_REG_START			0x00002c00
+#define		PACKET3_SET_SH_REG_END				0x00003000
+#define	PACKET3_SET_SH_REG_OFFSET			0x77
+#define	PACKET3_SET_QUEUE_REG				0x78
+#define	PACKET3_SET_UCONFIG_REG				0x79
+#define		PACKET3_SET_UCONFIG_REG_START			0x0000c000
+#define		PACKET3_SET_UCONFIG_REG_END			0x0000c400
+#define		PACKET3_SET_UCONFIG_REG_INDEX_TYPE		(2 << 28)
+#define	PACKET3_SCRATCH_RAM_WRITE			0x7D
+#define	PACKET3_SCRATCH_RAM_READ			0x7E
+#define	PACKET3_LOAD_CONST_RAM				0x80
+#define	PACKET3_WRITE_CONST_RAM				0x81
+#define	PACKET3_DUMP_CONST_RAM				0x83
+#define	PACKET3_INCREMENT_CE_COUNTER			0x84
+#define	PACKET3_INCREMENT_DE_COUNTER			0x85
+#define	PACKET3_WAIT_ON_CE_COUNTER			0x86
+#define	PACKET3_WAIT_ON_DE_COUNTER_DIFF			0x88
+#define	PACKET3_SWITCH_BUFFER				0x8B
+#define PACKET3_FRAME_CONTROL				0x90
+#			define FRAME_CMD(x) ((x) << 28)
+			/*
+			 * x=0: tmz_begin
+			 * x=1: tmz_end
+			 */
+
+#define	PACKET3_INVALIDATE_TLBS				0x98
+#              define PACKET3_INVALIDATE_TLBS_DST_SEL(x)     ((x) << 0)
+#              define PACKET3_INVALIDATE_TLBS_ALL_HUB(x)     ((x) << 4)
+#              define PACKET3_INVALIDATE_TLBS_PASID(x)       ((x) << 5)
+#              define PACKET3_INVALIDATE_TLBS_FLUSH_TYPE(x)  ((x) << 29)
+#define PACKET3_SET_RESOURCES				0xA0
+/* 1. header
+ * 2. CONTROL
+ * 3. QUEUE_MASK_LO [31:0]
+ * 4. QUEUE_MASK_HI [31:0]
+ * 5. GWS_MASK_LO [31:0]
+ * 6. GWS_MASK_HI [31:0]
+ * 7. OAC_MASK [15:0]
+ * 8. GDS_HEAP_SIZE [16:11] | GDS_HEAP_BASE [5:0]
+ */
+#              define PACKET3_SET_RESOURCES_VMID_MASK(x)     ((x) << 0)
+#              define PACKET3_SET_RESOURCES_UNMAP_LATENTY(x) ((x) << 16)
+#              define PACKET3_SET_RESOURCES_QUEUE_TYPE(x)    ((x) << 29)
+#define PACKET3_MAP_QUEUES				0xA2
+/* 1. header
+ * 2. CONTROL
+ * 3. CONTROL2
+ * 4. MQD_ADDR_LO [31:0]
+ * 5. MQD_ADDR_HI [31:0]
+ * 6. WPTR_ADDR_LO [31:0]
+ * 7. WPTR_ADDR_HI [31:0]
+ */
+/* CONTROL */
+#              define PACKET3_MAP_QUEUES_QUEUE_SEL(x)       ((x) << 4)
+#              define PACKET3_MAP_QUEUES_VMID(x)            ((x) << 8)
+#              define PACKET3_MAP_QUEUES_QUEUE(x)           ((x) << 13)
+#              define PACKET3_MAP_QUEUES_PIPE(x)            ((x) << 16)
+#              define PACKET3_MAP_QUEUES_ME(x)              ((x) << 18)
+#              define PACKET3_MAP_QUEUES_QUEUE_TYPE(x)      ((x) << 21)
+#              define PACKET3_MAP_QUEUES_ALLOC_FORMAT(x)    ((x) << 24)
+#              define PACKET3_MAP_QUEUES_ENGINE_SEL(x)      ((x) << 26)
+#              define PACKET3_MAP_QUEUES_NUM_QUEUES(x)      ((x) << 29)
+/* CONTROL2 */
+#              define PACKET3_MAP_QUEUES_CHECK_DISABLE(x)   ((x) << 1)
+#              define PACKET3_MAP_QUEUES_DOORBELL_OFFSET(x) ((x) << 2)
+#define	PACKET3_UNMAP_QUEUES				0xA3
+/* 1. header
+ * 2. CONTROL
+ * 3. CONTROL2
+ * 4. CONTROL3
+ * 5. CONTROL4
+ * 6. CONTROL5
+ */
+/* CONTROL */
+#              define PACKET3_UNMAP_QUEUES_ACTION(x)           ((x) << 0)
+		/* 0 - PREEMPT_QUEUES
+		 * 1 - RESET_QUEUES
+		 * 2 - DISABLE_PROCESS_QUEUES
+		 * 3 - PREEMPT_QUEUES_NO_UNMAP
+		 */
+#              define PACKET3_UNMAP_QUEUES_QUEUE_SEL(x)        ((x) << 4)
+#              define PACKET3_UNMAP_QUEUES_ENGINE_SEL(x)       ((x) << 26)
+#              define PACKET3_UNMAP_QUEUES_NUM_QUEUES(x)       ((x) << 29)
+/* CONTROL2a */
+#              define PACKET3_UNMAP_QUEUES_PASID(x)            ((x) << 0)
+/* CONTROL2b */
+#              define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(x) ((x) << 2)
+/* CONTROL3a */
+#              define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET1(x) ((x) << 2)
+/* CONTROL3b */
+#              define PACKET3_UNMAP_QUEUES_RB_WPTR(x)          ((x) << 0)
+/* CONTROL4 */
+#              define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET2(x) ((x) << 2)
+/* CONTROL5 */
+#              define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET3(x) ((x) << 2)
+#define	PACKET3_QUERY_STATUS				0xA4
+/* 1. header
+ * 2. CONTROL
+ * 3. CONTROL2
+ * 4. ADDR_LO [31:0]
+ * 5. ADDR_HI [31:0]
+ * 6. DATA_LO [31:0]
+ * 7. DATA_HI [31:0]
+ */
+/* CONTROL */
+#              define PACKET3_QUERY_STATUS_CONTEXT_ID(x)       ((x) << 0)
+#              define PACKET3_QUERY_STATUS_INTERRUPT_SEL(x)    ((x) << 28)
+#              define PACKET3_QUERY_STATUS_COMMAND(x)          ((x) << 30)
+/* CONTROL2a */
+#              define PACKET3_QUERY_STATUS_PASID(x)            ((x) << 0)
+/* CONTROL2b */
+#              define PACKET3_QUERY_STATUS_DOORBELL_OFFSET(x)  ((x) << 2)
+#              define PACKET3_QUERY_STATUS_ENG_SEL(x)          ((x) << 25)
+
+
+#define VCE_CMD_NO_OP		0x00000000
+#define VCE_CMD_END		0x00000001
+#define VCE_CMD_IB		0x00000002
+#define VCE_CMD_FENCE		0x00000003
+#define VCE_CMD_TRAP		0x00000004
+#define VCE_CMD_IB_AUTO 	0x00000005
+#define VCE_CMD_SEMAPHORE	0x00000006
+
+#define VCE_CMD_IB_VM           0x00000102
+#define VCE_CMD_WAIT_GE         0x00000106
+#define VCE_CMD_UPDATE_PTB      0x00000107
+#define VCE_CMD_FLUSH_TLB       0x00000108
+#define VCE_CMD_REG_WRITE       0x00000109
+#define VCE_CMD_REG_WAIT        0x0000010a
+
+#define HEVC_ENC_CMD_NO_OP		0x00000000
+#define HEVC_ENC_CMD_END		0x00000001
+#define HEVC_ENC_CMD_FENCE		0x00000003
+#define HEVC_ENC_CMD_TRAP		0x00000004
+#define HEVC_ENC_CMD_IB_VM		0x00000102
+#define HEVC_ENC_CMD_REG_WRITE		0x00000109
+#define HEVC_ENC_CMD_REG_WAIT		0x0000010a
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/tonga_ih.c b/sys/dev/pci/drm/amd/amdgpu/tonga_ih.c
new file mode 100644
index 00000000000..52853d8a8fd
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/tonga_ih.c
@@ -0,0 +1,527 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "vid.h"
+
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+
+#include "bif/bif_5_1_d.h"
+#include "bif/bif_5_1_sh_mask.h"
+
+/*
+ * Interrupts
+ * Starting with r6xx, interrupts are handled via a ring buffer.
+ * Ring buffers are areas of GPU accessible memory that the GPU
+ * writes interrupt vectors into and the host reads vectors out of.
+ * There is a rptr (read pointer) that determines where the
+ * host is currently reading, and a wptr (write pointer)
+ * which determines where the GPU has written.  When the
+ * pointers are equal, the ring is idle.  When the GPU
+ * writes vectors to the ring buffer, it increments the
+ * wptr.  When there is an interrupt, the host then starts
+ * fetching commands and processing them until the pointers are
+ * equal again at which point it updates the rptr.
+ */
+
+static void tonga_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * tonga_ih_enable_interrupts - Enable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enable the interrupt ring buffer (VI).
+ */
+static void tonga_ih_enable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 1);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	adev->irq.ih.enabled = true;
+}
+
+/**
+ * tonga_ih_disable_interrupts - Disable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable the interrupt ring buffer (VI).
+ */
+static void tonga_ih_disable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 0);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 0);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+	adev->irq.ih.enabled = false;
+	adev->irq.ih.rptr = 0;
+}
+
+/**
+ * tonga_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (VI).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int tonga_ih_irq_init(struct amdgpu_device *adev)
+{
+	int rb_bufsz;
+	u32 interrupt_cntl, ih_rb_cntl, ih_doorbell_rtpr;
+	u64 wptr_off;
+
+	/* disable irqs */
+	tonga_ih_disable_interrupts(adev);
+
+	/* setup interrupt control */
+	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
+	interrupt_cntl = RREG32(mmINTERRUPT_CNTL);
+	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_DUMMY_RD_OVERRIDE, 0);
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_REQ_NONSNOOP_EN, 0);
+	WREG32(mmINTERRUPT_CNTL, interrupt_cntl);
+
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	if (adev->irq.ih.use_bus_addr)
+		WREG32(mmIH_RB_BASE, adev->irq.ih.rb_dma_addr >> 8);
+	else
+		WREG32(mmIH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
+
+	rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
+	ih_rb_cntl = REG_SET_FIELD(0, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register value is written to memory */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
+
+	if (adev->irq.msi_enabled)
+		ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RPTR_REARM, 1);
+
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+
+	/* set the writeback address whether it's enabled or not */
+	if (adev->irq.ih.use_bus_addr)
+		wptr_off = adev->irq.ih.rb_dma_addr + (adev->irq.ih.wptr_offs * 4);
+	else
+		wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
+	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
+	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+
+	ih_doorbell_rtpr = RREG32(mmIH_DOORBELL_RPTR);
+	if (adev->irq.ih.use_doorbell) {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
+						 OFFSET, adev->irq.ih.doorbell_index);
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
+						 ENABLE, 1);
+	} else {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
+						 ENABLE, 0);
+	}
+	WREG32(mmIH_DOORBELL_RPTR, ih_doorbell_rtpr);
+
+	pci_set_master(adev->pdev);
+
+	/* enable interrupts */
+	tonga_ih_enable_interrupts(adev);
+
+	return 0;
+}
+
+/**
+ * tonga_ih_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw (VI).
+ */
+static void tonga_ih_irq_disable(struct amdgpu_device *adev)
+{
+	tonga_ih_disable_interrupts(adev);
+
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * tonga_ih_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer (VI).  Also check for
+ * ring buffer overflow and deal with it.
+ * Used by cz_irq_process(VI).
+ * Returns the value of the wptr.
+ */
+static u32 tonga_ih_get_wptr(struct amdgpu_device *adev)
+{
+	u32 wptr, tmp;
+
+	if (adev->irq.ih.use_bus_addr)
+		wptr = le32_to_cpu(adev->irq.ih.ring[adev->irq.ih.wptr_offs]);
+	else
+		wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
+
+	if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
+		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+		/* When a ring buffer overflow happen start parsing interrupt
+		 * from the last not overwritten vector (wptr + 16). Hopefully
+		 * this should allow us to catchup.
+		 */
+		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
+		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
+		tmp = RREG32(mmIH_RB_CNTL);
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+		WREG32(mmIH_RB_CNTL, tmp);
+	}
+	return (wptr & adev->irq.ih.ptr_mask);
+}
+
+/**
+ * tonga_ih_prescreen_iv - prescreen an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns true if the interrupt vector should be further processed.
+ */
+static bool tonga_ih_prescreen_iv(struct amdgpu_device *adev)
+{
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	u16 pasid;
+
+	switch (le32_to_cpu(adev->irq.ih.ring[ring_index]) & 0xff) {
+	case 146:
+	case 147:
+		pasid = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]) >> 16;
+		if (!pasid || amdgpu_vm_pasid_fault_credit(adev, pasid))
+			return true;
+		break;
+	default:
+		/* Not a VM fault */
+		return true;
+	}
+
+	adev->irq.ih.rptr += 16;
+	return false;
+}
+
+/**
+ * tonga_ih_decode_iv - decode an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Decodes the interrupt vector at the current rptr
+ * position and also advance the position.
+ */
+static void tonga_ih_decode_iv(struct amdgpu_device *adev,
+				 struct amdgpu_iv_entry *entry)
+{
+	/* wptr/rptr are in bytes! */
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	uint32_t dw[4];
+
+	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+
+	entry->client_id = AMDGPU_IH_CLIENTID_LEGACY;
+	entry->src_id = dw[0] & 0xff;
+	entry->src_data[0] = dw[1] & 0xfffffff;
+	entry->ring_id = dw[2] & 0xff;
+	entry->vmid = (dw[2] >> 8) & 0xff;
+	entry->pasid = (dw[2] >> 16) & 0xffff;
+
+	/* wptr/rptr are in bytes! */
+	adev->irq.ih.rptr += 16;
+}
+
+/**
+ * tonga_ih_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void tonga_ih_set_rptr(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih.use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		if (adev->irq.ih.use_bus_addr)
+			adev->irq.ih.ring[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
+		else
+			adev->wb.wb[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
+		WDOORBELL32(adev->irq.ih.doorbell_index, adev->irq.ih.rptr);
+	} else {
+		WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
+	}
+}
+
+static int tonga_ih_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret;
+
+	ret = amdgpu_irq_add_domain(adev);
+	if (ret)
+		return ret;
+
+	tonga_ih_set_interrupt_funcs(adev);
+
+	return 0;
+}
+
+static int tonga_ih_sw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_ih_ring_init(adev, 64 * 1024, true);
+	if (r)
+		return r;
+
+	adev->irq.ih.use_doorbell = true;
+	adev->irq.ih.doorbell_index = AMDGPU_DOORBELL_IH;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int tonga_ih_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_fini(adev);
+	amdgpu_ih_ring_fini(adev);
+	amdgpu_irq_remove_domain(adev);
+
+	return 0;
+}
+
+static int tonga_ih_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = tonga_ih_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int tonga_ih_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	tonga_ih_irq_disable(adev);
+
+	return 0;
+}
+
+static int tonga_ih_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return tonga_ih_hw_fini(adev);
+}
+
+static int tonga_ih_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return tonga_ih_hw_init(adev);
+}
+
+static bool tonga_ih_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
+		return false;
+
+	return true;
+}
+
+static int tonga_ih_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS);
+		if (!REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static bool tonga_ih_check_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset = 0;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET,
+						SOFT_RESET_IH, 1);
+
+	if (srbm_soft_reset) {
+		adev->irq.srbm_soft_reset = srbm_soft_reset;
+		return true;
+	} else {
+		adev->irq.srbm_soft_reset = 0;
+		return false;
+	}
+}
+
+static int tonga_ih_pre_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->irq.srbm_soft_reset)
+		return 0;
+
+	return tonga_ih_hw_fini(adev);
+}
+
+static int tonga_ih_post_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->irq.srbm_soft_reset)
+		return 0;
+
+	return tonga_ih_hw_init(adev);
+}
+
+static int tonga_ih_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset;
+
+	if (!adev->irq.srbm_soft_reset)
+		return 0;
+	srbm_soft_reset = adev->irq.srbm_soft_reset;
+
+	if (srbm_soft_reset) {
+		u32 tmp;
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int tonga_ih_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int tonga_ih_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static const struct amd_ip_funcs tonga_ih_ip_funcs = {
+	.name = "tonga_ih",
+	.early_init = tonga_ih_early_init,
+	.late_init = NULL,
+	.sw_init = tonga_ih_sw_init,
+	.sw_fini = tonga_ih_sw_fini,
+	.hw_init = tonga_ih_hw_init,
+	.hw_fini = tonga_ih_hw_fini,
+	.suspend = tonga_ih_suspend,
+	.resume = tonga_ih_resume,
+	.is_idle = tonga_ih_is_idle,
+	.wait_for_idle = tonga_ih_wait_for_idle,
+	.check_soft_reset = tonga_ih_check_soft_reset,
+	.pre_soft_reset = tonga_ih_pre_soft_reset,
+	.soft_reset = tonga_ih_soft_reset,
+	.post_soft_reset = tonga_ih_post_soft_reset,
+	.set_clockgating_state = tonga_ih_set_clockgating_state,
+	.set_powergating_state = tonga_ih_set_powergating_state,
+};
+
+static const struct amdgpu_ih_funcs tonga_ih_funcs = {
+	.get_wptr = tonga_ih_get_wptr,
+	.prescreen_iv = tonga_ih_prescreen_iv,
+	.decode_iv = tonga_ih_decode_iv,
+	.set_rptr = tonga_ih_set_rptr
+};
+
+static void tonga_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih_funcs == NULL)
+		adev->irq.ih_funcs = &tonga_ih_funcs;
+}
+
+const struct amdgpu_ip_block_version tonga_ih_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_IH,
+	.major = 3,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &tonga_ih_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/tonga_ih.h b/sys/dev/pci/drm/amd/amdgpu/tonga_ih.h
new file mode 100644
index 00000000000..499027eee5c
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/tonga_ih.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __TONGA_IH_H__
+#define __TONGA_IH_H__
+
+extern const struct amdgpu_ip_block_version tonga_ih_ip_block;
+
+#endif /* __TONGA_IH_H__ */
diff --git a/sys/dev/pci/drm/amd/amdgpu/tonga_sdma_pkt_open.h b/sys/dev/pci/drm/amd/amdgpu/tonga_sdma_pkt_open.h
new file mode 100644
index 00000000000..e5ebd084288
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/tonga_sdma_pkt_open.h
@@ -0,0 +1,2245 @@
+/*
+ * Copyright (C) 2014  Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __TONGA_SDMA_PKT_OPEN_H_
+#define __TONGA_SDMA_PKT_OPEN_H_
+
+#define SDMA_OP_NOP  0
+#define SDMA_OP_COPY  1
+#define SDMA_OP_WRITE  2
+#define SDMA_OP_INDIRECT  4
+#define SDMA_OP_FENCE  5
+#define SDMA_OP_TRAP  6
+#define SDMA_OP_SEM  7
+#define SDMA_OP_POLL_REGMEM  8
+#define SDMA_OP_COND_EXE  9
+#define SDMA_OP_ATOMIC  10
+#define SDMA_OP_CONST_FILL  11
+#define SDMA_OP_GEN_PTEPDE  12
+#define SDMA_OP_TIMESTAMP  13
+#define SDMA_OP_SRBM_WRITE  14
+#define SDMA_OP_PRE_EXE  15
+#define SDMA_SUBOP_TIMESTAMP_SET  0
+#define SDMA_SUBOP_TIMESTAMP_GET  1
+#define SDMA_SUBOP_TIMESTAMP_GET_GLOBAL  2
+#define SDMA_SUBOP_COPY_LINEAR  0
+#define SDMA_SUBOP_COPY_LINEAR_SUB_WIND  4
+#define SDMA_SUBOP_COPY_TILED  1
+#define SDMA_SUBOP_COPY_TILED_SUB_WIND  5
+#define SDMA_SUBOP_COPY_T2T_SUB_WIND  6
+#define SDMA_SUBOP_COPY_SOA  3
+#define SDMA_SUBOP_WRITE_LINEAR  0
+#define SDMA_SUBOP_WRITE_TILED  1
+
+/*define for op field*/
+#define SDMA_PKT_HEADER_op_offset 0
+#define SDMA_PKT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_HEADER_op_shift  0
+#define SDMA_PKT_HEADER_OP(x) (((x) & SDMA_PKT_HEADER_op_mask) << SDMA_PKT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_HEADER_sub_op_offset 0
+#define SDMA_PKT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_HEADER_sub_op_shift  8
+#define SDMA_PKT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_HEADER_sub_op_mask) << SDMA_PKT_HEADER_sub_op_shift)
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for dst_ha field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_shift  22
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_DST_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_shift  30
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_SRC_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_BROADCAST_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST2_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift)
+
+/*define for dst2_ha field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_shift  14
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST2_HA(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_shift)
+
+/*define for dst1_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift  16
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST1_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift)
+
+/*define for dst1_ha field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_shift  22
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST1_HA(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_shift  30
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_SRC_HA(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST1_ADDR_LO word*/
+/*define for dst1_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_offset 5
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_DST1_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift)
+
+/*define for DST1_ADDR_HI word*/
+/*define for dst1_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_offset 6
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_DST1_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift)
+
+/*define for DST2_ADDR_LO word*/
+/*define for dst2_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_offset 7
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_DST2_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift)
+
+/*define for DST2_ADDR_HI word*/
+/*define for dst2_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_offset 8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_DST2_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift)
+
+/*define for elementsize field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift  29
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_ELEMENTSIZE(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift)
+
+/*define for src_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_offset 6
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_offset 7
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_8 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift)
+
+/*define for DW_9 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift)
+
+/*define for dst_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift)
+
+/*define for DW_10 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_offset 10
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift)
+
+/*define for DW_12 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_RECT_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift)
+
+/*define for dst_ha field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_shift  22
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_DST_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_shift  30
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_SRC_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_op_mask) << SDMA_PKT_COPY_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_offset 3
+#define SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_shift  0
+#define SDMA_PKT_COPY_TILED_DW_3_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_mask) << SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_shift)
+
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_DW_3_height_offset 3
+#define SDMA_PKT_COPY_TILED_DW_3_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_3_height_shift  16
+#define SDMA_PKT_COPY_TILED_DW_3_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_DW_3_height_mask) << SDMA_PKT_COPY_TILED_DW_3_height_shift)
+
+/*define for DW_4 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_DW_4_slice_pitch_offset 4
+#define SDMA_PKT_COPY_TILED_DW_4_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_TILED_DW_4_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_DW_4_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_DW_4_slice_pitch_mask) << SDMA_PKT_COPY_TILED_DW_4_slice_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_element_size_mask) << SDMA_PKT_COPY_TILED_DW_5_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_TILED_DW_5_array_mode_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_DW_5_array_mode_shift  3
+#define SDMA_PKT_COPY_TILED_DW_5_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_array_mode_mask) << SDMA_PKT_COPY_TILED_DW_5_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_TILED_DW_5_mit_mode_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_mit_mode_shift  8
+#define SDMA_PKT_COPY_TILED_DW_5_MIT_MODE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_mit_mode_mask) << SDMA_PKT_COPY_TILED_DW_5_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_TILED_DW_5_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_mask) << SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_TILED_DW_5_bank_w_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_bank_w_shift  15
+#define SDMA_PKT_COPY_TILED_DW_5_BANK_W(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_bank_w_mask) << SDMA_PKT_COPY_TILED_DW_5_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_TILED_DW_5_bank_h_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_bank_h_shift  18
+#define SDMA_PKT_COPY_TILED_DW_5_BANK_H(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_bank_h_mask) << SDMA_PKT_COPY_TILED_DW_5_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_TILED_DW_5_num_bank_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_num_bank_shift  21
+#define SDMA_PKT_COPY_TILED_DW_5_NUM_BANK(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_num_bank_mask) << SDMA_PKT_COPY_TILED_DW_5_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_TILED_DW_5_mat_aspt_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_mat_aspt_shift  24
+#define SDMA_PKT_COPY_TILED_DW_5_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_mat_aspt_mask) << SDMA_PKT_COPY_TILED_DW_5_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_TILED_DW_5_pipe_config_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_DW_5_pipe_config_shift  26
+#define SDMA_PKT_COPY_TILED_DW_5_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_pipe_config_mask) << SDMA_PKT_COPY_TILED_DW_5_pipe_config_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_TILED_DW_6_x_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_x_shift  0
+#define SDMA_PKT_COPY_TILED_DW_6_X(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_x_mask) << SDMA_PKT_COPY_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_TILED_DW_6_y_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_y_shift  16
+#define SDMA_PKT_COPY_TILED_DW_6_Y(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_y_mask) << SDMA_PKT_COPY_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_TILED_DW_7_z_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_z_mask   0x00000FFF
+#define SDMA_PKT_COPY_TILED_DW_7_z_shift  0
+#define SDMA_PKT_COPY_TILED_DW_7_Z(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_z_mask) << SDMA_PKT_COPY_TILED_DW_7_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_DW_7_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_DW_7_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_offset 8
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_offset 9
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_TILED_COUNT_count_offset 11
+#define SDMA_PKT_COPY_TILED_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_COPY_TILED_COUNT_count_shift  0
+#define SDMA_PKT_COPY_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_TILED_COUNT_count_mask) << SDMA_PKT_COPY_TILED_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_L2T_BROADCAST packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift)
+
+/*define for videocopy field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift  26
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_VIDEOCOPY(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift)
+
+/*define for TILED_ADDR_LO_0 word*/
+/*define for tiled_addr0_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_offset 1
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_TILED_ADDR0_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift)
+
+/*define for TILED_ADDR_HI_0 word*/
+/*define for tiled_addr0_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_offset 2
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_TILED_ADDR0_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift)
+
+/*define for TILED_ADDR_LO_1 word*/
+/*define for tiled_addr1_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_offset 3
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_TILED_ADDR1_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift)
+
+/*define for TILED_ADDR_HI_1 word*/
+/*define for tiled_addr1_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_offset 4
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_TILED_ADDR1_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift)
+
+/*define for DW_5 word*/
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_offset 5
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_mask   0x000007FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_shift)
+
+/*define for height field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_offset 5
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_HEIGHT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_shift)
+
+/*define for DW_6 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_offset 6
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_shift)
+
+/*define for DW_7 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_shift  3
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_MIT_MODE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_shift  15
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_BANK_W(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_shift  18
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_BANK_H(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_shift  21
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_NUM_BANK(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_shift  24
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_shift  26
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_shift)
+
+/*define for DW_8 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_X(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_Y(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift)
+
+/*define for DW_9 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_offset 9
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask   0x00000FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_Z(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift)
+
+/*define for DW_10 word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_DST2_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift)
+
+/*define for dst2_ha field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_shift  14
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_DST2_HA(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift  24
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_TILE_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_offset 11
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_offset 12
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_offset 13
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_offset 14
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask) << SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_T2T packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_op_shift  0
+#define SDMA_PKT_COPY_T2T_HEADER_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_op_mask) << SDMA_PKT_COPY_T2T_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_T2T_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_sub_op_mask) << SDMA_PKT_COPY_T2T_HEADER_sub_op_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_x_mask) << SDMA_PKT_COPY_T2T_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_y_mask) << SDMA_PKT_COPY_T2T_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_z_mask) << SDMA_PKT_COPY_T2T_DW_4_src_z_shift)
+
+/*define for src_pitch_in_tile field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_mask   0x00000FFF
+#define SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_shift  16
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_mask) << SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_T2T_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_shift)
+
+/*define for DW_6 word*/
+/*define for src_element_size field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask) << SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift)
+
+/*define for src_array_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_array_mode_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_6_src_array_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_array_mode_mask) << SDMA_PKT_COPY_T2T_DW_6_src_array_mode_shift)
+
+/*define for src_mit_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_shift  8
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_MIT_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_mask) << SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_shift)
+
+/*define for src_tilesplit_size field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_mask) << SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_shift)
+
+/*define for src_bank_w field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_w_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_w_shift  15
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_BANK_W(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_bank_w_mask) << SDMA_PKT_COPY_T2T_DW_6_src_bank_w_shift)
+
+/*define for src_bank_h field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_h_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_h_shift  18
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_BANK_H(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_bank_h_mask) << SDMA_PKT_COPY_T2T_DW_6_src_bank_h_shift)
+
+/*define for src_num_bank field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_num_bank_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_num_bank_shift  21
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_NUM_BANK(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_num_bank_mask) << SDMA_PKT_COPY_T2T_DW_6_src_num_bank_shift)
+
+/*define for src_mat_aspt field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_shift  24
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_mask) << SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_shift)
+
+/*define for src_pipe_config field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_shift  26
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_mask) << SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_offset 7
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_offset 8
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_9_DST_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_x_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_9_DST_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_y_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_y_shift)
+
+/*define for DW_10 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_10_DST_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_z_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_z_shift)
+
+/*define for dst_pitch_in_tile field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_mask   0x00000FFF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_shift  16
+#define SDMA_PKT_COPY_T2T_DW_10_DST_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_shift)
+
+/*define for DW_11 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_offset 11
+#define SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_T2T_DW_11_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_mask) << SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for dst_array_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_12_DST_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_shift)
+
+/*define for dst_mit_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_shift  8
+#define SDMA_PKT_COPY_T2T_DW_12_DST_MIT_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_shift)
+
+/*define for dst_tilesplit_size field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_T2T_DW_12_DST_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_shift)
+
+/*define for dst_bank_w field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_shift  15
+#define SDMA_PKT_COPY_T2T_DW_12_DST_BANK_W(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_shift)
+
+/*define for dst_bank_h field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_shift  18
+#define SDMA_PKT_COPY_T2T_DW_12_DST_BANK_H(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_shift)
+
+/*define for dst_num_bank field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_shift  21
+#define SDMA_PKT_COPY_T2T_DW_12_DST_NUM_BANK(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_shift)
+
+/*define for dst_mat_aspt field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_shift  24
+#define SDMA_PKT_COPY_T2T_DW_12_DST_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_shift)
+
+/*define for dst_pipe_config field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_shift  26
+#define SDMA_PKT_COPY_T2T_DW_12_DST_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_shift)
+
+/*define for DW_13 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_x_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_y_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_y_shift)
+
+/*define for DW_14 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_14_RECT_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_rect_z_mask) << SDMA_PKT_COPY_T2T_DW_14_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift  16
+#define SDMA_PKT_COPY_T2T_DW_14_DST_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_shift  24
+#define SDMA_PKT_COPY_T2T_DW_14_SRC_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_src_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_src_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for tiled_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift)
+
+/*define for tiled_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift)
+
+/*define for DW_4 word*/
+/*define for tiled_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_TILED_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift)
+
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_mask   0x00000FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_shift)
+
+/*define for DW_5 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_shift)
+
+/*define for DW_6 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_shift  3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_MIT_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_shift  15
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_BANK_W(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_shift  18
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_BANK_H(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_shift  21
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_NUM_BANK(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_shift  26
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_offset 7
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_offset 8
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for linear_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift)
+
+/*define for linear_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift)
+
+/*define for DW_10 word*/
+/*define for linear_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift)
+
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift)
+
+/*define for DW_13 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_RECT_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_STRUCT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_shift  0
+#define SDMA_PKT_COPY_STRUCT_HEADER_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_STRUCT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_STRUCT_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_detile_mask) << SDMA_PKT_COPY_STRUCT_HEADER_detile_shift)
+
+/*define for SB_ADDR_LO word*/
+/*define for sb_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_offset 1
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_SB_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift)
+
+/*define for SB_ADDR_HI word*/
+/*define for sb_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_offset 2
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_SB_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift)
+
+/*define for START_INDEX word*/
+/*define for start_index field*/
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_offset 3
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift  0
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_START_INDEX(x) (((x) & SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask) << SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_offset 4
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_shift  0
+#define SDMA_PKT_COPY_STRUCT_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_STRUCT_COUNT_count_mask) << SDMA_PKT_COPY_STRUCT_COUNT_count_shift)
+
+/*define for DW_5 word*/
+/*define for stride field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_mask   0x000007FF
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_shift  0
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRIDE(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_stride_mask) << SDMA_PKT_COPY_STRUCT_DW_5_stride_shift)
+
+/*define for struct_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift  16
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRUCT_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift)
+
+/*define for struct_ha field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_shift  22
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRUCT_HA(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_mask) << SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift  24
+#define SDMA_PKT_COPY_STRUCT_DW_5_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift)
+
+/*define for linear_ha field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_shift  30
+#define SDMA_PKT_COPY_STRUCT_DW_5_LINEAR_HA(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_mask) << SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_offset 6
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_offset 7
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_UNTILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_UNTILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_mask   0x003FFFFF
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_count_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_count_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift  24
+#define SDMA_PKT_WRITE_UNTILED_DW_3_SW(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_offset 4
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask) << SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_TILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_offset 3
+#define SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_mask   0x000007FF
+#define SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_3_PITCH_IN_TILE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_mask) << SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_shift)
+
+/*define for height field*/
+#define SDMA_PKT_WRITE_TILED_DW_3_height_offset 3
+#define SDMA_PKT_WRITE_TILED_DW_3_height_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_3_height_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_3_HEIGHT(x) (((x) & SDMA_PKT_WRITE_TILED_DW_3_height_mask) << SDMA_PKT_WRITE_TILED_DW_3_height_shift)
+
+/*define for DW_4 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_offset 4
+#define SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_4_SLICE_PITCH(x) (((x) & SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_mask) << SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_element_size_mask) << SDMA_PKT_WRITE_TILED_DW_5_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_array_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_array_mode_mask   0x0000000F
+#define SDMA_PKT_WRITE_TILED_DW_5_array_mode_shift  3
+#define SDMA_PKT_WRITE_TILED_DW_5_ARRAY_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_array_mode_mask) << SDMA_PKT_WRITE_TILED_DW_5_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_mit_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_mit_mode_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_mit_mode_shift  8
+#define SDMA_PKT_WRITE_TILED_DW_5_MIT_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_mit_mode_mask) << SDMA_PKT_WRITE_TILED_DW_5_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_shift  11
+#define SDMA_PKT_WRITE_TILED_DW_5_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_mask) << SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_w_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_w_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_w_shift  15
+#define SDMA_PKT_WRITE_TILED_DW_5_BANK_W(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_bank_w_mask) << SDMA_PKT_WRITE_TILED_DW_5_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_h_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_h_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_h_shift  18
+#define SDMA_PKT_WRITE_TILED_DW_5_BANK_H(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_bank_h_mask) << SDMA_PKT_WRITE_TILED_DW_5_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_num_bank_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_num_bank_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_num_bank_shift  21
+#define SDMA_PKT_WRITE_TILED_DW_5_NUM_BANK(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_num_bank_mask) << SDMA_PKT_WRITE_TILED_DW_5_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_shift  24
+#define SDMA_PKT_WRITE_TILED_DW_5_MAT_ASPT(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_mask) << SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_pipe_config_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_pipe_config_mask   0x0000001F
+#define SDMA_PKT_WRITE_TILED_DW_5_pipe_config_shift  26
+#define SDMA_PKT_WRITE_TILED_DW_5_PIPE_CONFIG(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_pipe_config_mask) << SDMA_PKT_WRITE_TILED_DW_5_pipe_config_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_x_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_x_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_6_X(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_x_mask) << SDMA_PKT_WRITE_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_y_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_y_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_6_Y(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_y_mask) << SDMA_PKT_WRITE_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_z_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_z_mask   0x00000FFF
+#define SDMA_PKT_WRITE_TILED_DW_7_z_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_7_Z(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_z_mask) << SDMA_PKT_WRITE_TILED_DW_7_z_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_shift  24
+#define SDMA_PKT_WRITE_TILED_DW_7_SW(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_sw_mask) << SDMA_PKT_WRITE_TILED_DW_7_sw_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_TILED_COUNT_count_offset 8
+#define SDMA_PKT_WRITE_TILED_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_WRITE_TILED_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_TILED_COUNT_count_mask) << SDMA_PKT_WRITE_TILED_COUNT_count_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_offset 9
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_TILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_TILED_DATA0_data0_mask) << SDMA_PKT_WRITE_TILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_INCR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_INCR_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_INCR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_DW0 word*/
+/*define for mask_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_offset 3
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_MASK_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask) << SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift)
+
+/*define for MASK_DW1 word*/
+/*define for mask_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_offset 4
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_MASK_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask) << SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift)
+
+/*define for INIT_DW0 word*/
+/*define for init_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_offset 5
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_INIT_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask) << SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift)
+
+/*define for INIT_DW1 word*/
+/*define for init_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_offset 6
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_INIT_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask) << SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift)
+
+/*define for INCR_DW0 word*/
+/*define for incr_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_offset 7
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_INCR_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask) << SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift)
+
+/*define for INCR_DW1 word*/
+/*define for incr_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_offset 8
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_INCR_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask) << SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_INCR_COUNT_count_offset 9
+#define SDMA_PKT_WRITE_INCR_COUNT_count_mask   0x0007FFFF
+#define SDMA_PKT_WRITE_INCR_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_INCR_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_INCR_COUNT_count_mask) << SDMA_PKT_WRITE_INCR_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_INDIRECT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_INDIRECT_HEADER_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_op_shift  0
+#define SDMA_PKT_INDIRECT_HEADER_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_op_mask) << SDMA_PKT_INDIRECT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_shift  8
+#define SDMA_PKT_INDIRECT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_sub_op_mask) << SDMA_PKT_INDIRECT_HEADER_sub_op_shift)
+
+/*define for vmid field*/
+#define SDMA_PKT_INDIRECT_HEADER_vmid_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_vmid_mask   0x0000000F
+#define SDMA_PKT_INDIRECT_HEADER_vmid_shift  16
+#define SDMA_PKT_INDIRECT_HEADER_VMID(x) (((x) & SDMA_PKT_INDIRECT_HEADER_vmid_mask) << SDMA_PKT_INDIRECT_HEADER_vmid_shift)
+
+/*define for BASE_LO word*/
+/*define for ib_base_31_0 field*/
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_offset 1
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift  0
+#define SDMA_PKT_INDIRECT_BASE_LO_IB_BASE_31_0(x) (((x) & SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask) << SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift)
+
+/*define for BASE_HI word*/
+/*define for ib_base_63_32 field*/
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_offset 2
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift  0
+#define SDMA_PKT_INDIRECT_BASE_HI_IB_BASE_63_32(x) (((x) & SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask) << SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift)
+
+/*define for IB_SIZE word*/
+/*define for ib_size field*/
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_offset 3
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask   0x000FFFFF
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift  0
+#define SDMA_PKT_INDIRECT_IB_SIZE_IB_SIZE(x) (((x) & SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask) << SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift)
+
+/*define for CSA_ADDR_LO word*/
+/*define for csa_addr_31_0 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_offset 4
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_CSA_ADDR_31_0(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift)
+
+/*define for CSA_ADDR_HI word*/
+/*define for csa_addr_63_32 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_offset 5
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_CSA_ADDR_63_32(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SEMAPHORE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_op_shift  0
+#define SDMA_PKT_SEMAPHORE_HEADER_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SEMAPHORE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift)
+
+/*define for write_one field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_shift  29
+#define SDMA_PKT_SEMAPHORE_HEADER_WRITE_ONE(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_write_one_mask) << SDMA_PKT_SEMAPHORE_HEADER_write_one_shift)
+
+/*define for signal field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_shift  30
+#define SDMA_PKT_SEMAPHORE_HEADER_SIGNAL(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_signal_mask) << SDMA_PKT_SEMAPHORE_HEADER_signal_shift)
+
+/*define for mailbox field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift  31
+#define SDMA_PKT_SEMAPHORE_HEADER_MAILBOX(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask) << SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_FENCE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_FENCE_HEADER_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_op_shift  0
+#define SDMA_PKT_FENCE_HEADER_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_op_mask) << SDMA_PKT_FENCE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_FENCE_HEADER_sub_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_sub_op_shift  8
+#define SDMA_PKT_FENCE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_sub_op_mask) << SDMA_PKT_FENCE_HEADER_sub_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_FENCE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_FENCE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_FENCE_DATA_data_offset 3
+#define SDMA_PKT_FENCE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_DATA_data_shift  0
+#define SDMA_PKT_FENCE_DATA_DATA(x) (((x) & SDMA_PKT_FENCE_DATA_data_mask) << SDMA_PKT_FENCE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SRBM_WRITE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_shift  0
+#define SDMA_PKT_SRBM_WRITE_HEADER_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SRBM_WRITE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift)
+
+/*define for byte_en field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask   0x0000000F
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift  28
+#define SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask) << SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift)
+
+/*define for ADDR word*/
+/*define for addr field*/
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_offset 1
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_mask   0x0000FFFF
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_shift  0
+#define SDMA_PKT_SRBM_WRITE_ADDR_ADDR(x) (((x) & SDMA_PKT_SRBM_WRITE_ADDR_addr_mask) << SDMA_PKT_SRBM_WRITE_ADDR_addr_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_SRBM_WRITE_DATA_data_offset 2
+#define SDMA_PKT_SRBM_WRITE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_SRBM_WRITE_DATA_data_shift  0
+#define SDMA_PKT_SRBM_WRITE_DATA_DATA(x) (((x) & SDMA_PKT_SRBM_WRITE_DATA_data_mask) << SDMA_PKT_SRBM_WRITE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PRE_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_op_shift  0
+#define SDMA_PKT_PRE_EXE_HEADER_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_op_mask) << SDMA_PKT_PRE_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_PRE_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_sub_op_mask) << SDMA_PKT_PRE_EXE_HEADER_sub_op_shift)
+
+/*define for dev_sel field*/
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift  16
+#define SDMA_PKT_PRE_EXE_HEADER_DEV_SEL(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask) << SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_offset 1
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COND_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COND_EXE_HEADER_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_op_shift  0
+#define SDMA_PKT_COND_EXE_HEADER_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_op_mask) << SDMA_PKT_COND_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_COND_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_sub_op_mask) << SDMA_PKT_COND_EXE_HEADER_sub_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift)
+
+/*define for REFERENCE word*/
+/*define for reference field*/
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_offset 3
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_shift  0
+#define SDMA_PKT_COND_EXE_REFERENCE_REFERENCE(x) (((x) & SDMA_PKT_COND_EXE_REFERENCE_reference_mask) << SDMA_PKT_COND_EXE_REFERENCE_reference_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_offset 4
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_CONSTANT_FILL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_shift  0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift  8
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift  16
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SW(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift)
+
+/*define for fillsize field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift  30
+#define SDMA_PKT_CONSTANT_FILL_HEADER_FILLSIZE(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for src_data_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_offset 3
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DATA_SRC_DATA_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_offset 4
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_shift  0
+#define SDMA_PKT_CONSTANT_FILL_COUNT_COUNT(x) (((x) & SDMA_PKT_CONSTANT_FILL_COUNT_count_mask) << SDMA_PKT_CONSTANT_FILL_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_REGMEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_REGMEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_REGMEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift)
+
+/*define for hdp_flush field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift  26
+#define SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask) << SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift)
+
+/*define for func field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_mask   0x00000007
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_shift  28
+#define SDMA_PKT_POLL_REGMEM_HEADER_FUNC(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_func_mask) << SDMA_PKT_POLL_REGMEM_HEADER_func_shift)
+
+/*define for mem_poll field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift  31
+#define SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask) << SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift)
+
+/*define for VALUE word*/
+/*define for value field*/
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_offset 3
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_shift  0
+#define SDMA_PKT_POLL_REGMEM_VALUE_VALUE(x) (((x) & SDMA_PKT_POLL_REGMEM_VALUE_value_mask) << SDMA_PKT_POLL_REGMEM_VALUE_value_shift)
+
+/*define for MASK word*/
+/*define for mask field*/
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_offset 4
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_shift  0
+#define SDMA_PKT_POLL_REGMEM_MASK_MASK(x) (((x) & SDMA_PKT_POLL_REGMEM_MASK_mask_mask) << SDMA_PKT_POLL_REGMEM_MASK_mask_shift)
+
+/*define for DW5 word*/
+/*define for interval field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_mask   0x0000FFFF
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_shift  0
+#define SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_interval_mask) << SDMA_PKT_POLL_REGMEM_DW5_interval_shift)
+
+/*define for retry_count field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask   0x00000FFF
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift  16
+#define SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask) << SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_ATOMIC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_ATOMIC_HEADER_op_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_ATOMIC_HEADER_op_shift  0
+#define SDMA_PKT_ATOMIC_HEADER_OP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_op_mask) << SDMA_PKT_ATOMIC_HEADER_op_shift)
+
+/*define for loop field*/
+#define SDMA_PKT_ATOMIC_HEADER_loop_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_loop_mask   0x00000001
+#define SDMA_PKT_ATOMIC_HEADER_loop_shift  16
+#define SDMA_PKT_ATOMIC_HEADER_LOOP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_loop_mask) << SDMA_PKT_ATOMIC_HEADER_loop_shift)
+
+/*define for atomic_op field*/
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_mask   0x0000007F
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_shift  25
+#define SDMA_PKT_ATOMIC_HEADER_ATOMIC_OP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_atomic_op_mask) << SDMA_PKT_ATOMIC_HEADER_atomic_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_ATOMIC_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_mask) << SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_ATOMIC_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_mask) << SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_shift)
+
+/*define for SRC_DATA_LO word*/
+/*define for src_data_31_0 field*/
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_offset 3
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_shift  0
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_SRC_DATA_31_0(x) (((x) & SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_mask) << SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_shift)
+
+/*define for SRC_DATA_HI word*/
+/*define for src_data_63_32 field*/
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_offset 4
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_shift  0
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_SRC_DATA_63_32(x) (((x) & SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_mask) << SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_shift)
+
+/*define for CMP_DATA_LO word*/
+/*define for cmp_data_31_0 field*/
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_offset 5
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_shift  0
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_CMP_DATA_31_0(x) (((x) & SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_mask) << SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_shift)
+
+/*define for CMP_DATA_HI word*/
+/*define for cmp_data_63_32 field*/
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_offset 6
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_shift  0
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_CMP_DATA_63_32(x) (((x) & SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_mask) << SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_shift)
+
+/*define for LOOP_INTERVAL word*/
+/*define for loop_interval field*/
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_offset 7
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_mask   0x00001FFF
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_shift  0
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_LOOP_INTERVAL(x) (((x) & SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_mask) << SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_SET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift)
+
+/*define for INIT_DATA_LO word*/
+/*define for init_data_31_0 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_offset 1
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_INIT_DATA_31_0(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift)
+
+/*define for INIT_DATA_HI word*/
+/*define for init_data_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_INIT_DATA_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET_GLOBAL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TRAP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TRAP_HEADER_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_op_shift  0
+#define SDMA_PKT_TRAP_HEADER_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_op_mask) << SDMA_PKT_TRAP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TRAP_HEADER_sub_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_sub_op_shift  8
+#define SDMA_PKT_TRAP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_sub_op_mask) << SDMA_PKT_TRAP_HEADER_sub_op_shift)
+
+/*define for INT_CONTEXT word*/
+/*define for int_context field*/
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_offset 1
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask   0x0FFFFFFF
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift  0
+#define SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(x) (((x) & SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask) << SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift)
+
+
+/*
+** Definitions for SDMA_PKT_NOP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_NOP_HEADER_op_offset 0
+#define SDMA_PKT_NOP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_op_shift  0
+#define SDMA_PKT_NOP_HEADER_OP(x) (((x) & SDMA_PKT_NOP_HEADER_op_mask) << SDMA_PKT_NOP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_NOP_HEADER_sub_op_offset 0
+#define SDMA_PKT_NOP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_sub_op_shift  8
+#define SDMA_PKT_NOP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_NOP_HEADER_sub_op_mask) << SDMA_PKT_NOP_HEADER_sub_op_shift)
+
+/*define for count field*/
+#define SDMA_PKT_NOP_HEADER_count_offset 0
+#define SDMA_PKT_NOP_HEADER_count_mask   0x00003FFF
+#define SDMA_PKT_NOP_HEADER_count_shift  16
+#define SDMA_PKT_NOP_HEADER_COUNT(x) (((x) & SDMA_PKT_NOP_HEADER_count_mask) << SDMA_PKT_NOP_HEADER_count_shift)
+
+#endif /* __TONGA_SDMA_PKT_OPEN_H_ */
diff --git a/sys/dev/pci/drm/amd/amdgpu/uvd_v4_2.h b/sys/dev/pci/drm/amd/amdgpu/uvd_v4_2.h
new file mode 100644
index 00000000000..8a0444bb8b9
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/uvd_v4_2.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __UVD_V4_2_H__
+#define __UVD_V4_2_H__
+
+extern const struct amdgpu_ip_block_version uvd_v4_2_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/uvd_v5_0.c b/sys/dev/pci/drm/amd/amdgpu/uvd_v5_0.c
new file mode 100644
index 00000000000..eae693f8a0c
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/uvd_v5_0.c
@@ -0,0 +1,902 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_uvd.h"
+#include "vid.h"
+#include "uvd/uvd_5_0_d.h"
+#include "uvd/uvd_5_0_sh_mask.h"
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+#include "bif/bif_5_0_d.h"
+#include "vi.h"
+#include "smu/smu_7_1_2_d.h"
+#include "smu/smu_7_1_2_sh_mask.h"
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+static void uvd_v5_0_set_ring_funcs(struct amdgpu_device *adev);
+static void uvd_v5_0_set_irq_funcs(struct amdgpu_device *adev);
+static int uvd_v5_0_start(struct amdgpu_device *adev);
+static void uvd_v5_0_stop(struct amdgpu_device *adev);
+static int uvd_v5_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state);
+static void uvd_v5_0_enable_mgcg(struct amdgpu_device *adev,
+				 bool enable);
+/**
+ * uvd_v5_0_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t uvd_v5_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_RPTR);
+}
+
+/**
+ * uvd_v5_0_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t uvd_v5_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_WPTR);
+}
+
+/**
+ * uvd_v5_0_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void uvd_v5_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32(mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+}
+
+static int uvd_v5_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	adev->uvd.num_uvd_inst = 1;
+
+	uvd_v5_0_set_ring_funcs(adev);
+	uvd_v5_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int uvd_v5_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r;
+
+	/* UVD TRAP */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_UVD_SYSTEM_MESSAGE, &adev->uvd.inst->irq);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_sw_init(adev);
+	if (r)
+		return r;
+
+	ring = &adev->uvd.inst->ring;
+	snprintf(ring->name, sizeof(ring->name), "uvd");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_entity_init(adev);
+
+	return r;
+}
+
+static int uvd_v5_0_sw_fini(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_suspend(adev);
+	if (r)
+		return r;
+
+	return amdgpu_uvd_sw_fini(adev);
+}
+
+/**
+ * uvd_v5_0_hw_init - start and test UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int uvd_v5_0_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
+	uint32_t tmp;
+	int r;
+
+	amdgpu_asic_set_uvd_clocks(adev, 10000, 10000);
+	uvd_v5_0_set_clockgating_state(adev, AMD_CG_STATE_UNGATE);
+	uvd_v5_0_enable_mgcg(adev, true);
+
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		goto done;
+	}
+
+	r = amdgpu_ring_alloc(ring, 10);
+	if (r) {
+		DRM_ERROR("amdgpu: ring failed to lock UVD ring (%d).\n", r);
+		goto done;
+	}
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	/* Clear timeout status bits */
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_TIMEOUT_STATUS, 0));
+	amdgpu_ring_write(ring, 0x8);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_CNTL, 0));
+	amdgpu_ring_write(ring, 3);
+
+	amdgpu_ring_commit(ring);
+
+done:
+	if (!r)
+		DRM_INFO("UVD initialized successfully.\n");
+
+	return r;
+
+}
+
+/**
+ * uvd_v5_0_hw_fini - stop the hardware block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the UVD block, mark ring as not ready any more
+ */
+static int uvd_v5_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
+
+	if (RREG32(mmUVD_STATUS) != 0)
+		uvd_v5_0_stop(adev);
+
+	ring->ready = false;
+
+	return 0;
+}
+
+static int uvd_v5_0_suspend(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = uvd_v5_0_hw_fini(adev);
+	if (r)
+		return r;
+	uvd_v5_0_set_clockgating_state(adev, AMD_CG_STATE_GATE);
+
+	return amdgpu_uvd_suspend(adev);
+}
+
+static int uvd_v5_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	return uvd_v5_0_hw_init(adev);
+}
+
+/**
+ * uvd_v5_0_mc_resume - memory controller programming
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Let the UVD memory controller know it's offsets
+ */
+static void uvd_v5_0_mc_resume(struct amdgpu_device *adev)
+{
+	uint64_t offset;
+	uint32_t size;
+
+	/* programm memory controller bits 0-27 */
+	WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->uvd.inst->gpu_addr));
+	WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->uvd.inst->gpu_addr));
+
+	offset = AMDGPU_UVD_FIRMWARE_OFFSET;
+	size = AMDGPU_UVD_FIRMWARE_SIZE(adev);
+	WREG32(mmUVD_VCPU_CACHE_OFFSET0, offset >> 3);
+	WREG32(mmUVD_VCPU_CACHE_SIZE0, size);
+
+	offset += size;
+	size = AMDGPU_UVD_HEAP_SIZE;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET1, offset >> 3);
+	WREG32(mmUVD_VCPU_CACHE_SIZE1, size);
+
+	offset += size;
+	size = AMDGPU_UVD_STACK_SIZE +
+	       (AMDGPU_UVD_SESSION_SIZE * adev->uvd.max_handles);
+	WREG32(mmUVD_VCPU_CACHE_OFFSET2, offset >> 3);
+	WREG32(mmUVD_VCPU_CACHE_SIZE2, size);
+
+	WREG32(mmUVD_UDEC_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmUVD_UDEC_DB_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmUVD_UDEC_DBW_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+}
+
+/**
+ * uvd_v5_0_start - start UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the UVD block
+ */
+static int uvd_v5_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
+	uint32_t rb_bufsz, tmp;
+	uint32_t lmi_swap_cntl;
+	uint32_t mp_swap_cntl;
+	int i, j, r;
+
+	/*disable DPG */
+	WREG32_P(mmUVD_POWER_STATUS, 0, ~(1 << 2));
+
+	/* disable byte swapping */
+	lmi_swap_cntl = 0;
+	mp_swap_cntl = 0;
+
+	uvd_v5_0_mc_resume(adev);
+
+	/* disable interupt */
+	WREG32_P(mmUVD_MASTINT_EN, 0, ~(1 << 1));
+
+	/* stall UMC and register bus before resetting VCPU */
+	WREG32_P(mmUVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+	mdelay(1);
+
+	/* put LMI, VCPU, RBC etc... into reset */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK | UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__RBC_SOFT_RESET_MASK | UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__CXW_SOFT_RESET_MASK | UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* take UVD block out of reset */
+	WREG32_P(mmSRBM_SOFT_RESET, 0, ~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK);
+	mdelay(5);
+
+	/* initialize UVD memory controller */
+	WREG32(mmUVD_LMI_CTRL, 0x40 | (1 << 8) | (1 << 13) |
+			     (1 << 21) | (1 << 9) | (1 << 20));
+
+#ifdef __BIG_ENDIAN
+	/* swap (8 in 32) RB and IB */
+	lmi_swap_cntl = 0xa;
+	mp_swap_cntl = 0;
+#endif
+	WREG32(mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
+	WREG32(mmUVD_MP_SWAP_CNTL, mp_swap_cntl);
+
+	WREG32(mmUVD_MPC_SET_MUXA0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXA1, 0x0);
+	WREG32(mmUVD_MPC_SET_MUXB0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXB1, 0x0);
+	WREG32(mmUVD_MPC_SET_ALU, 0);
+	WREG32(mmUVD_MPC_SET_MUX, 0x88);
+
+	/* take all subblocks out of reset, except VCPU */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* enable VCPU clock */
+	WREG32(mmUVD_VCPU_CNTL,  1 << 9);
+
+	/* enable UMC */
+	WREG32_P(mmUVD_LMI_CTRL2, 0, ~(1 << 8));
+
+	/* boot up the VCPU */
+	WREG32(mmUVD_SOFT_RESET, 0);
+	mdelay(10);
+
+	for (i = 0; i < 10; ++i) {
+		uint32_t status;
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("UVD not responding, trying to reset the VCPU!!!\n");
+		WREG32_P(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
+				~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		WREG32_P(mmUVD_SOFT_RESET, 0, ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		r = -1;
+	}
+
+	if (r) {
+		DRM_ERROR("UVD not responding, giving up!!!\n");
+		return r;
+	}
+	/* enable master interrupt */
+	WREG32_P(mmUVD_MASTINT_EN, 3 << 1, ~(3 << 1));
+
+	/* clear the bit 4 of UVD_STATUS */
+	WREG32_P(mmUVD_STATUS, 0, ~(2 << 1));
+
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = 0;
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	/* force RBC into idle state */
+	WREG32(mmUVD_RBC_RB_CNTL, tmp);
+
+	/* set the write pointer delay */
+	WREG32(mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* set the wb address */
+	WREG32(mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2));
+
+	/* programm the RB_BASE for ring buffer */
+	WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+			lower_32_bits(ring->gpu_addr));
+	WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+			upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(mmUVD_RBC_RB_RPTR, 0);
+
+	ring->wptr = RREG32(mmUVD_RBC_RB_RPTR);
+	WREG32(mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+
+	WREG32_P(mmUVD_RBC_RB_CNTL, 0, ~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK);
+
+	return 0;
+}
+
+/**
+ * uvd_v5_0_stop - stop UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the UVD block
+ */
+static void uvd_v5_0_stop(struct amdgpu_device *adev)
+{
+	/* force RBC into idle state */
+	WREG32(mmUVD_RBC_RB_CNTL, 0x11010101);
+
+	/* Stall UMC and register bus before resetting VCPU */
+	WREG32_P(mmUVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+	mdelay(1);
+
+	/* put VCPU into reset */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* disable VCPU clock */
+	WREG32(mmUVD_VCPU_CNTL, 0x0);
+
+	/* Unstall UMC and register bus */
+	WREG32_P(mmUVD_LMI_CTRL2, 0, ~(1 << 8));
+
+	WREG32(mmUVD_STATUS, 0);
+}
+
+/**
+ * uvd_v5_0_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @fence: fence to emit
+ *
+ * Write a fence and a trap command to the ring.
+ */
+static void uvd_v5_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				     unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, addr & 0xffffffff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 2);
+}
+
+/**
+ * uvd_v5_0_ring_test_ring - register write test
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully write to the context register
+ */
+static int uvd_v5_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmUVD_CONTEXT_ID);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	return r;
+}
+
+/**
+ * uvd_v5_0_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @ib: indirect buffer to execute
+ *
+ * Write ring commands to execute the indirect buffer
+ */
+static void uvd_v5_0_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib,
+				  unsigned vmid, bool ctx_switch)
+{
+	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_LOW, 0));
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH, 0));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_SIZE, 0));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+static void uvd_v5_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKET0(mmUVD_NO_OP, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static bool uvd_v5_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return !(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK);
+}
+
+static int uvd_v5_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK))
+			return 0;
+	}
+	return -ETIMEDOUT;
+}
+
+static int uvd_v5_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	uvd_v5_0_stop(adev);
+
+	WREG32_P(mmSRBM_SOFT_RESET, SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK,
+			~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK);
+	mdelay(5);
+
+	return uvd_v5_0_start(adev);
+}
+
+static int uvd_v5_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	// TODO
+	return 0;
+}
+
+static int uvd_v5_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: UVD TRAP\n");
+	amdgpu_fence_process(&adev->uvd.inst->ring);
+	return 0;
+}
+
+static void uvd_v5_0_enable_clock_gating(struct amdgpu_device *adev, bool enable)
+{
+	uint32_t data1, data3, suvd_flags;
+
+	data1 = RREG32(mmUVD_SUVD_CGC_GATE);
+	data3 = RREG32(mmUVD_CGC_GATE);
+
+	suvd_flags = UVD_SUVD_CGC_GATE__SRE_MASK |
+		     UVD_SUVD_CGC_GATE__SIT_MASK |
+		     UVD_SUVD_CGC_GATE__SMP_MASK |
+		     UVD_SUVD_CGC_GATE__SCM_MASK |
+		     UVD_SUVD_CGC_GATE__SDB_MASK;
+
+	if (enable) {
+		data3 |= (UVD_CGC_GATE__SYS_MASK     |
+			UVD_CGC_GATE__UDEC_MASK      |
+			UVD_CGC_GATE__MPEG2_MASK     |
+			UVD_CGC_GATE__RBC_MASK       |
+			UVD_CGC_GATE__LMI_MC_MASK    |
+			UVD_CGC_GATE__IDCT_MASK      |
+			UVD_CGC_GATE__MPRD_MASK      |
+			UVD_CGC_GATE__MPC_MASK       |
+			UVD_CGC_GATE__LBSI_MASK      |
+			UVD_CGC_GATE__LRBBM_MASK     |
+			UVD_CGC_GATE__UDEC_RE_MASK   |
+			UVD_CGC_GATE__UDEC_CM_MASK   |
+			UVD_CGC_GATE__UDEC_IT_MASK   |
+			UVD_CGC_GATE__UDEC_DB_MASK   |
+			UVD_CGC_GATE__UDEC_MP_MASK   |
+			UVD_CGC_GATE__WCB_MASK       |
+			UVD_CGC_GATE__JPEG_MASK      |
+			UVD_CGC_GATE__SCPU_MASK);
+		/* only in pg enabled, we can gate clock to vcpu*/
+		if (adev->pg_flags & AMD_PG_SUPPORT_UVD)
+			data3 |= UVD_CGC_GATE__VCPU_MASK;
+		data3 &= ~UVD_CGC_GATE__REGS_MASK;
+		data1 |= suvd_flags;
+	} else {
+		data3 = 0;
+		data1 = 0;
+	}
+
+	WREG32(mmUVD_SUVD_CGC_GATE, data1);
+	WREG32(mmUVD_CGC_GATE, data3);
+}
+
+static void uvd_v5_0_set_sw_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data, data2;
+
+	data = RREG32(mmUVD_CGC_CTRL);
+	data2 = RREG32(mmUVD_SUVD_CGC_CTRL);
+
+
+	data &= ~(UVD_CGC_CTRL__CLK_OFF_DELAY_MASK |
+		  UVD_CGC_CTRL__CLK_GATE_DLY_TIMER_MASK);
+
+
+	data |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK |
+		(1 << REG_FIELD_SHIFT(UVD_CGC_CTRL, CLK_GATE_DLY_TIMER)) |
+		(4 << REG_FIELD_SHIFT(UVD_CGC_CTRL, CLK_OFF_DELAY));
+
+	data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_CM_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_IT_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_DB_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_MP_MODE_MASK |
+			UVD_CGC_CTRL__SYS_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_MODE_MASK |
+			UVD_CGC_CTRL__MPEG2_MODE_MASK |
+			UVD_CGC_CTRL__REGS_MODE_MASK |
+			UVD_CGC_CTRL__RBC_MODE_MASK |
+			UVD_CGC_CTRL__LMI_MC_MODE_MASK |
+			UVD_CGC_CTRL__LMI_UMC_MODE_MASK |
+			UVD_CGC_CTRL__IDCT_MODE_MASK |
+			UVD_CGC_CTRL__MPRD_MODE_MASK |
+			UVD_CGC_CTRL__MPC_MODE_MASK |
+			UVD_CGC_CTRL__LBSI_MODE_MASK |
+			UVD_CGC_CTRL__LRBBM_MODE_MASK |
+			UVD_CGC_CTRL__WCB_MODE_MASK |
+			UVD_CGC_CTRL__VCPU_MODE_MASK |
+			UVD_CGC_CTRL__JPEG_MODE_MASK |
+			UVD_CGC_CTRL__SCPU_MODE_MASK);
+	data2 &= ~(UVD_SUVD_CGC_CTRL__SRE_MODE_MASK |
+			UVD_SUVD_CGC_CTRL__SIT_MODE_MASK |
+			UVD_SUVD_CGC_CTRL__SMP_MODE_MASK |
+			UVD_SUVD_CGC_CTRL__SCM_MODE_MASK |
+			UVD_SUVD_CGC_CTRL__SDB_MODE_MASK);
+
+	WREG32(mmUVD_CGC_CTRL, data);
+	WREG32(mmUVD_SUVD_CGC_CTRL, data2);
+}
+
+#if 0
+static void uvd_v5_0_set_hw_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data, data1, cgc_flags, suvd_flags;
+
+	data = RREG32(mmUVD_CGC_GATE);
+	data1 = RREG32(mmUVD_SUVD_CGC_GATE);
+
+	cgc_flags = UVD_CGC_GATE__SYS_MASK |
+				UVD_CGC_GATE__UDEC_MASK |
+				UVD_CGC_GATE__MPEG2_MASK |
+				UVD_CGC_GATE__RBC_MASK |
+				UVD_CGC_GATE__LMI_MC_MASK |
+				UVD_CGC_GATE__IDCT_MASK |
+				UVD_CGC_GATE__MPRD_MASK |
+				UVD_CGC_GATE__MPC_MASK |
+				UVD_CGC_GATE__LBSI_MASK |
+				UVD_CGC_GATE__LRBBM_MASK |
+				UVD_CGC_GATE__UDEC_RE_MASK |
+				UVD_CGC_GATE__UDEC_CM_MASK |
+				UVD_CGC_GATE__UDEC_IT_MASK |
+				UVD_CGC_GATE__UDEC_DB_MASK |
+				UVD_CGC_GATE__UDEC_MP_MASK |
+				UVD_CGC_GATE__WCB_MASK |
+				UVD_CGC_GATE__VCPU_MASK |
+				UVD_CGC_GATE__SCPU_MASK;
+
+	suvd_flags = UVD_SUVD_CGC_GATE__SRE_MASK |
+				UVD_SUVD_CGC_GATE__SIT_MASK |
+				UVD_SUVD_CGC_GATE__SMP_MASK |
+				UVD_SUVD_CGC_GATE__SCM_MASK |
+				UVD_SUVD_CGC_GATE__SDB_MASK;
+
+	data |= cgc_flags;
+	data1 |= suvd_flags;
+
+	WREG32(mmUVD_CGC_GATE, data);
+	WREG32(mmUVD_SUVD_CGC_GATE, data1);
+}
+#endif
+
+static void uvd_v5_0_enable_mgcg(struct amdgpu_device *adev,
+				 bool enable)
+{
+	u32 orig, data;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_UVD_MGCG)) {
+		data = RREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL);
+		data |= 0xfff;
+		WREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL, data);
+
+		orig = data = RREG32(mmUVD_CGC_CTRL);
+		data |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+		if (orig != data)
+			WREG32(mmUVD_CGC_CTRL, data);
+	} else {
+		data = RREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL);
+		data &= ~0xfff;
+		WREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL, data);
+
+		orig = data = RREG32(mmUVD_CGC_CTRL);
+		data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+		if (orig != data)
+			WREG32(mmUVD_CGC_CTRL, data);
+	}
+}
+
+static int uvd_v5_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+
+	if (enable) {
+		/* wait for STATUS to clear */
+		if (uvd_v5_0_wait_for_idle(handle))
+			return -EBUSY;
+		uvd_v5_0_enable_clock_gating(adev, true);
+
+		/* enable HW gates because UVD is idle */
+/*		uvd_v5_0_set_hw_clock_gating(adev); */
+	} else {
+		uvd_v5_0_enable_clock_gating(adev, false);
+	}
+
+	uvd_v5_0_set_sw_clock_gating(adev);
+	return 0;
+}
+
+static int uvd_v5_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the UVD block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret = 0;
+
+	if (state == AMD_PG_STATE_GATE) {
+		uvd_v5_0_stop(adev);
+	} else {
+		ret = uvd_v5_0_start(adev);
+		if (ret)
+			goto out;
+	}
+
+out:
+	return ret;
+}
+
+static void uvd_v5_0_get_clockgating_state(void *handle, u32 *flags)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int data;
+
+	mutex_lock(&adev->pm.mutex);
+
+	if (RREG32_SMC(ixCURRENT_PG_STATUS) &
+				CURRENT_PG_STATUS__UVD_PG_STATUS_MASK) {
+		DRM_INFO("Cannot get clockgating state when UVD is powergated.\n");
+		goto out;
+	}
+
+	/* AMD_CG_SUPPORT_UVD_MGCG */
+	data = RREG32(mmUVD_CGC_CTRL);
+	if (data & UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK)
+		*flags |= AMD_CG_SUPPORT_UVD_MGCG;
+
+out:
+	mutex_unlock(&adev->pm.mutex);
+}
+
+static const struct amd_ip_funcs uvd_v5_0_ip_funcs = {
+	.name = "uvd_v5_0",
+	.early_init = uvd_v5_0_early_init,
+	.late_init = NULL,
+	.sw_init = uvd_v5_0_sw_init,
+	.sw_fini = uvd_v5_0_sw_fini,
+	.hw_init = uvd_v5_0_hw_init,
+	.hw_fini = uvd_v5_0_hw_fini,
+	.suspend = uvd_v5_0_suspend,
+	.resume = uvd_v5_0_resume,
+	.is_idle = uvd_v5_0_is_idle,
+	.wait_for_idle = uvd_v5_0_wait_for_idle,
+	.soft_reset = uvd_v5_0_soft_reset,
+	.set_clockgating_state = uvd_v5_0_set_clockgating_state,
+	.set_powergating_state = uvd_v5_0_set_powergating_state,
+	.get_clockgating_state = uvd_v5_0_get_clockgating_state,
+};
+
+static const struct amdgpu_ring_funcs uvd_v5_0_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_UVD,
+	.align_mask = 0xf,
+	.support_64bit_ptrs = false,
+	.get_rptr = uvd_v5_0_ring_get_rptr,
+	.get_wptr = uvd_v5_0_ring_get_wptr,
+	.set_wptr = uvd_v5_0_ring_set_wptr,
+	.parse_cs = amdgpu_uvd_ring_parse_cs,
+	.emit_frame_size =
+		14, /* uvd_v5_0_ring_emit_fence  x1 no user fence */
+	.emit_ib_size = 6, /* uvd_v5_0_ring_emit_ib */
+	.emit_ib = uvd_v5_0_ring_emit_ib,
+	.emit_fence = uvd_v5_0_ring_emit_fence,
+	.test_ring = uvd_v5_0_ring_test_ring,
+	.test_ib = amdgpu_uvd_ring_test_ib,
+	.insert_nop = uvd_v5_0_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
+};
+
+static void uvd_v5_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->uvd.inst->ring.funcs = &uvd_v5_0_ring_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs uvd_v5_0_irq_funcs = {
+	.set = uvd_v5_0_set_interrupt_state,
+	.process = uvd_v5_0_process_interrupt,
+};
+
+static void uvd_v5_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->uvd.inst->irq.num_types = 1;
+	adev->uvd.inst->irq.funcs = &uvd_v5_0_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version uvd_v5_0_ip_block =
+{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 5,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &uvd_v5_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/uvd_v5_0.h b/sys/dev/pci/drm/amd/amdgpu/uvd_v5_0.h
new file mode 100644
index 00000000000..2eaaea793ac
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/uvd_v5_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __UVD_V5_0_H__
+#define __UVD_V5_0_H__
+
+extern const struct amdgpu_ip_block_version uvd_v5_0_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/uvd_v6_0.c b/sys/dev/pci/drm/amd/amdgpu/uvd_v6_0.c
new file mode 100644
index 00000000000..b82d065e130
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/uvd_v6_0.c
@@ -0,0 +1,1667 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_uvd.h"
+#include "vid.h"
+#include "uvd/uvd_6_0_d.h"
+#include "uvd/uvd_6_0_sh_mask.h"
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+#include "smu/smu_7_1_3_d.h"
+#include "smu/smu_7_1_3_sh_mask.h"
+#include "bif/bif_5_1_d.h"
+#include "gmc/gmc_8_1_d.h"
+#include "vi.h"
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+/* Polaris10/11/12 firmware version */
+#define FW_1_130_16 ((1 << 24) | (130 << 16) | (16 << 8))
+
+static void uvd_v6_0_set_ring_funcs(struct amdgpu_device *adev);
+static void uvd_v6_0_set_enc_ring_funcs(struct amdgpu_device *adev);
+
+static void uvd_v6_0_set_irq_funcs(struct amdgpu_device *adev);
+static int uvd_v6_0_start(struct amdgpu_device *adev);
+static void uvd_v6_0_stop(struct amdgpu_device *adev);
+static void uvd_v6_0_set_sw_clock_gating(struct amdgpu_device *adev);
+static int uvd_v6_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state);
+static void uvd_v6_0_enable_mgcg(struct amdgpu_device *adev,
+				 bool enable);
+
+/**
+* uvd_v6_0_enc_support - get encode support status
+*
+* @adev: amdgpu_device pointer
+*
+* Returns the current hardware encode support status
+*/
+static inline bool uvd_v6_0_enc_support(struct amdgpu_device *adev)
+{
+	return ((adev->asic_type >= CHIP_POLARIS10) &&
+			(adev->asic_type <= CHIP_VEGAM) &&
+			(!adev->uvd.fw_version || adev->uvd.fw_version >= FW_1_130_16));
+}
+
+/**
+ * uvd_v6_0_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t uvd_v6_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_RPTR);
+}
+
+/**
+ * uvd_v6_0_enc_ring_get_rptr - get enc read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc read pointer
+ */
+static uint64_t uvd_v6_0_enc_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->uvd.inst->ring_enc[0])
+		return RREG32(mmUVD_RB_RPTR);
+	else
+		return RREG32(mmUVD_RB_RPTR2);
+}
+/**
+ * uvd_v6_0_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t uvd_v6_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_WPTR);
+}
+
+/**
+ * uvd_v6_0_enc_ring_get_wptr - get enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc write pointer
+ */
+static uint64_t uvd_v6_0_enc_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->uvd.inst->ring_enc[0])
+		return RREG32(mmUVD_RB_WPTR);
+	else
+		return RREG32(mmUVD_RB_WPTR2);
+}
+
+/**
+ * uvd_v6_0_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void uvd_v6_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32(mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+}
+
+/**
+ * uvd_v6_0_enc_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void uvd_v6_0_enc_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->uvd.inst->ring_enc[0])
+		WREG32(mmUVD_RB_WPTR,
+			lower_32_bits(ring->wptr));
+	else
+		WREG32(mmUVD_RB_WPTR2,
+			lower_32_bits(ring->wptr));
+}
+
+/**
+ * uvd_v6_0_enc_ring_test_ring - test if UVD ENC ring is working
+ *
+ * @ring: the engine to test on
+ *
+ */
+static int uvd_v6_0_enc_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t rptr = amdgpu_ring_get_rptr(ring);
+	unsigned i;
+	int r;
+
+	r = amdgpu_ring_alloc(ring, 16);
+	if (r) {
+		DRM_ERROR("amdgpu: uvd enc failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_END);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (amdgpu_ring_get_rptr(ring) != rptr)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed\n",
+			  ring->idx);
+		r = -ETIMEDOUT;
+	}
+
+	return r;
+}
+
+/**
+ * uvd_v6_0_enc_get_create_msg - generate a UVD ENC create msg
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: ring we should submit the msg to
+ * @handle: session handle to use
+ * @fence: optional fence to return
+ *
+ * Open up a stream for HW test
+ */
+static int uvd_v6_0_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+				       struct dma_fence **fence)
+{
+	const unsigned ib_size_dw = 16;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct dma_fence *f = NULL;
+	uint64_t dummy;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	if (r)
+		return r;
+
+	ib = &job->ibs[0];
+	dummy = ib->gpu_addr + 1024;
+
+	ib->length_dw = 0;
+	ib->ptr[ib->length_dw++] = 0x00000018;
+	ib->ptr[ib->length_dw++] = 0x00000001; /* session info */
+	ib->ptr[ib->length_dw++] = handle;
+	ib->ptr[ib->length_dw++] = 0x00010000;
+	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
+	ib->ptr[ib->length_dw++] = dummy;
+
+	ib->ptr[ib->length_dw++] = 0x00000014;
+	ib->ptr[ib->length_dw++] = 0x00000002; /* task info */
+	ib->ptr[ib->length_dw++] = 0x0000001c;
+	ib->ptr[ib->length_dw++] = 0x00000001;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+
+	ib->ptr[ib->length_dw++] = 0x00000008;
+	ib->ptr[ib->length_dw++] = 0x08000001; /* op initialize */
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err;
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+	return 0;
+
+err:
+	amdgpu_job_free(job);
+	return r;
+}
+
+/**
+ * uvd_v6_0_enc_get_destroy_msg - generate a UVD ENC destroy msg
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: ring we should submit the msg to
+ * @handle: session handle to use
+ * @fence: optional fence to return
+ *
+ * Close up a stream for HW test or if userspace failed to do so
+ */
+static int uvd_v6_0_enc_get_destroy_msg(struct amdgpu_ring *ring,
+					uint32_t handle,
+					bool direct, struct dma_fence **fence)
+{
+	const unsigned ib_size_dw = 16;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct dma_fence *f = NULL;
+	uint64_t dummy;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	if (r)
+		return r;
+
+	ib = &job->ibs[0];
+	dummy = ib->gpu_addr + 1024;
+
+	ib->length_dw = 0;
+	ib->ptr[ib->length_dw++] = 0x00000018;
+	ib->ptr[ib->length_dw++] = 0x00000001; /* session info */
+	ib->ptr[ib->length_dw++] = handle;
+	ib->ptr[ib->length_dw++] = 0x00010000;
+	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
+	ib->ptr[ib->length_dw++] = dummy;
+
+	ib->ptr[ib->length_dw++] = 0x00000014;
+	ib->ptr[ib->length_dw++] = 0x00000002; /* task info */
+	ib->ptr[ib->length_dw++] = 0x0000001c;
+	ib->ptr[ib->length_dw++] = 0x00000001;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+
+	ib->ptr[ib->length_dw++] = 0x00000008;
+	ib->ptr[ib->length_dw++] = 0x08000002; /* op close session */
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+
+	if (direct)
+		r = amdgpu_job_submit_direct(job, ring, &f);
+	else
+		r = amdgpu_job_submit(job, &ring->adev->vce.entity,
+				      AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+	if (r)
+		goto err;
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+	return 0;
+
+err:
+	amdgpu_job_free(job);
+	return r;
+}
+
+/**
+ * uvd_v6_0_enc_ring_test_ib - test if UVD ENC IBs are working
+ *
+ * @ring: the engine to test on
+ *
+ */
+static int uvd_v6_0_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct dma_fence *fence = NULL;
+	long r;
+
+	r = uvd_v6_0_enc_get_create_msg(ring, 1, NULL);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
+		goto error;
+	}
+
+	r = uvd_v6_0_enc_get_destroy_msg(ring, 1, true, &fence);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
+		goto error;
+	}
+
+	r = dma_fence_wait_timeout(fence, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out.\n");
+		r = -ETIMEDOUT;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+	} else {
+		DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
+	}
+error:
+	dma_fence_put(fence);
+	return r;
+}
+static int uvd_v6_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	adev->uvd.num_uvd_inst = 1;
+
+	if (!(adev->flags & AMD_IS_APU) &&
+	    (RREG32_SMC(ixCC_HARVEST_FUSES) & CC_HARVEST_FUSES__UVD_DISABLE_MASK))
+		return -ENOENT;
+
+	uvd_v6_0_set_ring_funcs(adev);
+
+	if (uvd_v6_0_enc_support(adev)) {
+		adev->uvd.num_enc_rings = 2;
+		uvd_v6_0_set_enc_ring_funcs(adev);
+	}
+
+	uvd_v6_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int uvd_v6_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int i, r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* UVD TRAP */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_UVD_SYSTEM_MESSAGE, &adev->uvd.inst->irq);
+	if (r)
+		return r;
+
+	/* UVD ENC TRAP */
+	if (uvd_v6_0_enc_support(adev)) {
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, i + VISLANDS30_IV_SRCID_UVD_ENC_GEN_PURP, &adev->uvd.inst->irq);
+			if (r)
+				return r;
+		}
+	}
+
+	r = amdgpu_uvd_sw_init(adev);
+	if (r)
+		return r;
+
+	if (!uvd_v6_0_enc_support(adev)) {
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i)
+			adev->uvd.inst->ring_enc[i].funcs = NULL;
+
+		adev->uvd.inst->irq.num_types = 1;
+		adev->uvd.num_enc_rings = 0;
+
+		DRM_INFO("UVD ENC is disabled\n");
+	}
+
+	ring = &adev->uvd.inst->ring;
+	snprintf(ring->name, sizeof(ring->name), "uvd");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	if (uvd_v6_0_enc_support(adev)) {
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			ring = &adev->uvd.inst->ring_enc[i];
+			snprintf(ring->name, sizeof(ring->name), "uvd_enc%d", i);
+			r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst->irq, 0);
+			if (r)
+				return r;
+		}
+	}
+
+	r = amdgpu_uvd_entity_init(adev);
+
+	return r;
+}
+
+static int uvd_v6_0_sw_fini(void *handle)
+{
+	int i, r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_suspend(adev);
+	if (r)
+		return r;
+
+	if (uvd_v6_0_enc_support(adev)) {
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i)
+			amdgpu_ring_fini(&adev->uvd.inst->ring_enc[i]);
+	}
+
+	return amdgpu_uvd_sw_fini(adev);
+}
+
+/**
+ * uvd_v6_0_hw_init - start and test UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int uvd_v6_0_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
+	uint32_t tmp;
+	int i, r;
+
+	amdgpu_asic_set_uvd_clocks(adev, 10000, 10000);
+	uvd_v6_0_set_clockgating_state(adev, AMD_CG_STATE_UNGATE);
+	uvd_v6_0_enable_mgcg(adev, true);
+
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		goto done;
+	}
+
+	r = amdgpu_ring_alloc(ring, 10);
+	if (r) {
+		DRM_ERROR("amdgpu: ring failed to lock UVD ring (%d).\n", r);
+		goto done;
+	}
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	/* Clear timeout status bits */
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_TIMEOUT_STATUS, 0));
+	amdgpu_ring_write(ring, 0x8);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_CNTL, 0));
+	amdgpu_ring_write(ring, 3);
+
+	amdgpu_ring_commit(ring);
+
+	if (uvd_v6_0_enc_support(adev)) {
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			ring = &adev->uvd.inst->ring_enc[i];
+			ring->ready = true;
+			r = amdgpu_ring_test_ring(ring);
+			if (r) {
+				ring->ready = false;
+				goto done;
+			}
+		}
+	}
+
+done:
+	if (!r) {
+		if (uvd_v6_0_enc_support(adev))
+			DRM_INFO("UVD and UVD ENC initialized successfully.\n");
+		else
+			DRM_INFO("UVD initialized successfully.\n");
+	}
+
+	return r;
+}
+
+/**
+ * uvd_v6_0_hw_fini - stop the hardware block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the UVD block, mark ring as not ready any more
+ */
+static int uvd_v6_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
+
+	if (RREG32(mmUVD_STATUS) != 0)
+		uvd_v6_0_stop(adev);
+
+	ring->ready = false;
+
+	return 0;
+}
+
+static int uvd_v6_0_suspend(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = uvd_v6_0_hw_fini(adev);
+	if (r)
+		return r;
+
+	return amdgpu_uvd_suspend(adev);
+}
+
+static int uvd_v6_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	return uvd_v6_0_hw_init(adev);
+}
+
+/**
+ * uvd_v6_0_mc_resume - memory controller programming
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Let the UVD memory controller know it's offsets
+ */
+static void uvd_v6_0_mc_resume(struct amdgpu_device *adev)
+{
+	uint64_t offset;
+	uint32_t size;
+
+	/* programm memory controller bits 0-27 */
+	WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->uvd.inst->gpu_addr));
+	WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->uvd.inst->gpu_addr));
+
+	offset = AMDGPU_UVD_FIRMWARE_OFFSET;
+	size = AMDGPU_UVD_FIRMWARE_SIZE(adev);
+	WREG32(mmUVD_VCPU_CACHE_OFFSET0, offset >> 3);
+	WREG32(mmUVD_VCPU_CACHE_SIZE0, size);
+
+	offset += size;
+	size = AMDGPU_UVD_HEAP_SIZE;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET1, offset >> 3);
+	WREG32(mmUVD_VCPU_CACHE_SIZE1, size);
+
+	offset += size;
+	size = AMDGPU_UVD_STACK_SIZE +
+	       (AMDGPU_UVD_SESSION_SIZE * adev->uvd.max_handles);
+	WREG32(mmUVD_VCPU_CACHE_OFFSET2, offset >> 3);
+	WREG32(mmUVD_VCPU_CACHE_SIZE2, size);
+
+	WREG32(mmUVD_UDEC_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmUVD_UDEC_DB_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmUVD_UDEC_DBW_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+
+	WREG32(mmUVD_GP_SCRATCH4, adev->uvd.max_handles);
+}
+
+#if 0
+static void cz_set_uvd_clock_gating_branches(struct amdgpu_device *adev,
+		bool enable)
+{
+	u32 data, data1;
+
+	data = RREG32(mmUVD_CGC_GATE);
+	data1 = RREG32(mmUVD_SUVD_CGC_GATE);
+	if (enable) {
+		data |= UVD_CGC_GATE__SYS_MASK |
+				UVD_CGC_GATE__UDEC_MASK |
+				UVD_CGC_GATE__MPEG2_MASK |
+				UVD_CGC_GATE__RBC_MASK |
+				UVD_CGC_GATE__LMI_MC_MASK |
+				UVD_CGC_GATE__IDCT_MASK |
+				UVD_CGC_GATE__MPRD_MASK |
+				UVD_CGC_GATE__MPC_MASK |
+				UVD_CGC_GATE__LBSI_MASK |
+				UVD_CGC_GATE__LRBBM_MASK |
+				UVD_CGC_GATE__UDEC_RE_MASK |
+				UVD_CGC_GATE__UDEC_CM_MASK |
+				UVD_CGC_GATE__UDEC_IT_MASK |
+				UVD_CGC_GATE__UDEC_DB_MASK |
+				UVD_CGC_GATE__UDEC_MP_MASK |
+				UVD_CGC_GATE__WCB_MASK |
+				UVD_CGC_GATE__VCPU_MASK |
+				UVD_CGC_GATE__SCPU_MASK;
+		data1 |= UVD_SUVD_CGC_GATE__SRE_MASK |
+				UVD_SUVD_CGC_GATE__SIT_MASK |
+				UVD_SUVD_CGC_GATE__SMP_MASK |
+				UVD_SUVD_CGC_GATE__SCM_MASK |
+				UVD_SUVD_CGC_GATE__SDB_MASK |
+				UVD_SUVD_CGC_GATE__SRE_H264_MASK |
+				UVD_SUVD_CGC_GATE__SRE_HEVC_MASK |
+				UVD_SUVD_CGC_GATE__SIT_H264_MASK |
+				UVD_SUVD_CGC_GATE__SIT_HEVC_MASK |
+				UVD_SUVD_CGC_GATE__SCM_H264_MASK |
+				UVD_SUVD_CGC_GATE__SCM_HEVC_MASK |
+				UVD_SUVD_CGC_GATE__SDB_H264_MASK |
+				UVD_SUVD_CGC_GATE__SDB_HEVC_MASK;
+	} else {
+		data &= ~(UVD_CGC_GATE__SYS_MASK |
+				UVD_CGC_GATE__UDEC_MASK |
+				UVD_CGC_GATE__MPEG2_MASK |
+				UVD_CGC_GATE__RBC_MASK |
+				UVD_CGC_GATE__LMI_MC_MASK |
+				UVD_CGC_GATE__LMI_UMC_MASK |
+				UVD_CGC_GATE__IDCT_MASK |
+				UVD_CGC_GATE__MPRD_MASK |
+				UVD_CGC_GATE__MPC_MASK |
+				UVD_CGC_GATE__LBSI_MASK |
+				UVD_CGC_GATE__LRBBM_MASK |
+				UVD_CGC_GATE__UDEC_RE_MASK |
+				UVD_CGC_GATE__UDEC_CM_MASK |
+				UVD_CGC_GATE__UDEC_IT_MASK |
+				UVD_CGC_GATE__UDEC_DB_MASK |
+				UVD_CGC_GATE__UDEC_MP_MASK |
+				UVD_CGC_GATE__WCB_MASK |
+				UVD_CGC_GATE__VCPU_MASK |
+				UVD_CGC_GATE__SCPU_MASK);
+		data1 &= ~(UVD_SUVD_CGC_GATE__SRE_MASK |
+				UVD_SUVD_CGC_GATE__SIT_MASK |
+				UVD_SUVD_CGC_GATE__SMP_MASK |
+				UVD_SUVD_CGC_GATE__SCM_MASK |
+				UVD_SUVD_CGC_GATE__SDB_MASK |
+				UVD_SUVD_CGC_GATE__SRE_H264_MASK |
+				UVD_SUVD_CGC_GATE__SRE_HEVC_MASK |
+				UVD_SUVD_CGC_GATE__SIT_H264_MASK |
+				UVD_SUVD_CGC_GATE__SIT_HEVC_MASK |
+				UVD_SUVD_CGC_GATE__SCM_H264_MASK |
+				UVD_SUVD_CGC_GATE__SCM_HEVC_MASK |
+				UVD_SUVD_CGC_GATE__SDB_H264_MASK |
+				UVD_SUVD_CGC_GATE__SDB_HEVC_MASK);
+	}
+	WREG32(mmUVD_CGC_GATE, data);
+	WREG32(mmUVD_SUVD_CGC_GATE, data1);
+}
+#endif
+
+/**
+ * uvd_v6_0_start - start UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the UVD block
+ */
+static int uvd_v6_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->uvd.inst->ring;
+	uint32_t rb_bufsz, tmp;
+	uint32_t lmi_swap_cntl;
+	uint32_t mp_swap_cntl;
+	int i, j, r;
+
+	/* disable DPG */
+	WREG32_P(mmUVD_POWER_STATUS, 0, ~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+
+	/* disable byte swapping */
+	lmi_swap_cntl = 0;
+	mp_swap_cntl = 0;
+
+	uvd_v6_0_mc_resume(adev);
+
+	/* disable interupt */
+	WREG32_FIELD(UVD_MASTINT_EN, VCPU_EN, 0);
+
+	/* stall UMC and register bus before resetting VCPU */
+	WREG32_FIELD(UVD_LMI_CTRL2, STALL_ARB_UMC, 1);
+	mdelay(1);
+
+	/* put LMI, VCPU, RBC etc... into reset */
+	WREG32(mmUVD_SOFT_RESET,
+		UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__RBC_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__CXW_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* take UVD block out of reset */
+	WREG32_FIELD(SRBM_SOFT_RESET, SOFT_RESET_UVD, 0);
+	mdelay(5);
+
+	/* initialize UVD memory controller */
+	WREG32(mmUVD_LMI_CTRL,
+		(0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		UVD_LMI_CTRL__DISABLE_ON_FWV_FAIL_MASK);
+
+#ifdef __BIG_ENDIAN
+	/* swap (8 in 32) RB and IB */
+	lmi_swap_cntl = 0xa;
+	mp_swap_cntl = 0;
+#endif
+	WREG32(mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
+	WREG32(mmUVD_MP_SWAP_CNTL, mp_swap_cntl);
+
+	WREG32(mmUVD_MPC_SET_MUXA0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXA1, 0x0);
+	WREG32(mmUVD_MPC_SET_MUXB0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXB1, 0x0);
+	WREG32(mmUVD_MPC_SET_ALU, 0);
+	WREG32(mmUVD_MPC_SET_MUX, 0x88);
+
+	/* take all subblocks out of reset, except VCPU */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* enable VCPU clock */
+	WREG32(mmUVD_VCPU_CNTL, UVD_VCPU_CNTL__CLK_EN_MASK);
+
+	/* enable UMC */
+	WREG32_FIELD(UVD_LMI_CTRL2, STALL_ARB_UMC, 0);
+
+	/* boot up the VCPU */
+	WREG32(mmUVD_SOFT_RESET, 0);
+	mdelay(10);
+
+	for (i = 0; i < 10; ++i) {
+		uint32_t status;
+
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("UVD not responding, trying to reset the VCPU!!!\n");
+		WREG32_FIELD(UVD_SOFT_RESET, VCPU_SOFT_RESET, 1);
+		mdelay(10);
+		WREG32_FIELD(UVD_SOFT_RESET, VCPU_SOFT_RESET, 0);
+		mdelay(10);
+		r = -1;
+	}
+
+	if (r) {
+		DRM_ERROR("UVD not responding, giving up!!!\n");
+		return r;
+	}
+	/* enable master interrupt */
+	WREG32_P(mmUVD_MASTINT_EN,
+		(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
+		~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK));
+
+	/* clear the bit 4 of UVD_STATUS */
+	WREG32_P(mmUVD_STATUS, 0, ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+
+	/* force RBC into idle state */
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	WREG32(mmUVD_RBC_RB_CNTL, tmp);
+
+	/* set the write pointer delay */
+	WREG32(mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* set the wb address */
+	WREG32(mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2));
+
+	/* programm the RB_BASE for ring buffer */
+	WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+			lower_32_bits(ring->gpu_addr));
+	WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+			upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(mmUVD_RBC_RB_RPTR, 0);
+
+	ring->wptr = RREG32(mmUVD_RBC_RB_RPTR);
+	WREG32(mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+
+	WREG32_FIELD(UVD_RBC_RB_CNTL, RB_NO_FETCH, 0);
+
+	if (uvd_v6_0_enc_support(adev)) {
+		ring = &adev->uvd.inst->ring_enc[0];
+		WREG32(mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+		WREG32(mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+		WREG32(mmUVD_RB_BASE_LO, ring->gpu_addr);
+		WREG32(mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+		WREG32(mmUVD_RB_SIZE, ring->ring_size / 4);
+
+		ring = &adev->uvd.inst->ring_enc[1];
+		WREG32(mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+		WREG32(mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+		WREG32(mmUVD_RB_BASE_LO2, ring->gpu_addr);
+		WREG32(mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+		WREG32(mmUVD_RB_SIZE2, ring->ring_size / 4);
+	}
+
+	return 0;
+}
+
+/**
+ * uvd_v6_0_stop - stop UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the UVD block
+ */
+static void uvd_v6_0_stop(struct amdgpu_device *adev)
+{
+	/* force RBC into idle state */
+	WREG32(mmUVD_RBC_RB_CNTL, 0x11010101);
+
+	/* Stall UMC and register bus before resetting VCPU */
+	WREG32_P(mmUVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+	mdelay(1);
+
+	/* put VCPU into reset */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* disable VCPU clock */
+	WREG32(mmUVD_VCPU_CNTL, 0x0);
+
+	/* Unstall UMC and register bus */
+	WREG32_P(mmUVD_LMI_CTRL2, 0, ~(1 << 8));
+
+	WREG32(mmUVD_STATUS, 0);
+}
+
+/**
+ * uvd_v6_0_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @fence: fence to emit
+ *
+ * Write a fence and a trap command to the ring.
+ */
+static void uvd_v6_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				     unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, addr & 0xffffffff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 2);
+}
+
+/**
+ * uvd_v6_0_enc_ring_emit_fence - emit an enc fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @fence: fence to emit
+ *
+ * Write enc a fence and a trap command to the ring.
+ */
+static void uvd_v6_0_enc_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+			u64 seq, unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_FENCE);
+	amdgpu_ring_write(ring, addr);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_TRAP);
+}
+
+/**
+ * uvd_v6_0_ring_emit_hdp_flush - skip HDP flushing
+ *
+ * @ring: amdgpu_ring pointer
+ */
+static void uvd_v6_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	/* The firmware doesn't seem to like touching registers at this point. */
+}
+
+/**
+ * uvd_v6_0_ring_test_ring - register write test
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully write to the context register
+ */
+static int uvd_v6_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmUVD_CONTEXT_ID);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	return r;
+}
+
+/**
+ * uvd_v6_0_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @ib: indirect buffer to execute
+ *
+ * Write ring commands to execute the indirect buffer
+ */
+static void uvd_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib,
+				  unsigned vmid, bool ctx_switch)
+{
+	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_VMID, 0));
+	amdgpu_ring_write(ring, vmid);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_LOW, 0));
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH, 0));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_SIZE, 0));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+/**
+ * uvd_v6_0_enc_ring_emit_ib - enc execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @ib: indirect buffer to execute
+ *
+ * Write enc ring commands to execute the indirect buffer
+ */
+static void uvd_v6_0_enc_ring_emit_ib(struct amdgpu_ring *ring,
+		struct amdgpu_ib *ib, unsigned int vmid, bool ctx_switch)
+{
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_IB_VM);
+	amdgpu_ring_write(ring, vmid);
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+static void uvd_v6_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				    uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, val);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0x8);
+}
+
+static void uvd_v6_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned vmid, uint64_t pd_addr)
+{
+	amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST << 2);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GP_SCRATCH8, 0));
+	amdgpu_ring_write(ring, 1 << vmid); /* mask */
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0xC);
+}
+
+static void uvd_v6_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GP_SCRATCH8, 0));
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GP_SCRATCH9, 0));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0xE);
+}
+
+static void uvd_v6_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKET0(mmUVD_NO_OP, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static void uvd_v6_0_enc_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_WAIT_GE);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq);
+}
+
+static void uvd_v6_0_enc_ring_insert_end(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_END);
+}
+
+static void uvd_v6_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					    unsigned int vmid, uint64_t pd_addr)
+{
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_UPDATE_PTB);
+	amdgpu_ring_write(ring, vmid);
+	amdgpu_ring_write(ring, pd_addr >> 12);
+
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_FLUSH_TLB);
+	amdgpu_ring_write(ring, vmid);
+}
+
+static bool uvd_v6_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return !(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK);
+}
+
+static int uvd_v6_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (uvd_v6_0_is_idle(handle))
+			return 0;
+	}
+	return -ETIMEDOUT;
+}
+
+#define AMDGPU_UVD_STATUS_BUSY_MASK    0xfd
+static bool uvd_v6_0_check_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset = 0;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (REG_GET_FIELD(tmp, SRBM_STATUS, UVD_RQ_PENDING) ||
+	    REG_GET_FIELD(tmp, SRBM_STATUS, UVD_BUSY) ||
+	    (RREG32(mmUVD_STATUS) & AMDGPU_UVD_STATUS_BUSY_MASK))
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_UVD, 1);
+
+	if (srbm_soft_reset) {
+		adev->uvd.inst->srbm_soft_reset = srbm_soft_reset;
+		return true;
+	} else {
+		adev->uvd.inst->srbm_soft_reset = 0;
+		return false;
+	}
+}
+
+static int uvd_v6_0_pre_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->uvd.inst->srbm_soft_reset)
+		return 0;
+
+	uvd_v6_0_stop(adev);
+	return 0;
+}
+
+static int uvd_v6_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset;
+
+	if (!adev->uvd.inst->srbm_soft_reset)
+		return 0;
+	srbm_soft_reset = adev->uvd.inst->srbm_soft_reset;
+
+	if (srbm_soft_reset) {
+		u32 tmp;
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int uvd_v6_0_post_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->uvd.inst->srbm_soft_reset)
+		return 0;
+
+	mdelay(5);
+
+	return uvd_v6_0_start(adev);
+}
+
+static int uvd_v6_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	// TODO
+	return 0;
+}
+
+static int uvd_v6_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	bool int_handled = true;
+	DRM_DEBUG("IH: UVD TRAP\n");
+
+	switch (entry->src_id) {
+	case 124:
+		amdgpu_fence_process(&adev->uvd.inst->ring);
+		break;
+	case 119:
+		if (likely(uvd_v6_0_enc_support(adev)))
+			amdgpu_fence_process(&adev->uvd.inst->ring_enc[0]);
+		else
+			int_handled = false;
+		break;
+	case 120:
+		if (likely(uvd_v6_0_enc_support(adev)))
+			amdgpu_fence_process(&adev->uvd.inst->ring_enc[1]);
+		else
+			int_handled = false;
+		break;
+	}
+
+	if (false == int_handled)
+			DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+
+	return 0;
+}
+
+static void uvd_v6_0_enable_clock_gating(struct amdgpu_device *adev, bool enable)
+{
+	uint32_t data1, data3;
+
+	data1 = RREG32(mmUVD_SUVD_CGC_GATE);
+	data3 = RREG32(mmUVD_CGC_GATE);
+
+	data1 |= UVD_SUVD_CGC_GATE__SRE_MASK |
+		     UVD_SUVD_CGC_GATE__SIT_MASK |
+		     UVD_SUVD_CGC_GATE__SMP_MASK |
+		     UVD_SUVD_CGC_GATE__SCM_MASK |
+		     UVD_SUVD_CGC_GATE__SDB_MASK |
+		     UVD_SUVD_CGC_GATE__SRE_H264_MASK |
+		     UVD_SUVD_CGC_GATE__SRE_HEVC_MASK |
+		     UVD_SUVD_CGC_GATE__SIT_H264_MASK |
+		     UVD_SUVD_CGC_GATE__SIT_HEVC_MASK |
+		     UVD_SUVD_CGC_GATE__SCM_H264_MASK |
+		     UVD_SUVD_CGC_GATE__SCM_HEVC_MASK |
+		     UVD_SUVD_CGC_GATE__SDB_H264_MASK |
+		     UVD_SUVD_CGC_GATE__SDB_HEVC_MASK;
+
+	if (enable) {
+		data3 |= (UVD_CGC_GATE__SYS_MASK       |
+			UVD_CGC_GATE__UDEC_MASK      |
+			UVD_CGC_GATE__MPEG2_MASK     |
+			UVD_CGC_GATE__RBC_MASK       |
+			UVD_CGC_GATE__LMI_MC_MASK    |
+			UVD_CGC_GATE__LMI_UMC_MASK   |
+			UVD_CGC_GATE__IDCT_MASK      |
+			UVD_CGC_GATE__MPRD_MASK      |
+			UVD_CGC_GATE__MPC_MASK       |
+			UVD_CGC_GATE__LBSI_MASK      |
+			UVD_CGC_GATE__LRBBM_MASK     |
+			UVD_CGC_GATE__UDEC_RE_MASK   |
+			UVD_CGC_GATE__UDEC_CM_MASK   |
+			UVD_CGC_GATE__UDEC_IT_MASK   |
+			UVD_CGC_GATE__UDEC_DB_MASK   |
+			UVD_CGC_GATE__UDEC_MP_MASK   |
+			UVD_CGC_GATE__WCB_MASK       |
+			UVD_CGC_GATE__JPEG_MASK      |
+			UVD_CGC_GATE__SCPU_MASK      |
+			UVD_CGC_GATE__JPEG2_MASK);
+		/* only in pg enabled, we can gate clock to vcpu*/
+		if (adev->pg_flags & AMD_PG_SUPPORT_UVD)
+			data3 |= UVD_CGC_GATE__VCPU_MASK;
+
+		data3 &= ~UVD_CGC_GATE__REGS_MASK;
+	} else {
+		data3 = 0;
+	}
+
+	WREG32(mmUVD_SUVD_CGC_GATE, data1);
+	WREG32(mmUVD_CGC_GATE, data3);
+}
+
+static void uvd_v6_0_set_sw_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data, data2;
+
+	data = RREG32(mmUVD_CGC_CTRL);
+	data2 = RREG32(mmUVD_SUVD_CGC_CTRL);
+
+
+	data &= ~(UVD_CGC_CTRL__CLK_OFF_DELAY_MASK |
+		  UVD_CGC_CTRL__CLK_GATE_DLY_TIMER_MASK);
+
+
+	data |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK |
+		(1 << REG_FIELD_SHIFT(UVD_CGC_CTRL, CLK_GATE_DLY_TIMER)) |
+		(4 << REG_FIELD_SHIFT(UVD_CGC_CTRL, CLK_OFF_DELAY));
+
+	data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_CM_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_IT_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_DB_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_MP_MODE_MASK |
+			UVD_CGC_CTRL__SYS_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_MODE_MASK |
+			UVD_CGC_CTRL__MPEG2_MODE_MASK |
+			UVD_CGC_CTRL__REGS_MODE_MASK |
+			UVD_CGC_CTRL__RBC_MODE_MASK |
+			UVD_CGC_CTRL__LMI_MC_MODE_MASK |
+			UVD_CGC_CTRL__LMI_UMC_MODE_MASK |
+			UVD_CGC_CTRL__IDCT_MODE_MASK |
+			UVD_CGC_CTRL__MPRD_MODE_MASK |
+			UVD_CGC_CTRL__MPC_MODE_MASK |
+			UVD_CGC_CTRL__LBSI_MODE_MASK |
+			UVD_CGC_CTRL__LRBBM_MODE_MASK |
+			UVD_CGC_CTRL__WCB_MODE_MASK |
+			UVD_CGC_CTRL__VCPU_MODE_MASK |
+			UVD_CGC_CTRL__JPEG_MODE_MASK |
+			UVD_CGC_CTRL__SCPU_MODE_MASK |
+			UVD_CGC_CTRL__JPEG2_MODE_MASK);
+	data2 &= ~(UVD_SUVD_CGC_CTRL__SRE_MODE_MASK |
+			UVD_SUVD_CGC_CTRL__SIT_MODE_MASK |
+			UVD_SUVD_CGC_CTRL__SMP_MODE_MASK |
+			UVD_SUVD_CGC_CTRL__SCM_MODE_MASK |
+			UVD_SUVD_CGC_CTRL__SDB_MODE_MASK);
+
+	WREG32(mmUVD_CGC_CTRL, data);
+	WREG32(mmUVD_SUVD_CGC_CTRL, data2);
+}
+
+#if 0
+static void uvd_v6_0_set_hw_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data, data1, cgc_flags, suvd_flags;
+
+	data = RREG32(mmUVD_CGC_GATE);
+	data1 = RREG32(mmUVD_SUVD_CGC_GATE);
+
+	cgc_flags = UVD_CGC_GATE__SYS_MASK |
+		UVD_CGC_GATE__UDEC_MASK |
+		UVD_CGC_GATE__MPEG2_MASK |
+		UVD_CGC_GATE__RBC_MASK |
+		UVD_CGC_GATE__LMI_MC_MASK |
+		UVD_CGC_GATE__IDCT_MASK |
+		UVD_CGC_GATE__MPRD_MASK |
+		UVD_CGC_GATE__MPC_MASK |
+		UVD_CGC_GATE__LBSI_MASK |
+		UVD_CGC_GATE__LRBBM_MASK |
+		UVD_CGC_GATE__UDEC_RE_MASK |
+		UVD_CGC_GATE__UDEC_CM_MASK |
+		UVD_CGC_GATE__UDEC_IT_MASK |
+		UVD_CGC_GATE__UDEC_DB_MASK |
+		UVD_CGC_GATE__UDEC_MP_MASK |
+		UVD_CGC_GATE__WCB_MASK |
+		UVD_CGC_GATE__VCPU_MASK |
+		UVD_CGC_GATE__SCPU_MASK |
+		UVD_CGC_GATE__JPEG_MASK |
+		UVD_CGC_GATE__JPEG2_MASK;
+
+	suvd_flags = UVD_SUVD_CGC_GATE__SRE_MASK |
+				UVD_SUVD_CGC_GATE__SIT_MASK |
+				UVD_SUVD_CGC_GATE__SMP_MASK |
+				UVD_SUVD_CGC_GATE__SCM_MASK |
+				UVD_SUVD_CGC_GATE__SDB_MASK;
+
+	data |= cgc_flags;
+	data1 |= suvd_flags;
+
+	WREG32(mmUVD_CGC_GATE, data);
+	WREG32(mmUVD_SUVD_CGC_GATE, data1);
+}
+#endif
+
+static void uvd_v6_0_enable_mgcg(struct amdgpu_device *adev,
+				 bool enable)
+{
+	u32 orig, data;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_UVD_MGCG)) {
+		data = RREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL);
+		data |= 0xfff;
+		WREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL, data);
+
+		orig = data = RREG32(mmUVD_CGC_CTRL);
+		data |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+		if (orig != data)
+			WREG32(mmUVD_CGC_CTRL, data);
+	} else {
+		data = RREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL);
+		data &= ~0xfff;
+		WREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL, data);
+
+		orig = data = RREG32(mmUVD_CGC_CTRL);
+		data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+		if (orig != data)
+			WREG32(mmUVD_CGC_CTRL, data);
+	}
+}
+
+static int uvd_v6_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+
+	if (enable) {
+		/* wait for STATUS to clear */
+		if (uvd_v6_0_wait_for_idle(handle))
+			return -EBUSY;
+		uvd_v6_0_enable_clock_gating(adev, true);
+		/* enable HW gates because UVD is idle */
+/*		uvd_v6_0_set_hw_clock_gating(adev); */
+	} else {
+		/* disable HW gating and enable Sw gating */
+		uvd_v6_0_enable_clock_gating(adev, false);
+	}
+	uvd_v6_0_set_sw_clock_gating(adev);
+	return 0;
+}
+
+static int uvd_v6_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the UVD block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret = 0;
+
+	WREG32(mmUVD_POWER_STATUS, UVD_POWER_STATUS__UVD_PG_EN_MASK);
+
+	if (state == AMD_PG_STATE_GATE) {
+		uvd_v6_0_stop(adev);
+	} else {
+		ret = uvd_v6_0_start(adev);
+		if (ret)
+			goto out;
+	}
+
+out:
+	return ret;
+}
+
+static void uvd_v6_0_get_clockgating_state(void *handle, u32 *flags)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int data;
+
+	mutex_lock(&adev->pm.mutex);
+
+	if (adev->flags & AMD_IS_APU)
+		data = RREG32_SMC(ixCURRENT_PG_STATUS_APU);
+	else
+		data = RREG32_SMC(ixCURRENT_PG_STATUS);
+
+	if (data & CURRENT_PG_STATUS__UVD_PG_STATUS_MASK) {
+		DRM_INFO("Cannot get clockgating state when UVD is powergated.\n");
+		goto out;
+	}
+
+	/* AMD_CG_SUPPORT_UVD_MGCG */
+	data = RREG32(mmUVD_CGC_CTRL);
+	if (data & UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK)
+		*flags |= AMD_CG_SUPPORT_UVD_MGCG;
+
+out:
+	mutex_unlock(&adev->pm.mutex);
+}
+
+static const struct amd_ip_funcs uvd_v6_0_ip_funcs = {
+	.name = "uvd_v6_0",
+	.early_init = uvd_v6_0_early_init,
+	.late_init = NULL,
+	.sw_init = uvd_v6_0_sw_init,
+	.sw_fini = uvd_v6_0_sw_fini,
+	.hw_init = uvd_v6_0_hw_init,
+	.hw_fini = uvd_v6_0_hw_fini,
+	.suspend = uvd_v6_0_suspend,
+	.resume = uvd_v6_0_resume,
+	.is_idle = uvd_v6_0_is_idle,
+	.wait_for_idle = uvd_v6_0_wait_for_idle,
+	.check_soft_reset = uvd_v6_0_check_soft_reset,
+	.pre_soft_reset = uvd_v6_0_pre_soft_reset,
+	.soft_reset = uvd_v6_0_soft_reset,
+	.post_soft_reset = uvd_v6_0_post_soft_reset,
+	.set_clockgating_state = uvd_v6_0_set_clockgating_state,
+	.set_powergating_state = uvd_v6_0_set_powergating_state,
+	.get_clockgating_state = uvd_v6_0_get_clockgating_state,
+};
+
+static const struct amdgpu_ring_funcs uvd_v6_0_ring_phys_funcs = {
+	.type = AMDGPU_RING_TYPE_UVD,
+	.align_mask = 0xf,
+	.support_64bit_ptrs = false,
+	.get_rptr = uvd_v6_0_ring_get_rptr,
+	.get_wptr = uvd_v6_0_ring_get_wptr,
+	.set_wptr = uvd_v6_0_ring_set_wptr,
+	.parse_cs = amdgpu_uvd_ring_parse_cs,
+	.emit_frame_size =
+		6 + /* hdp invalidate */
+		10 + /* uvd_v6_0_ring_emit_pipeline_sync */
+		14, /* uvd_v6_0_ring_emit_fence x1 no user fence */
+	.emit_ib_size = 8, /* uvd_v6_0_ring_emit_ib */
+	.emit_ib = uvd_v6_0_ring_emit_ib,
+	.emit_fence = uvd_v6_0_ring_emit_fence,
+	.emit_hdp_flush = uvd_v6_0_ring_emit_hdp_flush,
+	.test_ring = uvd_v6_0_ring_test_ring,
+	.test_ib = amdgpu_uvd_ring_test_ib,
+	.insert_nop = uvd_v6_0_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
+	.emit_wreg = uvd_v6_0_ring_emit_wreg,
+};
+
+static const struct amdgpu_ring_funcs uvd_v6_0_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_UVD,
+	.align_mask = 0xf,
+	.support_64bit_ptrs = false,
+	.get_rptr = uvd_v6_0_ring_get_rptr,
+	.get_wptr = uvd_v6_0_ring_get_wptr,
+	.set_wptr = uvd_v6_0_ring_set_wptr,
+	.emit_frame_size =
+		6 + /* hdp invalidate */
+		10 + /* uvd_v6_0_ring_emit_pipeline_sync */
+		VI_FLUSH_GPU_TLB_NUM_WREG * 6 + 8 + /* uvd_v6_0_ring_emit_vm_flush */
+		14 + 14, /* uvd_v6_0_ring_emit_fence x2 vm fence */
+	.emit_ib_size = 8, /* uvd_v6_0_ring_emit_ib */
+	.emit_ib = uvd_v6_0_ring_emit_ib,
+	.emit_fence = uvd_v6_0_ring_emit_fence,
+	.emit_vm_flush = uvd_v6_0_ring_emit_vm_flush,
+	.emit_pipeline_sync = uvd_v6_0_ring_emit_pipeline_sync,
+	.emit_hdp_flush = uvd_v6_0_ring_emit_hdp_flush,
+	.test_ring = uvd_v6_0_ring_test_ring,
+	.test_ib = amdgpu_uvd_ring_test_ib,
+	.insert_nop = uvd_v6_0_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
+	.emit_wreg = uvd_v6_0_ring_emit_wreg,
+};
+
+static const struct amdgpu_ring_funcs uvd_v6_0_enc_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_UVD_ENC,
+	.align_mask = 0x3f,
+	.nop = HEVC_ENC_CMD_NO_OP,
+	.support_64bit_ptrs = false,
+	.get_rptr = uvd_v6_0_enc_ring_get_rptr,
+	.get_wptr = uvd_v6_0_enc_ring_get_wptr,
+	.set_wptr = uvd_v6_0_enc_ring_set_wptr,
+	.emit_frame_size =
+		4 + /* uvd_v6_0_enc_ring_emit_pipeline_sync */
+		5 + /* uvd_v6_0_enc_ring_emit_vm_flush */
+		5 + 5 + /* uvd_v6_0_enc_ring_emit_fence x2 vm fence */
+		1, /* uvd_v6_0_enc_ring_insert_end */
+	.emit_ib_size = 5, /* uvd_v6_0_enc_ring_emit_ib */
+	.emit_ib = uvd_v6_0_enc_ring_emit_ib,
+	.emit_fence = uvd_v6_0_enc_ring_emit_fence,
+	.emit_vm_flush = uvd_v6_0_enc_ring_emit_vm_flush,
+	.emit_pipeline_sync = uvd_v6_0_enc_ring_emit_pipeline_sync,
+	.test_ring = uvd_v6_0_enc_ring_test_ring,
+	.test_ib = uvd_v6_0_enc_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = uvd_v6_0_enc_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
+};
+
+static void uvd_v6_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	if (adev->asic_type >= CHIP_POLARIS10) {
+		adev->uvd.inst->ring.funcs = &uvd_v6_0_ring_vm_funcs;
+		DRM_INFO("UVD is enabled in VM mode\n");
+	} else {
+		adev->uvd.inst->ring.funcs = &uvd_v6_0_ring_phys_funcs;
+		DRM_INFO("UVD is enabled in physical mode\n");
+	}
+}
+
+static void uvd_v6_0_set_enc_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->uvd.num_enc_rings; ++i)
+		adev->uvd.inst->ring_enc[i].funcs = &uvd_v6_0_enc_ring_vm_funcs;
+
+	DRM_INFO("UVD ENC is enabled in VM mode\n");
+}
+
+static const struct amdgpu_irq_src_funcs uvd_v6_0_irq_funcs = {
+	.set = uvd_v6_0_set_interrupt_state,
+	.process = uvd_v6_0_process_interrupt,
+};
+
+static void uvd_v6_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	if (uvd_v6_0_enc_support(adev))
+		adev->uvd.inst->irq.num_types = adev->uvd.num_enc_rings + 1;
+	else
+		adev->uvd.inst->irq.num_types = 1;
+
+	adev->uvd.inst->irq.funcs = &uvd_v6_0_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version uvd_v6_0_ip_block =
+{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 6,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &uvd_v6_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version uvd_v6_2_ip_block =
+{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 6,
+		.minor = 2,
+		.rev = 0,
+		.funcs = &uvd_v6_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version uvd_v6_3_ip_block =
+{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 6,
+		.minor = 3,
+		.rev = 0,
+		.funcs = &uvd_v6_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/uvd_v6_0.h b/sys/dev/pci/drm/amd/amdgpu/uvd_v6_0.h
new file mode 100644
index 00000000000..d3d48c6428c
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/uvd_v6_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __UVD_V6_0_H__
+#define __UVD_V6_0_H__
+
+extern const struct amdgpu_ip_block_version uvd_v6_0_ip_block;
+extern const struct amdgpu_ip_block_version uvd_v6_2_ip_block;
+extern const struct amdgpu_ip_block_version uvd_v6_3_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/uvd_v7_0.c b/sys/dev/pci/drm/amd/amdgpu/uvd_v7_0.c
new file mode 100644
index 00000000000..2b2f70f8346
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/uvd_v7_0.c
@@ -0,0 +1,1885 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_uvd.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "soc15_common.h"
+#include "mmsch_v1_0.h"
+
+#include "uvd/uvd_7_0_offset.h"
+#include "uvd/uvd_7_0_sh_mask.h"
+#include "vce/vce_4_0_offset.h"
+#include "vce/vce_4_0_default.h"
+#include "vce/vce_4_0_sh_mask.h"
+#include "nbif/nbif_6_1_offset.h"
+#include "hdp/hdp_4_0_offset.h"
+#include "mmhub/mmhub_1_0_offset.h"
+#include "mmhub/mmhub_1_0_sh_mask.h"
+#include "ivsrcid/uvd/irqsrcs_uvd_7_0.h"
+
+#define mmUVD_PG0_CC_UVD_HARVESTING                                                                    0x00c7
+#define mmUVD_PG0_CC_UVD_HARVESTING_BASE_IDX                                                           1
+//UVD_PG0_CC_UVD_HARVESTING
+#define UVD_PG0_CC_UVD_HARVESTING__UVD_DISABLE__SHIFT                                                         0x1
+#define UVD_PG0_CC_UVD_HARVESTING__UVD_DISABLE_MASK                                                           0x00000002L
+
+#define UVD7_MAX_HW_INSTANCES_VEGA20			2
+
+static void uvd_v7_0_set_ring_funcs(struct amdgpu_device *adev);
+static void uvd_v7_0_set_enc_ring_funcs(struct amdgpu_device *adev);
+static void uvd_v7_0_set_irq_funcs(struct amdgpu_device *adev);
+static int uvd_v7_0_start(struct amdgpu_device *adev);
+static void uvd_v7_0_stop(struct amdgpu_device *adev);
+static int uvd_v7_0_sriov_start(struct amdgpu_device *adev);
+
+static int amdgpu_ih_clientid_uvds[] = {
+	SOC15_IH_CLIENTID_UVD,
+	SOC15_IH_CLIENTID_UVD1
+};
+
+/**
+ * uvd_v7_0_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t uvd_v7_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(UVD, ring->me, mmUVD_RBC_RB_RPTR);
+}
+
+/**
+ * uvd_v7_0_enc_ring_get_rptr - get enc read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc read pointer
+ */
+static uint64_t uvd_v7_0_enc_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->uvd.inst[ring->me].ring_enc[0])
+		return RREG32_SOC15(UVD, ring->me, mmUVD_RB_RPTR);
+	else
+		return RREG32_SOC15(UVD, ring->me, mmUVD_RB_RPTR2);
+}
+
+/**
+ * uvd_v7_0_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t uvd_v7_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(UVD, ring->me, mmUVD_RBC_RB_WPTR);
+}
+
+/**
+ * uvd_v7_0_enc_ring_get_wptr - get enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc write pointer
+ */
+static uint64_t uvd_v7_0_enc_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return adev->wb.wb[ring->wptr_offs];
+
+	if (ring == &adev->uvd.inst[ring->me].ring_enc[0])
+		return RREG32_SOC15(UVD, ring->me, mmUVD_RB_WPTR);
+	else
+		return RREG32_SOC15(UVD, ring->me, mmUVD_RB_WPTR2);
+}
+
+/**
+ * uvd_v7_0_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void uvd_v7_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32_SOC15(UVD, ring->me, mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+}
+
+/**
+ * uvd_v7_0_enc_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void uvd_v7_0_enc_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+		return;
+	}
+
+	if (ring == &adev->uvd.inst[ring->me].ring_enc[0])
+		WREG32_SOC15(UVD, ring->me, mmUVD_RB_WPTR,
+			lower_32_bits(ring->wptr));
+	else
+		WREG32_SOC15(UVD, ring->me, mmUVD_RB_WPTR2,
+			lower_32_bits(ring->wptr));
+}
+
+/**
+ * uvd_v7_0_enc_ring_test_ring - test if UVD ENC ring is working
+ *
+ * @ring: the engine to test on
+ *
+ */
+static int uvd_v7_0_enc_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t rptr = amdgpu_ring_get_rptr(ring);
+	unsigned i;
+	int r;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	r = amdgpu_ring_alloc(ring, 16);
+	if (r) {
+		DRM_ERROR("amdgpu: uvd enc failed to lock (%d)ring %d (%d).\n",
+			  ring->me, ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_END);
+	amdgpu_ring_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (amdgpu_ring_get_rptr(ring) != rptr)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("(%d)ring test on %d succeeded in %d usecs\n",
+			 ring->me, ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: (%d)ring %d test failed\n",
+			  ring->me, ring->idx);
+		r = -ETIMEDOUT;
+	}
+
+	return r;
+}
+
+/**
+ * uvd_v7_0_enc_get_create_msg - generate a UVD ENC create msg
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: ring we should submit the msg to
+ * @handle: session handle to use
+ * @fence: optional fence to return
+ *
+ * Open up a stream for HW test
+ */
+static int uvd_v7_0_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+				       struct dma_fence **fence)
+{
+	const unsigned ib_size_dw = 16;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct dma_fence *f = NULL;
+	uint64_t dummy;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	if (r)
+		return r;
+
+	ib = &job->ibs[0];
+	dummy = ib->gpu_addr + 1024;
+
+	ib->length_dw = 0;
+	ib->ptr[ib->length_dw++] = 0x00000018;
+	ib->ptr[ib->length_dw++] = 0x00000001; /* session info */
+	ib->ptr[ib->length_dw++] = handle;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
+	ib->ptr[ib->length_dw++] = dummy;
+
+	ib->ptr[ib->length_dw++] = 0x00000014;
+	ib->ptr[ib->length_dw++] = 0x00000002; /* task info */
+	ib->ptr[ib->length_dw++] = 0x0000001c;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+
+	ib->ptr[ib->length_dw++] = 0x00000008;
+	ib->ptr[ib->length_dw++] = 0x08000001; /* op initialize */
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+
+	r = amdgpu_job_submit_direct(job, ring, &f);
+	if (r)
+		goto err;
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+	return 0;
+
+err:
+	amdgpu_job_free(job);
+	return r;
+}
+
+/**
+ * uvd_v7_0_enc_get_destroy_msg - generate a UVD ENC destroy msg
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: ring we should submit the msg to
+ * @handle: session handle to use
+ * @fence: optional fence to return
+ *
+ * Close up a stream for HW test or if userspace failed to do so
+ */
+int uvd_v7_0_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+				 bool direct, struct dma_fence **fence)
+{
+	const unsigned ib_size_dw = 16;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct dma_fence *f = NULL;
+	uint64_t dummy;
+	int i, r;
+
+	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
+	if (r)
+		return r;
+
+	ib = &job->ibs[0];
+	dummy = ib->gpu_addr + 1024;
+
+	ib->length_dw = 0;
+	ib->ptr[ib->length_dw++] = 0x00000018;
+	ib->ptr[ib->length_dw++] = 0x00000001;
+	ib->ptr[ib->length_dw++] = handle;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
+	ib->ptr[ib->length_dw++] = dummy;
+
+	ib->ptr[ib->length_dw++] = 0x00000014;
+	ib->ptr[ib->length_dw++] = 0x00000002;
+	ib->ptr[ib->length_dw++] = 0x0000001c;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+
+	ib->ptr[ib->length_dw++] = 0x00000008;
+	ib->ptr[ib->length_dw++] = 0x08000002; /* op close session */
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+
+	if (direct)
+		r = amdgpu_job_submit_direct(job, ring, &f);
+	else
+		r = amdgpu_job_submit(job, &ring->adev->vce.entity,
+				      AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+	if (r)
+		goto err;
+
+	if (fence)
+		*fence = dma_fence_get(f);
+	dma_fence_put(f);
+	return 0;
+
+err:
+	amdgpu_job_free(job);
+	return r;
+}
+
+/**
+ * uvd_v7_0_enc_ring_test_ib - test if UVD ENC IBs are working
+ *
+ * @ring: the engine to test on
+ *
+ */
+static int uvd_v7_0_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct dma_fence *fence = NULL;
+	long r;
+
+	r = uvd_v7_0_enc_get_create_msg(ring, 1, NULL);
+	if (r) {
+		DRM_ERROR("amdgpu: (%d)failed to get create msg (%ld).\n", ring->me, r);
+		goto error;
+	}
+
+	r = uvd_v7_0_enc_get_destroy_msg(ring, 1, true, &fence);
+	if (r) {
+		DRM_ERROR("amdgpu: (%d)failed to get destroy ib (%ld).\n", ring->me, r);
+		goto error;
+	}
+
+	r = dma_fence_wait_timeout(fence, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: (%d)IB test timed out.\n", ring->me);
+		r = -ETIMEDOUT;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: (%d)fence wait failed (%ld).\n", ring->me, r);
+	} else {
+		DRM_DEBUG("ib test on (%d)ring %d succeeded\n", ring->me, ring->idx);
+		r = 0;
+	}
+error:
+	dma_fence_put(fence);
+	return r;
+}
+
+static int uvd_v7_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->asic_type == CHIP_VEGA20) {
+		u32 harvest;
+		int i;
+
+		adev->uvd.num_uvd_inst = UVD7_MAX_HW_INSTANCES_VEGA20;
+		for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+			harvest = RREG32_SOC15(UVD, i, mmUVD_PG0_CC_UVD_HARVESTING);
+			if (harvest & UVD_PG0_CC_UVD_HARVESTING__UVD_DISABLE_MASK) {
+				adev->uvd.harvest_config |= 1 << i;
+			}
+		}
+		if (adev->uvd.harvest_config == (AMDGPU_UVD_HARVEST_UVD0 |
+						 AMDGPU_UVD_HARVEST_UVD1))
+			/* both instances are harvested, disable the block */
+			return -ENOENT;
+	} else {
+		adev->uvd.num_uvd_inst = 1;
+	}
+
+	if (amdgpu_sriov_vf(adev))
+		adev->uvd.num_enc_rings = 1;
+	else
+		adev->uvd.num_enc_rings = 2;
+	uvd_v7_0_set_ring_funcs(adev);
+	uvd_v7_0_set_enc_ring_funcs(adev);
+	uvd_v7_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int uvd_v7_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+
+	int i, j, r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (j = 0; j < adev->uvd.num_uvd_inst; j++) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		/* UVD TRAP */
+		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_uvds[j], UVD_7_0__SRCID__UVD_SYSTEM_MESSAGE_INTERRUPT, &adev->uvd.inst[j].irq);
+		if (r)
+			return r;
+
+		/* UVD ENC TRAP */
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_uvds[j], i + UVD_7_0__SRCID__UVD_ENC_GEN_PURP, &adev->uvd.inst[j].irq);
+			if (r)
+				return r;
+		}
+	}
+
+	r = amdgpu_uvd_sw_init(adev);
+	if (r)
+		return r;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		const struct common_firmware_header *hdr;
+		hdr = (const struct common_firmware_header *)adev->uvd.fw->data;
+		adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].ucode_id = AMDGPU_UCODE_ID_UVD;
+		adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].fw = adev->uvd.fw;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(hdr->ucode_size_bytes), PAGE_SIZE);
+		DRM_INFO("PSP loading UVD firmware\n");
+	}
+
+	for (j = 0; j < adev->uvd.num_uvd_inst; j++) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		if (!amdgpu_sriov_vf(adev)) {
+			ring = &adev->uvd.inst[j].ring;
+			snprintf(ring->name, sizeof(ring->name), "uvd<%d>", j);
+			r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst[j].irq, 0);
+			if (r)
+				return r;
+		}
+
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			ring = &adev->uvd.inst[j].ring_enc[i];
+			snprintf(ring->name, sizeof(ring->name), "uvd_enc%d<%d>", i, j);
+			if (amdgpu_sriov_vf(adev)) {
+				ring->use_doorbell = true;
+
+				/* currently only use the first enconding ring for
+				 * sriov, so set unused location for other unused rings.
+				 */
+				if (i == 0)
+					ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING0_1 * 2;
+				else
+					ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING2_3 * 2 + 1;
+			}
+			r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.inst[j].irq, 0);
+			if (r)
+				return r;
+		}
+	}
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_entity_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_virt_alloc_mm_table(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int uvd_v7_0_sw_fini(void *handle)
+{
+	int i, j, r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_virt_free_mm_table(adev);
+
+	r = amdgpu_uvd_suspend(adev);
+	if (r)
+		return r;
+
+	for (j = 0; j < adev->uvd.num_uvd_inst; ++j) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i)
+			amdgpu_ring_fini(&adev->uvd.inst[j].ring_enc[i]);
+	}
+	return amdgpu_uvd_sw_fini(adev);
+}
+
+/**
+ * uvd_v7_0_hw_init - start and test UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int uvd_v7_0_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring;
+	uint32_t tmp;
+	int i, j, r;
+
+	if (amdgpu_sriov_vf(adev))
+		r = uvd_v7_0_sriov_start(adev);
+	else
+		r = uvd_v7_0_start(adev);
+	if (r)
+		goto done;
+
+	for (j = 0; j < adev->uvd.num_uvd_inst; ++j) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		ring = &adev->uvd.inst[j].ring;
+
+		if (!amdgpu_sriov_vf(adev)) {
+			ring->ready = true;
+			r = amdgpu_ring_test_ring(ring);
+			if (r) {
+				ring->ready = false;
+				goto done;
+			}
+
+			r = amdgpu_ring_alloc(ring, 10);
+			if (r) {
+				DRM_ERROR("amdgpu: (%d)ring failed to lock UVD ring (%d).\n", j, r);
+				goto done;
+			}
+
+			tmp = PACKET0(SOC15_REG_OFFSET(UVD, j,
+				mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL), 0);
+			amdgpu_ring_write(ring, tmp);
+			amdgpu_ring_write(ring, 0xFFFFF);
+
+			tmp = PACKET0(SOC15_REG_OFFSET(UVD, j,
+				mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL), 0);
+			amdgpu_ring_write(ring, tmp);
+			amdgpu_ring_write(ring, 0xFFFFF);
+
+			tmp = PACKET0(SOC15_REG_OFFSET(UVD, j,
+				mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL), 0);
+			amdgpu_ring_write(ring, tmp);
+			amdgpu_ring_write(ring, 0xFFFFF);
+
+			/* Clear timeout status bits */
+			amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, j,
+				mmUVD_SEMA_TIMEOUT_STATUS), 0));
+			amdgpu_ring_write(ring, 0x8);
+
+			amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, j,
+				mmUVD_SEMA_CNTL), 0));
+			amdgpu_ring_write(ring, 3);
+
+			amdgpu_ring_commit(ring);
+		}
+
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			ring = &adev->uvd.inst[j].ring_enc[i];
+			ring->ready = true;
+			r = amdgpu_ring_test_ring(ring);
+			if (r) {
+				ring->ready = false;
+				goto done;
+			}
+		}
+	}
+done:
+	if (!r)
+		DRM_INFO("UVD and UVD ENC initialized successfully.\n");
+
+	return r;
+}
+
+/**
+ * uvd_v7_0_hw_fini - stop the hardware block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the UVD block, mark ring as not ready any more
+ */
+static int uvd_v7_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	if (!amdgpu_sriov_vf(adev))
+		uvd_v7_0_stop(adev);
+	else {
+		/* full access mode, so don't touch any UVD register */
+		DRM_DEBUG("For SRIOV client, shouldn't do anything.\n");
+	}
+
+	for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		adev->uvd.inst[i].ring.ready = false;
+	}
+
+	return 0;
+}
+
+static int uvd_v7_0_suspend(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = uvd_v7_0_hw_fini(adev);
+	if (r)
+		return r;
+
+	return amdgpu_uvd_suspend(adev);
+}
+
+static int uvd_v7_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	return uvd_v7_0_hw_init(adev);
+}
+
+/**
+ * uvd_v7_0_mc_resume - memory controller programming
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Let the UVD memory controller know it's offsets
+ */
+static void uvd_v7_0_mc_resume(struct amdgpu_device *adev)
+{
+	uint32_t size = AMDGPU_UVD_FIRMWARE_SIZE(adev);
+	uint32_t offset;
+	int i;
+
+	for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+			WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+				lower_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr));
+			WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+				upper_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr));
+			offset = 0;
+		} else {
+			WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+				lower_32_bits(adev->uvd.inst[i].gpu_addr));
+			WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+				upper_32_bits(adev->uvd.inst[i].gpu_addr));
+			offset = size;
+		}
+
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_OFFSET0,
+					AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_SIZE0, size);
+
+		WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+				lower_32_bits(adev->uvd.inst[i].gpu_addr + offset));
+		WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+				upper_32_bits(adev->uvd.inst[i].gpu_addr + offset));
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_OFFSET1, (1 << 21));
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_UVD_HEAP_SIZE);
+
+		WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+				lower_32_bits(adev->uvd.inst[i].gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE));
+		WREG32_SOC15(UVD, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+				upper_32_bits(adev->uvd.inst[i].gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE));
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_OFFSET2, (2 << 21));
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CACHE_SIZE2,
+				AMDGPU_UVD_STACK_SIZE + (AMDGPU_UVD_SESSION_SIZE * 40));
+
+		WREG32_SOC15(UVD, i, mmUVD_UDEC_ADDR_CONFIG,
+				adev->gfx.config.gb_addr_config);
+		WREG32_SOC15(UVD, i, mmUVD_UDEC_DB_ADDR_CONFIG,
+				adev->gfx.config.gb_addr_config);
+		WREG32_SOC15(UVD, i, mmUVD_UDEC_DBW_ADDR_CONFIG,
+				adev->gfx.config.gb_addr_config);
+
+		WREG32_SOC15(UVD, i, mmUVD_GP_SCRATCH4, adev->uvd.max_handles);
+	}
+}
+
+static int uvd_v7_0_mmsch_start(struct amdgpu_device *adev,
+				struct amdgpu_mm_table *table)
+{
+	uint32_t data = 0, loop;
+	uint64_t addr = table->gpu_addr;
+	struct mmsch_v1_0_init_header *header = (struct mmsch_v1_0_init_header *)table->cpu_addr;
+	uint32_t size;
+	int i;
+
+	size = header->header_size + header->vce_table_size + header->uvd_table_size;
+
+	/* 1, write to vce_mmsch_vf_ctx_addr_lo/hi register with GPU mc addr of memory descriptor location */
+	WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_CTX_ADDR_LO, lower_32_bits(addr));
+	WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_CTX_ADDR_HI, upper_32_bits(addr));
+
+	/* 2, update vmid of descriptor */
+	data = RREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_VMID);
+	data &= ~VCE_MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+	data |= (0 << VCE_MMSCH_VF_VMID__VF_CTX_VMID__SHIFT); /* use domain0 for MM scheduler */
+	WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_VMID, data);
+
+	/* 3, notify mmsch about the size of this descriptor */
+	WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_CTX_SIZE, size);
+
+	/* 4, set resp to zero */
+	WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP, 0);
+
+	for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		WDOORBELL32(adev->uvd.inst[i].ring_enc[0].doorbell_index, 0);
+		adev->wb.wb[adev->uvd.inst[i].ring_enc[0].wptr_offs] = 0;
+		adev->uvd.inst[i].ring_enc[0].wptr = 0;
+		adev->uvd.inst[i].ring_enc[0].wptr_old = 0;
+	}
+	/* 5, kick off the initialization and wait until VCE_MMSCH_VF_MAILBOX_RESP becomes non-zero */
+	WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_HOST, 0x10000001);
+
+	data = RREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP);
+	loop = 1000;
+	while ((data & 0x10000002) != 0x10000002) {
+		udelay(10);
+		data = RREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP);
+		loop--;
+		if (!loop)
+			break;
+	}
+
+	if (!loop) {
+		dev_err(adev->dev, "failed to init MMSCH, mmVCE_MMSCH_VF_MAILBOX_RESP = %x\n", data);
+		return -EBUSY;
+	}
+
+	return 0;
+}
+
+static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	uint32_t offset, size, tmp;
+	uint32_t table_size = 0;
+	struct mmsch_v1_0_cmd_direct_write direct_wt = { {0} };
+	struct mmsch_v1_0_cmd_direct_read_modify_write direct_rd_mod_wt = { {0} };
+	struct mmsch_v1_0_cmd_direct_polling direct_poll = { {0} };
+	struct mmsch_v1_0_cmd_end end = { {0} };
+	uint32_t *init_table = adev->virt.mm_table.cpu_addr;
+	struct mmsch_v1_0_init_header *header = (struct mmsch_v1_0_init_header *)init_table;
+	uint8_t i = 0;
+
+	direct_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_WRITE;
+	direct_rd_mod_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE;
+	direct_poll.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_POLLING;
+	end.cmd_header.command_type = MMSCH_COMMAND__END;
+
+	if (header->uvd_table_offset == 0 && header->uvd_table_size == 0) {
+		header->version = MMSCH_VERSION;
+		header->header_size = sizeof(struct mmsch_v1_0_init_header) >> 2;
+
+		if (header->vce_table_offset == 0 && header->vce_table_size == 0)
+			header->uvd_table_offset = header->header_size;
+		else
+			header->uvd_table_offset = header->vce_table_size + header->vce_table_offset;
+
+		init_table += header->uvd_table_offset;
+
+		for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
+			if (adev->uvd.harvest_config & (1 << i))
+				continue;
+			ring = &adev->uvd.inst[i].ring;
+			ring->wptr = 0;
+			size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4);
+
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_STATUS),
+							   0xFFFFFFFF, 0x00000004);
+			/* mc resume*/
+			if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+				MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+							    lower_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr));
+				MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+							    upper_32_bits(adev->firmware.ucode[AMDGPU_UCODE_ID_UVD].mc_addr));
+				offset = 0;
+			} else {
+				MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+							    lower_32_bits(adev->uvd.inst[i].gpu_addr));
+				MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+							    upper_32_bits(adev->uvd.inst[i].gpu_addr));
+				offset = size;
+			}
+
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET0),
+						    AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE0), size);
+
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+						    lower_32_bits(adev->uvd.inst[i].gpu_addr + offset));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+						    upper_32_bits(adev->uvd.inst[i].gpu_addr + offset));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET1), (1 << 21));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE1), AMDGPU_UVD_HEAP_SIZE);
+
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+						    lower_32_bits(adev->uvd.inst[i].gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+						    upper_32_bits(adev->uvd.inst[i].gpu_addr + offset + AMDGPU_UVD_HEAP_SIZE));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET2), (2 << 21));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE2),
+						    AMDGPU_UVD_STACK_SIZE + (AMDGPU_UVD_SESSION_SIZE * 40));
+
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_GP_SCRATCH4), adev->uvd.max_handles);
+			/* mc resume end*/
+
+			/* disable clock gating */
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_CGC_CTRL),
+							   ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK, 0);
+
+			/* disable interupt */
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_MASTINT_EN),
+							   ~UVD_MASTINT_EN__VCPU_EN_MASK, 0);
+
+			/* stall UMC and register bus before resetting VCPU */
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL2),
+							   ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
+							   UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+			/* put LMI, VCPU, RBC etc... into reset */
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_SOFT_RESET),
+						    (uint32_t)(UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__RBC_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__CXW_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
+							       UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK));
+
+			/* initialize UVD memory controller */
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL),
+						    (uint32_t)((0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+							       UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+							       UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+							       UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+							       UVD_LMI_CTRL__REQ_MODE_MASK |
+							       0x00100000L));
+
+			/* take all subblocks out of reset, except VCPU */
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_SOFT_RESET),
+						    UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+
+			/* enable VCPU clock */
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CNTL),
+						    UVD_VCPU_CNTL__CLK_EN_MASK);
+
+			/* enable master interrupt */
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_MASTINT_EN),
+							   ~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
+							   (UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK));
+
+			/* clear the bit 4 of UVD_STATUS */
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_STATUS),
+							   ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT), 0);
+
+			/* force RBC into idle state */
+			size = order_base_2(ring->ring_size);
+			tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, size);
+			tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_RBC_RB_CNTL), tmp);
+
+			ring = &adev->uvd.inst[i].ring_enc[0];
+			ring->wptr = 0;
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_RB_BASE_LO), ring->gpu_addr);
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_RB_BASE_HI), upper_32_bits(ring->gpu_addr));
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_RB_SIZE), ring->ring_size / 4);
+
+			/* boot up the VCPU */
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_SOFT_RESET), 0);
+
+			/* enable UMC */
+			MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL2),
+											   ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, 0);
+
+			MMSCH_V1_0_INSERT_DIRECT_POLL(SOC15_REG_OFFSET(UVD, i, mmUVD_STATUS), 0x02, 0x02);
+		}
+		/* add end packet */
+		memcpy((void *)init_table, &end, sizeof(struct mmsch_v1_0_cmd_end));
+		table_size += sizeof(struct mmsch_v1_0_cmd_end) / 4;
+		header->uvd_table_size = table_size;
+
+	}
+	return uvd_v7_0_mmsch_start(adev, &adev->virt.mm_table);
+}
+
+/**
+ * uvd_v7_0_start - start UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the UVD block
+ */
+static int uvd_v7_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	uint32_t rb_bufsz, tmp;
+	uint32_t lmi_swap_cntl;
+	uint32_t mp_swap_cntl;
+	int i, j, k, r;
+
+	for (k = 0; k < adev->uvd.num_uvd_inst; ++k) {
+		if (adev->uvd.harvest_config & (1 << k))
+			continue;
+		/* disable DPG */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_POWER_STATUS), 0,
+				~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+	}
+
+	/* disable byte swapping */
+	lmi_swap_cntl = 0;
+	mp_swap_cntl = 0;
+
+	uvd_v7_0_mc_resume(adev);
+
+	for (k = 0; k < adev->uvd.num_uvd_inst; ++k) {
+		if (adev->uvd.harvest_config & (1 << k))
+			continue;
+		ring = &adev->uvd.inst[k].ring;
+		/* disable clock gating */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_CGC_CTRL), 0,
+				~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK);
+
+		/* disable interupt */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_MASTINT_EN), 0,
+				~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+		/* stall UMC and register bus before resetting VCPU */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_LMI_CTRL2),
+				UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
+				~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+		mdelay(1);
+
+		/* put LMI, VCPU, RBC etc... into reset */
+		WREG32_SOC15(UVD, k, mmUVD_SOFT_RESET,
+			UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__RBC_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__CXW_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
+			UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+		mdelay(5);
+
+		/* initialize UVD memory controller */
+		WREG32_SOC15(UVD, k, mmUVD_LMI_CTRL,
+			(0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+			UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+			UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+			UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+			UVD_LMI_CTRL__REQ_MODE_MASK |
+			0x00100000L);
+
+#ifdef __BIG_ENDIAN
+		/* swap (8 in 32) RB and IB */
+		lmi_swap_cntl = 0xa;
+		mp_swap_cntl = 0;
+#endif
+		WREG32_SOC15(UVD, k, mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
+		WREG32_SOC15(UVD, k, mmUVD_MP_SWAP_CNTL, mp_swap_cntl);
+
+		WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUXA0, 0x40c2040);
+		WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUXA1, 0x0);
+		WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUXB0, 0x40c2040);
+		WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUXB1, 0x0);
+		WREG32_SOC15(UVD, k, mmUVD_MPC_SET_ALU, 0);
+		WREG32_SOC15(UVD, k, mmUVD_MPC_SET_MUX, 0x88);
+
+		/* take all subblocks out of reset, except VCPU */
+		WREG32_SOC15(UVD, k, mmUVD_SOFT_RESET,
+				UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(5);
+
+		/* enable VCPU clock */
+		WREG32_SOC15(UVD, k, mmUVD_VCPU_CNTL,
+				UVD_VCPU_CNTL__CLK_EN_MASK);
+
+		/* enable UMC */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_LMI_CTRL2), 0,
+				~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+		/* boot up the VCPU */
+		WREG32_SOC15(UVD, k, mmUVD_SOFT_RESET, 0);
+		mdelay(10);
+
+		for (i = 0; i < 10; ++i) {
+			uint32_t status;
+
+			for (j = 0; j < 100; ++j) {
+				status = RREG32_SOC15(UVD, k, mmUVD_STATUS);
+				if (status & 2)
+					break;
+				mdelay(10);
+			}
+			r = 0;
+			if (status & 2)
+				break;
+
+			DRM_ERROR("UVD(%d) not responding, trying to reset the VCPU!!!\n", k);
+			WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_SOFT_RESET),
+					UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
+					~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+			mdelay(10);
+			WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_SOFT_RESET), 0,
+					~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+			mdelay(10);
+			r = -1;
+		}
+
+		if (r) {
+			DRM_ERROR("UVD(%d) not responding, giving up!!!\n", k);
+			return r;
+		}
+		/* enable master interrupt */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_MASTINT_EN),
+			(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
+			~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK));
+
+		/* clear the bit 4 of UVD_STATUS */
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_STATUS), 0,
+				~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+
+		/* force RBC into idle state */
+		rb_bufsz = order_base_2(ring->ring_size);
+		tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+		WREG32_SOC15(UVD, k, mmUVD_RBC_RB_CNTL, tmp);
+
+		/* set the write pointer delay */
+		WREG32_SOC15(UVD, k, mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+		/* set the wb address */
+		WREG32_SOC15(UVD, k, mmUVD_RBC_RB_RPTR_ADDR,
+				(upper_32_bits(ring->gpu_addr) >> 2));
+
+		/* programm the RB_BASE for ring buffer */
+		WREG32_SOC15(UVD, k, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+				lower_32_bits(ring->gpu_addr));
+		WREG32_SOC15(UVD, k, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+				upper_32_bits(ring->gpu_addr));
+
+		/* Initialize the ring buffer's read and write pointers */
+		WREG32_SOC15(UVD, k, mmUVD_RBC_RB_RPTR, 0);
+
+		ring->wptr = RREG32_SOC15(UVD, k, mmUVD_RBC_RB_RPTR);
+		WREG32_SOC15(UVD, k, mmUVD_RBC_RB_WPTR,
+				lower_32_bits(ring->wptr));
+
+		WREG32_P(SOC15_REG_OFFSET(UVD, k, mmUVD_RBC_RB_CNTL), 0,
+				~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK);
+
+		ring = &adev->uvd.inst[k].ring_enc[0];
+		WREG32_SOC15(UVD, k, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+		WREG32_SOC15(UVD, k, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+		WREG32_SOC15(UVD, k, mmUVD_RB_BASE_LO, ring->gpu_addr);
+		WREG32_SOC15(UVD, k, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+		WREG32_SOC15(UVD, k, mmUVD_RB_SIZE, ring->ring_size / 4);
+
+		ring = &adev->uvd.inst[k].ring_enc[1];
+		WREG32_SOC15(UVD, k, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+		WREG32_SOC15(UVD, k, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+		WREG32_SOC15(UVD, k, mmUVD_RB_BASE_LO2, ring->gpu_addr);
+		WREG32_SOC15(UVD, k, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+		WREG32_SOC15(UVD, k, mmUVD_RB_SIZE2, ring->ring_size / 4);
+	}
+	return 0;
+}
+
+/**
+ * uvd_v7_0_stop - stop UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the UVD block
+ */
+static void uvd_v7_0_stop(struct amdgpu_device *adev)
+{
+	uint8_t i = 0;
+
+	for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		/* force RBC into idle state */
+		WREG32_SOC15(UVD, i, mmUVD_RBC_RB_CNTL, 0x11010101);
+
+		/* Stall UMC and register bus before resetting VCPU */
+		WREG32_P(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL2),
+				UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
+				~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+		mdelay(1);
+
+		/* put VCPU into reset */
+		WREG32_SOC15(UVD, i, mmUVD_SOFT_RESET,
+				UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(5);
+
+		/* disable VCPU clock */
+		WREG32_SOC15(UVD, i, mmUVD_VCPU_CNTL, 0x0);
+
+		/* Unstall UMC and register bus */
+		WREG32_P(SOC15_REG_OFFSET(UVD, i, mmUVD_LMI_CTRL2), 0,
+				~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+	}
+}
+
+/**
+ * uvd_v7_0_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @fence: fence to emit
+ *
+ * Write a fence and a trap command to the ring.
+ */
+static void uvd_v7_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				     unsigned flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_CONTEXT_ID), 0));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA0), 0));
+	amdgpu_ring_write(ring, addr & 0xffffffff);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA1), 0));
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xff);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_CMD), 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA0), 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA1), 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_CMD), 0));
+	amdgpu_ring_write(ring, 2);
+}
+
+/**
+ * uvd_v7_0_enc_ring_emit_fence - emit an enc fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @fence: fence to emit
+ *
+ * Write enc a fence and a trap command to the ring.
+ */
+static void uvd_v7_0_enc_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+			u64 seq, unsigned flags)
+{
+
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_FENCE);
+	amdgpu_ring_write(ring, addr);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_TRAP);
+}
+
+/**
+ * uvd_v7_0_ring_emit_hdp_flush - skip HDP flushing
+ *
+ * @ring: amdgpu_ring pointer
+ */
+static void uvd_v7_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	/* The firmware doesn't seem to like touching registers at this point. */
+}
+
+/**
+ * uvd_v7_0_ring_test_ring - register write test
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully write to the context register
+ */
+static int uvd_v7_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32_SOC15(UVD, ring->me, mmUVD_CONTEXT_ID, 0xCAFEDEAD);
+	r = amdgpu_ring_alloc(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: (%d)cp failed to lock ring %d (%d).\n",
+			  ring->me, ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_CONTEXT_ID), 0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_commit(ring);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32_SOC15(UVD, ring->me, mmUVD_CONTEXT_ID);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_DEBUG("(%d)ring test on %d succeeded in %d usecs\n",
+			 ring->me, ring->idx, i);
+	} else {
+		DRM_ERROR("(%d)amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->me, ring->idx, tmp);
+		r = -EINVAL;
+	}
+	return r;
+}
+
+/**
+ * uvd_v7_0_ring_patch_cs_in_place - Patch the IB for command submission.
+ *
+ * @p: the CS parser with the IBs
+ * @ib_idx: which IB to patch
+ *
+ */
+static int uvd_v7_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p,
+					   uint32_t ib_idx)
+{
+	struct amdgpu_ib *ib = &p->job->ibs[ib_idx];
+	unsigned i;
+
+	/* No patching necessary for the first instance */
+	if (!p->ring->me)
+		return 0;
+
+	for (i = 0; i < ib->length_dw; i += 2) {
+		uint32_t reg = amdgpu_get_ib_value(p, ib_idx, i);
+
+		reg -= p->adev->reg_offset[UVD_HWIP][0][1];
+		reg += p->adev->reg_offset[UVD_HWIP][1][1];
+
+		amdgpu_set_ib_value(p, ib_idx, i, reg);
+	}
+	return 0;
+}
+
+/**
+ * uvd_v7_0_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @ib: indirect buffer to execute
+ *
+ * Write ring commands to execute the indirect buffer
+ */
+static void uvd_v7_0_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib,
+				  unsigned vmid, bool ctx_switch)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_LMI_RBC_IB_VMID), 0));
+	amdgpu_ring_write(ring, vmid);
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_LMI_RBC_IB_64BIT_BAR_LOW), 0));
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH), 0));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_RBC_IB_SIZE), 0));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+/**
+ * uvd_v7_0_enc_ring_emit_ib - enc execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @ib: indirect buffer to execute
+ *
+ * Write enc ring commands to execute the indirect buffer
+ */
+static void uvd_v7_0_enc_ring_emit_ib(struct amdgpu_ring *ring,
+		struct amdgpu_ib *ib, unsigned int vmid, bool ctx_switch)
+{
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_IB_VM);
+	amdgpu_ring_write(ring, vmid);
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+static void uvd_v7_0_ring_emit_wreg(struct amdgpu_ring *ring,
+				    uint32_t reg, uint32_t val)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA0), 0));
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA1), 0));
+	amdgpu_ring_write(ring, val);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_CMD), 0));
+	amdgpu_ring_write(ring, 8);
+}
+
+static void uvd_v7_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+					uint32_t val, uint32_t mask)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA0), 0));
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_DATA1), 0));
+	amdgpu_ring_write(ring, val);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GP_SCRATCH8), 0));
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_GPCOM_VCPU_CMD), 0));
+	amdgpu_ring_write(ring, 12);
+}
+
+static void uvd_v7_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	uint32_t data0, data1, mask;
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for reg writes */
+	data0 = hub->ctx0_ptb_addr_lo32 + vmid * 2;
+	data1 = lower_32_bits(pd_addr);
+	mask = 0xffffffff;
+	uvd_v7_0_ring_emit_reg_wait(ring, data0, data1, mask);
+}
+
+static void uvd_v7_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_NO_OP), 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static void uvd_v7_0_enc_ring_insert_end(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_END);
+}
+
+static void uvd_v7_0_enc_ring_emit_reg_wait(struct amdgpu_ring *ring,
+					    uint32_t reg, uint32_t val,
+					    uint32_t mask)
+{
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_REG_WAIT);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, val);
+}
+
+static void uvd_v7_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					    unsigned int vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for reg writes */
+	uvd_v7_0_enc_ring_emit_reg_wait(ring, hub->ctx0_ptb_addr_lo32 + vmid * 2,
+					lower_32_bits(pd_addr), 0xffffffff);
+}
+
+static void uvd_v7_0_enc_ring_emit_wreg(struct amdgpu_ring *ring,
+					uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, HEVC_ENC_CMD_REG_WRITE);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, val);
+}
+
+#if 0
+static bool uvd_v7_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return !(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK);
+}
+
+static int uvd_v7_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (uvd_v7_0_is_idle(handle))
+			return 0;
+	}
+	return -ETIMEDOUT;
+}
+
+#define AMDGPU_UVD_STATUS_BUSY_MASK    0xfd
+static bool uvd_v7_0_check_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset = 0;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (REG_GET_FIELD(tmp, SRBM_STATUS, UVD_RQ_PENDING) ||
+	    REG_GET_FIELD(tmp, SRBM_STATUS, UVD_BUSY) ||
+	    (RREG32_SOC15(UVD, ring->me, mmUVD_STATUS) &
+		    AMDGPU_UVD_STATUS_BUSY_MASK))
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+				SRBM_SOFT_RESET, SOFT_RESET_UVD, 1);
+
+	if (srbm_soft_reset) {
+		adev->uvd.inst[ring->me].srbm_soft_reset = srbm_soft_reset;
+		return true;
+	} else {
+		adev->uvd.inst[ring->me].srbm_soft_reset = 0;
+		return false;
+	}
+}
+
+static int uvd_v7_0_pre_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->uvd.inst[ring->me].srbm_soft_reset)
+		return 0;
+
+	uvd_v7_0_stop(adev);
+	return 0;
+}
+
+static int uvd_v7_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset;
+
+	if (!adev->uvd.inst[ring->me].srbm_soft_reset)
+		return 0;
+	srbm_soft_reset = adev->uvd.inst[ring->me].srbm_soft_reset;
+
+	if (srbm_soft_reset) {
+		u32 tmp;
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int uvd_v7_0_post_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->uvd.inst[ring->me].srbm_soft_reset)
+		return 0;
+
+	mdelay(5);
+
+	return uvd_v7_0_start(adev);
+}
+#endif
+
+static int uvd_v7_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	// TODO
+	return 0;
+}
+
+static int uvd_v7_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	uint32_t ip_instance;
+
+	switch (entry->client_id) {
+	case SOC15_IH_CLIENTID_UVD:
+		ip_instance = 0;
+		break;
+	case SOC15_IH_CLIENTID_UVD1:
+		ip_instance = 1;
+		break;
+	default:
+		DRM_ERROR("Unhandled client id: %d\n", entry->client_id);
+		return 0;
+	}
+
+	DRM_DEBUG("IH: UVD TRAP\n");
+
+	switch (entry->src_id) {
+	case 124:
+		amdgpu_fence_process(&adev->uvd.inst[ip_instance].ring);
+		break;
+	case 119:
+		amdgpu_fence_process(&adev->uvd.inst[ip_instance].ring_enc[0]);
+		break;
+	case 120:
+		if (!amdgpu_sriov_vf(adev))
+			amdgpu_fence_process(&adev->uvd.inst[ip_instance].ring_enc[1]);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+#if 0
+static void uvd_v7_0_set_sw_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data, data1, data2, suvd_flags;
+
+	data = RREG32_SOC15(UVD, ring->me, mmUVD_CGC_CTRL);
+	data1 = RREG32_SOC15(UVD, ring->me, mmUVD_SUVD_CGC_GATE);
+	data2 = RREG32_SOC15(UVD, ring->me, mmUVD_SUVD_CGC_CTRL);
+
+	data &= ~(UVD_CGC_CTRL__CLK_OFF_DELAY_MASK |
+		  UVD_CGC_CTRL__CLK_GATE_DLY_TIMER_MASK);
+
+	suvd_flags = UVD_SUVD_CGC_GATE__SRE_MASK |
+		     UVD_SUVD_CGC_GATE__SIT_MASK |
+		     UVD_SUVD_CGC_GATE__SMP_MASK |
+		     UVD_SUVD_CGC_GATE__SCM_MASK |
+		     UVD_SUVD_CGC_GATE__SDB_MASK;
+
+	data |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK |
+		(1 << REG_FIELD_SHIFT(UVD_CGC_CTRL, CLK_GATE_DLY_TIMER)) |
+		(4 << REG_FIELD_SHIFT(UVD_CGC_CTRL, CLK_OFF_DELAY));
+
+	data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_CM_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_IT_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_DB_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_MP_MODE_MASK |
+			UVD_CGC_CTRL__SYS_MODE_MASK |
+			UVD_CGC_CTRL__UDEC_MODE_MASK |
+			UVD_CGC_CTRL__MPEG2_MODE_MASK |
+			UVD_CGC_CTRL__REGS_MODE_MASK |
+			UVD_CGC_CTRL__RBC_MODE_MASK |
+			UVD_CGC_CTRL__LMI_MC_MODE_MASK |
+			UVD_CGC_CTRL__LMI_UMC_MODE_MASK |
+			UVD_CGC_CTRL__IDCT_MODE_MASK |
+			UVD_CGC_CTRL__MPRD_MODE_MASK |
+			UVD_CGC_CTRL__MPC_MODE_MASK |
+			UVD_CGC_CTRL__LBSI_MODE_MASK |
+			UVD_CGC_CTRL__LRBBM_MODE_MASK |
+			UVD_CGC_CTRL__WCB_MODE_MASK |
+			UVD_CGC_CTRL__VCPU_MODE_MASK |
+			UVD_CGC_CTRL__JPEG_MODE_MASK |
+			UVD_CGC_CTRL__JPEG2_MODE_MASK |
+			UVD_CGC_CTRL__SCPU_MODE_MASK);
+	data2 &= ~(UVD_SUVD_CGC_CTRL__SRE_MODE_MASK |
+			UVD_SUVD_CGC_CTRL__SIT_MODE_MASK |
+			UVD_SUVD_CGC_CTRL__SMP_MODE_MASK |
+			UVD_SUVD_CGC_CTRL__SCM_MODE_MASK |
+			UVD_SUVD_CGC_CTRL__SDB_MODE_MASK);
+	data1 |= suvd_flags;
+
+	WREG32_SOC15(UVD, ring->me, mmUVD_CGC_CTRL, data);
+	WREG32_SOC15(UVD, ring->me, mmUVD_CGC_GATE, 0);
+	WREG32_SOC15(UVD, ring->me, mmUVD_SUVD_CGC_GATE, data1);
+	WREG32_SOC15(UVD, ring->me, mmUVD_SUVD_CGC_CTRL, data2);
+}
+
+static void uvd_v7_0_set_hw_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data, data1, cgc_flags, suvd_flags;
+
+	data = RREG32_SOC15(UVD, ring->me, mmUVD_CGC_GATE);
+	data1 = RREG32_SOC15(UVD, ring->me, mmUVD_SUVD_CGC_GATE);
+
+	cgc_flags = UVD_CGC_GATE__SYS_MASK |
+		UVD_CGC_GATE__UDEC_MASK |
+		UVD_CGC_GATE__MPEG2_MASK |
+		UVD_CGC_GATE__RBC_MASK |
+		UVD_CGC_GATE__LMI_MC_MASK |
+		UVD_CGC_GATE__IDCT_MASK |
+		UVD_CGC_GATE__MPRD_MASK |
+		UVD_CGC_GATE__MPC_MASK |
+		UVD_CGC_GATE__LBSI_MASK |
+		UVD_CGC_GATE__LRBBM_MASK |
+		UVD_CGC_GATE__UDEC_RE_MASK |
+		UVD_CGC_GATE__UDEC_CM_MASK |
+		UVD_CGC_GATE__UDEC_IT_MASK |
+		UVD_CGC_GATE__UDEC_DB_MASK |
+		UVD_CGC_GATE__UDEC_MP_MASK |
+		UVD_CGC_GATE__WCB_MASK |
+		UVD_CGC_GATE__VCPU_MASK |
+		UVD_CGC_GATE__SCPU_MASK |
+		UVD_CGC_GATE__JPEG_MASK |
+		UVD_CGC_GATE__JPEG2_MASK;
+
+	suvd_flags = UVD_SUVD_CGC_GATE__SRE_MASK |
+				UVD_SUVD_CGC_GATE__SIT_MASK |
+				UVD_SUVD_CGC_GATE__SMP_MASK |
+				UVD_SUVD_CGC_GATE__SCM_MASK |
+				UVD_SUVD_CGC_GATE__SDB_MASK;
+
+	data |= cgc_flags;
+	data1 |= suvd_flags;
+
+	WREG32_SOC15(UVD, ring->me, mmUVD_CGC_GATE, data);
+	WREG32_SOC15(UVD, ring->me, mmUVD_SUVD_CGC_GATE, data1);
+}
+
+static void uvd_v7_0_set_bypass_mode(struct amdgpu_device *adev, bool enable)
+{
+	u32 tmp = RREG32_SMC(ixGCK_DFS_BYPASS_CNTL);
+
+	if (enable)
+		tmp |= (GCK_DFS_BYPASS_CNTL__BYPASSDCLK_MASK |
+			GCK_DFS_BYPASS_CNTL__BYPASSVCLK_MASK);
+	else
+		tmp &= ~(GCK_DFS_BYPASS_CNTL__BYPASSDCLK_MASK |
+			 GCK_DFS_BYPASS_CNTL__BYPASSVCLK_MASK);
+
+	WREG32_SMC(ixGCK_DFS_BYPASS_CNTL, tmp);
+}
+
+
+static int uvd_v7_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+
+	uvd_v7_0_set_bypass_mode(adev, enable);
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_UVD_MGCG))
+		return 0;
+
+	if (enable) {
+		/* disable HW gating and enable Sw gating */
+		uvd_v7_0_set_sw_clock_gating(adev);
+	} else {
+		/* wait for STATUS to clear */
+		if (uvd_v7_0_wait_for_idle(handle))
+			return -EBUSY;
+
+		/* enable HW gates because UVD is idle */
+		/* uvd_v7_0_set_hw_clock_gating(adev); */
+	}
+
+	return 0;
+}
+
+static int uvd_v7_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the UVD block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!(adev->pg_flags & AMD_PG_SUPPORT_UVD))
+		return 0;
+
+	WREG32_SOC15(UVD, ring->me, mmUVD_POWER_STATUS, UVD_POWER_STATUS__UVD_PG_EN_MASK);
+
+	if (state == AMD_PG_STATE_GATE) {
+		uvd_v7_0_stop(adev);
+		return 0;
+	} else {
+		return uvd_v7_0_start(adev);
+	}
+}
+#endif
+
+static int uvd_v7_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	/* needed for driver unload*/
+	return 0;
+}
+
+const struct amd_ip_funcs uvd_v7_0_ip_funcs = {
+	.name = "uvd_v7_0",
+	.early_init = uvd_v7_0_early_init,
+	.late_init = NULL,
+	.sw_init = uvd_v7_0_sw_init,
+	.sw_fini = uvd_v7_0_sw_fini,
+	.hw_init = uvd_v7_0_hw_init,
+	.hw_fini = uvd_v7_0_hw_fini,
+	.suspend = uvd_v7_0_suspend,
+	.resume = uvd_v7_0_resume,
+	.is_idle = NULL /* uvd_v7_0_is_idle */,
+	.wait_for_idle = NULL /* uvd_v7_0_wait_for_idle */,
+	.check_soft_reset = NULL /* uvd_v7_0_check_soft_reset */,
+	.pre_soft_reset = NULL /* uvd_v7_0_pre_soft_reset */,
+	.soft_reset = NULL /* uvd_v7_0_soft_reset */,
+	.post_soft_reset = NULL /* uvd_v7_0_post_soft_reset */,
+	.set_clockgating_state = uvd_v7_0_set_clockgating_state,
+	.set_powergating_state = NULL /* uvd_v7_0_set_powergating_state */,
+};
+
+static const struct amdgpu_ring_funcs uvd_v7_0_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_UVD,
+	.align_mask = 0xf,
+	.support_64bit_ptrs = false,
+	.vmhub = AMDGPU_MMHUB,
+	.get_rptr = uvd_v7_0_ring_get_rptr,
+	.get_wptr = uvd_v7_0_ring_get_wptr,
+	.set_wptr = uvd_v7_0_ring_set_wptr,
+	.patch_cs_in_place = uvd_v7_0_ring_patch_cs_in_place,
+	.emit_frame_size =
+		6 + /* hdp invalidate */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* uvd_v7_0_ring_emit_vm_flush */
+		14 + 14, /* uvd_v7_0_ring_emit_fence x2 vm fence */
+	.emit_ib_size = 8, /* uvd_v7_0_ring_emit_ib */
+	.emit_ib = uvd_v7_0_ring_emit_ib,
+	.emit_fence = uvd_v7_0_ring_emit_fence,
+	.emit_vm_flush = uvd_v7_0_ring_emit_vm_flush,
+	.emit_hdp_flush = uvd_v7_0_ring_emit_hdp_flush,
+	.test_ring = uvd_v7_0_ring_test_ring,
+	.test_ib = amdgpu_uvd_ring_test_ib,
+	.insert_nop = uvd_v7_0_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
+	.emit_wreg = uvd_v7_0_ring_emit_wreg,
+	.emit_reg_wait = uvd_v7_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+};
+
+static const struct amdgpu_ring_funcs uvd_v7_0_enc_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_UVD_ENC,
+	.align_mask = 0x3f,
+	.nop = HEVC_ENC_CMD_NO_OP,
+	.support_64bit_ptrs = false,
+	.vmhub = AMDGPU_MMHUB,
+	.get_rptr = uvd_v7_0_enc_ring_get_rptr,
+	.get_wptr = uvd_v7_0_enc_ring_get_wptr,
+	.set_wptr = uvd_v7_0_enc_ring_set_wptr,
+	.emit_frame_size =
+		3 + 3 + /* hdp flush / invalidate */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* uvd_v7_0_enc_ring_emit_vm_flush */
+		5 + 5 + /* uvd_v7_0_enc_ring_emit_fence x2 vm fence */
+		1, /* uvd_v7_0_enc_ring_insert_end */
+	.emit_ib_size = 5, /* uvd_v7_0_enc_ring_emit_ib */
+	.emit_ib = uvd_v7_0_enc_ring_emit_ib,
+	.emit_fence = uvd_v7_0_enc_ring_emit_fence,
+	.emit_vm_flush = uvd_v7_0_enc_ring_emit_vm_flush,
+	.test_ring = uvd_v7_0_enc_ring_test_ring,
+	.test_ib = uvd_v7_0_enc_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = uvd_v7_0_enc_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
+	.emit_wreg = uvd_v7_0_enc_ring_emit_wreg,
+	.emit_reg_wait = uvd_v7_0_enc_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+};
+
+static void uvd_v7_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		adev->uvd.inst[i].ring.funcs = &uvd_v7_0_ring_vm_funcs;
+		adev->uvd.inst[i].ring.me = i;
+		DRM_INFO("UVD(%d) is enabled in VM mode\n", i);
+	}
+}
+
+static void uvd_v7_0_set_enc_ring_funcs(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	for (j = 0; j < adev->uvd.num_uvd_inst; j++) {
+		if (adev->uvd.harvest_config & (1 << j))
+			continue;
+		for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
+			adev->uvd.inst[j].ring_enc[i].funcs = &uvd_v7_0_enc_ring_vm_funcs;
+			adev->uvd.inst[j].ring_enc[i].me = j;
+		}
+
+		DRM_INFO("UVD(%d) ENC is enabled in VM mode\n", j);
+	}
+}
+
+static const struct amdgpu_irq_src_funcs uvd_v7_0_irq_funcs = {
+	.set = uvd_v7_0_set_interrupt_state,
+	.process = uvd_v7_0_process_interrupt,
+};
+
+static void uvd_v7_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
+		if (adev->uvd.harvest_config & (1 << i))
+			continue;
+		adev->uvd.inst[i].irq.num_types = adev->uvd.num_enc_rings + 1;
+		adev->uvd.inst[i].irq.funcs = &uvd_v7_0_irq_funcs;
+	}
+}
+
+const struct amdgpu_ip_block_version uvd_v7_0_ip_block =
+{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 7,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &uvd_v7_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/uvd_v7_0.h b/sys/dev/pci/drm/amd/amdgpu/uvd_v7_0.h
new file mode 100644
index 00000000000..cbe82ab3224
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/uvd_v7_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __UVD_V7_0_H__
+#define __UVD_V7_0_H__
+
+extern const struct amdgpu_ip_block_version uvd_v7_0_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/vce_v2_0.h b/sys/dev/pci/drm/amd/amdgpu/vce_v2_0.h
new file mode 100644
index 00000000000..4d15167654a
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vce_v2_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCE_V2_0_H__
+#define __VCE_V2_0_H__
+
+extern const struct amdgpu_ip_block_version vce_v2_0_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/vce_v3_0.c b/sys/dev/pci/drm/amd/amdgpu/vce_v3_0.c
new file mode 100644
index 00000000000..55480177a8c
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vce_v3_0.c
@@ -0,0 +1,998 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * Authors: Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_vce.h"
+#include "vid.h"
+#include "vce/vce_3_0_d.h"
+#include "vce/vce_3_0_sh_mask.h"
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+#include "gca/gfx_8_0_d.h"
+#include "smu/smu_7_1_2_d.h"
+#include "smu/smu_7_1_2_sh_mask.h"
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_sh_mask.h"
+#include "ivsrcid/ivsrcid_vislands30.h"
+
+
+#define GRBM_GFX_INDEX__VCE_INSTANCE__SHIFT	0x04
+#define GRBM_GFX_INDEX__VCE_INSTANCE_MASK	0x10
+#define GRBM_GFX_INDEX__VCE_ALL_PIPE		0x07
+
+#define mmVCE_LMI_VCPU_CACHE_40BIT_BAR0	0x8616
+#define mmVCE_LMI_VCPU_CACHE_40BIT_BAR1	0x8617
+#define mmVCE_LMI_VCPU_CACHE_40BIT_BAR2	0x8618
+#define mmGRBM_GFX_INDEX_DEFAULT 0xE0000000
+
+#define VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK	0x02
+
+#define VCE_V3_0_FW_SIZE	(384 * 1024)
+#define VCE_V3_0_STACK_SIZE	(64 * 1024)
+#define VCE_V3_0_DATA_SIZE	((16 * 1024 * AMDGPU_MAX_VCE_HANDLES) + (52 * 1024))
+
+#define FW_52_8_3	((52 << 24) | (8 << 16) | (3 << 8))
+
+#define GET_VCE_INSTANCE(i)  ((i) << GRBM_GFX_INDEX__VCE_INSTANCE__SHIFT \
+					| GRBM_GFX_INDEX__VCE_ALL_PIPE)
+
+static void vce_v3_0_mc_resume(struct amdgpu_device *adev, int idx);
+static void vce_v3_0_set_ring_funcs(struct amdgpu_device *adev);
+static void vce_v3_0_set_irq_funcs(struct amdgpu_device *adev);
+static int vce_v3_0_wait_for_idle(void *handle);
+static int vce_v3_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state);
+/**
+ * vce_v3_0_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t vce_v3_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 v;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	if (adev->vce.harvest_config == 0 ||
+		adev->vce.harvest_config == AMDGPU_VCE_HARVEST_VCE1)
+		WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(0));
+	else if (adev->vce.harvest_config == AMDGPU_VCE_HARVEST_VCE0)
+		WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(1));
+
+	if (ring->me == 0)
+		v = RREG32(mmVCE_RB_RPTR);
+	else if (ring->me == 1)
+		v = RREG32(mmVCE_RB_RPTR2);
+	else
+		v = RREG32(mmVCE_RB_RPTR3);
+
+	WREG32(mmGRBM_GFX_INDEX, mmGRBM_GFX_INDEX_DEFAULT);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return v;
+}
+
+/**
+ * vce_v3_0_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t vce_v3_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 v;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	if (adev->vce.harvest_config == 0 ||
+		adev->vce.harvest_config == AMDGPU_VCE_HARVEST_VCE1)
+		WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(0));
+	else if (adev->vce.harvest_config == AMDGPU_VCE_HARVEST_VCE0)
+		WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(1));
+
+	if (ring->me == 0)
+		v = RREG32(mmVCE_RB_WPTR);
+	else if (ring->me == 1)
+		v = RREG32(mmVCE_RB_WPTR2);
+	else
+		v = RREG32(mmVCE_RB_WPTR3);
+
+	WREG32(mmGRBM_GFX_INDEX, mmGRBM_GFX_INDEX_DEFAULT);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return v;
+}
+
+/**
+ * vce_v3_0_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void vce_v3_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	if (adev->vce.harvest_config == 0 ||
+		adev->vce.harvest_config == AMDGPU_VCE_HARVEST_VCE1)
+		WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(0));
+	else if (adev->vce.harvest_config == AMDGPU_VCE_HARVEST_VCE0)
+		WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(1));
+
+	if (ring->me == 0)
+		WREG32(mmVCE_RB_WPTR, lower_32_bits(ring->wptr));
+	else if (ring->me == 1)
+		WREG32(mmVCE_RB_WPTR2, lower_32_bits(ring->wptr));
+	else
+		WREG32(mmVCE_RB_WPTR3, lower_32_bits(ring->wptr));
+
+	WREG32(mmGRBM_GFX_INDEX, mmGRBM_GFX_INDEX_DEFAULT);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+static void vce_v3_0_override_vce_clock_gating(struct amdgpu_device *adev, bool override)
+{
+	WREG32_FIELD(VCE_RB_ARB_CTRL, VCE_CGTT_OVERRIDE, override ? 1 : 0);
+}
+
+static void vce_v3_0_set_vce_sw_clock_gating(struct amdgpu_device *adev,
+					     bool gated)
+{
+	u32 data;
+
+	/* Set Override to disable Clock Gating */
+	vce_v3_0_override_vce_clock_gating(adev, true);
+
+	/* This function enables MGCG which is controlled by firmware.
+	   With the clocks in the gated state the core is still
+	   accessible but the firmware will throttle the clocks on the
+	   fly as necessary.
+	*/
+	if (!gated) {
+		data = RREG32(mmVCE_CLOCK_GATING_B);
+		data |= 0x1ff;
+		data &= ~0xef0000;
+		WREG32(mmVCE_CLOCK_GATING_B, data);
+
+		data = RREG32(mmVCE_UENC_CLOCK_GATING);
+		data |= 0x3ff000;
+		data &= ~0xffc00000;
+		WREG32(mmVCE_UENC_CLOCK_GATING, data);
+
+		data = RREG32(mmVCE_UENC_CLOCK_GATING_2);
+		data |= 0x2;
+		data &= ~0x00010000;
+		WREG32(mmVCE_UENC_CLOCK_GATING_2, data);
+
+		data = RREG32(mmVCE_UENC_REG_CLOCK_GATING);
+		data |= 0x37f;
+		WREG32(mmVCE_UENC_REG_CLOCK_GATING, data);
+
+		data = RREG32(mmVCE_UENC_DMA_DCLK_CTRL);
+		data |= VCE_UENC_DMA_DCLK_CTRL__WRDMCLK_FORCEON_MASK |
+			VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK |
+			VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK  |
+			0x8;
+		WREG32(mmVCE_UENC_DMA_DCLK_CTRL, data);
+	} else {
+		data = RREG32(mmVCE_CLOCK_GATING_B);
+		data &= ~0x80010;
+		data |= 0xe70008;
+		WREG32(mmVCE_CLOCK_GATING_B, data);
+
+		data = RREG32(mmVCE_UENC_CLOCK_GATING);
+		data |= 0xffc00000;
+		WREG32(mmVCE_UENC_CLOCK_GATING, data);
+
+		data = RREG32(mmVCE_UENC_CLOCK_GATING_2);
+		data |= 0x10000;
+		WREG32(mmVCE_UENC_CLOCK_GATING_2, data);
+
+		data = RREG32(mmVCE_UENC_REG_CLOCK_GATING);
+		data &= ~0x3ff;
+		WREG32(mmVCE_UENC_REG_CLOCK_GATING, data);
+
+		data = RREG32(mmVCE_UENC_DMA_DCLK_CTRL);
+		data &= ~(VCE_UENC_DMA_DCLK_CTRL__WRDMCLK_FORCEON_MASK |
+			  VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK |
+			  VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK  |
+			  0x8);
+		WREG32(mmVCE_UENC_DMA_DCLK_CTRL, data);
+	}
+	vce_v3_0_override_vce_clock_gating(adev, false);
+}
+
+static int vce_v3_0_firmware_loaded(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	for (i = 0; i < 10; ++i) {
+		for (j = 0; j < 100; ++j) {
+			uint32_t status = RREG32(mmVCE_STATUS);
+
+			if (status & VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK)
+				return 0;
+			mdelay(10);
+		}
+
+		DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n");
+		WREG32_FIELD(VCE_SOFT_RESET, ECPU_SOFT_RESET, 1);
+		mdelay(10);
+		WREG32_FIELD(VCE_SOFT_RESET, ECPU_SOFT_RESET, 0);
+		mdelay(10);
+	}
+
+	return -ETIMEDOUT;
+}
+
+/**
+ * vce_v3_0_start - start VCE block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the VCE block
+ */
+static int vce_v3_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	int idx, r;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (idx = 0; idx < 2; ++idx) {
+		if (adev->vce.harvest_config & (1 << idx))
+			continue;
+
+		WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(idx));
+
+		/* Program instance 0 reg space for two instances or instance 0 case
+		program instance 1 reg space for only instance 1 available case */
+		if (idx != 1 || adev->vce.harvest_config == AMDGPU_VCE_HARVEST_VCE0) {
+			ring = &adev->vce.ring[0];
+			WREG32(mmVCE_RB_RPTR, lower_32_bits(ring->wptr));
+			WREG32(mmVCE_RB_WPTR, lower_32_bits(ring->wptr));
+			WREG32(mmVCE_RB_BASE_LO, ring->gpu_addr);
+			WREG32(mmVCE_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+			WREG32(mmVCE_RB_SIZE, ring->ring_size / 4);
+
+			ring = &adev->vce.ring[1];
+			WREG32(mmVCE_RB_RPTR2, lower_32_bits(ring->wptr));
+			WREG32(mmVCE_RB_WPTR2, lower_32_bits(ring->wptr));
+			WREG32(mmVCE_RB_BASE_LO2, ring->gpu_addr);
+			WREG32(mmVCE_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+			WREG32(mmVCE_RB_SIZE2, ring->ring_size / 4);
+
+			ring = &adev->vce.ring[2];
+			WREG32(mmVCE_RB_RPTR3, lower_32_bits(ring->wptr));
+			WREG32(mmVCE_RB_WPTR3, lower_32_bits(ring->wptr));
+			WREG32(mmVCE_RB_BASE_LO3, ring->gpu_addr);
+			WREG32(mmVCE_RB_BASE_HI3, upper_32_bits(ring->gpu_addr));
+			WREG32(mmVCE_RB_SIZE3, ring->ring_size / 4);
+		}
+
+		vce_v3_0_mc_resume(adev, idx);
+		WREG32_FIELD(VCE_STATUS, JOB_BUSY, 1);
+
+		if (adev->asic_type >= CHIP_STONEY)
+			WREG32_P(mmVCE_VCPU_CNTL, 1, ~0x200001);
+		else
+			WREG32_FIELD(VCE_VCPU_CNTL, CLK_EN, 1);
+
+		WREG32_FIELD(VCE_SOFT_RESET, ECPU_SOFT_RESET, 0);
+		mdelay(100);
+
+		r = vce_v3_0_firmware_loaded(adev);
+
+		/* clear BUSY flag */
+		WREG32_FIELD(VCE_STATUS, JOB_BUSY, 0);
+
+		if (r) {
+			DRM_ERROR("VCE not responding, giving up!!!\n");
+			mutex_unlock(&adev->grbm_idx_mutex);
+			return r;
+		}
+	}
+
+	WREG32(mmGRBM_GFX_INDEX, mmGRBM_GFX_INDEX_DEFAULT);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+static int vce_v3_0_stop(struct amdgpu_device *adev)
+{
+	int idx;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (idx = 0; idx < 2; ++idx) {
+		if (adev->vce.harvest_config & (1 << idx))
+			continue;
+
+		WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(idx));
+
+		if (adev->asic_type >= CHIP_STONEY)
+			WREG32_P(mmVCE_VCPU_CNTL, 0, ~0x200001);
+		else
+			WREG32_FIELD(VCE_VCPU_CNTL, CLK_EN, 0);
+
+		/* hold on ECPU */
+		WREG32_FIELD(VCE_SOFT_RESET, ECPU_SOFT_RESET, 1);
+
+		/* clear VCE STATUS */
+		WREG32(mmVCE_STATUS, 0);
+	}
+
+	WREG32(mmGRBM_GFX_INDEX, mmGRBM_GFX_INDEX_DEFAULT);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+#define ixVCE_HARVEST_FUSE_MACRO__ADDRESS     0xC0014074
+#define VCE_HARVEST_FUSE_MACRO__SHIFT       27
+#define VCE_HARVEST_FUSE_MACRO__MASK        0x18000000
+
+static unsigned vce_v3_0_get_harvest_config(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	if ((adev->asic_type == CHIP_FIJI) ||
+	    (adev->asic_type == CHIP_STONEY))
+		return AMDGPU_VCE_HARVEST_VCE1;
+
+	if (adev->flags & AMD_IS_APU)
+		tmp = (RREG32_SMC(ixVCE_HARVEST_FUSE_MACRO__ADDRESS) &
+		       VCE_HARVEST_FUSE_MACRO__MASK) >>
+			VCE_HARVEST_FUSE_MACRO__SHIFT;
+	else
+		tmp = (RREG32_SMC(ixCC_HARVEST_FUSES) &
+		       CC_HARVEST_FUSES__VCE_DISABLE_MASK) >>
+			CC_HARVEST_FUSES__VCE_DISABLE__SHIFT;
+
+	switch (tmp) {
+	case 1:
+		return AMDGPU_VCE_HARVEST_VCE0;
+	case 2:
+		return AMDGPU_VCE_HARVEST_VCE1;
+	case 3:
+		return AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1;
+	default:
+		if ((adev->asic_type == CHIP_POLARIS10) ||
+		    (adev->asic_type == CHIP_POLARIS11) ||
+		    (adev->asic_type == CHIP_POLARIS12) ||
+		    (adev->asic_type == CHIP_VEGAM))
+			return AMDGPU_VCE_HARVEST_VCE1;
+
+		return 0;
+	}
+}
+
+static int vce_v3_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->vce.harvest_config = vce_v3_0_get_harvest_config(adev);
+
+	if ((adev->vce.harvest_config &
+	     (AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1)) ==
+	    (AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1))
+		return -ENOENT;
+
+	adev->vce.num_rings = 3;
+
+	vce_v3_0_set_ring_funcs(adev);
+	vce_v3_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int vce_v3_0_sw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring;
+	int r, i;
+
+	/* VCE */
+	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, VISLANDS30_IV_SRCID_VCE_TRAP, &adev->vce.irq);
+	if (r)
+		return r;
+
+	r = amdgpu_vce_sw_init(adev, VCE_V3_0_FW_SIZE +
+		(VCE_V3_0_STACK_SIZE + VCE_V3_0_DATA_SIZE) * 2);
+	if (r)
+		return r;
+
+	/* 52.8.3 required for 3 ring support */
+	if (adev->vce.fw_version < FW_52_8_3)
+		adev->vce.num_rings = 2;
+
+	r = amdgpu_vce_resume(adev);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->vce.num_rings; i++) {
+		ring = &adev->vce.ring[i];
+		snprintf(ring->name, sizeof(ring->name), "vce%d", i);
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_vce_entity_init(adev);
+
+	return r;
+}
+
+static int vce_v3_0_sw_fini(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_vce_suspend(adev);
+	if (r)
+		return r;
+
+	return amdgpu_vce_sw_fini(adev);
+}
+
+static int vce_v3_0_hw_init(void *handle)
+{
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	vce_v3_0_override_vce_clock_gating(adev, true);
+
+	amdgpu_asic_set_vce_clocks(adev, 10000, 10000);
+
+	for (i = 0; i < adev->vce.num_rings; i++)
+		adev->vce.ring[i].ready = false;
+
+	for (i = 0; i < adev->vce.num_rings; i++) {
+		r = amdgpu_ring_test_ring(&adev->vce.ring[i]);
+		if (r)
+			return r;
+		else
+			adev->vce.ring[i].ready = true;
+	}
+
+	DRM_INFO("VCE initialized successfully.\n");
+
+	return 0;
+}
+
+static int vce_v3_0_hw_fini(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = vce_v3_0_wait_for_idle(handle);
+	if (r)
+		return r;
+
+	vce_v3_0_stop(adev);
+	return vce_v3_0_set_clockgating_state(adev, AMD_CG_STATE_GATE);
+}
+
+static int vce_v3_0_suspend(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = vce_v3_0_hw_fini(adev);
+	if (r)
+		return r;
+
+	return amdgpu_vce_suspend(adev);
+}
+
+static int vce_v3_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_vce_resume(adev);
+	if (r)
+		return r;
+
+	return vce_v3_0_hw_init(adev);
+}
+
+static void vce_v3_0_mc_resume(struct amdgpu_device *adev, int idx)
+{
+	uint32_t offset, size;
+
+	WREG32_P(mmVCE_CLOCK_GATING_A, 0, ~(1 << 16));
+	WREG32_P(mmVCE_UENC_CLOCK_GATING, 0x1FF000, ~0xFF9FF000);
+	WREG32_P(mmVCE_UENC_REG_CLOCK_GATING, 0x3F, ~0x3F);
+	WREG32(mmVCE_CLOCK_GATING_B, 0x1FF);
+
+	WREG32(mmVCE_LMI_CTRL, 0x00398000);
+	WREG32_P(mmVCE_LMI_CACHE_CTRL, 0x0, ~0x1);
+	WREG32(mmVCE_LMI_SWAP_CNTL, 0);
+	WREG32(mmVCE_LMI_SWAP_CNTL1, 0);
+	WREG32(mmVCE_LMI_VM_CTRL, 0);
+	WREG32_OR(mmVCE_VCPU_CNTL, 0x00100000);
+
+	if (adev->asic_type >= CHIP_STONEY) {
+		WREG32(mmVCE_LMI_VCPU_CACHE_40BIT_BAR0, (adev->vce.gpu_addr >> 8));
+		WREG32(mmVCE_LMI_VCPU_CACHE_40BIT_BAR1, (adev->vce.gpu_addr >> 8));
+		WREG32(mmVCE_LMI_VCPU_CACHE_40BIT_BAR2, (adev->vce.gpu_addr >> 8));
+	} else
+		WREG32(mmVCE_LMI_VCPU_CACHE_40BIT_BAR, (adev->vce.gpu_addr >> 8));
+	offset = AMDGPU_VCE_FIRMWARE_OFFSET;
+	size = VCE_V3_0_FW_SIZE;
+	WREG32(mmVCE_VCPU_CACHE_OFFSET0, offset & 0x7fffffff);
+	WREG32(mmVCE_VCPU_CACHE_SIZE0, size);
+
+	if (idx == 0) {
+		offset += size;
+		size = VCE_V3_0_STACK_SIZE;
+		WREG32(mmVCE_VCPU_CACHE_OFFSET1, offset & 0x7fffffff);
+		WREG32(mmVCE_VCPU_CACHE_SIZE1, size);
+		offset += size;
+		size = VCE_V3_0_DATA_SIZE;
+		WREG32(mmVCE_VCPU_CACHE_OFFSET2, offset & 0x7fffffff);
+		WREG32(mmVCE_VCPU_CACHE_SIZE2, size);
+	} else {
+		offset += size + VCE_V3_0_STACK_SIZE + VCE_V3_0_DATA_SIZE;
+		size = VCE_V3_0_STACK_SIZE;
+		WREG32(mmVCE_VCPU_CACHE_OFFSET1, offset & 0xfffffff);
+		WREG32(mmVCE_VCPU_CACHE_SIZE1, size);
+		offset += size;
+		size = VCE_V3_0_DATA_SIZE;
+		WREG32(mmVCE_VCPU_CACHE_OFFSET2, offset & 0xfffffff);
+		WREG32(mmVCE_VCPU_CACHE_SIZE2, size);
+	}
+
+	WREG32_P(mmVCE_LMI_CTRL2, 0x0, ~0x100);
+	WREG32_FIELD(VCE_SYS_INT_EN, VCE_SYS_INT_TRAP_INTERRUPT_EN, 1);
+}
+
+static bool vce_v3_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 mask = 0;
+
+	mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE0) ? 0 : SRBM_STATUS2__VCE0_BUSY_MASK;
+	mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE1) ? 0 : SRBM_STATUS2__VCE1_BUSY_MASK;
+
+	return !(RREG32(mmSRBM_STATUS2) & mask);
+}
+
+static int vce_v3_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++)
+		if (vce_v3_0_is_idle(handle))
+			return 0;
+
+	return -ETIMEDOUT;
+}
+
+#define  VCE_STATUS_VCPU_REPORT_AUTO_BUSY_MASK  0x00000008L   /* AUTO_BUSY */
+#define  VCE_STATUS_VCPU_REPORT_RB0_BUSY_MASK   0x00000010L   /* RB0_BUSY */
+#define  VCE_STATUS_VCPU_REPORT_RB1_BUSY_MASK   0x00000020L   /* RB1_BUSY */
+#define  AMDGPU_VCE_STATUS_BUSY_MASK (VCE_STATUS_VCPU_REPORT_AUTO_BUSY_MASK | \
+				      VCE_STATUS_VCPU_REPORT_RB0_BUSY_MASK)
+
+static bool vce_v3_0_check_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset = 0;
+
+	/* According to VCE team , we should use VCE_STATUS instead
+	 * SRBM_STATUS.VCE_BUSY bit for busy status checking.
+	 * GRBM_GFX_INDEX.INSTANCE_INDEX is used to specify which VCE
+	 * instance's registers are accessed
+	 * (0 for 1st instance, 10 for 2nd instance).
+	 *
+	 *VCE_STATUS
+	 *|UENC|ACPI|AUTO ACTIVE|RB1 |RB0 |RB2 |          |FW_LOADED|JOB |
+	 *|----+----+-----------+----+----+----+----------+---------+----|
+	 *|bit8|bit7|    bit6   |bit5|bit4|bit3|   bit2   |  bit1   |bit0|
+	 *
+	 * VCE team suggest use bit 3--bit 6 for busy status check
+	 */
+	mutex_lock(&adev->grbm_idx_mutex);
+	WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(0));
+	if (RREG32(mmVCE_STATUS) & AMDGPU_VCE_STATUS_BUSY_MASK) {
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE0, 1);
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE1, 1);
+	}
+	WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(1));
+	if (RREG32(mmVCE_STATUS) & AMDGPU_VCE_STATUS_BUSY_MASK) {
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE0, 1);
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE1, 1);
+	}
+	WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(0));
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	if (srbm_soft_reset) {
+		adev->vce.srbm_soft_reset = srbm_soft_reset;
+		return true;
+	} else {
+		adev->vce.srbm_soft_reset = 0;
+		return false;
+	}
+}
+
+static int vce_v3_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset;
+
+	if (!adev->vce.srbm_soft_reset)
+		return 0;
+	srbm_soft_reset = adev->vce.srbm_soft_reset;
+
+	if (srbm_soft_reset) {
+		u32 tmp;
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int vce_v3_0_pre_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->vce.srbm_soft_reset)
+		return 0;
+
+	mdelay(5);
+
+	return vce_v3_0_suspend(adev);
+}
+
+
+static int vce_v3_0_post_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->vce.srbm_soft_reset)
+		return 0;
+
+	mdelay(5);
+
+	return vce_v3_0_resume(adev);
+}
+
+static int vce_v3_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	uint32_t val = 0;
+
+	if (state == AMDGPU_IRQ_STATE_ENABLE)
+		val |= VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK;
+
+	WREG32_P(mmVCE_SYS_INT_EN, val, ~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
+	return 0;
+}
+
+static int vce_v3_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: VCE\n");
+
+	WREG32_FIELD(VCE_SYS_INT_STATUS, VCE_SYS_INT_TRAP_INTERRUPT_INT, 1);
+
+	switch (entry->src_data[0]) {
+	case 0:
+	case 1:
+	case 2:
+		amdgpu_fence_process(&adev->vce.ring[entry->src_data[0]]);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static int vce_v3_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+	int i;
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG))
+		return 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < 2; i++) {
+		/* Program VCE Instance 0 or 1 if not harvested */
+		if (adev->vce.harvest_config & (1 << i))
+			continue;
+
+		WREG32(mmGRBM_GFX_INDEX, GET_VCE_INSTANCE(i));
+
+		if (!enable) {
+			/* initialize VCE_CLOCK_GATING_A: Clock ON/OFF delay */
+			uint32_t data = RREG32(mmVCE_CLOCK_GATING_A);
+			data &= ~(0xf | 0xff0);
+			data |= ((0x0 << 0) | (0x04 << 4));
+			WREG32(mmVCE_CLOCK_GATING_A, data);
+
+			/* initialize VCE_UENC_CLOCK_GATING: Clock ON/OFF delay */
+			data = RREG32(mmVCE_UENC_CLOCK_GATING);
+			data &= ~(0xf | 0xff0);
+			data |= ((0x0 << 0) | (0x04 << 4));
+			WREG32(mmVCE_UENC_CLOCK_GATING, data);
+		}
+
+		vce_v3_0_set_vce_sw_clock_gating(adev, enable);
+	}
+
+	WREG32(mmGRBM_GFX_INDEX, mmGRBM_GFX_INDEX_DEFAULT);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+static int vce_v3_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the VCE block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret = 0;
+
+	if (state == AMD_PG_STATE_GATE) {
+		ret = vce_v3_0_stop(adev);
+		if (ret)
+			goto out;
+	} else {
+		ret = vce_v3_0_start(adev);
+		if (ret)
+			goto out;
+	}
+
+out:
+	return ret;
+}
+
+static void vce_v3_0_get_clockgating_state(void *handle, u32 *flags)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int data;
+
+	mutex_lock(&adev->pm.mutex);
+
+	if (adev->flags & AMD_IS_APU)
+		data = RREG32_SMC(ixCURRENT_PG_STATUS_APU);
+	else
+		data = RREG32_SMC(ixCURRENT_PG_STATUS);
+
+	if (data & CURRENT_PG_STATUS__VCE_PG_STATUS_MASK) {
+		DRM_INFO("Cannot get clockgating state when VCE is powergated.\n");
+		goto out;
+	}
+
+	WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, 0);
+
+	/* AMD_CG_SUPPORT_VCE_MGCG */
+	data = RREG32(mmVCE_CLOCK_GATING_A);
+	if (data & (0x04 << 4))
+		*flags |= AMD_CG_SUPPORT_VCE_MGCG;
+
+out:
+	mutex_unlock(&adev->pm.mutex);
+}
+
+static void vce_v3_0_ring_emit_ib(struct amdgpu_ring *ring,
+		struct amdgpu_ib *ib, unsigned int vmid, bool ctx_switch)
+{
+	amdgpu_ring_write(ring, VCE_CMD_IB_VM);
+	amdgpu_ring_write(ring, vmid);
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+static void vce_v3_0_emit_vm_flush(struct amdgpu_ring *ring,
+				   unsigned int vmid, uint64_t pd_addr)
+{
+	amdgpu_ring_write(ring, VCE_CMD_UPDATE_PTB);
+	amdgpu_ring_write(ring, vmid);
+	amdgpu_ring_write(ring, pd_addr >> 12);
+
+	amdgpu_ring_write(ring, VCE_CMD_FLUSH_TLB);
+	amdgpu_ring_write(ring, vmid);
+	amdgpu_ring_write(ring, VCE_CMD_END);
+}
+
+static void vce_v3_0_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	amdgpu_ring_write(ring, VCE_CMD_WAIT_GE);
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq);
+}
+
+static const struct amd_ip_funcs vce_v3_0_ip_funcs = {
+	.name = "vce_v3_0",
+	.early_init = vce_v3_0_early_init,
+	.late_init = NULL,
+	.sw_init = vce_v3_0_sw_init,
+	.sw_fini = vce_v3_0_sw_fini,
+	.hw_init = vce_v3_0_hw_init,
+	.hw_fini = vce_v3_0_hw_fini,
+	.suspend = vce_v3_0_suspend,
+	.resume = vce_v3_0_resume,
+	.is_idle = vce_v3_0_is_idle,
+	.wait_for_idle = vce_v3_0_wait_for_idle,
+	.check_soft_reset = vce_v3_0_check_soft_reset,
+	.pre_soft_reset = vce_v3_0_pre_soft_reset,
+	.soft_reset = vce_v3_0_soft_reset,
+	.post_soft_reset = vce_v3_0_post_soft_reset,
+	.set_clockgating_state = vce_v3_0_set_clockgating_state,
+	.set_powergating_state = vce_v3_0_set_powergating_state,
+	.get_clockgating_state = vce_v3_0_get_clockgating_state,
+};
+
+static const struct amdgpu_ring_funcs vce_v3_0_ring_phys_funcs = {
+	.type = AMDGPU_RING_TYPE_VCE,
+	.align_mask = 0xf,
+	.nop = VCE_CMD_NO_OP,
+	.support_64bit_ptrs = false,
+	.get_rptr = vce_v3_0_ring_get_rptr,
+	.get_wptr = vce_v3_0_ring_get_wptr,
+	.set_wptr = vce_v3_0_ring_set_wptr,
+	.parse_cs = amdgpu_vce_ring_parse_cs,
+	.emit_frame_size =
+		4 + /* vce_v3_0_emit_pipeline_sync */
+		6, /* amdgpu_vce_ring_emit_fence x1 no user fence */
+	.emit_ib_size = 4, /* amdgpu_vce_ring_emit_ib */
+	.emit_ib = amdgpu_vce_ring_emit_ib,
+	.emit_fence = amdgpu_vce_ring_emit_fence,
+	.test_ring = amdgpu_vce_ring_test_ring,
+	.test_ib = amdgpu_vce_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vce_ring_begin_use,
+	.end_use = amdgpu_vce_ring_end_use,
+};
+
+static const struct amdgpu_ring_funcs vce_v3_0_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCE,
+	.align_mask = 0xf,
+	.nop = VCE_CMD_NO_OP,
+	.support_64bit_ptrs = false,
+	.get_rptr = vce_v3_0_ring_get_rptr,
+	.get_wptr = vce_v3_0_ring_get_wptr,
+	.set_wptr = vce_v3_0_ring_set_wptr,
+	.parse_cs = amdgpu_vce_ring_parse_cs_vm,
+	.emit_frame_size =
+		6 + /* vce_v3_0_emit_vm_flush */
+		4 + /* vce_v3_0_emit_pipeline_sync */
+		6 + 6, /* amdgpu_vce_ring_emit_fence x2 vm fence */
+	.emit_ib_size = 5, /* vce_v3_0_ring_emit_ib */
+	.emit_ib = vce_v3_0_ring_emit_ib,
+	.emit_vm_flush = vce_v3_0_emit_vm_flush,
+	.emit_pipeline_sync = vce_v3_0_emit_pipeline_sync,
+	.emit_fence = amdgpu_vce_ring_emit_fence,
+	.test_ring = amdgpu_vce_ring_test_ring,
+	.test_ib = amdgpu_vce_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vce_ring_begin_use,
+	.end_use = amdgpu_vce_ring_end_use,
+};
+
+static void vce_v3_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (adev->asic_type >= CHIP_STONEY) {
+		for (i = 0; i < adev->vce.num_rings; i++) {
+			adev->vce.ring[i].funcs = &vce_v3_0_ring_vm_funcs;
+			adev->vce.ring[i].me = i;
+		}
+		DRM_INFO("VCE enabled in VM mode\n");
+	} else {
+		for (i = 0; i < adev->vce.num_rings; i++) {
+			adev->vce.ring[i].funcs = &vce_v3_0_ring_phys_funcs;
+			adev->vce.ring[i].me = i;
+		}
+		DRM_INFO("VCE enabled in physical mode\n");
+	}
+}
+
+static const struct amdgpu_irq_src_funcs vce_v3_0_irq_funcs = {
+	.set = vce_v3_0_set_interrupt_state,
+	.process = vce_v3_0_process_interrupt,
+};
+
+static void vce_v3_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->vce.irq.num_types = 1;
+	adev->vce.irq.funcs = &vce_v3_0_irq_funcs;
+};
+
+const struct amdgpu_ip_block_version vce_v3_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_VCE,
+	.major = 3,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &vce_v3_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version vce_v3_1_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_VCE,
+	.major = 3,
+	.minor = 1,
+	.rev = 0,
+	.funcs = &vce_v3_0_ip_funcs,
+};
+
+const struct amdgpu_ip_block_version vce_v3_4_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_VCE,
+	.major = 3,
+	.minor = 4,
+	.rev = 0,
+	.funcs = &vce_v3_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/vce_v3_0.h b/sys/dev/pci/drm/amd/amdgpu/vce_v3_0.h
new file mode 100644
index 00000000000..08b908c7de0
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vce_v3_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCE_V3_0_H__
+#define __VCE_V3_0_H__
+
+extern const struct amdgpu_ip_block_version vce_v3_0_ip_block;
+extern const struct amdgpu_ip_block_version vce_v3_1_ip_block;
+extern const struct amdgpu_ip_block_version vce_v3_4_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/vce_v4_0.c b/sys/dev/pci/drm/amd/amdgpu/vce_v4_0.c
new file mode 100644
index 00000000000..b9b801e70f3
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vce_v4_0.c
@@ -0,0 +1,1124 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_vce.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "soc15_common.h"
+#include "mmsch_v1_0.h"
+
+#include "vce/vce_4_0_offset.h"
+#include "vce/vce_4_0_default.h"
+#include "vce/vce_4_0_sh_mask.h"
+#include "mmhub/mmhub_1_0_offset.h"
+#include "mmhub/mmhub_1_0_sh_mask.h"
+
+#include "ivsrcid/vce/irqsrcs_vce_4_0.h"
+
+#define VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK	0x02
+
+#define VCE_V4_0_FW_SIZE	(384 * 1024)
+#define VCE_V4_0_STACK_SIZE	(64 * 1024)
+#define VCE_V4_0_DATA_SIZE	((16 * 1024 * AMDGPU_MAX_VCE_HANDLES) + (52 * 1024))
+
+static void vce_v4_0_mc_resume(struct amdgpu_device *adev);
+static void vce_v4_0_set_ring_funcs(struct amdgpu_device *adev);
+static void vce_v4_0_set_irq_funcs(struct amdgpu_device *adev);
+
+/**
+ * vce_v4_0_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t vce_v4_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->me == 0)
+		return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR));
+	else if (ring->me == 1)
+		return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR2));
+	else
+		return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR3));
+}
+
+/**
+ * vce_v4_0_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t vce_v4_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell)
+		return adev->wb.wb[ring->wptr_offs];
+
+	if (ring->me == 0)
+		return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR));
+	else if (ring->me == 1)
+		return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR2));
+	else
+		return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR3));
+}
+
+/**
+ * vce_v4_0_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void vce_v4_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
+		WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+		return;
+	}
+
+	if (ring->me == 0)
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR),
+			lower_32_bits(ring->wptr));
+	else if (ring->me == 1)
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR2),
+			lower_32_bits(ring->wptr));
+	else
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR3),
+			lower_32_bits(ring->wptr));
+}
+
+static int vce_v4_0_firmware_loaded(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	for (i = 0; i < 10; ++i) {
+		for (j = 0; j < 100; ++j) {
+			uint32_t status =
+				RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS));
+
+			if (status & VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK)
+				return 0;
+			mdelay(10);
+		}
+
+		DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n");
+		WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SOFT_RESET),
+				VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK,
+				~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+		mdelay(10);
+		WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SOFT_RESET), 0,
+				~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+		mdelay(10);
+
+	}
+
+	return -ETIMEDOUT;
+}
+
+static int vce_v4_0_mmsch_start(struct amdgpu_device *adev,
+				struct amdgpu_mm_table *table)
+{
+	uint32_t data = 0, loop;
+	uint64_t addr = table->gpu_addr;
+	struct mmsch_v1_0_init_header *header = (struct mmsch_v1_0_init_header *)table->cpu_addr;
+	uint32_t size;
+
+	size = header->header_size + header->vce_table_size + header->uvd_table_size;
+
+	/* 1, write to vce_mmsch_vf_ctx_addr_lo/hi register with GPU mc addr of memory descriptor location */
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_CTX_ADDR_LO), lower_32_bits(addr));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_CTX_ADDR_HI), upper_32_bits(addr));
+
+	/* 2, update vmid of descriptor */
+	data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_VMID));
+	data &= ~VCE_MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+	data |= (0 << VCE_MMSCH_VF_VMID__VF_CTX_VMID__SHIFT); /* use domain0 for MM scheduler */
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_VMID), data);
+
+	/* 3, notify mmsch about the size of this descriptor */
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_CTX_SIZE), size);
+
+	/* 4, set resp to zero */
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP), 0);
+
+	WDOORBELL32(adev->vce.ring[0].doorbell_index, 0);
+	adev->wb.wb[adev->vce.ring[0].wptr_offs] = 0;
+	adev->vce.ring[0].wptr = 0;
+	adev->vce.ring[0].wptr_old = 0;
+
+	/* 5, kick off the initialization and wait until VCE_MMSCH_VF_MAILBOX_RESP becomes non-zero */
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_HOST), 0x10000001);
+
+	data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP));
+	loop = 1000;
+	while ((data & 0x10000002) != 0x10000002) {
+		udelay(10);
+		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP));
+		loop--;
+		if (!loop)
+			break;
+	}
+
+	if (!loop) {
+		dev_err(adev->dev, "failed to init MMSCH, mmVCE_MMSCH_VF_MAILBOX_RESP = %x\n", data);
+		return -EBUSY;
+	}
+
+	return 0;
+}
+
+static int vce_v4_0_sriov_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	uint32_t offset, size;
+	uint32_t table_size = 0;
+	struct mmsch_v1_0_cmd_direct_write direct_wt = { { 0 } };
+	struct mmsch_v1_0_cmd_direct_read_modify_write direct_rd_mod_wt = { { 0 } };
+	struct mmsch_v1_0_cmd_direct_polling direct_poll = { { 0 } };
+	struct mmsch_v1_0_cmd_end end = { { 0 } };
+	uint32_t *init_table = adev->virt.mm_table.cpu_addr;
+	struct mmsch_v1_0_init_header *header = (struct mmsch_v1_0_init_header *)init_table;
+
+	direct_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_WRITE;
+	direct_rd_mod_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE;
+	direct_poll.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_POLLING;
+	end.cmd_header.command_type = MMSCH_COMMAND__END;
+
+	if (header->vce_table_offset == 0 && header->vce_table_size == 0) {
+		header->version = MMSCH_VERSION;
+		header->header_size = sizeof(struct mmsch_v1_0_init_header) >> 2;
+
+		if (header->uvd_table_offset == 0 && header->uvd_table_size == 0)
+			header->vce_table_offset = header->header_size;
+		else
+			header->vce_table_offset = header->uvd_table_size + header->uvd_table_offset;
+
+		init_table += header->vce_table_offset;
+
+		ring = &adev->vce.ring[0];
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_LO),
+					    lower_32_bits(ring->gpu_addr));
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_HI),
+					    upper_32_bits(ring->gpu_addr));
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_SIZE),
+					    ring->ring_size / 4);
+
+		/* BEGING OF MC_RESUME */
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_CTRL), 0x398000);
+		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_CACHE_CTRL), ~0x1, 0);
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_SWAP_CNTL), 0);
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_SWAP_CNTL1), 0);
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VM_CTRL), 0);
+
+		if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_40BIT_BAR0),
+						adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].mc_addr >> 8);
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_64BIT_BAR0),
+						(adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].mc_addr >> 40) & 0xff);
+		} else {
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_40BIT_BAR0),
+						adev->vce.gpu_addr >> 8);
+			MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_64BIT_BAR0),
+						(adev->vce.gpu_addr >> 40) & 0xff);
+		}
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_40BIT_BAR1),
+						adev->vce.gpu_addr >> 8);
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_64BIT_BAR1),
+						(adev->vce.gpu_addr >> 40) & 0xff);
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_40BIT_BAR2),
+						adev->vce.gpu_addr >> 8);
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0,
+						mmVCE_LMI_VCPU_CACHE_64BIT_BAR2),
+						(adev->vce.gpu_addr >> 40) & 0xff);
+
+		offset = AMDGPU_VCE_FIRMWARE_OFFSET;
+		size = VCE_V4_0_FW_SIZE;
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET0),
+					offset & ~0x0f000000);
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE0), size);
+
+		offset = (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) ? offset + size : 0;
+		size = VCE_V4_0_STACK_SIZE;
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET1),
+					(offset & ~0x0f000000) | (1 << 24));
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE1), size);
+
+		offset += size;
+		size = VCE_V4_0_DATA_SIZE;
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET2),
+					(offset & ~0x0f000000) | (2 << 24));
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE2), size);
+
+		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_CTRL2), ~0x100, 0);
+		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_EN),
+						   VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK,
+						   VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
+
+		/* end of MC_RESUME */
+		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS),
+						   VCE_STATUS__JOB_BUSY_MASK, ~VCE_STATUS__JOB_BUSY_MASK);
+		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CNTL),
+						   ~0x200001, VCE_VCPU_CNTL__CLK_EN_MASK);
+		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_SOFT_RESET),
+						   ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK, 0);
+
+		MMSCH_V1_0_INSERT_DIRECT_POLL(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS),
+					      VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK,
+					      VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK);
+
+		/* clear BUSY flag */
+		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS),
+						   ~VCE_STATUS__JOB_BUSY_MASK, 0);
+
+		/* add end packet */
+		memcpy((void *)init_table, &end, sizeof(struct mmsch_v1_0_cmd_end));
+		table_size += sizeof(struct mmsch_v1_0_cmd_end) / 4;
+		header->vce_table_size = table_size;
+	}
+
+	return vce_v4_0_mmsch_start(adev, &adev->virt.mm_table);
+}
+
+/**
+ * vce_v4_0_start - start VCE block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the VCE block
+ */
+static int vce_v4_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	int r;
+
+	ring = &adev->vce.ring[0];
+
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR), lower_32_bits(ring->wptr));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR), lower_32_bits(ring->wptr));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_LO), ring->gpu_addr);
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_HI), upper_32_bits(ring->gpu_addr));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_SIZE), ring->ring_size / 4);
+
+	ring = &adev->vce.ring[1];
+
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR2), lower_32_bits(ring->wptr));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR2), lower_32_bits(ring->wptr));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_LO2), ring->gpu_addr);
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_HI2), upper_32_bits(ring->gpu_addr));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_SIZE2), ring->ring_size / 4);
+
+	ring = &adev->vce.ring[2];
+
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR3), lower_32_bits(ring->wptr));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR3), lower_32_bits(ring->wptr));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_LO3), ring->gpu_addr);
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_HI3), upper_32_bits(ring->gpu_addr));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_SIZE3), ring->ring_size / 4);
+
+	vce_v4_0_mc_resume(adev);
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS), VCE_STATUS__JOB_BUSY_MASK,
+			~VCE_STATUS__JOB_BUSY_MASK);
+
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CNTL), 1, ~0x200001);
+
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SOFT_RESET), 0,
+			~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+	mdelay(100);
+
+	r = vce_v4_0_firmware_loaded(adev);
+
+	/* clear BUSY flag */
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS), 0, ~VCE_STATUS__JOB_BUSY_MASK);
+
+	if (r) {
+		DRM_ERROR("VCE not responding, giving up!!!\n");
+		return r;
+	}
+
+	return 0;
+}
+
+static int vce_v4_0_stop(struct amdgpu_device *adev)
+{
+
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CNTL), 0, ~0x200001);
+
+	/* hold on ECPU */
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SOFT_RESET),
+			VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK,
+			~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+
+	/* clear BUSY flag */
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS), 0, ~VCE_STATUS__JOB_BUSY_MASK);
+
+	/* Set Clock-Gating off */
+	/* if (adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG)
+		vce_v4_0_set_vce_sw_clock_gating(adev, false);
+	*/
+
+	return 0;
+}
+
+static int vce_v4_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev)) /* currently only VCN0 support SRIOV */
+		adev->vce.num_rings = 1;
+	else
+		adev->vce.num_rings = 3;
+
+	vce_v4_0_set_ring_funcs(adev);
+	vce_v4_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int vce_v4_0_sw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring;
+
+	unsigned size;
+	int r, i;
+
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCE0, 167, &adev->vce.irq);
+	if (r)
+		return r;
+
+	size  = VCE_V4_0_STACK_SIZE + VCE_V4_0_DATA_SIZE;
+	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
+		size += VCE_V4_0_FW_SIZE;
+
+	r = amdgpu_vce_sw_init(adev, size);
+	if (r)
+		return r;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		const struct common_firmware_header *hdr;
+		unsigned size = amdgpu_bo_size(adev->vce.vcpu_bo);
+
+		adev->vce.saved_bo = kvmalloc(size, GFP_KERNEL);
+		if (!adev->vce.saved_bo)
+			return -ENOMEM;
+
+		hdr = (const struct common_firmware_header *)adev->vce.fw->data;
+		adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].ucode_id = AMDGPU_UCODE_ID_VCE;
+		adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].fw = adev->vce.fw;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(hdr->ucode_size_bytes), PAGE_SIZE);
+		DRM_INFO("PSP loading VCE firmware\n");
+	} else {
+		r = amdgpu_vce_resume(adev);
+		if (r)
+			return r;
+	}
+
+	for (i = 0; i < adev->vce.num_rings; i++) {
+		ring = &adev->vce.ring[i];
+		snprintf(ring->name, sizeof(ring->name), "vce%d", i);
+		if (amdgpu_sriov_vf(adev)) {
+			/* DOORBELL only works under SRIOV */
+			ring->use_doorbell = true;
+
+			/* currently only use the first encoding ring for sriov,
+			 * so set unused location for other unused rings.
+			 */
+			if (i == 0)
+				ring->doorbell_index = AMDGPU_DOORBELL64_VCE_RING0_1 * 2;
+			else
+				ring->doorbell_index = AMDGPU_DOORBELL64_VCE_RING2_3 * 2 + 1;
+		}
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0);
+		if (r)
+			return r;
+	}
+
+
+	r = amdgpu_vce_entity_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_virt_alloc_mm_table(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int vce_v4_0_sw_fini(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* free MM table */
+	amdgpu_virt_free_mm_table(adev);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		kvfree(adev->vce.saved_bo);
+		adev->vce.saved_bo = NULL;
+	}
+
+	r = amdgpu_vce_suspend(adev);
+	if (r)
+		return r;
+
+	return amdgpu_vce_sw_fini(adev);
+}
+
+static int vce_v4_0_hw_init(void *handle)
+{
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		r = vce_v4_0_sriov_start(adev);
+	else
+		r = vce_v4_0_start(adev);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->vce.num_rings; i++)
+		adev->vce.ring[i].ready = false;
+
+	for (i = 0; i < adev->vce.num_rings; i++) {
+		r = amdgpu_ring_test_ring(&adev->vce.ring[i]);
+		if (r)
+			return r;
+		else
+			adev->vce.ring[i].ready = true;
+	}
+
+	DRM_INFO("VCE initialized successfully.\n");
+
+	return 0;
+}
+
+static int vce_v4_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		/* vce_v4_0_wait_for_idle(handle); */
+		vce_v4_0_stop(adev);
+	} else {
+		/* full access mode, so don't touch any VCE register */
+		DRM_DEBUG("For SRIOV client, shouldn't do anything.\n");
+	}
+
+	for (i = 0; i < adev->vce.num_rings; i++)
+		adev->vce.ring[i].ready = false;
+
+	return 0;
+}
+
+static int vce_v4_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r;
+
+	if (adev->vce.vcpu_bo == NULL)
+		return 0;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		unsigned size = amdgpu_bo_size(adev->vce.vcpu_bo);
+		void *ptr = adev->vce.cpu_addr;
+
+		memcpy_fromio(adev->vce.saved_bo, ptr, size);
+	}
+
+	r = vce_v4_0_hw_fini(adev);
+	if (r)
+		return r;
+
+	return amdgpu_vce_suspend(adev);
+}
+
+static int vce_v4_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r;
+
+	if (adev->vce.vcpu_bo == NULL)
+		return -EINVAL;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		unsigned size = amdgpu_bo_size(adev->vce.vcpu_bo);
+		void *ptr = adev->vce.cpu_addr;
+
+		memcpy_toio(ptr, adev->vce.saved_bo, size);
+	} else {
+		r = amdgpu_vce_resume(adev);
+		if (r)
+			return r;
+	}
+
+	return vce_v4_0_hw_init(adev);
+}
+
+static void vce_v4_0_mc_resume(struct amdgpu_device *adev)
+{
+	uint32_t offset, size;
+
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_A), 0, ~(1 << 16));
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING), 0x1FF000, ~0xFF9FF000);
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING), 0x3F, ~0x3F);
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B), 0x1FF);
+
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_CTRL), 0x00398000);
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_CACHE_CTRL), 0x0, ~0x1);
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_SWAP_CNTL), 0);
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_SWAP_CNTL1), 0);
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VM_CTRL), 0);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR0),
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].mc_addr >> 8));
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_64BIT_BAR0),
+			(adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].mc_addr >> 40) & 0xff);
+	} else {
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR0),
+			(adev->vce.gpu_addr >> 8));
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_64BIT_BAR0),
+			(adev->vce.gpu_addr >> 40) & 0xff);
+	}
+
+	offset = AMDGPU_VCE_FIRMWARE_OFFSET;
+	size = VCE_V4_0_FW_SIZE;
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET0), offset & ~0x0f000000);
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE0), size);
+
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR1), (adev->vce.gpu_addr >> 8));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_64BIT_BAR1), (adev->vce.gpu_addr >> 40) & 0xff);
+	offset = (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) ? offset + size : 0;
+	size = VCE_V4_0_STACK_SIZE;
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET1), (offset & ~0x0f000000) | (1 << 24));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE1), size);
+
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR2), (adev->vce.gpu_addr >> 8));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_64BIT_BAR2), (adev->vce.gpu_addr >> 40) & 0xff);
+	offset += size;
+	size = VCE_V4_0_DATA_SIZE;
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET2), (offset & ~0x0f000000) | (2 << 24));
+	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE2), size);
+
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_CTRL2), 0x0, ~0x100);
+	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_EN),
+			VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK,
+			~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
+}
+
+static int vce_v4_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	/* needed for driver unload*/
+	return 0;
+}
+
+#if 0
+static bool vce_v4_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 mask = 0;
+
+	mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE0) ? 0 : SRBM_STATUS2__VCE0_BUSY_MASK;
+	mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE1) ? 0 : SRBM_STATUS2__VCE1_BUSY_MASK;
+
+	return !(RREG32(mmSRBM_STATUS2) & mask);
+}
+
+static int vce_v4_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++)
+		if (vce_v4_0_is_idle(handle))
+			return 0;
+
+	return -ETIMEDOUT;
+}
+
+#define  VCE_STATUS_VCPU_REPORT_AUTO_BUSY_MASK  0x00000008L   /* AUTO_BUSY */
+#define  VCE_STATUS_VCPU_REPORT_RB0_BUSY_MASK   0x00000010L   /* RB0_BUSY */
+#define  VCE_STATUS_VCPU_REPORT_RB1_BUSY_MASK   0x00000020L   /* RB1_BUSY */
+#define  AMDGPU_VCE_STATUS_BUSY_MASK (VCE_STATUS_VCPU_REPORT_AUTO_BUSY_MASK | \
+				      VCE_STATUS_VCPU_REPORT_RB0_BUSY_MASK)
+
+static bool vce_v4_0_check_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset = 0;
+
+	/* According to VCE team , we should use VCE_STATUS instead
+	 * SRBM_STATUS.VCE_BUSY bit for busy status checking.
+	 * GRBM_GFX_INDEX.INSTANCE_INDEX is used to specify which VCE
+	 * instance's registers are accessed
+	 * (0 for 1st instance, 10 for 2nd instance).
+	 *
+	 *VCE_STATUS
+	 *|UENC|ACPI|AUTO ACTIVE|RB1 |RB0 |RB2 |          |FW_LOADED|JOB |
+	 *|----+----+-----------+----+----+----+----------+---------+----|
+	 *|bit8|bit7|    bit6   |bit5|bit4|bit3|   bit2   |  bit1   |bit0|
+	 *
+	 * VCE team suggest use bit 3--bit 6 for busy status check
+	 */
+	mutex_lock(&adev->grbm_idx_mutex);
+	WREG32_FIELD(GRBM_GFX_INDEX, INSTANCE_INDEX, 0);
+	if (RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS) & AMDGPU_VCE_STATUS_BUSY_MASK) {
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE0, 1);
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE1, 1);
+	}
+	WREG32_FIELD(GRBM_GFX_INDEX, INSTANCE_INDEX, 0x10);
+	if (RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS) & AMDGPU_VCE_STATUS_BUSY_MASK) {
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE0, 1);
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE1, 1);
+	}
+	WREG32_FIELD(GRBM_GFX_INDEX, INSTANCE_INDEX, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	if (srbm_soft_reset) {
+		adev->vce.srbm_soft_reset = srbm_soft_reset;
+		return true;
+	} else {
+		adev->vce.srbm_soft_reset = 0;
+		return false;
+	}
+}
+
+static int vce_v4_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 srbm_soft_reset;
+
+	if (!adev->vce.srbm_soft_reset)
+		return 0;
+	srbm_soft_reset = adev->vce.srbm_soft_reset;
+
+	if (srbm_soft_reset) {
+		u32 tmp;
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+	}
+
+	return 0;
+}
+
+static int vce_v4_0_pre_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->vce.srbm_soft_reset)
+		return 0;
+
+	mdelay(5);
+
+	return vce_v4_0_suspend(adev);
+}
+
+
+static int vce_v4_0_post_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!adev->vce.srbm_soft_reset)
+		return 0;
+
+	mdelay(5);
+
+	return vce_v4_0_resume(adev);
+}
+
+static void vce_v4_0_override_vce_clock_gating(struct amdgpu_device *adev, bool override)
+{
+	u32 tmp, data;
+
+	tmp = data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_ARB_CTRL));
+	if (override)
+		data |= VCE_RB_ARB_CTRL__VCE_CGTT_OVERRIDE_MASK;
+	else
+		data &= ~VCE_RB_ARB_CTRL__VCE_CGTT_OVERRIDE_MASK;
+
+	if (tmp != data)
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_ARB_CTRL), data);
+}
+
+static void vce_v4_0_set_vce_sw_clock_gating(struct amdgpu_device *adev,
+					     bool gated)
+{
+	u32 data;
+
+	/* Set Override to disable Clock Gating */
+	vce_v4_0_override_vce_clock_gating(adev, true);
+
+	/* This function enables MGCG which is controlled by firmware.
+	   With the clocks in the gated state the core is still
+	   accessible but the firmware will throttle the clocks on the
+	   fly as necessary.
+	*/
+	if (gated) {
+		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B));
+		data |= 0x1ff;
+		data &= ~0xef0000;
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B), data);
+
+		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING));
+		data |= 0x3ff000;
+		data &= ~0xffc00000;
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING), data);
+
+		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING_2));
+		data |= 0x2;
+		data &= ~0x00010000;
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING_2), data);
+
+		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING));
+		data |= 0x37f;
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING), data);
+
+		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_DMA_DCLK_CTRL));
+		data |= VCE_UENC_DMA_DCLK_CTRL__WRDMCLK_FORCEON_MASK |
+			VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK |
+			VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK  |
+			0x8;
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_DMA_DCLK_CTRL), data);
+	} else {
+		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B));
+		data &= ~0x80010;
+		data |= 0xe70008;
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B), data);
+
+		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING));
+		data |= 0xffc00000;
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING), data);
+
+		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING_2));
+		data |= 0x10000;
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING_2), data);
+
+		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING));
+		data &= ~0xffc00000;
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING), data);
+
+		data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_DMA_DCLK_CTRL));
+		data &= ~(VCE_UENC_DMA_DCLK_CTRL__WRDMCLK_FORCEON_MASK |
+			  VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK |
+			  VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK  |
+			  0x8);
+		WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_DMA_DCLK_CTRL), data);
+	}
+	vce_v4_0_override_vce_clock_gating(adev, false);
+}
+
+static void vce_v4_0_set_bypass_mode(struct amdgpu_device *adev, bool enable)
+{
+	u32 tmp = RREG32_SMC(ixGCK_DFS_BYPASS_CNTL);
+
+	if (enable)
+		tmp |= GCK_DFS_BYPASS_CNTL__BYPASSECLK_MASK;
+	else
+		tmp &= ~GCK_DFS_BYPASS_CNTL__BYPASSECLK_MASK;
+
+	WREG32_SMC(ixGCK_DFS_BYPASS_CNTL, tmp);
+}
+
+static int vce_v4_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+	int i;
+
+	if ((adev->asic_type == CHIP_POLARIS10) ||
+		(adev->asic_type == CHIP_TONGA) ||
+		(adev->asic_type == CHIP_FIJI))
+		vce_v4_0_set_bypass_mode(adev, enable);
+
+	if (!(adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG))
+		return 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < 2; i++) {
+		/* Program VCE Instance 0 or 1 if not harvested */
+		if (adev->vce.harvest_config & (1 << i))
+			continue;
+
+		WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, i);
+
+		if (enable) {
+			/* initialize VCE_CLOCK_GATING_A: Clock ON/OFF delay */
+			uint32_t data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_A);
+			data &= ~(0xf | 0xff0);
+			data |= ((0x0 << 0) | (0x04 << 4));
+			WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_A, data);
+
+			/* initialize VCE_UENC_CLOCK_GATING: Clock ON/OFF delay */
+			data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING);
+			data &= ~(0xf | 0xff0);
+			data |= ((0x0 << 0) | (0x04 << 4));
+			WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING, data);
+		}
+
+		vce_v4_0_set_vce_sw_clock_gating(adev, enable);
+	}
+
+	WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, 0);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+static int vce_v4_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the VCE block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!(adev->pg_flags & AMD_PG_SUPPORT_VCE))
+		return 0;
+
+	if (state == AMD_PG_STATE_GATE)
+		/* XXX do we need a vce_v4_0_stop()? */
+		return 0;
+	else
+		return vce_v4_0_start(adev);
+}
+#endif
+
+static void vce_v4_0_ring_emit_ib(struct amdgpu_ring *ring,
+		struct amdgpu_ib *ib, unsigned int vmid, bool ctx_switch)
+{
+	amdgpu_ring_write(ring, VCE_CMD_IB_VM);
+	amdgpu_ring_write(ring, vmid);
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+static void vce_v4_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+			u64 seq, unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, VCE_CMD_FENCE);
+	amdgpu_ring_write(ring, addr);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, VCE_CMD_TRAP);
+}
+
+static void vce_v4_0_ring_insert_end(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, VCE_CMD_END);
+}
+
+static void vce_v4_0_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+				   uint32_t val, uint32_t mask)
+{
+	amdgpu_ring_write(ring, VCE_CMD_REG_WAIT);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, val);
+}
+
+static void vce_v4_0_emit_vm_flush(struct amdgpu_ring *ring,
+				   unsigned int vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for reg writes */
+	vce_v4_0_emit_reg_wait(ring, hub->ctx0_ptb_addr_lo32 + vmid * 2,
+			       lower_32_bits(pd_addr), 0xffffffff);
+}
+
+static void vce_v4_0_emit_wreg(struct amdgpu_ring *ring,
+			       uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, VCE_CMD_REG_WRITE);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, val);
+}
+
+static int vce_v4_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	uint32_t val = 0;
+
+	if (!amdgpu_sriov_vf(adev)) {
+		if (state == AMDGPU_IRQ_STATE_ENABLE)
+			val |= VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK;
+
+		WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_EN), val,
+				~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
+	}
+	return 0;
+}
+
+static int vce_v4_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: VCE\n");
+
+	switch (entry->src_data[0]) {
+	case 0:
+	case 1:
+	case 2:
+		amdgpu_fence_process(&adev->vce.ring[entry->src_data[0]]);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+const struct amd_ip_funcs vce_v4_0_ip_funcs = {
+	.name = "vce_v4_0",
+	.early_init = vce_v4_0_early_init,
+	.late_init = NULL,
+	.sw_init = vce_v4_0_sw_init,
+	.sw_fini = vce_v4_0_sw_fini,
+	.hw_init = vce_v4_0_hw_init,
+	.hw_fini = vce_v4_0_hw_fini,
+	.suspend = vce_v4_0_suspend,
+	.resume = vce_v4_0_resume,
+	.is_idle = NULL /* vce_v4_0_is_idle */,
+	.wait_for_idle = NULL /* vce_v4_0_wait_for_idle */,
+	.check_soft_reset = NULL /* vce_v4_0_check_soft_reset */,
+	.pre_soft_reset = NULL /* vce_v4_0_pre_soft_reset */,
+	.soft_reset = NULL /* vce_v4_0_soft_reset */,
+	.post_soft_reset = NULL /* vce_v4_0_post_soft_reset */,
+	.set_clockgating_state = vce_v4_0_set_clockgating_state,
+	.set_powergating_state = NULL /* vce_v4_0_set_powergating_state */,
+};
+
+static const struct amdgpu_ring_funcs vce_v4_0_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCE,
+	.align_mask = 0x3f,
+	.nop = VCE_CMD_NO_OP,
+	.support_64bit_ptrs = false,
+	.vmhub = AMDGPU_MMHUB,
+	.get_rptr = vce_v4_0_ring_get_rptr,
+	.get_wptr = vce_v4_0_ring_get_wptr,
+	.set_wptr = vce_v4_0_ring_set_wptr,
+	.parse_cs = amdgpu_vce_ring_parse_cs_vm,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* vce_v4_0_emit_vm_flush */
+		5 + 5 + /* amdgpu_vce_ring_emit_fence x2 vm fence */
+		1, /* vce_v4_0_ring_insert_end */
+	.emit_ib_size = 5, /* vce_v4_0_ring_emit_ib */
+	.emit_ib = vce_v4_0_ring_emit_ib,
+	.emit_vm_flush = vce_v4_0_emit_vm_flush,
+	.emit_fence = vce_v4_0_ring_emit_fence,
+	.test_ring = amdgpu_vce_ring_test_ring,
+	.test_ib = amdgpu_vce_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = vce_v4_0_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vce_ring_begin_use,
+	.end_use = amdgpu_vce_ring_end_use,
+	.emit_wreg = vce_v4_0_emit_wreg,
+	.emit_reg_wait = vce_v4_0_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+};
+
+static void vce_v4_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vce.num_rings; i++) {
+		adev->vce.ring[i].funcs = &vce_v4_0_ring_vm_funcs;
+		adev->vce.ring[i].me = i;
+	}
+	DRM_INFO("VCE enabled in VM mode\n");
+}
+
+static const struct amdgpu_irq_src_funcs vce_v4_0_irq_funcs = {
+	.set = vce_v4_0_set_interrupt_state,
+	.process = vce_v4_0_process_interrupt,
+};
+
+static void vce_v4_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->vce.irq.num_types = 1;
+	adev->vce.irq.funcs = &vce_v4_0_irq_funcs;
+};
+
+const struct amdgpu_ip_block_version vce_v4_0_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_VCE,
+	.major = 4,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &vce_v4_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/vce_v4_0.h b/sys/dev/pci/drm/amd/amdgpu/vce_v4_0.h
new file mode 100644
index 00000000000..a32beda6a47
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vce_v4_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCE_V4_0_H__
+#define __VCE_V4_0_H__
+
+extern const struct amdgpu_ip_block_version vce_v4_0_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/vcn_v1_0.c b/sys/dev/pci/drm/amd/amdgpu/vcn_v1_0.c
new file mode 100644
index 00000000000..8fd919f683f
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vcn_v1_0.c
@@ -0,0 +1,1792 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_vcn.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "soc15_common.h"
+
+#include "vcn/vcn_1_0_offset.h"
+#include "vcn/vcn_1_0_sh_mask.h"
+#include "hdp/hdp_4_0_offset.h"
+#include "mmhub/mmhub_9_1_offset.h"
+#include "mmhub/mmhub_9_1_sh_mask.h"
+
+#include "ivsrcid/vcn/irqsrcs_vcn_1_0.h"
+
+static int vcn_v1_0_stop(struct amdgpu_device *adev);
+static void vcn_v1_0_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v1_0_set_enc_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v1_0_set_jpeg_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v1_0_set_irq_funcs(struct amdgpu_device *adev);
+static void vcn_v1_0_jpeg_ring_set_patch_ring(struct amdgpu_ring *ring, uint32_t ptr);
+static int vcn_v1_0_set_powergating_state(void *handle, enum amd_powergating_state state);
+
+/**
+ * vcn_v1_0_early_init - set function pointers
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Set ring and irq function pointers
+ */
+static int vcn_v1_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->vcn.num_enc_rings = 2;
+
+	vcn_v1_0_set_dec_ring_funcs(adev);
+	vcn_v1_0_set_enc_ring_funcs(adev);
+	vcn_v1_0_set_jpeg_ring_funcs(adev);
+	vcn_v1_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+/**
+ * vcn_v1_0_sw_init - sw init for VCN block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Load firmware and sw initialization
+ */
+static int vcn_v1_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int i, r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* VCN DEC TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, VCN_1_0__SRCID__UVD_SYSTEM_MESSAGE_INTERRUPT, &adev->vcn.irq);
+	if (r)
+		return r;
+
+	/* VCN ENC TRAP */
+	for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
+		r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, i + VCN_1_0__SRCID__UVD_ENC_GENERAL_PURPOSE,
+					&adev->vcn.irq);
+		if (r)
+			return r;
+	}
+
+	/* VCN JPEG TRAP */
+	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, 126, &adev->vcn.irq);
+	if (r)
+		return r;
+
+	r = amdgpu_vcn_sw_init(adev);
+	if (r)
+		return r;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		const struct common_firmware_header *hdr;
+		hdr = (const struct common_firmware_header *)adev->vcn.fw->data;
+		adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].ucode_id = AMDGPU_UCODE_ID_VCN;
+		adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].fw = adev->vcn.fw;
+		adev->firmware.fw_size +=
+			roundup2(le32_to_cpu(hdr->ucode_size_bytes), PAGE_SIZE);
+		DRM_INFO("PSP loading VCN firmware\n");
+	}
+
+	r = amdgpu_vcn_resume(adev);
+	if (r)
+		return r;
+
+	ring = &adev->vcn.ring_dec;
+	snprintf(ring->name, sizeof(ring->name), "vcn_dec");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.irq, 0);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
+		ring = &adev->vcn.ring_enc[i];
+		snprintf(ring->name, sizeof(ring->name), "vcn_enc%d", i);
+		r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.irq, 0);
+		if (r)
+			return r;
+	}
+
+	ring = &adev->vcn.ring_jpeg;
+	snprintf(ring->name, sizeof(ring->name), "vcn_jpeg");
+	r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.irq, 0);
+	if (r)
+		return r;
+
+	return r;
+}
+
+/**
+ * vcn_v1_0_sw_fini - sw fini for VCN block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * VCN suspend and free up sw allocation
+ */
+static int vcn_v1_0_sw_fini(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_vcn_suspend(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_vcn_sw_fini(adev);
+
+	return r;
+}
+
+/**
+ * vcn_v1_0_hw_init - start and test VCN block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int vcn_v1_0_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->vcn.ring_dec;
+	int i, r;
+
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		goto done;
+	}
+
+	for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
+		ring = &adev->vcn.ring_enc[i];
+		ring->ready = true;
+		r = amdgpu_ring_test_ring(ring);
+		if (r) {
+			ring->ready = false;
+			goto done;
+		}
+	}
+
+	ring = &adev->vcn.ring_jpeg;
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		goto done;
+	}
+
+done:
+	if (!r)
+		DRM_INFO("VCN decode and encode initialized successfully.\n");
+
+	return r;
+}
+
+/**
+ * vcn_v1_0_hw_fini - stop the hardware block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Stop the VCN block, mark ring as not ready any more
+ */
+static int vcn_v1_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->vcn.ring_dec;
+
+	if (RREG32_SOC15(VCN, 0, mmUVD_STATUS))
+		vcn_v1_0_set_powergating_state(adev, AMD_PG_STATE_GATE);
+
+	ring->ready = false;
+
+	return 0;
+}
+
+/**
+ * vcn_v1_0_suspend - suspend VCN block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * HW fini and suspend VCN block
+ */
+static int vcn_v1_0_suspend(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = vcn_v1_0_hw_fini(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_vcn_suspend(adev);
+
+	return r;
+}
+
+/**
+ * vcn_v1_0_resume - resume VCN block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Resume firmware and hw init VCN block
+ */
+static int vcn_v1_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_vcn_resume(adev);
+	if (r)
+		return r;
+
+	r = vcn_v1_0_hw_init(adev);
+
+	return r;
+}
+
+/**
+ * vcn_v1_0_mc_resume - memory controller programming
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Let the VCN memory controller know it's offsets
+ */
+static void vcn_v1_0_mc_resume(struct amdgpu_device *adev)
+{
+	uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.fw->size + 4);
+	uint32_t offset;
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			     (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_lo));
+		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			     (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_hi));
+		WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0, 0);
+		offset = 0;
+	} else {
+		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->vcn.gpu_addr));
+		WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->vcn.gpu_addr));
+		offset = size;
+		WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET0,
+			     AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+	}
+
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE0, size);
+
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+		     lower_32_bits(adev->vcn.gpu_addr + offset));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+		     upper_32_bits(adev->vcn.gpu_addr + offset));
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET1, 0);
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_HEAP_SIZE);
+
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+		     lower_32_bits(adev->vcn.gpu_addr + offset + AMDGPU_VCN_HEAP_SIZE));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+		     upper_32_bits(adev->vcn.gpu_addr + offset + AMDGPU_VCN_HEAP_SIZE));
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET2, 0);
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE2,
+			AMDGPU_VCN_STACK_SIZE + (AMDGPU_VCN_SESSION_SIZE * 40));
+
+	WREG32_SOC15(UVD, 0, mmUVD_UDEC_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(UVD, 0, mmUVD_UDEC_DB_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+	WREG32_SOC15(UVD, 0, mmUVD_UDEC_DBW_ADDR_CONFIG,
+			adev->gfx.config.gb_addr_config);
+}
+
+/**
+ * vcn_v1_0_disable_clock_gating - disable VCN clock gating
+ *
+ * @adev: amdgpu_device pointer
+ * @sw: enable SW clock gating
+ *
+ * Disable clock gating for VCN block
+ */
+static void vcn_v1_0_disable_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data;
+
+	/* JPEG disable CGC */
+	data = RREG32_SOC15(VCN, 0, mmJPEG_CGC_CTRL);
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data &= ~JPEG_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, 0, mmJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, 0, mmJPEG_CGC_GATE);
+	data &= ~(JPEG_CGC_GATE__JPEG_MASK | JPEG_CGC_GATE__JPEG2_MASK);
+	WREG32_SOC15(VCN, 0, mmJPEG_CGC_GATE, data);
+
+	/* UVD disable CGC */
+	data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		data |= 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data &= ~ UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_CGC_GATE);
+	data &= ~(UVD_CGC_GATE__SYS_MASK
+		| UVD_CGC_GATE__UDEC_MASK
+		| UVD_CGC_GATE__MPEG2_MASK
+		| UVD_CGC_GATE__REGS_MASK
+		| UVD_CGC_GATE__RBC_MASK
+		| UVD_CGC_GATE__LMI_MC_MASK
+		| UVD_CGC_GATE__LMI_UMC_MASK
+		| UVD_CGC_GATE__IDCT_MASK
+		| UVD_CGC_GATE__MPRD_MASK
+		| UVD_CGC_GATE__MPC_MASK
+		| UVD_CGC_GATE__LBSI_MASK
+		| UVD_CGC_GATE__LRBBM_MASK
+		| UVD_CGC_GATE__UDEC_RE_MASK
+		| UVD_CGC_GATE__UDEC_CM_MASK
+		| UVD_CGC_GATE__UDEC_IT_MASK
+		| UVD_CGC_GATE__UDEC_DB_MASK
+		| UVD_CGC_GATE__UDEC_MP_MASK
+		| UVD_CGC_GATE__WCB_MASK
+		| UVD_CGC_GATE__VCPU_MASK
+		| UVD_CGC_GATE__SCPU_MASK);
+	WREG32_SOC15(VCN, 0, mmUVD_CGC_GATE, data);
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL);
+	data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_CM_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_IT_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_DB_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MP_MODE_MASK
+		| UVD_CGC_CTRL__SYS_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MODE_MASK
+		| UVD_CGC_CTRL__MPEG2_MODE_MASK
+		| UVD_CGC_CTRL__REGS_MODE_MASK
+		| UVD_CGC_CTRL__RBC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		| UVD_CGC_CTRL__IDCT_MODE_MASK
+		| UVD_CGC_CTRL__MPRD_MODE_MASK
+		| UVD_CGC_CTRL__MPC_MODE_MASK
+		| UVD_CGC_CTRL__LBSI_MODE_MASK
+		| UVD_CGC_CTRL__LRBBM_MODE_MASK
+		| UVD_CGC_CTRL__WCB_MODE_MASK
+		| UVD_CGC_CTRL__VCPU_MODE_MASK
+		| UVD_CGC_CTRL__SCPU_MODE_MASK);
+	WREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL, data);
+
+	/* turn on */
+	data = RREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_GATE);
+	data |= (UVD_SUVD_CGC_GATE__SRE_MASK
+		| UVD_SUVD_CGC_GATE__SIT_MASK
+		| UVD_SUVD_CGC_GATE__SMP_MASK
+		| UVD_SUVD_CGC_GATE__SCM_MASK
+		| UVD_SUVD_CGC_GATE__SDB_MASK
+		| UVD_SUVD_CGC_GATE__SRE_H264_MASK
+		| UVD_SUVD_CGC_GATE__SRE_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SIT_H264_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SCM_H264_MASK
+		| UVD_SUVD_CGC_GATE__SCM_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SDB_H264_MASK
+		| UVD_SUVD_CGC_GATE__SDB_HEVC_MASK
+		| UVD_SUVD_CGC_GATE__SCLR_MASK
+		| UVD_SUVD_CGC_GATE__UVD_SC_MASK
+		| UVD_SUVD_CGC_GATE__ENT_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_DEC_MASK
+		| UVD_SUVD_CGC_GATE__SIT_HEVC_ENC_MASK
+		| UVD_SUVD_CGC_GATE__SITE_MASK
+		| UVD_SUVD_CGC_GATE__SRE_VP9_MASK
+		| UVD_SUVD_CGC_GATE__SCM_VP9_MASK
+		| UVD_SUVD_CGC_GATE__SIT_VP9_DEC_MASK
+		| UVD_SUVD_CGC_GATE__SDB_VP9_MASK
+		| UVD_SUVD_CGC_GATE__IME_HEVC_MASK);
+	WREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_GATE, data);
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_CTRL);
+	data &= ~(UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCLR_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__UVD_SC_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SITE_MODE_MASK);
+	WREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_CTRL, data);
+}
+
+/**
+ * vcn_v1_0_enable_clock_gating - enable VCN clock gating
+ *
+ * @adev: amdgpu_device pointer
+ * @sw: enable SW clock gating
+ *
+ * Enable clock gating for VCN block
+ */
+static void vcn_v1_0_enable_clock_gating(struct amdgpu_device *adev)
+{
+	uint32_t data = 0;
+
+	/* enable JPEG CGC */
+	data = RREG32_SOC15(VCN, 0, mmJPEG_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data |= 0 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, 0, mmJPEG_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, 0, mmJPEG_CGC_GATE);
+	data |= (JPEG_CGC_GATE__JPEG_MASK | JPEG_CGC_GATE__JPEG2_MASK);
+	WREG32_SOC15(VCN, 0, mmJPEG_CGC_GATE, data);
+
+	/* enable UVD CGC */
+	data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL);
+	if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
+		data |= 1 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	else
+		data |= 0 << UVD_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+	data |= 1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+	data |= 4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+	WREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL);
+	data |= (UVD_CGC_CTRL__UDEC_RE_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_CM_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_IT_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_DB_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MP_MODE_MASK
+		| UVD_CGC_CTRL__SYS_MODE_MASK
+		| UVD_CGC_CTRL__UDEC_MODE_MASK
+		| UVD_CGC_CTRL__MPEG2_MODE_MASK
+		| UVD_CGC_CTRL__REGS_MODE_MASK
+		| UVD_CGC_CTRL__RBC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_MC_MODE_MASK
+		| UVD_CGC_CTRL__LMI_UMC_MODE_MASK
+		| UVD_CGC_CTRL__IDCT_MODE_MASK
+		| UVD_CGC_CTRL__MPRD_MODE_MASK
+		| UVD_CGC_CTRL__MPC_MODE_MASK
+		| UVD_CGC_CTRL__LBSI_MODE_MASK
+		| UVD_CGC_CTRL__LRBBM_MODE_MASK
+		| UVD_CGC_CTRL__WCB_MODE_MASK
+		| UVD_CGC_CTRL__VCPU_MODE_MASK
+		| UVD_CGC_CTRL__SCPU_MODE_MASK);
+	WREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL, data);
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_CTRL);
+	data |= (UVD_SUVD_CGC_CTRL__SRE_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SIT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SMP_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCM_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SDB_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SCLR_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__UVD_SC_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__ENT_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__IME_MODE_MASK
+		| UVD_SUVD_CGC_CTRL__SITE_MODE_MASK);
+	WREG32_SOC15(VCN, 0, mmUVD_SUVD_CGC_CTRL, data);
+}
+
+static void vcn_1_0_disable_static_power_gating(struct amdgpu_device *adev)
+{
+	uint32_t data = 0;
+	int ret;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIL_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIR_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDW_PWR_CONFIG__SHIFT);
+
+		WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON, 0xFFFFFF, ret);
+	} else {
+		data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDIL_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDIR_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT
+			| 1 << UVD_PGFSM_CONFIG__UVDW_PWR_CONFIG__SHIFT);
+		WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data);
+		SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, 0,  0xFFFFFFFF, ret);
+	}
+
+	/* polling UVD_PGFSM_STATUS to confirm UVDM_PWR_STATUS , UVDU_PWR_STATUS are 0 (power on) */
+
+	data = RREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS);
+	data &= ~0x103;
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN)
+		data |= UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON | UVD_POWER_STATUS__UVD_PG_EN_MASK;
+
+	WREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS, data);
+}
+
+static void vcn_1_0_enable_static_power_gating(struct amdgpu_device *adev)
+{
+	uint32_t data = 0;
+	int ret;
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+		/* Before power off, this indicator has to be turned on */
+		data = RREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS);
+		data &= ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK;
+		data |= UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF;
+		WREG32_SOC15(VCN, 0, mmUVD_POWER_STATUS, data);
+
+
+		data = (2 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDF_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDC_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDB_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIL_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDIR_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTD_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT
+			| 2 << UVD_PGFSM_CONFIG__UVDW_PWR_CONFIG__SHIFT);
+
+		WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data);
+
+		data = (2 << UVD_PGFSM_STATUS__UVDM_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDU_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDF_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDC_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDB_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDIL_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDIR_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTD_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDTE_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDE_PWR_STATUS__SHIFT
+			| 2 << UVD_PGFSM_STATUS__UVDW_PWR_STATUS__SHIFT);
+		SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, data, 0xFFFFFFFF, ret);
+	}
+}
+
+/**
+ * vcn_v1_0_start - start VCN block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the VCN block
+ */
+static int vcn_v1_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->vcn.ring_dec;
+	uint32_t rb_bufsz, tmp;
+	uint32_t lmi_swap_cntl;
+	int i, j, r;
+
+	/* disable byte swapping */
+	lmi_swap_cntl = 0;
+
+	vcn_1_0_disable_static_power_gating(adev);
+	/* disable clock gating */
+	vcn_v1_0_disable_clock_gating(adev);
+
+	vcn_v1_0_mc_resume(adev);
+
+	/* disable interupt */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN), 0,
+			~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+	/* stall UMC and register bus before resetting VCPU */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2),
+			UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
+			~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+	mdelay(1);
+
+	/* put LMI, VCPU, RBC etc... into reset */
+	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET,
+		UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__RBC_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__CXW_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* initialize VCN memory controller */
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_CTRL,
+		(0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		0x00100000L);
+
+#ifdef __BIG_ENDIAN
+	/* swap (8 in 32) RB and IB */
+	lmi_swap_cntl = 0xa;
+#endif
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
+
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXA0, 0x40c2040);
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXA1, 0x0);
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXB0, 0x40c2040);
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUXB1, 0x0);
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_ALU, 0);
+	WREG32_SOC15(UVD, 0, mmUVD_MPC_SET_MUX, 0x88);
+
+	/* take all subblocks out of reset, except VCPU */
+	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET,
+			UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* enable VCPU clock */
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CNTL,
+			UVD_VCPU_CNTL__CLK_EN_MASK);
+
+	/* enable UMC */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), 0,
+			~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+	/* boot up the VCPU */
+	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET, 0);
+	mdelay(10);
+
+	for (i = 0; i < 10; ++i) {
+		uint32_t status;
+
+		for (j = 0; j < 100; ++j) {
+			status = RREG32_SOC15(UVD, 0, mmUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("VCN decode not responding, trying to reset the VCPU!!!\n");
+		WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
+				UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
+				~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), 0,
+				~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		r = -1;
+	}
+
+	if (r) {
+		DRM_ERROR("VCN decode not responding, giving up!!!\n");
+		return r;
+	}
+	/* enable master interrupt */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN),
+		(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
+		~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK));
+
+	/* clear the bit 4 of VCN_STATUS */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), 0,
+			~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+
+	/* force RBC into idle state */
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, tmp);
+
+	/* set the write pointer delay */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* set the wb address */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR,
+			(upper_32_bits(ring->gpu_addr) >> 2));
+
+	/* programm the RB_BASE for ring buffer */
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+			lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+			upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR, 0);
+
+	ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR);
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
+			lower_32_bits(ring->wptr));
+
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_CNTL), 0,
+			~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK);
+
+	ring = &adev->vcn.ring_enc[0];
+	WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+	WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO, ring->gpu_addr);
+	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE, ring->ring_size / 4);
+
+	ring = &adev->vcn.ring_enc[1];
+	WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+	WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO2, ring->gpu_addr);
+	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4);
+
+	ring = &adev->vcn.ring_jpeg;
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_VMID, 0);
+	WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_LOW, lower_32_bits(ring->gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_HIGH, upper_32_bits(ring->gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_RPTR, 0);
+	WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR, 0);
+	WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_CNTL, 0x00000002L);
+
+	/* initialize wptr */
+	ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR);
+
+	/* copy patch commands to the jpeg ring */
+	vcn_v1_0_jpeg_ring_set_patch_ring(ring,
+		(ring->wptr + ring->max_dw * amdgpu_sched_hw_submission));
+
+	return 0;
+}
+
+/**
+ * vcn_v1_0_stop - stop VCN block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the VCN block
+ */
+static int vcn_v1_0_stop(struct amdgpu_device *adev)
+{
+	/* force RBC into idle state */
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, 0x11010101);
+
+	/* Stall UMC and register bus before resetting VCPU */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2),
+			UVD_LMI_CTRL2__STALL_ARB_UMC_MASK,
+			~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+	mdelay(1);
+
+	/* put VCPU into reset */
+	WREG32_SOC15(UVD, 0, mmUVD_SOFT_RESET,
+			UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* disable VCPU clock */
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CNTL, 0x0);
+
+	/* Unstall UMC and register bus */
+	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2), 0,
+			~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+	WREG32_SOC15(VCN, 0, mmUVD_STATUS, 0);
+
+	vcn_v1_0_enable_clock_gating(adev);
+	vcn_1_0_enable_static_power_gating(adev);
+	return 0;
+}
+
+static bool vcn_v1_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return (RREG32_SOC15(VCN, 0, mmUVD_STATUS) == 0x2);
+}
+
+static int vcn_v1_0_wait_for_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret = 0;
+
+	SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_STATUS, 0x2, 0x2, ret);
+
+	return ret;
+}
+
+static int vcn_v1_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+
+	if (enable) {
+		/* wait for STATUS to clear */
+		if (vcn_v1_0_is_idle(handle))
+			return -EBUSY;
+		vcn_v1_0_enable_clock_gating(adev);
+	} else {
+		/* disable HW gating and enable Sw gating */
+		vcn_v1_0_disable_clock_gating(adev);
+	}
+	return 0;
+}
+
+/**
+ * vcn_v1_0_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t vcn_v1_0_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR);
+}
+
+/**
+ * vcn_v1_0_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t vcn_v1_0_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR);
+}
+
+/**
+ * vcn_v1_0_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void vcn_v1_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR, lower_32_bits(ring->wptr));
+}
+
+/**
+ * vcn_v1_0_dec_ring_insert_start - insert a start command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a start command to the ring.
+ */
+static void vcn_v1_0_dec_ring_insert_start(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
+	amdgpu_ring_write(ring, VCN_DEC_CMD_PACKET_START << 1);
+}
+
+/**
+ * vcn_v1_0_dec_ring_insert_end - insert a end command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a end command to the ring.
+ */
+static void vcn_v1_0_dec_ring_insert_end(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
+	amdgpu_ring_write(ring, VCN_DEC_CMD_PACKET_END << 1);
+}
+
+/**
+ * vcn_v1_0_dec_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @fence: fence to emit
+ *
+ * Write a fence and a trap command to the ring.
+ */
+static void vcn_v1_0_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				     unsigned flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_CONTEXT_ID), 0));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
+	amdgpu_ring_write(ring, addr & 0xffffffff);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0));
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xff);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
+	amdgpu_ring_write(ring, VCN_DEC_CMD_FENCE << 1);
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
+	amdgpu_ring_write(ring, VCN_DEC_CMD_TRAP << 1);
+}
+
+/**
+ * vcn_v1_0_dec_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @ib: indirect buffer to execute
+ *
+ * Write ring commands to execute the indirect buffer
+ */
+static void vcn_v1_0_dec_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib,
+				  unsigned vmid, bool ctx_switch)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_IB_VMID), 0));
+	amdgpu_ring_write(ring, vmid);
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_IB_64BIT_BAR_LOW), 0));
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH), 0));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_IB_SIZE), 0));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+static void vcn_v1_0_dec_ring_emit_reg_wait(struct amdgpu_ring *ring,
+					    uint32_t reg, uint32_t val,
+					    uint32_t mask)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0));
+	amdgpu_ring_write(ring, val);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GP_SCRATCH8), 0));
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
+	amdgpu_ring_write(ring, VCN_DEC_CMD_REG_READ_COND_WAIT << 1);
+}
+
+static void vcn_v1_0_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					    unsigned vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	uint32_t data0, data1, mask;
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for register write */
+	data0 = hub->ctx0_ptb_addr_lo32 + vmid * 2;
+	data1 = lower_32_bits(pd_addr);
+	mask = 0xffffffff;
+	vcn_v1_0_dec_ring_emit_reg_wait(ring, data0, data1, mask);
+}
+
+static void vcn_v1_0_dec_ring_emit_wreg(struct amdgpu_ring *ring,
+					uint32_t reg, uint32_t val)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0));
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0));
+	amdgpu_ring_write(ring, val);
+	amdgpu_ring_write(ring,
+		PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0));
+	amdgpu_ring_write(ring, VCN_DEC_CMD_WRITE_REG << 1);
+}
+
+/**
+ * vcn_v1_0_enc_ring_get_rptr - get enc read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc read pointer
+ */
+static uint64_t vcn_v1_0_enc_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vcn.ring_enc[0])
+		return RREG32_SOC15(UVD, 0, mmUVD_RB_RPTR);
+	else
+		return RREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2);
+}
+
+ /**
+ * vcn_v1_0_enc_ring_get_wptr - get enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware enc write pointer
+ */
+static uint64_t vcn_v1_0_enc_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vcn.ring_enc[0])
+		return RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR);
+	else
+		return RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2);
+}
+
+ /**
+ * vcn_v1_0_enc_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void vcn_v1_0_enc_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vcn.ring_enc[0])
+		WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR,
+			lower_32_bits(ring->wptr));
+	else
+		WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2,
+			lower_32_bits(ring->wptr));
+}
+
+/**
+ * vcn_v1_0_enc_ring_emit_fence - emit an enc fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @fence: fence to emit
+ *
+ * Write enc a fence and a trap command to the ring.
+ */
+static void vcn_v1_0_enc_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
+			u64 seq, unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, VCN_ENC_CMD_FENCE);
+	amdgpu_ring_write(ring, addr);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, VCN_ENC_CMD_TRAP);
+}
+
+static void vcn_v1_0_enc_ring_insert_end(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, VCN_ENC_CMD_END);
+}
+
+/**
+ * vcn_v1_0_enc_ring_emit_ib - enc execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @ib: indirect buffer to execute
+ *
+ * Write enc ring commands to execute the indirect buffer
+ */
+static void vcn_v1_0_enc_ring_emit_ib(struct amdgpu_ring *ring,
+		struct amdgpu_ib *ib, unsigned int vmid, bool ctx_switch)
+{
+	amdgpu_ring_write(ring, VCN_ENC_CMD_IB);
+	amdgpu_ring_write(ring, vmid);
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+static void vcn_v1_0_enc_ring_emit_reg_wait(struct amdgpu_ring *ring,
+					    uint32_t reg, uint32_t val,
+					    uint32_t mask)
+{
+	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WAIT);
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, mask);
+	amdgpu_ring_write(ring, val);
+}
+
+static void vcn_v1_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					    unsigned int vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for reg writes */
+	vcn_v1_0_enc_ring_emit_reg_wait(ring, hub->ctx0_ptb_addr_lo32 + vmid * 2,
+					lower_32_bits(pd_addr), 0xffffffff);
+}
+
+static void vcn_v1_0_enc_ring_emit_wreg(struct amdgpu_ring *ring,
+					uint32_t reg, uint32_t val)
+{
+	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WRITE);
+	amdgpu_ring_write(ring,	reg << 2);
+	amdgpu_ring_write(ring, val);
+}
+
+
+/**
+ * vcn_v1_0_jpeg_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t vcn_v1_0_jpeg_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_RPTR);
+}
+
+/**
+ * vcn_v1_0_jpeg_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t vcn_v1_0_jpeg_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR);
+}
+
+/**
+ * vcn_v1_0_jpeg_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void vcn_v1_0_jpeg_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR, lower_32_bits(ring->wptr));
+}
+
+/**
+ * vcn_v1_0_jpeg_ring_insert_start - insert a start command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a start command to the ring.
+ */
+static void vcn_v1_0_jpeg_ring_insert_start(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x68e04);
+
+	amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x80010000);
+}
+
+/**
+ * vcn_v1_0_jpeg_ring_insert_end - insert a end command
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Write a end command to the ring.
+ */
+static void vcn_v1_0_jpeg_ring_insert_end(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x68e04);
+
+	amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x00010000);
+}
+
+/**
+ * vcn_v1_0_jpeg_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @fence: fence to emit
+ *
+ * Write a fence and a trap command to the ring.
+ */
+static void vcn_v1_0_jpeg_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				     unsigned flags)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_GPCOM_DATA0), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, seq);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_GPCOM_DATA1), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, seq);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_LOW), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_HIGH), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_GPCOM_CMD), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x8);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_GPCOM_CMD), 0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE4));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_COND_RD_TIMER), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x01400200);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_REF_DATA), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, seq);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(0, 0, PACKETJ_CONDITION_CHECK3, PACKETJ_TYPE2));
+	amdgpu_ring_write(ring, 0xffffffff);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x3fbc);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(0, 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x1);
+}
+
+/**
+ * vcn_v1_0_jpeg_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @ib: indirect buffer to execute
+ *
+ * Write ring commands to execute the indirect buffer.
+ */
+static void vcn_v1_0_jpeg_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib,
+				  unsigned vmid, bool ctx_switch)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JRBC_IB_VMID), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, (vmid | (vmid << 4)));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JPEG_VMID), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, (vmid | (vmid << 4)));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JRBC_IB_64BIT_BAR_LOW), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JRBC_IB_64BIT_BAR_HIGH), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_IB_SIZE), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, ib->length_dw);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, lower_32_bits(ring->gpu_addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, upper_32_bits(ring->gpu_addr));
+
+	amdgpu_ring_write(ring,
+		PACKETJ(0, 0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE2));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_COND_RD_TIMER), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x01400200);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_REF_DATA), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x2);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_STATUS), 0, PACKETJ_CONDITION_CHECK3, PACKETJ_TYPE3));
+	amdgpu_ring_write(ring, 0x2);
+}
+
+static void vcn_v1_0_jpeg_ring_emit_reg_wait(struct amdgpu_ring *ring,
+					    uint32_t reg, uint32_t val,
+					    uint32_t mask)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reg_offset = (reg << 2);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_COND_RD_TIMER), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, 0x01400200);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_REF_DATA), 0, 0, PACKETJ_TYPE0));
+	amdgpu_ring_write(ring, val);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0));
+	if (((reg_offset >= 0x1f800) && (reg_offset <= 0x21fff)) ||
+		((reg_offset >= 0x1e000) && (reg_offset <= 0x1e1ff))) {
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring,
+			PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE3));
+	} else {
+		amdgpu_ring_write(ring, reg_offset);
+		amdgpu_ring_write(ring,
+			PACKETJ(0, 0, 0, PACKETJ_TYPE3));
+	}
+	amdgpu_ring_write(ring, mask);
+}
+
+static void vcn_v1_0_jpeg_ring_emit_vm_flush(struct amdgpu_ring *ring,
+		unsigned vmid, uint64_t pd_addr)
+{
+	struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
+	uint32_t data0, data1, mask;
+
+	pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
+
+	/* wait for register write */
+	data0 = hub->ctx0_ptb_addr_lo32 + vmid * 2;
+	data1 = lower_32_bits(pd_addr);
+	mask = 0xffffffff;
+	vcn_v1_0_jpeg_ring_emit_reg_wait(ring, data0, data1, mask);
+}
+
+static void vcn_v1_0_jpeg_ring_emit_wreg(struct amdgpu_ring *ring,
+					uint32_t reg, uint32_t val)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t reg_offset = (reg << 2);
+
+	amdgpu_ring_write(ring,
+		PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0));
+	if (((reg_offset >= 0x1f800) && (reg_offset <= 0x21fff)) ||
+			((reg_offset >= 0x1e000) && (reg_offset <= 0x1e1ff))) {
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring,
+			PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE0));
+	} else {
+		amdgpu_ring_write(ring, reg_offset);
+		amdgpu_ring_write(ring,
+			PACKETJ(0, 0, 0, PACKETJ_TYPE0));
+	}
+	amdgpu_ring_write(ring, val);
+}
+
+static void vcn_v1_0_jpeg_ring_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE6));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static void vcn_v1_0_jpeg_ring_patch_wreg(struct amdgpu_ring *ring, uint32_t *ptr, uint32_t reg_offset, uint32_t val)
+{
+	struct amdgpu_device *adev = ring->adev;
+	ring->ring[(*ptr)++] = PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0);
+	if (((reg_offset >= 0x1f800) && (reg_offset <= 0x21fff)) ||
+		((reg_offset >= 0x1e000) && (reg_offset <= 0x1e1ff))) {
+		ring->ring[(*ptr)++] = 0;
+		ring->ring[(*ptr)++] = PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE0);
+	} else {
+		ring->ring[(*ptr)++] = reg_offset;
+		ring->ring[(*ptr)++] = PACKETJ(0, 0, 0, PACKETJ_TYPE0);
+	}
+	ring->ring[(*ptr)++] = val;
+}
+
+static void vcn_v1_0_jpeg_ring_set_patch_ring(struct amdgpu_ring *ring, uint32_t ptr)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	uint32_t reg, reg_offset, val, mask, i;
+
+	// 1st: program mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW
+	reg = SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW);
+	reg_offset = (reg << 2);
+	val = lower_32_bits(ring->gpu_addr);
+	vcn_v1_0_jpeg_ring_patch_wreg(ring, &ptr, reg_offset, val);
+
+	// 2nd: program mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH
+	reg = SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH);
+	reg_offset = (reg << 2);
+	val = upper_32_bits(ring->gpu_addr);
+	vcn_v1_0_jpeg_ring_patch_wreg(ring, &ptr, reg_offset, val);
+
+	// 3rd to 5th: issue MEM_READ commands
+	for (i = 0; i <= 2; i++) {
+		ring->ring[ptr++] = PACKETJ(0, 0, 0, PACKETJ_TYPE2);
+		ring->ring[ptr++] = 0;
+	}
+
+	// 6th: program mmUVD_JRBC_RB_CNTL register to enable NO_FETCH and RPTR write ability
+	reg = SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_CNTL);
+	reg_offset = (reg << 2);
+	val = 0x13;
+	vcn_v1_0_jpeg_ring_patch_wreg(ring, &ptr, reg_offset, val);
+
+	// 7th: program mmUVD_JRBC_RB_REF_DATA
+	reg = SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_REF_DATA);
+	reg_offset = (reg << 2);
+	val = 0x1;
+	vcn_v1_0_jpeg_ring_patch_wreg(ring, &ptr, reg_offset, val);
+
+	// 8th: issue conditional register read mmUVD_JRBC_RB_CNTL
+	reg = SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_CNTL);
+	reg_offset = (reg << 2);
+	val = 0x1;
+	mask = 0x1;
+
+	ring->ring[ptr++] = PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_COND_RD_TIMER), 0, 0, PACKETJ_TYPE0);
+	ring->ring[ptr++] = 0x01400200;
+	ring->ring[ptr++] = PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_REF_DATA), 0, 0, PACKETJ_TYPE0);
+	ring->ring[ptr++] = val;
+	ring->ring[ptr++] = PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_EXTERNAL_REG_BASE), 0, 0, PACKETJ_TYPE0);
+	if (((reg_offset >= 0x1f800) && (reg_offset <= 0x21fff)) ||
+		((reg_offset >= 0x1e000) && (reg_offset <= 0x1e1ff))) {
+		ring->ring[ptr++] = 0;
+		ring->ring[ptr++] = PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE3);
+	} else {
+		ring->ring[ptr++] = reg_offset;
+		ring->ring[ptr++] = PACKETJ(0, 0, 0, PACKETJ_TYPE3);
+	}
+	ring->ring[ptr++] = mask;
+
+	//9th to 21st: insert no-op
+	for (i = 0; i <= 12; i++) {
+		ring->ring[ptr++] = PACKETJ(0, 0, 0, PACKETJ_TYPE6);
+		ring->ring[ptr++] = 0;
+	}
+
+	//22nd: reset mmUVD_JRBC_RB_RPTR
+	reg = SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_RPTR);
+	reg_offset = (reg << 2);
+	val = 0;
+	vcn_v1_0_jpeg_ring_patch_wreg(ring, &ptr, reg_offset, val);
+
+	//23rd: program mmUVD_JRBC_RB_CNTL to disable no_fetch
+	reg = SOC15_REG_OFFSET(UVD, 0, mmUVD_JRBC_RB_CNTL);
+	reg_offset = (reg << 2);
+	val = 0x12;
+	vcn_v1_0_jpeg_ring_patch_wreg(ring, &ptr, reg_offset, val);
+}
+
+static int vcn_v1_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	return 0;
+}
+
+static int vcn_v1_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: VCN TRAP\n");
+
+	switch (entry->src_id) {
+	case 124:
+		amdgpu_fence_process(&adev->vcn.ring_dec);
+		break;
+	case 119:
+		amdgpu_fence_process(&adev->vcn.ring_enc[0]);
+		break;
+	case 120:
+		amdgpu_fence_process(&adev->vcn.ring_enc[1]);
+		break;
+	case 126:
+		amdgpu_fence_process(&adev->vcn.ring_jpeg);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data[0]);
+		break;
+	}
+
+	return 0;
+}
+
+static void vcn_v1_0_dec_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int i;
+
+	WARN_ON(ring->wptr % 2 || count % 2);
+
+	for (i = 0; i < count / 2; i++) {
+		amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP), 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static int vcn_v1_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the VCN block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_PG_STATE_GATE)
+		return vcn_v1_0_stop(adev);
+	else
+		return vcn_v1_0_start(adev);
+}
+
+static const struct amd_ip_funcs vcn_v1_0_ip_funcs = {
+	.name = "vcn_v1_0",
+	.early_init = vcn_v1_0_early_init,
+	.late_init = NULL,
+	.sw_init = vcn_v1_0_sw_init,
+	.sw_fini = vcn_v1_0_sw_fini,
+	.hw_init = vcn_v1_0_hw_init,
+	.hw_fini = vcn_v1_0_hw_fini,
+	.suspend = vcn_v1_0_suspend,
+	.resume = vcn_v1_0_resume,
+	.is_idle = vcn_v1_0_is_idle,
+	.wait_for_idle = vcn_v1_0_wait_for_idle,
+	.check_soft_reset = NULL /* vcn_v1_0_check_soft_reset */,
+	.pre_soft_reset = NULL /* vcn_v1_0_pre_soft_reset */,
+	.soft_reset = NULL /* vcn_v1_0_soft_reset */,
+	.post_soft_reset = NULL /* vcn_v1_0_post_soft_reset */,
+	.set_clockgating_state = vcn_v1_0_set_clockgating_state,
+	.set_powergating_state = vcn_v1_0_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs vcn_v1_0_dec_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_DEC,
+	.align_mask = 0xf,
+	.support_64bit_ptrs = false,
+	.vmhub = AMDGPU_MMHUB,
+	.get_rptr = vcn_v1_0_dec_ring_get_rptr,
+	.get_wptr = vcn_v1_0_dec_ring_get_wptr,
+	.set_wptr = vcn_v1_0_dec_ring_set_wptr,
+	.emit_frame_size =
+		6 + 6 + /* hdp invalidate / flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* vcn_v1_0_dec_ring_emit_vm_flush */
+		14 + 14 + /* vcn_v1_0_dec_ring_emit_fence x2 vm fence */
+		6,
+	.emit_ib_size = 8, /* vcn_v1_0_dec_ring_emit_ib */
+	.emit_ib = vcn_v1_0_dec_ring_emit_ib,
+	.emit_fence = vcn_v1_0_dec_ring_emit_fence,
+	.emit_vm_flush = vcn_v1_0_dec_ring_emit_vm_flush,
+	.test_ring = amdgpu_vcn_dec_ring_test_ring,
+	.test_ib = amdgpu_vcn_dec_ring_test_ib,
+	.insert_nop = vcn_v1_0_dec_ring_insert_nop,
+	.insert_start = vcn_v1_0_dec_ring_insert_start,
+	.insert_end = vcn_v1_0_dec_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_v1_0_dec_ring_emit_wreg,
+	.emit_reg_wait = vcn_v1_0_dec_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+};
+
+static const struct amdgpu_ring_funcs vcn_v1_0_enc_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_ENC,
+	.align_mask = 0x3f,
+	.nop = VCN_ENC_CMD_NO_OP,
+	.support_64bit_ptrs = false,
+	.vmhub = AMDGPU_MMHUB,
+	.get_rptr = vcn_v1_0_enc_ring_get_rptr,
+	.get_wptr = vcn_v1_0_enc_ring_get_wptr,
+	.set_wptr = vcn_v1_0_enc_ring_set_wptr,
+	.emit_frame_size =
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+		4 + /* vcn_v1_0_enc_ring_emit_vm_flush */
+		5 + 5 + /* vcn_v1_0_enc_ring_emit_fence x2 vm fence */
+		1, /* vcn_v1_0_enc_ring_insert_end */
+	.emit_ib_size = 5, /* vcn_v1_0_enc_ring_emit_ib */
+	.emit_ib = vcn_v1_0_enc_ring_emit_ib,
+	.emit_fence = vcn_v1_0_enc_ring_emit_fence,
+	.emit_vm_flush = vcn_v1_0_enc_ring_emit_vm_flush,
+	.test_ring = amdgpu_vcn_enc_ring_test_ring,
+	.test_ib = amdgpu_vcn_enc_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.insert_end = vcn_v1_0_enc_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_v1_0_enc_ring_emit_wreg,
+	.emit_reg_wait = vcn_v1_0_enc_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+};
+
+static const struct amdgpu_ring_funcs vcn_v1_0_jpeg_ring_vm_funcs = {
+	.type = AMDGPU_RING_TYPE_VCN_JPEG,
+	.align_mask = 0xf,
+	.nop = PACKET0(0x81ff, 0),
+	.support_64bit_ptrs = false,
+	.vmhub = AMDGPU_MMHUB,
+	.extra_dw = 64,
+	.get_rptr = vcn_v1_0_jpeg_ring_get_rptr,
+	.get_wptr = vcn_v1_0_jpeg_ring_get_wptr,
+	.set_wptr = vcn_v1_0_jpeg_ring_set_wptr,
+	.emit_frame_size =
+		6 + 6 + /* hdp invalidate / flush */
+		SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+		8 + /* vcn_v1_0_dec_ring_emit_vm_flush */
+		14 + 14 + /* vcn_v1_0_dec_ring_emit_fence x2 vm fence */
+		6,
+	.emit_ib_size = 22, /* vcn_v1_0_dec_ring_emit_ib */
+	.emit_ib = vcn_v1_0_jpeg_ring_emit_ib,
+	.emit_fence = vcn_v1_0_jpeg_ring_emit_fence,
+	.emit_vm_flush = vcn_v1_0_jpeg_ring_emit_vm_flush,
+	.test_ring = amdgpu_vcn_jpeg_ring_test_ring,
+	.test_ib = amdgpu_vcn_jpeg_ring_test_ib,
+	.insert_nop = vcn_v1_0_jpeg_ring_nop,
+	.insert_start = vcn_v1_0_jpeg_ring_insert_start,
+	.insert_end = vcn_v1_0_jpeg_ring_insert_end,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vcn_ring_begin_use,
+	.end_use = amdgpu_vcn_ring_end_use,
+	.emit_wreg = vcn_v1_0_jpeg_ring_emit_wreg,
+	.emit_reg_wait = vcn_v1_0_jpeg_ring_emit_reg_wait,
+};
+
+static void vcn_v1_0_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->vcn.ring_dec.funcs = &vcn_v1_0_dec_ring_vm_funcs;
+	DRM_INFO("VCN decode is enabled in VM mode\n");
+}
+
+static void vcn_v1_0_set_enc_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->vcn.num_enc_rings; ++i)
+		adev->vcn.ring_enc[i].funcs = &vcn_v1_0_enc_ring_vm_funcs;
+
+	DRM_INFO("VCN encode is enabled in VM mode\n");
+}
+
+static void vcn_v1_0_set_jpeg_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->vcn.ring_jpeg.funcs = &vcn_v1_0_jpeg_ring_vm_funcs;
+	DRM_INFO("VCN jpeg decode is enabled in VM mode\n");
+}
+
+static const struct amdgpu_irq_src_funcs vcn_v1_0_irq_funcs = {
+	.set = vcn_v1_0_set_interrupt_state,
+	.process = vcn_v1_0_process_interrupt,
+};
+
+static void vcn_v1_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->vcn.irq.num_types = adev->vcn.num_enc_rings + 1;
+	adev->vcn.irq.funcs = &vcn_v1_0_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version vcn_v1_0_ip_block =
+{
+		.type = AMD_IP_BLOCK_TYPE_VCN,
+		.major = 1,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vcn_v1_0_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/vcn_v1_0.h b/sys/dev/pci/drm/amd/amdgpu/vcn_v1_0.h
new file mode 100644
index 00000000000..2a497a7a484
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vcn_v1_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCN_V1_0_H__
+#define __VCN_V1_0_H__
+
+extern const struct amdgpu_ip_block_version vcn_v1_0_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/vega10_ih.c b/sys/dev/pci/drm/amd/amdgpu/vega10_ih.c
new file mode 100644
index 00000000000..560815f32e4
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vega10_ih.c
@@ -0,0 +1,530 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "soc15.h"
+
+#include "oss/osssys_4_0_offset.h"
+#include "oss/osssys_4_0_sh_mask.h"
+
+#include "soc15_common.h"
+#include "vega10_ih.h"
+
+
+
+static void vega10_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * vega10_ih_enable_interrupts - Enable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enable the interrupt ring buffer (VEGA10).
+ */
+static void vega10_ih_enable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 1);
+	WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL, ih_rb_cntl);
+	adev->irq.ih.enabled = true;
+}
+
+/**
+ * vega10_ih_disable_interrupts - Disable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable the interrupt ring buffer (VEGA10).
+ */
+static void vega10_ih_disable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 0);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 0);
+	WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL, ih_rb_cntl);
+	/* set rptr, wptr to 0 */
+	WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR, 0);
+	WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR, 0);
+	adev->irq.ih.enabled = false;
+	adev->irq.ih.rptr = 0;
+}
+
+/**
+ * vega10_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (VI).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int vega10_ih_irq_init(struct amdgpu_device *adev)
+{
+	int ret = 0;
+	int rb_bufsz;
+	u32 ih_rb_cntl, ih_doorbell_rtpr;
+	u32 tmp;
+	u64 wptr_off;
+
+	/* disable irqs */
+	vega10_ih_disable_interrupts(adev);
+
+	adev->nbio_funcs->ih_control(adev);
+
+	ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL);
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	if (adev->irq.ih.use_bus_addr) {
+		WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE, adev->irq.ih.rb_dma_addr >> 8);
+		WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE_HI, ((u64)adev->irq.ih.rb_dma_addr >> 40) & 0xff);
+		ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SPACE, 1);
+	} else {
+		WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
+		WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE_HI, (adev->irq.ih.gpu_addr >> 40) & 0xff);
+		ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SPACE, 4);
+	}
+	rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register value is written to memory */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SNOOP, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_RO, 0);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
+
+	if (adev->irq.msi_enabled)
+		ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RPTR_REARM, 1);
+
+	WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL, ih_rb_cntl);
+
+	/* set the writeback address whether it's enabled or not */
+	if (adev->irq.ih.use_bus_addr)
+		wptr_off = adev->irq.ih.rb_dma_addr + (adev->irq.ih.wptr_offs * 4);
+	else
+		wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
+	WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
+	WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFFFF);
+
+	/* set rptr, wptr to 0 */
+	WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR, 0);
+	WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR, 0);
+
+	ih_doorbell_rtpr = RREG32_SOC15(OSSSYS, 0, mmIH_DOORBELL_RPTR);
+	if (adev->irq.ih.use_doorbell) {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
+						 OFFSET, adev->irq.ih.doorbell_index);
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
+						 ENABLE, 1);
+	} else {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
+						 ENABLE, 0);
+	}
+	WREG32_SOC15(OSSSYS, 0, mmIH_DOORBELL_RPTR, ih_doorbell_rtpr);
+	adev->nbio_funcs->ih_doorbell_range(adev, adev->irq.ih.use_doorbell,
+					    adev->irq.ih.doorbell_index);
+
+	tmp = RREG32_SOC15(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL);
+	tmp = REG_SET_FIELD(tmp, IH_STORM_CLIENT_LIST_CNTL,
+			    CLIENT18_IS_STORM_CLIENT, 1);
+	WREG32_SOC15(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL, tmp);
+
+	tmp = RREG32_SOC15(OSSSYS, 0, mmIH_INT_FLOOD_CNTL);
+	tmp = REG_SET_FIELD(tmp, IH_INT_FLOOD_CNTL, FLOOD_CNTL_ENABLE, 1);
+	WREG32_SOC15(OSSSYS, 0, mmIH_INT_FLOOD_CNTL, tmp);
+
+	pci_set_master(adev->pdev);
+
+	/* enable interrupts */
+	vega10_ih_enable_interrupts(adev);
+
+	return ret;
+}
+
+/**
+ * vega10_ih_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw (VEGA10).
+ */
+static void vega10_ih_irq_disable(struct amdgpu_device *adev)
+{
+	vega10_ih_disable_interrupts(adev);
+
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * vega10_ih_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer (VEGA10).  Also check for
+ * ring buffer overflow and deal with it.
+ * Returns the value of the wptr.
+ */
+static u32 vega10_ih_get_wptr(struct amdgpu_device *adev)
+{
+	u32 wptr, tmp;
+
+	if (adev->irq.ih.use_bus_addr)
+		wptr = le32_to_cpu(adev->irq.ih.ring[adev->irq.ih.wptr_offs]);
+	else
+		wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
+
+	if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
+		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+
+		/* When a ring buffer overflow happen start parsing interrupt
+		 * from the last not overwritten vector (wptr + 32). Hopefully
+		 * this should allow us to catchup.
+		 */
+		tmp = (wptr + 32) & adev->irq.ih.ptr_mask;
+		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+			wptr, adev->irq.ih.rptr, tmp);
+		adev->irq.ih.rptr = tmp;
+
+		tmp = RREG32_NO_KIQ(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL));
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+		WREG32_NO_KIQ(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL), tmp);
+	}
+	return (wptr & adev->irq.ih.ptr_mask);
+}
+
+/**
+ * vega10_ih_prescreen_iv - prescreen an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns true if the interrupt vector should be further processed.
+ */
+static bool vega10_ih_prescreen_iv(struct amdgpu_device *adev)
+{
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	u32 dw0, dw3, dw4, dw5;
+	u16 pasid;
+	u64 addr, key;
+	struct amdgpu_vm *vm;
+	struct amdgpu_vm_fault *vmf;
+	int r;
+
+	dw0 = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
+	dw3 = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+	dw4 = le32_to_cpu(adev->irq.ih.ring[ring_index + 4]);
+	dw5 = le32_to_cpu(adev->irq.ih.ring[ring_index + 5]);
+
+	/* Filter retry page faults, let only the first one pass. If
+	 * there are too many outstanding faults, ignore them until
+	 * some faults get cleared.
+	 */
+	switch (dw0 & 0xff) {
+	case SOC15_IH_CLIENTID_VMC:
+	case SOC15_IH_CLIENTID_UTCL2:
+		break;
+	default:
+		/* Not a VM fault */
+		return true;
+	}
+
+	pasid = dw3 & 0xffff;
+	/* No PASID, can't identify faulting process */
+	if (!pasid)
+		return true;
+
+	/* Not a retry fault, check fault credit */
+	if (!(dw5 & 0x80)) {
+		if (!amdgpu_vm_pasid_fault_credit(adev, pasid))
+			goto ignore_iv;
+		return true;
+	}
+
+	addr = ((u64)(dw5 & 0xf) << 44) | ((u64)dw4 << 12);
+	key = AMDGPU_VM_FAULT(pasid, addr);
+	r = amdgpu_ih_add_fault(adev, key);
+
+	/* Hash table is full or the fault is already being processed,
+	 * ignore further page faults
+	 */
+	if (r != 0)
+		goto ignore_iv;
+
+	/* Track retry faults in per-VM fault FIFO. */
+	spin_lock(&adev->vm_manager.pasid_lock);
+	vm = idr_find(&adev->vm_manager.pasid_idr, pasid);
+	if (!vm) {
+		/* VM not found, process it normally */
+		spin_unlock(&adev->vm_manager.pasid_lock);
+		amdgpu_ih_clear_fault(adev, key);
+		return true;
+	}
+	/* No locking required with single writer and single reader */
+#ifdef __linux__
+	r = kfifo_put(&vm->faults, key);
+	if (!r) {
+		/* FIFO is full. Ignore it until there is space */
+		spin_unlock(&adev->vm_manager.pasid_lock);
+		amdgpu_ih_clear_fault(adev, key);
+		goto ignore_iv;
+	}
+#else
+	vmf = malloc(sizeof(*vmf), M_DRM, M_NOWAIT);
+	if (vmf == NULL) {
+		spin_unlock(&adev->vm_manager.pasid_lock);
+		amdgpu_ih_clear_fault(adev, key);
+		goto ignore_iv;
+	}
+	vmf->val = key;
+	SIMPLEQ_INSERT_TAIL(&vm->faults, vmf, vm_fault_entry);
+#endif
+	spin_unlock(&adev->vm_manager.pasid_lock);
+
+	/* It's the first fault for this address, process it normally */
+	return true;
+
+ignore_iv:
+	adev->irq.ih.rptr += 32;
+	return false;
+}
+
+/**
+ * vega10_ih_decode_iv - decode an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Decodes the interrupt vector at the current rptr
+ * position and also advance the position.
+ */
+static void vega10_ih_decode_iv(struct amdgpu_device *adev,
+				 struct amdgpu_iv_entry *entry)
+{
+	/* wptr/rptr are in bytes! */
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	uint32_t dw[8];
+
+	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+	dw[4] = le32_to_cpu(adev->irq.ih.ring[ring_index + 4]);
+	dw[5] = le32_to_cpu(adev->irq.ih.ring[ring_index + 5]);
+	dw[6] = le32_to_cpu(adev->irq.ih.ring[ring_index + 6]);
+	dw[7] = le32_to_cpu(adev->irq.ih.ring[ring_index + 7]);
+
+	entry->client_id = dw[0] & 0xff;
+	entry->src_id = (dw[0] >> 8) & 0xff;
+	entry->ring_id = (dw[0] >> 16) & 0xff;
+	entry->vmid = (dw[0] >> 24) & 0xf;
+	entry->vmid_src = (dw[0] >> 31);
+	entry->timestamp = dw[1] | ((u64)(dw[2] & 0xffff) << 32);
+	entry->timestamp_src = dw[2] >> 31;
+	entry->pasid = dw[3] & 0xffff;
+	entry->pasid_src = dw[3] >> 31;
+	entry->src_data[0] = dw[4];
+	entry->src_data[1] = dw[5];
+	entry->src_data[2] = dw[6];
+	entry->src_data[3] = dw[7];
+
+
+	/* wptr/rptr are in bytes! */
+	adev->irq.ih.rptr += 32;
+}
+
+/**
+ * vega10_ih_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void vega10_ih_set_rptr(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih.use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		if (adev->irq.ih.use_bus_addr)
+			adev->irq.ih.ring[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
+		else
+			adev->wb.wb[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
+		WDOORBELL32(adev->irq.ih.doorbell_index, adev->irq.ih.rptr);
+	} else {
+		WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR, adev->irq.ih.rptr);
+	}
+}
+
+static int vega10_ih_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	vega10_ih_set_interrupt_funcs(adev);
+	return 0;
+}
+
+static int vega10_ih_sw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_ih_ring_init(adev, 256 * 1024, true);
+	if (r)
+		return r;
+
+	adev->irq.ih.use_doorbell = true;
+	adev->irq.ih.doorbell_index = AMDGPU_DOORBELL64_IH << 1;
+
+	adev->irq.ih.faults = kmalloc(sizeof(*adev->irq.ih.faults), GFP_KERNEL);
+	if (!adev->irq.ih.faults)
+		return -ENOMEM;
+	INIT_CHASH_TABLE(adev->irq.ih.faults->hash,
+			 AMDGPU_PAGEFAULT_HASH_BITS, 8, 0);
+	mtx_init(&adev->irq.ih.faults->lock, IPL_TTY);
+	adev->irq.ih.faults->count = 0;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int vega10_ih_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_fini(adev);
+	amdgpu_ih_ring_fini(adev);
+
+	kfree(adev->irq.ih.faults);
+	adev->irq.ih.faults = NULL;
+
+	return 0;
+}
+
+static int vega10_ih_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = vega10_ih_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int vega10_ih_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	vega10_ih_irq_disable(adev);
+
+	return 0;
+}
+
+static int vega10_ih_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return vega10_ih_hw_fini(adev);
+}
+
+static int vega10_ih_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return vega10_ih_hw_init(adev);
+}
+
+static bool vega10_ih_is_idle(void *handle)
+{
+	/* todo */
+	return true;
+}
+
+static int vega10_ih_wait_for_idle(void *handle)
+{
+	/* todo */
+	return -ETIMEDOUT;
+}
+
+static int vega10_ih_soft_reset(void *handle)
+{
+	/* todo */
+
+	return 0;
+}
+
+static int vega10_ih_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int vega10_ih_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs vega10_ih_ip_funcs = {
+	.name = "vega10_ih",
+	.early_init = vega10_ih_early_init,
+	.late_init = NULL,
+	.sw_init = vega10_ih_sw_init,
+	.sw_fini = vega10_ih_sw_fini,
+	.hw_init = vega10_ih_hw_init,
+	.hw_fini = vega10_ih_hw_fini,
+	.suspend = vega10_ih_suspend,
+	.resume = vega10_ih_resume,
+	.is_idle = vega10_ih_is_idle,
+	.wait_for_idle = vega10_ih_wait_for_idle,
+	.soft_reset = vega10_ih_soft_reset,
+	.set_clockgating_state = vega10_ih_set_clockgating_state,
+	.set_powergating_state = vega10_ih_set_powergating_state,
+};
+
+static const struct amdgpu_ih_funcs vega10_ih_funcs = {
+	.get_wptr = vega10_ih_get_wptr,
+	.prescreen_iv = vega10_ih_prescreen_iv,
+	.decode_iv = vega10_ih_decode_iv,
+	.set_rptr = vega10_ih_set_rptr
+};
+
+static void vega10_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih_funcs == NULL)
+		adev->irq.ih_funcs = &vega10_ih_funcs;
+}
+
+const struct amdgpu_ip_block_version vega10_ih_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_IH,
+	.major = 4,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &vega10_ih_ip_funcs,
+};
diff --git a/sys/dev/pci/drm/amd/amdgpu/vega10_ih.h b/sys/dev/pci/drm/amd/amdgpu/vega10_ih.h
new file mode 100644
index 00000000000..82edd28b997
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vega10_ih.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2016 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VEGA10_IH_H__
+#define __VEGA10_IH_H__
+
+extern const struct amd_ip_funcs vega10_ih_ip_funcs;
+extern const struct amdgpu_ip_block_version vega10_ih_ip_block;
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/vega10_reg_init.c b/sys/dev/pci/drm/amd/amdgpu/vega10_reg_init.c
new file mode 100644
index 00000000000..c5c9b2bc190
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vega10_reg_init.c
@@ -0,0 +1,59 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "soc15.h"
+
+#include "soc15_common.h"
+#include "soc15_hw_ip.h"
+#include "vega10_ip_offset.h"
+
+int vega10_reg_base_init(struct amdgpu_device *adev)
+{
+	/* HW has more IP blocks,  only initialized the blocke beend by our driver  */
+	uint32_t i;
+	for (i = 0 ; i < MAX_INSTANCE ; ++i) {
+		adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i]));
+		adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i]));
+		adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i]));
+		adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i]));
+		adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i]));
+		adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i]));
+		adev->reg_offset[UVD_HWIP][i] = (uint32_t *)(&(UVD_BASE.instance[i]));
+		adev->reg_offset[VCE_HWIP][i] = (uint32_t *)(&(VCE_BASE.instance[i]));
+		adev->reg_offset[VCN_HWIP][i] = (uint32_t *)(&(VCN_BASE.instance[i]));
+		adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i]));
+		adev->reg_offset[DCE_HWIP][i] = (uint32_t *)(&(DCE_BASE.instance[i]));
+		adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i]));
+		adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(SDMA0_BASE.instance[i]));
+		adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(SDMA1_BASE.instance[i]));
+		adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i]));
+		adev->reg_offset[PWR_HWIP][i] = (uint32_t *)(&(PWR_BASE.instance[i]));
+		adev->reg_offset[NBIF_HWIP][i] = (uint32_t *)(&(NBIF_BASE.instance[i]));
+		adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i]));
+		adev->reg_offset[CLK_HWIP][i] = (uint32_t *)(&(CLK_BASE.instance[i]));
+	}
+	return 0;
+}
+
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/vega10_sdma_pkt_open.h b/sys/dev/pci/drm/amd/amdgpu/vega10_sdma_pkt_open.h
new file mode 100644
index 00000000000..8de4ccce5e3
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vega10_sdma_pkt_open.h
@@ -0,0 +1,3335 @@
+/*
+ * Copyright (C) 2016  Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VEGA10_SDMA_PKT_OPEN_H_
+#define __VEGA10_SDMA_PKT_OPEN_H_
+
+#define SDMA_OP_NOP  0
+#define SDMA_OP_COPY  1
+#define SDMA_OP_WRITE  2
+#define SDMA_OP_INDIRECT  4
+#define SDMA_OP_FENCE  5
+#define SDMA_OP_TRAP  6
+#define SDMA_OP_SEM  7
+#define SDMA_OP_POLL_REGMEM  8
+#define SDMA_OP_COND_EXE  9
+#define SDMA_OP_ATOMIC  10
+#define SDMA_OP_CONST_FILL  11
+#define SDMA_OP_PTEPDE  12
+#define SDMA_OP_TIMESTAMP  13
+#define SDMA_OP_SRBM_WRITE  14
+#define SDMA_OP_PRE_EXE  15
+#define SDMA_OP_DUMMY_TRAP  16
+#define SDMA_SUBOP_TIMESTAMP_SET  0
+#define SDMA_SUBOP_TIMESTAMP_GET  1
+#define SDMA_SUBOP_TIMESTAMP_GET_GLOBAL  2
+#define SDMA_SUBOP_COPY_LINEAR  0
+#define SDMA_SUBOP_COPY_LINEAR_SUB_WIND  4
+#define SDMA_SUBOP_COPY_TILED  1
+#define SDMA_SUBOP_COPY_TILED_SUB_WIND  5
+#define SDMA_SUBOP_COPY_T2T_SUB_WIND  6
+#define SDMA_SUBOP_COPY_SOA  3
+#define SDMA_SUBOP_COPY_DIRTY_PAGE  7
+#define SDMA_SUBOP_COPY_LINEAR_PHY  8
+#define SDMA_SUBOP_WRITE_LINEAR  0
+#define SDMA_SUBOP_WRITE_TILED  1
+#define SDMA_SUBOP_PTEPDE_GEN  0
+#define SDMA_SUBOP_PTEPDE_COPY  1
+#define SDMA_SUBOP_PTEPDE_RMW  2
+#define SDMA_SUBOP_PTEPDE_COPY_BACKWARDS  3
+#define SDMA_SUBOP_DATA_FILL_MULTI  1
+#define SDMA_SUBOP_POLL_REG_WRITE_MEM  1
+#define SDMA_SUBOP_POLL_DBIT_WRITE_MEM  2
+#define SDMA_SUBOP_POLL_MEM_VERIFY  3
+#define HEADER_AGENT_DISPATCH  4
+#define HEADER_BARRIER  5
+#define SDMA_OP_AQL_COPY  0
+#define SDMA_OP_AQL_BARRIER_OR  0
+
+/*define for op field*/
+#define SDMA_PKT_HEADER_op_offset 0
+#define SDMA_PKT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_HEADER_op_shift  0
+#define SDMA_PKT_HEADER_OP(x) (((x) & SDMA_PKT_HEADER_op_mask) << SDMA_PKT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_HEADER_sub_op_offset 0
+#define SDMA_PKT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_HEADER_sub_op_shift  8
+#define SDMA_PKT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_HEADER_sub_op_mask) << SDMA_PKT_HEADER_sub_op_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_encrypt_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_encrypt_shift  16
+#define SDMA_PKT_COPY_LINEAR_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_encrypt_mask) << SDMA_PKT_COPY_LINEAR_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_LINEAR_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_tmz_mask) << SDMA_PKT_COPY_LINEAR_HEADER_tmz_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_DIRTY_PAGE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_OP(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_tmz_shift)
+
+/*define for all field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_offset 0
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_shift  31
+#define SDMA_PKT_COPY_DIRTY_PAGE_HEADER_ALL(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_mask) << SDMA_PKT_COPY_DIRTY_PAGE_HEADER_all_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_offset 1
+#define SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_mask) << SDMA_PKT_COPY_DIRTY_PAGE_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sw_shift)
+
+/*define for dst_gcc field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_shift  19
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_GCC(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gcc_shift)
+
+/*define for dst_sys field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_shift  20
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_SYS(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_sys_shift)
+
+/*define for dst_snoop field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_shift  22
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_SNOOP(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_snoop_shift)
+
+/*define for dst_gpa field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_shift  23
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_DST_GPA(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_dst_gpa_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sw_shift)
+
+/*define for src_sys field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_shift  28
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_SYS(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_sys_shift)
+
+/*define for src_snoop field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_shift  30
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_SNOOP(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_snoop_shift)
+
+/*define for src_gpa field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_offset 2
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_mask   0x00000001
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_shift  31
+#define SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_SRC_GPA(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_mask) << SDMA_PKT_COPY_DIRTY_PAGE_PARAMETER_src_gpa_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_DIRTY_PAGE_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_DIRTY_PAGE_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_PHYSICAL_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_HEADER_tmz_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for dst_gcc field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_shift  19
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_GCC(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gcc_shift)
+
+/*define for dst_sys field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_shift  20
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_SYS(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_sys_shift)
+
+/*define for dst_log field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_shift  21
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_LOG(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_log_shift)
+
+/*define for dst_snoop field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_shift  22
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_SNOOP(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_snoop_shift)
+
+/*define for dst_gpa field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_shift  23
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_DST_GPA(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_dst_gpa_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_gcc field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_shift  27
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_GCC(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gcc_shift)
+
+/*define for src_sys field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_shift  28
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_SYS(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_sys_shift)
+
+/*define for src_snoop field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_shift  30
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_SNOOP(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_snoop_shift)
+
+/*define for src_gpa field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_offset 2
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_mask   0x00000001
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_shift  31
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_SRC_GPA(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_PARAMETER_src_gpa_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_PHYSICAL_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_BROADCAST_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_shift  16
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_tmz_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST2_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift)
+
+/*define for dst1_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift  16
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST1_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST1_ADDR_LO word*/
+/*define for dst1_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_offset 5
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_DST1_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift)
+
+/*define for DST1_ADDR_HI word*/
+/*define for dst1_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_offset 6
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_DST1_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift)
+
+/*define for DST2_ADDR_LO word*/
+/*define for dst2_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_offset 7
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_DST2_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift)
+
+/*define for DST2_ADDR_HI word*/
+/*define for dst2_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_offset 8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_DST2_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_tmz_shift)
+
+/*define for elementsize field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift  29
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_ELEMENTSIZE(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift)
+
+/*define for src_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift  13
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_offset 6
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_offset 7
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_8 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift)
+
+/*define for DW_9 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift)
+
+/*define for dst_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift  13
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift)
+
+/*define for DW_10 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_offset 10
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift)
+
+/*define for DW_12 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_RECT_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_op_mask) << SDMA_PKT_COPY_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_COPY_TILED_HEADER_encrypt_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_encrypt_shift  16
+#define SDMA_PKT_COPY_TILED_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_encrypt_mask) << SDMA_PKT_COPY_TILED_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_TILED_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_TILED_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_tmz_mask) << SDMA_PKT_COPY_TILED_HEADER_tmz_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_COPY_TILED_HEADER_mip_max_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_HEADER_mip_max_shift  20
+#define SDMA_PKT_COPY_TILED_HEADER_MIP_MAX(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_mip_max_mask) << SDMA_PKT_COPY_TILED_HEADER_mip_max_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for width field*/
+#define SDMA_PKT_COPY_TILED_DW_3_width_offset 3
+#define SDMA_PKT_COPY_TILED_DW_3_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_3_width_shift  0
+#define SDMA_PKT_COPY_TILED_DW_3_WIDTH(x) (((x) & SDMA_PKT_COPY_TILED_DW_3_width_mask) << SDMA_PKT_COPY_TILED_DW_3_width_shift)
+
+/*define for DW_4 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_DW_4_height_offset 4
+#define SDMA_PKT_COPY_TILED_DW_4_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_4_height_shift  0
+#define SDMA_PKT_COPY_TILED_DW_4_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_DW_4_height_mask) << SDMA_PKT_COPY_TILED_DW_4_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_TILED_DW_4_depth_offset 4
+#define SDMA_PKT_COPY_TILED_DW_4_depth_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_DW_4_depth_shift  16
+#define SDMA_PKT_COPY_TILED_DW_4_DEPTH(x) (((x) & SDMA_PKT_COPY_TILED_DW_4_depth_mask) << SDMA_PKT_COPY_TILED_DW_4_depth_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_element_size_mask) << SDMA_PKT_COPY_TILED_DW_5_element_size_shift)
+
+/*define for swizzle_mode field*/
+#define SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_TILED_DW_5_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_mask) << SDMA_PKT_COPY_TILED_DW_5_swizzle_mode_shift)
+
+/*define for dimension field*/
+#define SDMA_PKT_COPY_TILED_DW_5_dimension_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_dimension_shift  9
+#define SDMA_PKT_COPY_TILED_DW_5_DIMENSION(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_dimension_mask) << SDMA_PKT_COPY_TILED_DW_5_dimension_shift)
+
+/*define for epitch field*/
+#define SDMA_PKT_COPY_TILED_DW_5_epitch_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_epitch_mask   0x0000FFFF
+#define SDMA_PKT_COPY_TILED_DW_5_epitch_shift  16
+#define SDMA_PKT_COPY_TILED_DW_5_EPITCH(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_epitch_mask) << SDMA_PKT_COPY_TILED_DW_5_epitch_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_TILED_DW_6_x_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_x_shift  0
+#define SDMA_PKT_COPY_TILED_DW_6_X(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_x_mask) << SDMA_PKT_COPY_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_TILED_DW_6_y_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_y_shift  16
+#define SDMA_PKT_COPY_TILED_DW_6_Y(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_y_mask) << SDMA_PKT_COPY_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_TILED_DW_7_z_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_DW_7_z_shift  0
+#define SDMA_PKT_COPY_TILED_DW_7_Z(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_z_mask) << SDMA_PKT_COPY_TILED_DW_7_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_DW_7_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_DW_7_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_offset 8
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_offset 9
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for LINEAR_SLICE_PITCH word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_LINEAR_SLICE_PITCH_linear_slice_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_TILED_COUNT_count_offset 12
+#define SDMA_PKT_COPY_TILED_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_COPY_TILED_COUNT_count_shift  0
+#define SDMA_PKT_COPY_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_TILED_COUNT_count_mask) << SDMA_PKT_COPY_TILED_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_L2T_BROADCAST packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_tmz_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_mip_max_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_mip_max_shift  20
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_MIP_MAX(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_mip_max_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_mip_max_shift)
+
+/*define for videocopy field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift  26
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_VIDEOCOPY(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift)
+
+/*define for TILED_ADDR_LO_0 word*/
+/*define for tiled_addr0_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_offset 1
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_TILED_ADDR0_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift)
+
+/*define for TILED_ADDR_HI_0 word*/
+/*define for tiled_addr0_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_offset 2
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_TILED_ADDR0_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift)
+
+/*define for TILED_ADDR_LO_1 word*/
+/*define for tiled_addr1_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_offset 3
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_TILED_ADDR1_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift)
+
+/*define for TILED_ADDR_HI_1 word*/
+/*define for tiled_addr1_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_offset 4
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_TILED_ADDR1_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift)
+
+/*define for DW_5 word*/
+/*define for width field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_offset 5
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_WIDTH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_5_width_shift)
+
+/*define for DW_6 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_offset 6
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_HEIGHT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_6_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_offset 6
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_mask   0x000007FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_DEPTH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_6_depth_shift)
+
+/*define for DW_7 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift)
+
+/*define for swizzle_mode field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_swizzle_mode_shift)
+
+/*define for dimension field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_shift  9
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_DIMENSION(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_dimension_shift)
+
+/*define for epitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_epitch_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_epitch_mask   0x0000FFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_epitch_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_EPITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_epitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_epitch_shift)
+
+/*define for DW_8 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_X(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_Y(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift)
+
+/*define for DW_9 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_offset 9
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask   0x000007FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_Z(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift)
+
+/*define for DW_10 word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_DST2_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift  24
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_TILE_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_offset 11
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_offset 12
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_offset 13
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for LINEAR_SLICE_PITCH word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_offset 14
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_SLICE_PITCH_linear_slice_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_offset 15
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask) << SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_T2T packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_op_shift  0
+#define SDMA_PKT_COPY_T2T_HEADER_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_op_mask) << SDMA_PKT_COPY_T2T_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_T2T_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_sub_op_mask) << SDMA_PKT_COPY_T2T_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_T2T_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_T2T_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_T2T_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_tmz_mask) << SDMA_PKT_COPY_T2T_HEADER_tmz_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_COPY_T2T_HEADER_mip_max_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_HEADER_mip_max_shift  20
+#define SDMA_PKT_COPY_T2T_HEADER_MIP_MAX(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_mip_max_mask) << SDMA_PKT_COPY_T2T_HEADER_mip_max_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_x_mask) << SDMA_PKT_COPY_T2T_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_y_mask) << SDMA_PKT_COPY_T2T_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_z_mask) << SDMA_PKT_COPY_T2T_DW_4_src_z_shift)
+
+/*define for src_width field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_width_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_4_src_width_shift  16
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_WIDTH(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_width_mask) << SDMA_PKT_COPY_T2T_DW_4_src_width_shift)
+
+/*define for DW_5 word*/
+/*define for src_height field*/
+#define SDMA_PKT_COPY_T2T_DW_5_src_height_offset 5
+#define SDMA_PKT_COPY_T2T_DW_5_src_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_5_src_height_shift  0
+#define SDMA_PKT_COPY_T2T_DW_5_SRC_HEIGHT(x) (((x) & SDMA_PKT_COPY_T2T_DW_5_src_height_mask) << SDMA_PKT_COPY_T2T_DW_5_src_height_shift)
+
+/*define for src_depth field*/
+#define SDMA_PKT_COPY_T2T_DW_5_src_depth_offset 5
+#define SDMA_PKT_COPY_T2T_DW_5_src_depth_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_5_src_depth_shift  16
+#define SDMA_PKT_COPY_T2T_DW_5_SRC_DEPTH(x) (((x) & SDMA_PKT_COPY_T2T_DW_5_src_depth_mask) << SDMA_PKT_COPY_T2T_DW_5_src_depth_shift)
+
+/*define for DW_6 word*/
+/*define for src_element_size field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask) << SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift)
+
+/*define for src_swizzle_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_mask) << SDMA_PKT_COPY_T2T_DW_6_src_swizzle_mode_shift)
+
+/*define for src_dimension field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_dimension_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_dimension_shift  9
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_DIMENSION(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_dimension_mask) << SDMA_PKT_COPY_T2T_DW_6_src_dimension_shift)
+
+/*define for src_epitch field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_epitch_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_epitch_mask   0x0000FFFF
+#define SDMA_PKT_COPY_T2T_DW_6_src_epitch_shift  16
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_EPITCH(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_epitch_mask) << SDMA_PKT_COPY_T2T_DW_6_src_epitch_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_offset 7
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_offset 8
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_9_DST_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_x_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_9_DST_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_y_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_y_shift)
+
+/*define for DW_10 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_10_DST_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_z_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_z_shift)
+
+/*define for dst_width field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_width_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_width_shift  16
+#define SDMA_PKT_COPY_T2T_DW_10_DST_WIDTH(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_width_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_width_shift)
+
+/*define for DW_11 word*/
+/*define for dst_height field*/
+#define SDMA_PKT_COPY_T2T_DW_11_dst_height_offset 11
+#define SDMA_PKT_COPY_T2T_DW_11_dst_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_11_dst_height_shift  0
+#define SDMA_PKT_COPY_T2T_DW_11_DST_HEIGHT(x) (((x) & SDMA_PKT_COPY_T2T_DW_11_dst_height_mask) << SDMA_PKT_COPY_T2T_DW_11_dst_height_shift)
+
+/*define for dst_depth field*/
+#define SDMA_PKT_COPY_T2T_DW_11_dst_depth_offset 11
+#define SDMA_PKT_COPY_T2T_DW_11_dst_depth_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_11_dst_depth_shift  16
+#define SDMA_PKT_COPY_T2T_DW_11_DST_DEPTH(x) (((x) & SDMA_PKT_COPY_T2T_DW_11_dst_depth_mask) << SDMA_PKT_COPY_T2T_DW_11_dst_depth_shift)
+
+/*define for DW_12 word*/
+/*define for dst_element_size field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_element_size_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_12_dst_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_DW_12_DST_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_element_size_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_element_size_shift)
+
+/*define for dst_swizzle_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_12_DST_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_swizzle_mode_shift)
+
+/*define for dst_dimension field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_dimension_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_dimension_shift  9
+#define SDMA_PKT_COPY_T2T_DW_12_DST_DIMENSION(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_dimension_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_dimension_shift)
+
+/*define for dst_epitch field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_epitch_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_epitch_mask   0x0000FFFF
+#define SDMA_PKT_COPY_T2T_DW_12_dst_epitch_shift  16
+#define SDMA_PKT_COPY_T2T_DW_12_DST_EPITCH(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_epitch_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_epitch_shift)
+
+/*define for DW_13 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_x_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_y_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_y_shift)
+
+/*define for DW_14 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_14_RECT_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_rect_z_mask) << SDMA_PKT_COPY_T2T_DW_14_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift  16
+#define SDMA_PKT_COPY_T2T_DW_14_DST_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_shift  24
+#define SDMA_PKT_COPY_T2T_DW_14_SRC_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_src_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_src_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_tmz_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_mip_max_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_mip_max_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_mip_max_shift  20
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_MIP_MAX(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_mip_max_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_mip_max_shift)
+
+/*define for mip_id field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_mip_id_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_mip_id_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_mip_id_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_MIP_ID(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_mip_id_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_mip_id_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for tiled_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift)
+
+/*define for tiled_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift)
+
+/*define for DW_4 word*/
+/*define for tiled_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_TILED_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift)
+
+/*define for width field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_WIDTH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_width_shift)
+
+/*define for DW_5 word*/
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_5_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_DEPTH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_5_depth_shift)
+
+/*define for DW_6 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift)
+
+/*define for swizzle_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_shift  3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_SWIZZLE_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_swizzle_mode_shift)
+
+/*define for dimension field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_shift  9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_DIMENSION(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_dimension_shift)
+
+/*define for epitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_epitch_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_epitch_mask   0x0000FFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_epitch_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_EPITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_epitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_epitch_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_offset 7
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_offset 8
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for linear_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift)
+
+/*define for linear_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift)
+
+/*define for DW_10 word*/
+/*define for linear_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift)
+
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift)
+
+/*define for DW_13 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_RECT_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_STRUCT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_shift  0
+#define SDMA_PKT_COPY_STRUCT_HEADER_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_STRUCT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_tmz_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_HEADER_tmz_shift  18
+#define SDMA_PKT_COPY_STRUCT_HEADER_TMZ(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_tmz_mask) << SDMA_PKT_COPY_STRUCT_HEADER_tmz_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_STRUCT_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_detile_mask) << SDMA_PKT_COPY_STRUCT_HEADER_detile_shift)
+
+/*define for SB_ADDR_LO word*/
+/*define for sb_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_offset 1
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_SB_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift)
+
+/*define for SB_ADDR_HI word*/
+/*define for sb_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_offset 2
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_SB_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift)
+
+/*define for START_INDEX word*/
+/*define for start_index field*/
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_offset 3
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift  0
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_START_INDEX(x) (((x) & SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask) << SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_offset 4
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_shift  0
+#define SDMA_PKT_COPY_STRUCT_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_STRUCT_COUNT_count_mask) << SDMA_PKT_COPY_STRUCT_COUNT_count_shift)
+
+/*define for DW_5 word*/
+/*define for stride field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_mask   0x000007FF
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_shift  0
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRIDE(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_stride_mask) << SDMA_PKT_COPY_STRUCT_DW_5_stride_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift  16
+#define SDMA_PKT_COPY_STRUCT_DW_5_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift)
+
+/*define for struct_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift  24
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRUCT_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_offset 6
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_offset 7
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_UNTILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_UNTILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_shift  16
+#define SDMA_PKT_WRITE_UNTILED_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_tmz_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_WRITE_UNTILED_HEADER_tmz_shift  18
+#define SDMA_PKT_WRITE_UNTILED_HEADER_TMZ(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_tmz_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_tmz_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_mask   0x000FFFFF
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_count_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_count_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift  24
+#define SDMA_PKT_WRITE_UNTILED_DW_3_SW(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_offset 4
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask) << SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_TILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift)
+
+/*define for encrypt field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_encrypt_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_encrypt_mask   0x00000001
+#define SDMA_PKT_WRITE_TILED_HEADER_encrypt_shift  16
+#define SDMA_PKT_WRITE_TILED_HEADER_ENCRYPT(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_encrypt_mask) << SDMA_PKT_WRITE_TILED_HEADER_encrypt_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_tmz_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_WRITE_TILED_HEADER_tmz_shift  18
+#define SDMA_PKT_WRITE_TILED_HEADER_TMZ(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_tmz_mask) << SDMA_PKT_WRITE_TILED_HEADER_tmz_shift)
+
+/*define for mip_max field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_mip_max_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_mip_max_mask   0x0000000F
+#define SDMA_PKT_WRITE_TILED_HEADER_mip_max_shift  20
+#define SDMA_PKT_WRITE_TILED_HEADER_MIP_MAX(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_mip_max_mask) << SDMA_PKT_WRITE_TILED_HEADER_mip_max_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for width field*/
+#define SDMA_PKT_WRITE_TILED_DW_3_width_offset 3
+#define SDMA_PKT_WRITE_TILED_DW_3_width_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_3_width_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_3_WIDTH(x) (((x) & SDMA_PKT_WRITE_TILED_DW_3_width_mask) << SDMA_PKT_WRITE_TILED_DW_3_width_shift)
+
+/*define for DW_4 word*/
+/*define for height field*/
+#define SDMA_PKT_WRITE_TILED_DW_4_height_offset 4
+#define SDMA_PKT_WRITE_TILED_DW_4_height_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_4_height_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_4_HEIGHT(x) (((x) & SDMA_PKT_WRITE_TILED_DW_4_height_mask) << SDMA_PKT_WRITE_TILED_DW_4_height_shift)
+
+/*define for depth field*/
+#define SDMA_PKT_WRITE_TILED_DW_4_depth_offset 4
+#define SDMA_PKT_WRITE_TILED_DW_4_depth_mask   0x000007FF
+#define SDMA_PKT_WRITE_TILED_DW_4_depth_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_4_DEPTH(x) (((x) & SDMA_PKT_WRITE_TILED_DW_4_depth_mask) << SDMA_PKT_WRITE_TILED_DW_4_depth_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_element_size_mask) << SDMA_PKT_WRITE_TILED_DW_5_element_size_shift)
+
+/*define for swizzle_mode field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_mask   0x0000001F
+#define SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_shift  3
+#define SDMA_PKT_WRITE_TILED_DW_5_SWIZZLE_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_mask) << SDMA_PKT_WRITE_TILED_DW_5_swizzle_mode_shift)
+
+/*define for dimension field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_dimension_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_dimension_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_dimension_shift  9
+#define SDMA_PKT_WRITE_TILED_DW_5_DIMENSION(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_dimension_mask) << SDMA_PKT_WRITE_TILED_DW_5_dimension_shift)
+
+/*define for epitch field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_epitch_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_epitch_mask   0x0000FFFF
+#define SDMA_PKT_WRITE_TILED_DW_5_epitch_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_5_EPITCH(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_epitch_mask) << SDMA_PKT_WRITE_TILED_DW_5_epitch_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_x_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_x_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_6_X(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_x_mask) << SDMA_PKT_WRITE_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_y_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_y_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_6_Y(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_y_mask) << SDMA_PKT_WRITE_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_z_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_z_mask   0x000007FF
+#define SDMA_PKT_WRITE_TILED_DW_7_z_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_7_Z(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_z_mask) << SDMA_PKT_WRITE_TILED_DW_7_z_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_shift  24
+#define SDMA_PKT_WRITE_TILED_DW_7_SW(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_sw_mask) << SDMA_PKT_WRITE_TILED_DW_7_sw_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_TILED_COUNT_count_offset 8
+#define SDMA_PKT_WRITE_TILED_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_WRITE_TILED_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_TILED_COUNT_count_mask) << SDMA_PKT_WRITE_TILED_COUNT_count_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_offset 9
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_TILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_TILED_DATA0_data0_mask) << SDMA_PKT_WRITE_TILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PTEPDE_COPY packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PTEPDE_COPY_HEADER_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_HEADER_op_shift  0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_HEADER_op_mask) << SDMA_PKT_PTEPDE_COPY_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_shift  8
+#define SDMA_PKT_PTEPDE_COPY_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_mask) << SDMA_PKT_PTEPDE_COPY_HEADER_sub_op_shift)
+
+/*define for ptepde_op field*/
+#define SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_mask   0x00000001
+#define SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_shift  31
+#define SDMA_PKT_PTEPDE_COPY_HEADER_PTEPDE_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_mask) << SDMA_PKT_PTEPDE_COPY_HEADER_ptepde_op_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_PTEPDE_COPY_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_PTEPDE_COPY_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_offset 3
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_PTEPDE_COPY_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_offset 4
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_PTEPDE_COPY_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_DW0 word*/
+/*define for mask_dw0 field*/
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_offset 5
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW0_MASK_DW0(x) (((x) & SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_mask) << SDMA_PKT_PTEPDE_COPY_MASK_DW0_mask_dw0_shift)
+
+/*define for MASK_DW1 word*/
+/*define for mask_dw1 field*/
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_offset 6
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_shift  0
+#define SDMA_PKT_PTEPDE_COPY_MASK_DW1_MASK_DW1(x) (((x) & SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_mask) << SDMA_PKT_PTEPDE_COPY_MASK_DW1_mask_dw1_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_PTEPDE_COPY_COUNT_count_offset 7
+#define SDMA_PKT_PTEPDE_COPY_COUNT_count_mask   0x0007FFFF
+#define SDMA_PKT_PTEPDE_COPY_COUNT_count_shift  0
+#define SDMA_PKT_PTEPDE_COPY_COUNT_COUNT(x) (((x) & SDMA_PKT_PTEPDE_COPY_COUNT_count_mask) << SDMA_PKT_PTEPDE_COPY_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PTEPDE_COPY_BACKWARDS packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_shift  8
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_sub_op_shift)
+
+/*define for pte_size field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_mask   0x00000003
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_shift  28
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_PTE_SIZE(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_pte_size_shift)
+
+/*define for direction field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_mask   0x00000001
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_shift  30
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_DIRECTION(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_direction_shift)
+
+/*define for ptepde_op field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_offset 0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_mask   0x00000001
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_shift  31
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_PTEPDE_OP(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_HEADER_ptepde_op_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_offset 3
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_offset 4
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_BIT_FOR_DW word*/
+/*define for mask_first_xfer field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_offset 5
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_MASK_FIRST_XFER(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_first_xfer_shift)
+
+/*define for mask_last_xfer field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_offset 5
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_shift  8
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_MASK_LAST_XFER(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_MASK_BIT_FOR_DW_mask_last_xfer_shift)
+
+/*define for COUNT_IN_32B_XFER word*/
+/*define for count field*/
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_offset 6
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_mask   0x0001FFFF
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_shift  0
+#define SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_COUNT(x) (((x) & SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_mask) << SDMA_PKT_PTEPDE_COPY_BACKWARDS_COUNT_IN_32B_XFER_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PTEPDE_RMW packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_op_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_RMW_HEADER_op_shift  0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_OP(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_op_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_shift  8
+#define SDMA_PKT_PTEPDE_RMW_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_sub_op_shift)
+
+/*define for gcc field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gcc_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gcc_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gcc_shift  19
+#define SDMA_PKT_PTEPDE_RMW_HEADER_GCC(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_gcc_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_gcc_shift)
+
+/*define for sys field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sys_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sys_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_sys_shift  20
+#define SDMA_PKT_PTEPDE_RMW_HEADER_SYS(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_sys_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_sys_shift)
+
+/*define for snp field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_snp_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_snp_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_snp_shift  22
+#define SDMA_PKT_PTEPDE_RMW_HEADER_SNP(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_snp_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_snp_shift)
+
+/*define for gpa field*/
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gpa_offset 0
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gpa_mask   0x00000001
+#define SDMA_PKT_PTEPDE_RMW_HEADER_gpa_shift  23
+#define SDMA_PKT_PTEPDE_RMW_HEADER_GPA(x) (((x) & SDMA_PKT_PTEPDE_RMW_HEADER_gpa_mask) << SDMA_PKT_PTEPDE_RMW_HEADER_gpa_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_PTEPDE_RMW_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_mask) << SDMA_PKT_PTEPDE_RMW_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_PTEPDE_RMW_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_mask) << SDMA_PKT_PTEPDE_RMW_ADDR_HI_addr_63_32_shift)
+
+/*define for MASK_LO word*/
+/*define for mask_31_0 field*/
+#define SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_offset 3
+#define SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_shift  0
+#define SDMA_PKT_PTEPDE_RMW_MASK_LO_MASK_31_0(x) (((x) & SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_mask) << SDMA_PKT_PTEPDE_RMW_MASK_LO_mask_31_0_shift)
+
+/*define for MASK_HI word*/
+/*define for mask_63_32 field*/
+#define SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_offset 4
+#define SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_shift  0
+#define SDMA_PKT_PTEPDE_RMW_MASK_HI_MASK_63_32(x) (((x) & SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_mask) << SDMA_PKT_PTEPDE_RMW_MASK_HI_mask_63_32_shift)
+
+/*define for VALUE_LO word*/
+/*define for value_31_0 field*/
+#define SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_offset 5
+#define SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_shift  0
+#define SDMA_PKT_PTEPDE_RMW_VALUE_LO_VALUE_31_0(x) (((x) & SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_mask) << SDMA_PKT_PTEPDE_RMW_VALUE_LO_value_31_0_shift)
+
+/*define for VALUE_HI word*/
+/*define for value_63_32 field*/
+#define SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_offset 6
+#define SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_shift  0
+#define SDMA_PKT_PTEPDE_RMW_VALUE_HI_VALUE_63_32(x) (((x) & SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_mask) << SDMA_PKT_PTEPDE_RMW_VALUE_HI_value_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_INCR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_INCR_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_INCR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_DW0 word*/
+/*define for mask_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_offset 3
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_MASK_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask) << SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift)
+
+/*define for MASK_DW1 word*/
+/*define for mask_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_offset 4
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_MASK_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask) << SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift)
+
+/*define for INIT_DW0 word*/
+/*define for init_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_offset 5
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_INIT_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask) << SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift)
+
+/*define for INIT_DW1 word*/
+/*define for init_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_offset 6
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_INIT_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask) << SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift)
+
+/*define for INCR_DW0 word*/
+/*define for incr_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_offset 7
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_INCR_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask) << SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift)
+
+/*define for INCR_DW1 word*/
+/*define for incr_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_offset 8
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_INCR_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask) << SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_INCR_COUNT_count_offset 9
+#define SDMA_PKT_WRITE_INCR_COUNT_count_mask   0x0007FFFF
+#define SDMA_PKT_WRITE_INCR_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_INCR_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_INCR_COUNT_count_mask) << SDMA_PKT_WRITE_INCR_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_INDIRECT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_INDIRECT_HEADER_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_op_shift  0
+#define SDMA_PKT_INDIRECT_HEADER_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_op_mask) << SDMA_PKT_INDIRECT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_shift  8
+#define SDMA_PKT_INDIRECT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_sub_op_mask) << SDMA_PKT_INDIRECT_HEADER_sub_op_shift)
+
+/*define for vmid field*/
+#define SDMA_PKT_INDIRECT_HEADER_vmid_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_vmid_mask   0x0000000F
+#define SDMA_PKT_INDIRECT_HEADER_vmid_shift  16
+#define SDMA_PKT_INDIRECT_HEADER_VMID(x) (((x) & SDMA_PKT_INDIRECT_HEADER_vmid_mask) << SDMA_PKT_INDIRECT_HEADER_vmid_shift)
+
+/*define for BASE_LO word*/
+/*define for ib_base_31_0 field*/
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_offset 1
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift  0
+#define SDMA_PKT_INDIRECT_BASE_LO_IB_BASE_31_0(x) (((x) & SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask) << SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift)
+
+/*define for BASE_HI word*/
+/*define for ib_base_63_32 field*/
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_offset 2
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift  0
+#define SDMA_PKT_INDIRECT_BASE_HI_IB_BASE_63_32(x) (((x) & SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask) << SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift)
+
+/*define for IB_SIZE word*/
+/*define for ib_size field*/
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_offset 3
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask   0x000FFFFF
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift  0
+#define SDMA_PKT_INDIRECT_IB_SIZE_IB_SIZE(x) (((x) & SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask) << SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift)
+
+/*define for CSA_ADDR_LO word*/
+/*define for csa_addr_31_0 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_offset 4
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_CSA_ADDR_31_0(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift)
+
+/*define for CSA_ADDR_HI word*/
+/*define for csa_addr_63_32 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_offset 5
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_CSA_ADDR_63_32(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SEMAPHORE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_op_shift  0
+#define SDMA_PKT_SEMAPHORE_HEADER_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SEMAPHORE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift)
+
+/*define for write_one field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_shift  29
+#define SDMA_PKT_SEMAPHORE_HEADER_WRITE_ONE(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_write_one_mask) << SDMA_PKT_SEMAPHORE_HEADER_write_one_shift)
+
+/*define for signal field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_shift  30
+#define SDMA_PKT_SEMAPHORE_HEADER_SIGNAL(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_signal_mask) << SDMA_PKT_SEMAPHORE_HEADER_signal_shift)
+
+/*define for mailbox field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift  31
+#define SDMA_PKT_SEMAPHORE_HEADER_MAILBOX(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask) << SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_FENCE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_FENCE_HEADER_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_op_shift  0
+#define SDMA_PKT_FENCE_HEADER_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_op_mask) << SDMA_PKT_FENCE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_FENCE_HEADER_sub_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_sub_op_shift  8
+#define SDMA_PKT_FENCE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_sub_op_mask) << SDMA_PKT_FENCE_HEADER_sub_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_FENCE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_FENCE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_FENCE_DATA_data_offset 3
+#define SDMA_PKT_FENCE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_DATA_data_shift  0
+#define SDMA_PKT_FENCE_DATA_DATA(x) (((x) & SDMA_PKT_FENCE_DATA_data_mask) << SDMA_PKT_FENCE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SRBM_WRITE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_shift  0
+#define SDMA_PKT_SRBM_WRITE_HEADER_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SRBM_WRITE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift)
+
+/*define for byte_en field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask   0x0000000F
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift  28
+#define SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask) << SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift)
+
+/*define for ADDR word*/
+/*define for addr field*/
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_offset 1
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_mask   0x0003FFFF
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_shift  0
+#define SDMA_PKT_SRBM_WRITE_ADDR_ADDR(x) (((x) & SDMA_PKT_SRBM_WRITE_ADDR_addr_mask) << SDMA_PKT_SRBM_WRITE_ADDR_addr_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_SRBM_WRITE_DATA_data_offset 2
+#define SDMA_PKT_SRBM_WRITE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_SRBM_WRITE_DATA_data_shift  0
+#define SDMA_PKT_SRBM_WRITE_DATA_DATA(x) (((x) & SDMA_PKT_SRBM_WRITE_DATA_data_mask) << SDMA_PKT_SRBM_WRITE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PRE_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_op_shift  0
+#define SDMA_PKT_PRE_EXE_HEADER_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_op_mask) << SDMA_PKT_PRE_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_PRE_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_sub_op_mask) << SDMA_PKT_PRE_EXE_HEADER_sub_op_shift)
+
+/*define for dev_sel field*/
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift  16
+#define SDMA_PKT_PRE_EXE_HEADER_DEV_SEL(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask) << SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_offset 1
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COND_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COND_EXE_HEADER_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_op_shift  0
+#define SDMA_PKT_COND_EXE_HEADER_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_op_mask) << SDMA_PKT_COND_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_COND_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_sub_op_mask) << SDMA_PKT_COND_EXE_HEADER_sub_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift)
+
+/*define for REFERENCE word*/
+/*define for reference field*/
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_offset 3
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_shift  0
+#define SDMA_PKT_COND_EXE_REFERENCE_REFERENCE(x) (((x) & SDMA_PKT_COND_EXE_REFERENCE_reference_mask) << SDMA_PKT_COND_EXE_REFERENCE_reference_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_offset 4
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_CONSTANT_FILL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_shift  0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift  8
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift  16
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SW(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift)
+
+/*define for fillsize field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift  30
+#define SDMA_PKT_CONSTANT_FILL_HEADER_FILLSIZE(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for src_data_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_offset 3
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DATA_SRC_DATA_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_offset 4
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_shift  0
+#define SDMA_PKT_CONSTANT_FILL_COUNT_COUNT(x) (((x) & SDMA_PKT_CONSTANT_FILL_COUNT_count_mask) << SDMA_PKT_CONSTANT_FILL_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_DATA_FILL_MULTI packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_op_offset 0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_op_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_OP(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_HEADER_op_mask) << SDMA_PKT_DATA_FILL_MULTI_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_offset 0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_shift  8
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_SUB_OP(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_mask) << SDMA_PKT_DATA_FILL_MULTI_HEADER_sub_op_shift)
+
+/*define for memlog_clr field*/
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_offset 0
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_mask   0x00000001
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_shift  31
+#define SDMA_PKT_DATA_FILL_MULTI_HEADER_MEMLOG_CLR(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_mask) << SDMA_PKT_DATA_FILL_MULTI_HEADER_memlog_clr_shift)
+
+/*define for BYTE_STRIDE word*/
+/*define for byte_stride field*/
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_offset 1
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_mask   0xFFFFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_BYTE_STRIDE(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_mask) << SDMA_PKT_DATA_FILL_MULTI_BYTE_STRIDE_byte_stride_shift)
+
+/*define for DMA_COUNT word*/
+/*define for dma_count field*/
+#define SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_offset 2
+#define SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_mask   0xFFFFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_DMA_COUNT(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_mask) << SDMA_PKT_DATA_FILL_MULTI_DMA_COUNT_dma_count_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_offset 3
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_offset 4
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_DATA_FILL_MULTI_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for BYTE_COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_offset 5
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_mask   0x03FFFFFF
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_shift  0
+#define SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_COUNT(x) (((x) & SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_mask) << SDMA_PKT_DATA_FILL_MULTI_BYTE_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_REGMEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_REGMEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_REGMEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift)
+
+/*define for hdp_flush field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift  26
+#define SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask) << SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift)
+
+/*define for func field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_mask   0x00000007
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_shift  28
+#define SDMA_PKT_POLL_REGMEM_HEADER_FUNC(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_func_mask) << SDMA_PKT_POLL_REGMEM_HEADER_func_shift)
+
+/*define for mem_poll field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift  31
+#define SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask) << SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift)
+
+/*define for VALUE word*/
+/*define for value field*/
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_offset 3
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_shift  0
+#define SDMA_PKT_POLL_REGMEM_VALUE_VALUE(x) (((x) & SDMA_PKT_POLL_REGMEM_VALUE_value_mask) << SDMA_PKT_POLL_REGMEM_VALUE_value_shift)
+
+/*define for MASK word*/
+/*define for mask field*/
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_offset 4
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_shift  0
+#define SDMA_PKT_POLL_REGMEM_MASK_MASK(x) (((x) & SDMA_PKT_POLL_REGMEM_MASK_mask_mask) << SDMA_PKT_POLL_REGMEM_MASK_mask_shift)
+
+/*define for DW5 word*/
+/*define for interval field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_mask   0x0000FFFF
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_shift  0
+#define SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_interval_mask) << SDMA_PKT_POLL_REGMEM_DW5_interval_shift)
+
+/*define for retry_count field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask   0x00000FFF
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift  16
+#define SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask) << SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_REG_WRITE_MEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_HEADER_sub_op_shift)
+
+/*define for SRC_ADDR word*/
+/*define for addr_31_2 field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_offset 1
+#define SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_mask   0x3FFFFFFF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_shift  2
+#define SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_ADDR_31_2(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_SRC_ADDR_addr_31_2_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_offset 2
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_LO_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_offset 3
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_REG_WRITE_MEM_DST_ADDR_HI_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_DBIT_WRITE_MEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_sub_op_shift)
+
+/*define for ea field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_offset 0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_mask   0x00000003
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_shift  16
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_EA(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_HEADER_ea_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_LO_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_DST_ADDR_HI_addr_63_32_shift)
+
+/*define for START_PAGE word*/
+/*define for addr_31_4 field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_offset 3
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_mask   0x0FFFFFFF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_shift  4
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_ADDR_31_4(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_START_PAGE_addr_31_4_shift)
+
+/*define for PAGE_NUM word*/
+/*define for page_num_31_0 field*/
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_offset 4
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_shift  0
+#define SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_PAGE_NUM_31_0(x) (((x) & SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_mask) << SDMA_PKT_POLL_DBIT_WRITE_MEM_PAGE_NUM_page_num_31_0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_MEM_VERIFY packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_offset 0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_OP(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_mask) << SDMA_PKT_POLL_MEM_VERIFY_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_mask) << SDMA_PKT_POLL_MEM_VERIFY_HEADER_sub_op_shift)
+
+/*define for mode field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_offset 0
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_mask   0x00000001
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_shift  31
+#define SDMA_PKT_POLL_MEM_VERIFY_HEADER_MODE(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_mask) << SDMA_PKT_POLL_MEM_VERIFY_HEADER_mode_shift)
+
+/*define for PATTERN word*/
+/*define for pattern field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_offset 1
+#define SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_PATTERN_PATTERN(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_mask) << SDMA_PKT_POLL_MEM_VERIFY_PATTERN_pattern_shift)
+
+/*define for CMP0_ADDR_START_LO word*/
+/*define for cmp0_start_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_offset 2
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_CMP0_START_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_LO_cmp0_start_31_0_shift)
+
+/*define for CMP0_ADDR_START_HI word*/
+/*define for cmp0_start_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_offset 3
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_CMP0_START_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_START_HI_cmp0_start_63_32_shift)
+
+/*define for CMP0_ADDR_END_LO word*/
+/*define for cmp1_end_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp1_end_31_0_offset 4
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp1_end_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp1_end_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_CMP1_END_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp1_end_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_LO_cmp1_end_31_0_shift)
+
+/*define for CMP0_ADDR_END_HI word*/
+/*define for cmp1_end_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp1_end_63_32_offset 5
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp1_end_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp1_end_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_CMP1_END_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp1_end_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP0_ADDR_END_HI_cmp1_end_63_32_shift)
+
+/*define for CMP1_ADDR_START_LO word*/
+/*define for cmp1_start_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_offset 6
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_CMP1_START_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_LO_cmp1_start_31_0_shift)
+
+/*define for CMP1_ADDR_START_HI word*/
+/*define for cmp1_start_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_offset 7
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_CMP1_START_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_START_HI_cmp1_start_63_32_shift)
+
+/*define for CMP1_ADDR_END_LO word*/
+/*define for cmp1_end_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_offset 8
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_CMP1_END_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_LO_cmp1_end_31_0_shift)
+
+/*define for CMP1_ADDR_END_HI word*/
+/*define for cmp1_end_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_offset 9
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_CMP1_END_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_CMP1_ADDR_END_HI_cmp1_end_63_32_shift)
+
+/*define for REC_ADDR_LO word*/
+/*define for rec_31_0 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_offset 10
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_REC_31_0(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_mask) << SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_LO_rec_31_0_shift)
+
+/*define for REC_ADDR_HI word*/
+/*define for rec_63_32 field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_offset 11
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_REC_63_32(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_mask) << SDMA_PKT_POLL_MEM_VERIFY_REC_ADDR_HI_rec_63_32_shift)
+
+/*define for RESERVED word*/
+/*define for reserved field*/
+#define SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_offset 12
+#define SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_shift  0
+#define SDMA_PKT_POLL_MEM_VERIFY_RESERVED_RESERVED(x) (((x) & SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_mask) << SDMA_PKT_POLL_MEM_VERIFY_RESERVED_reserved_shift)
+
+
+/*
+** Definitions for SDMA_PKT_ATOMIC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_ATOMIC_HEADER_op_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_ATOMIC_HEADER_op_shift  0
+#define SDMA_PKT_ATOMIC_HEADER_OP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_op_mask) << SDMA_PKT_ATOMIC_HEADER_op_shift)
+
+/*define for loop field*/
+#define SDMA_PKT_ATOMIC_HEADER_loop_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_loop_mask   0x00000001
+#define SDMA_PKT_ATOMIC_HEADER_loop_shift  16
+#define SDMA_PKT_ATOMIC_HEADER_LOOP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_loop_mask) << SDMA_PKT_ATOMIC_HEADER_loop_shift)
+
+/*define for tmz field*/
+#define SDMA_PKT_ATOMIC_HEADER_tmz_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_tmz_mask   0x00000001
+#define SDMA_PKT_ATOMIC_HEADER_tmz_shift  18
+#define SDMA_PKT_ATOMIC_HEADER_TMZ(x) (((x) & SDMA_PKT_ATOMIC_HEADER_tmz_mask) << SDMA_PKT_ATOMIC_HEADER_tmz_shift)
+
+/*define for atomic_op field*/
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_mask   0x0000007F
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_shift  25
+#define SDMA_PKT_ATOMIC_HEADER_ATOMIC_OP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_atomic_op_mask) << SDMA_PKT_ATOMIC_HEADER_atomic_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_ATOMIC_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_mask) << SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_ATOMIC_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_mask) << SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_shift)
+
+/*define for SRC_DATA_LO word*/
+/*define for src_data_31_0 field*/
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_offset 3
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_shift  0
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_SRC_DATA_31_0(x) (((x) & SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_mask) << SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_shift)
+
+/*define for SRC_DATA_HI word*/
+/*define for src_data_63_32 field*/
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_offset 4
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_shift  0
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_SRC_DATA_63_32(x) (((x) & SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_mask) << SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_shift)
+
+/*define for CMP_DATA_LO word*/
+/*define for cmp_data_31_0 field*/
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_offset 5
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_shift  0
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_CMP_DATA_31_0(x) (((x) & SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_mask) << SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_shift)
+
+/*define for CMP_DATA_HI word*/
+/*define for cmp_data_63_32 field*/
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_offset 6
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_shift  0
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_CMP_DATA_63_32(x) (((x) & SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_mask) << SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_shift)
+
+/*define for LOOP_INTERVAL word*/
+/*define for loop_interval field*/
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_offset 7
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_mask   0x00001FFF
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_shift  0
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_LOOP_INTERVAL(x) (((x) & SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_mask) << SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_SET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift)
+
+/*define for INIT_DATA_LO word*/
+/*define for init_data_31_0 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_offset 1
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_INIT_DATA_31_0(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift)
+
+/*define for INIT_DATA_HI word*/
+/*define for init_data_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_INIT_DATA_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET_GLOBAL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TRAP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TRAP_HEADER_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_op_shift  0
+#define SDMA_PKT_TRAP_HEADER_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_op_mask) << SDMA_PKT_TRAP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TRAP_HEADER_sub_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_sub_op_shift  8
+#define SDMA_PKT_TRAP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_sub_op_mask) << SDMA_PKT_TRAP_HEADER_sub_op_shift)
+
+/*define for INT_CONTEXT word*/
+/*define for int_context field*/
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_offset 1
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask   0x0FFFFFFF
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift  0
+#define SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(x) (((x) & SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask) << SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift)
+
+
+/*
+** Definitions for SDMA_PKT_DUMMY_TRAP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_DUMMY_TRAP_HEADER_op_offset 0
+#define SDMA_PKT_DUMMY_TRAP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_DUMMY_TRAP_HEADER_op_shift  0
+#define SDMA_PKT_DUMMY_TRAP_HEADER_OP(x) (((x) & SDMA_PKT_DUMMY_TRAP_HEADER_op_mask) << SDMA_PKT_DUMMY_TRAP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_offset 0
+#define SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_shift  8
+#define SDMA_PKT_DUMMY_TRAP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_mask) << SDMA_PKT_DUMMY_TRAP_HEADER_sub_op_shift)
+
+/*define for INT_CONTEXT word*/
+/*define for int_context field*/
+#define SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_offset 1
+#define SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_mask   0x0FFFFFFF
+#define SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_shift  0
+#define SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_INT_CONTEXT(x) (((x) & SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_mask) << SDMA_PKT_DUMMY_TRAP_INT_CONTEXT_int_context_shift)
+
+
+/*
+** Definitions for SDMA_PKT_NOP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_NOP_HEADER_op_offset 0
+#define SDMA_PKT_NOP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_op_shift  0
+#define SDMA_PKT_NOP_HEADER_OP(x) (((x) & SDMA_PKT_NOP_HEADER_op_mask) << SDMA_PKT_NOP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_NOP_HEADER_sub_op_offset 0
+#define SDMA_PKT_NOP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_sub_op_shift  8
+#define SDMA_PKT_NOP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_NOP_HEADER_sub_op_mask) << SDMA_PKT_NOP_HEADER_sub_op_shift)
+
+/*define for count field*/
+#define SDMA_PKT_NOP_HEADER_count_offset 0
+#define SDMA_PKT_NOP_HEADER_count_mask   0x00003FFF
+#define SDMA_PKT_NOP_HEADER_count_shift  16
+#define SDMA_PKT_NOP_HEADER_COUNT(x) (((x) & SDMA_PKT_NOP_HEADER_count_mask) << SDMA_PKT_NOP_HEADER_count_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_NOP_DATA0_data0_offset 1
+#define SDMA_PKT_NOP_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_NOP_DATA0_data0_shift  0
+#define SDMA_PKT_NOP_DATA0_DATA0(x) (((x) & SDMA_PKT_NOP_DATA0_data0_mask) << SDMA_PKT_NOP_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_AQL_PKT_HEADER packet
+*/
+
+/*define for HEADER word*/
+/*define for format field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_format_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_format_mask   0x000000FF
+#define SDMA_AQL_PKT_HEADER_HEADER_format_shift  0
+#define SDMA_AQL_PKT_HEADER_HEADER_FORMAT(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_format_mask) << SDMA_AQL_PKT_HEADER_HEADER_format_shift)
+
+/*define for barrier field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_barrier_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_barrier_mask   0x00000001
+#define SDMA_AQL_PKT_HEADER_HEADER_barrier_shift  8
+#define SDMA_AQL_PKT_HEADER_HEADER_BARRIER(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_barrier_mask) << SDMA_AQL_PKT_HEADER_HEADER_barrier_shift)
+
+/*define for acquire_fence_scope field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_shift  9
+#define SDMA_AQL_PKT_HEADER_HEADER_ACQUIRE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_mask) << SDMA_AQL_PKT_HEADER_HEADER_acquire_fence_scope_shift)
+
+/*define for release_fence_scope field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_shift  11
+#define SDMA_AQL_PKT_HEADER_HEADER_RELEASE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_mask) << SDMA_AQL_PKT_HEADER_HEADER_release_fence_scope_shift)
+
+/*define for reserved field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_reserved_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_reserved_mask   0x00000007
+#define SDMA_AQL_PKT_HEADER_HEADER_reserved_shift  13
+#define SDMA_AQL_PKT_HEADER_HEADER_RESERVED(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_reserved_mask) << SDMA_AQL_PKT_HEADER_HEADER_reserved_shift)
+
+/*define for op field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_op_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_op_mask   0x0000000F
+#define SDMA_AQL_PKT_HEADER_HEADER_op_shift  16
+#define SDMA_AQL_PKT_HEADER_HEADER_OP(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_op_mask) << SDMA_AQL_PKT_HEADER_HEADER_op_shift)
+
+/*define for subop field*/
+#define SDMA_AQL_PKT_HEADER_HEADER_subop_offset 0
+#define SDMA_AQL_PKT_HEADER_HEADER_subop_mask   0x00000007
+#define SDMA_AQL_PKT_HEADER_HEADER_subop_shift  20
+#define SDMA_AQL_PKT_HEADER_HEADER_SUBOP(x) (((x) & SDMA_AQL_PKT_HEADER_HEADER_subop_mask) << SDMA_AQL_PKT_HEADER_HEADER_subop_shift)
+
+
+/*
+** Definitions for SDMA_AQL_PKT_COPY_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for format field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_mask   0x000000FF
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_FORMAT(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_format_shift)
+
+/*define for barrier field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_mask   0x00000001
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_shift  8
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_BARRIER(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_barrier_shift)
+
+/*define for acquire_fence_scope field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_shift  9
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_ACQUIRE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_acquire_fence_scope_shift)
+
+/*define for release_fence_scope field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_shift  11
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_RELEASE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_release_fence_scope_shift)
+
+/*define for reserved field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_mask   0x00000007
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_shift  13
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_RESERVED(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_reserved_shift)
+
+/*define for op field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_mask   0x0000000F
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_shift  16
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_OP(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_op_shift)
+
+/*define for subop field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_offset 0
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_mask   0x00000007
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_shift  20
+#define SDMA_AQL_PKT_COPY_LINEAR_HEADER_SUBOP(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_mask) << SDMA_AQL_PKT_COPY_LINEAR_HEADER_subop_shift)
+
+/*define for RESERVED_DW1 word*/
+/*define for reserved_dw1 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_offset 1
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_RESERVED_DW1(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW1_reserved_dw1_shift)
+
+/*define for RETURN_ADDR_LO word*/
+/*define for return_addr_31_0 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_offset 2
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_RETURN_ADDR_31_0(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_mask) << SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_LO_return_addr_31_0_shift)
+
+/*define for RETURN_ADDR_HI word*/
+/*define for return_addr_63_32 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_offset 3
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_RETURN_ADDR_63_32(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_mask) << SDMA_AQL_PKT_COPY_LINEAR_RETURN_ADDR_HI_return_addr_63_32_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_offset 4
+#define SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_COUNT_COUNT(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_mask) << SDMA_AQL_PKT_COPY_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_offset 5
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift  16
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask) << SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_offset 5
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_mask) << SDMA_AQL_PKT_COPY_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 6
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 7
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_AQL_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 8
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 9
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_AQL_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for RESERVED_DW10 word*/
+/*define for reserved_dw10 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_offset 10
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_RESERVED_DW10(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW10_reserved_dw10_shift)
+
+/*define for RESERVED_DW11 word*/
+/*define for reserved_dw11 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_offset 11
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_RESERVED_DW11(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW11_reserved_dw11_shift)
+
+/*define for RESERVED_DW12 word*/
+/*define for reserved_dw12 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_offset 12
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_RESERVED_DW12(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW12_reserved_dw12_shift)
+
+/*define for RESERVED_DW13 word*/
+/*define for reserved_dw13 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_offset 13
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_RESERVED_DW13(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_mask) << SDMA_AQL_PKT_COPY_LINEAR_RESERVED_DW13_reserved_dw13_shift)
+
+/*define for COMPLETION_SIGNAL_LO word*/
+/*define for completion_signal_31_0 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_offset 14
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_COMPLETION_SIGNAL_31_0(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_mask) << SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_LO_completion_signal_31_0_shift)
+
+/*define for COMPLETION_SIGNAL_HI word*/
+/*define for completion_signal_63_32 field*/
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_offset 15
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_shift  0
+#define SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_COMPLETION_SIGNAL_63_32(x) (((x) & SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_mask) << SDMA_AQL_PKT_COPY_LINEAR_COMPLETION_SIGNAL_HI_completion_signal_63_32_shift)
+
+
+/*
+** Definitions for SDMA_AQL_PKT_BARRIER_OR packet
+*/
+
+/*define for HEADER word*/
+/*define for format field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_format_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_format_mask   0x000000FF
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_format_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_FORMAT(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_format_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_format_shift)
+
+/*define for barrier field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_mask   0x00000001
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_shift  8
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_BARRIER(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_barrier_shift)
+
+/*define for acquire_fence_scope field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_shift  9
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_ACQUIRE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_acquire_fence_scope_shift)
+
+/*define for release_fence_scope field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_mask   0x00000003
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_shift  11
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_RELEASE_FENCE_SCOPE(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_release_fence_scope_shift)
+
+/*define for reserved field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_mask   0x00000007
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_shift  13
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_RESERVED(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_reserved_shift)
+
+/*define for op field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_op_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_op_mask   0x0000000F
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_op_shift  16
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_OP(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_op_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_op_shift)
+
+/*define for subop field*/
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_offset 0
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_mask   0x00000007
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_shift  20
+#define SDMA_AQL_PKT_BARRIER_OR_HEADER_SUBOP(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_mask) << SDMA_AQL_PKT_BARRIER_OR_HEADER_subop_shift)
+
+/*define for RESERVED_DW1 word*/
+/*define for reserved_dw1 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_offset 1
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_RESERVED_DW1(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_mask) << SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW1_reserved_dw1_shift)
+
+/*define for DEPENDENT_ADDR_0_LO word*/
+/*define for dependent_addr_0_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_offset 2
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_DEPENDENT_ADDR_0_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_LO_dependent_addr_0_31_0_shift)
+
+/*define for DEPENDENT_ADDR_0_HI word*/
+/*define for dependent_addr_0_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_offset 3
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_DEPENDENT_ADDR_0_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_0_HI_dependent_addr_0_63_32_shift)
+
+/*define for DEPENDENT_ADDR_1_LO word*/
+/*define for dependent_addr_1_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_offset 4
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_DEPENDENT_ADDR_1_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_LO_dependent_addr_1_31_0_shift)
+
+/*define for DEPENDENT_ADDR_1_HI word*/
+/*define for dependent_addr_1_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_offset 5
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_DEPENDENT_ADDR_1_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_1_HI_dependent_addr_1_63_32_shift)
+
+/*define for DEPENDENT_ADDR_2_LO word*/
+/*define for dependent_addr_2_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_offset 6
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_DEPENDENT_ADDR_2_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_LO_dependent_addr_2_31_0_shift)
+
+/*define for DEPENDENT_ADDR_2_HI word*/
+/*define for dependent_addr_2_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_offset 7
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_DEPENDENT_ADDR_2_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_2_HI_dependent_addr_2_63_32_shift)
+
+/*define for DEPENDENT_ADDR_3_LO word*/
+/*define for dependent_addr_3_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_offset 8
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_DEPENDENT_ADDR_3_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_LO_dependent_addr_3_31_0_shift)
+
+/*define for DEPENDENT_ADDR_3_HI word*/
+/*define for dependent_addr_3_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_offset 9
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_DEPENDENT_ADDR_3_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_3_HI_dependent_addr_3_63_32_shift)
+
+/*define for DEPENDENT_ADDR_4_LO word*/
+/*define for dependent_addr_4_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_offset 10
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_DEPENDENT_ADDR_4_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_LO_dependent_addr_4_31_0_shift)
+
+/*define for DEPENDENT_ADDR_4_HI word*/
+/*define for dependent_addr_4_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_offset 11
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_DEPENDENT_ADDR_4_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_DEPENDENT_ADDR_4_HI_dependent_addr_4_63_32_shift)
+
+/*define for RESERVED_DW12 word*/
+/*define for reserved_dw12 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW12_reserved_dw12_offset 12
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW12_reserved_dw12_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW12_reserved_dw12_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW12_RESERVED_DW12(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW12_reserved_dw12_mask) << SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW12_reserved_dw12_shift)
+
+/*define for RESERVED_DW13 word*/
+/*define for reserved_dw13 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_offset 13
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_RESERVED_DW13(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_mask) << SDMA_AQL_PKT_BARRIER_OR_RESERVED_DW13_reserved_dw13_shift)
+
+/*define for COMPLETION_SIGNAL_LO word*/
+/*define for completion_signal_31_0 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_offset 14
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_COMPLETION_SIGNAL_31_0(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_mask) << SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_LO_completion_signal_31_0_shift)
+
+/*define for COMPLETION_SIGNAL_HI word*/
+/*define for completion_signal_63_32 field*/
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_offset 15
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_mask   0xFFFFFFFF
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_shift  0
+#define SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_COMPLETION_SIGNAL_63_32(x) (((x) & SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_mask) << SDMA_AQL_PKT_BARRIER_OR_COMPLETION_SIGNAL_HI_completion_signal_63_32_shift)
+
+
+#endif /* __SDMA_PKT_OPEN_H_ */
diff --git a/sys/dev/pci/drm/amd/amdgpu/vega20_reg_init.c b/sys/dev/pci/drm/amd/amdgpu/vega20_reg_init.c
new file mode 100644
index 00000000000..2d4473557b0
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vega20_reg_init.c
@@ -0,0 +1,56 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "soc15.h"
+
+#include "soc15_common.h"
+#include "soc15_hw_ip.h"
+#include "vega20_ip_offset.h"
+
+int vega20_reg_base_init(struct amdgpu_device *adev)
+{
+	/* HW has more IP blocks,  only initialized the blocke beend by our driver  */
+	uint32_t i;
+	for (i = 0 ; i < MAX_INSTANCE ; ++i) {
+		adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
+		adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i]));
+		adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i]));
+		adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i]));
+		adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i]));
+		adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i]));
+		adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i]));
+		adev->reg_offset[UVD_HWIP][i] = (uint32_t *)(&(UVD_BASE.instance[i]));
+		adev->reg_offset[VCE_HWIP][i] = (uint32_t *)(&(VCE_BASE.instance[i]));
+		adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i]));
+		adev->reg_offset[DCE_HWIP][i] = (uint32_t *)(&(DCE_BASE.instance[i]));
+		adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i]));
+		adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(SDMA0_BASE.instance[i]));
+		adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(SDMA1_BASE.instance[i]));
+		adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i]));
+		adev->reg_offset[NBIF_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i]));
+		adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i]));
+	}
+	return 0;
+}
+
+
diff --git a/sys/dev/pci/drm/amd/amdgpu/vi.c b/sys/dev/pci/drm/amd/amdgpu/vi.c
new file mode 100644
index 00000000000..a1a04a3de9c
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vi.c
@@ -0,0 +1,1714 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/slab.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+#include "amdgpu_ucode.h"
+#include "atom.h"
+#include "amd_pcie.h"
+
+#include "gmc/gmc_8_1_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+
+#include "bif/bif_5_0_d.h"
+#include "bif/bif_5_0_sh_mask.h"
+
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_sh_mask.h"
+
+#include "smu/smu_7_1_1_d.h"
+#include "smu/smu_7_1_1_sh_mask.h"
+
+#include "uvd/uvd_5_0_d.h"
+#include "uvd/uvd_5_0_sh_mask.h"
+
+#include "vce/vce_3_0_d.h"
+#include "vce/vce_3_0_sh_mask.h"
+
+#include "dce/dce_10_0_d.h"
+#include "dce/dce_10_0_sh_mask.h"
+
+#include "vid.h"
+#include "vi.h"
+#include "vi_dpm.h"
+#include "gmc_v8_0.h"
+#include "gmc_v7_0.h"
+#include "gfx_v8_0.h"
+#include "sdma_v2_4.h"
+#include "sdma_v3_0.h"
+#include "dce_v10_0.h"
+#include "dce_v11_0.h"
+#include "iceland_ih.h"
+#include "tonga_ih.h"
+#include "cz_ih.h"
+#include "uvd_v5_0.h"
+#include "uvd_v6_0.h"
+#include "vce_v3_0.h"
+#if defined(CONFIG_DRM_AMD_ACP)
+#include "amdgpu_acp.h"
+#endif
+#include "dce_virtual.h"
+#include "mxgpu_vi.h"
+#include "amdgpu_dm.h"
+
+/*
+ * Indirect registers accessor
+ */
+static u32 vi_pcie_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(mmPCIE_INDEX, reg);
+	(void)RREG32(mmPCIE_INDEX);
+	r = RREG32(mmPCIE_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+	return r;
+}
+
+static void vi_pcie_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(mmPCIE_INDEX, reg);
+	(void)RREG32(mmPCIE_INDEX);
+	WREG32(mmPCIE_DATA, v);
+	(void)RREG32(mmPCIE_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+static u32 vi_smc_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32_NO_KIQ(mmSMC_IND_INDEX_11, (reg));
+	r = RREG32_NO_KIQ(mmSMC_IND_DATA_11);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return r;
+}
+
+static void vi_smc_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmSMC_IND_INDEX_11, (reg));
+	WREG32(mmSMC_IND_DATA_11, (v));
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+}
+
+/* smu_8_0_d.h */
+#define mmMP0PUB_IND_INDEX                                                      0x180
+#define mmMP0PUB_IND_DATA                                                       0x181
+
+static u32 cz_smc_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmMP0PUB_IND_INDEX, (reg));
+	r = RREG32(mmMP0PUB_IND_DATA);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return r;
+}
+
+static void cz_smc_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmMP0PUB_IND_INDEX, (reg));
+	WREG32(mmMP0PUB_IND_DATA, (v));
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+}
+
+static u32 vi_uvd_ctx_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->uvd_ctx_idx_lock, flags);
+	WREG32(mmUVD_CTX_INDEX, ((reg) & 0x1ff));
+	r = RREG32(mmUVD_CTX_DATA);
+	spin_unlock_irqrestore(&adev->uvd_ctx_idx_lock, flags);
+	return r;
+}
+
+static void vi_uvd_ctx_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->uvd_ctx_idx_lock, flags);
+	WREG32(mmUVD_CTX_INDEX, ((reg) & 0x1ff));
+	WREG32(mmUVD_CTX_DATA, (v));
+	spin_unlock_irqrestore(&adev->uvd_ctx_idx_lock, flags);
+}
+
+static u32 vi_didt_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(mmDIDT_IND_INDEX, (reg));
+	r = RREG32(mmDIDT_IND_DATA);
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+	return r;
+}
+
+static void vi_didt_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(mmDIDT_IND_INDEX, (reg));
+	WREG32(mmDIDT_IND_DATA, (v));
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+}
+
+static u32 vi_gc_cac_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->gc_cac_idx_lock, flags);
+	WREG32(mmGC_CAC_IND_INDEX, (reg));
+	r = RREG32(mmGC_CAC_IND_DATA);
+	spin_unlock_irqrestore(&adev->gc_cac_idx_lock, flags);
+	return r;
+}
+
+static void vi_gc_cac_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->gc_cac_idx_lock, flags);
+	WREG32(mmGC_CAC_IND_INDEX, (reg));
+	WREG32(mmGC_CAC_IND_DATA, (v));
+	spin_unlock_irqrestore(&adev->gc_cac_idx_lock, flags);
+}
+
+
+static const u32 tonga_mgcg_cgcg_init[] =
+{
+	mmCGTT_DRM_CLK_CTRL0, 0xffffffff, 0x00600100,
+	mmPCIE_INDEX, 0xffffffff, 0x0140001c,
+	mmPCIE_DATA, 0x000f0000, 0x00000000,
+	mmSMC_IND_INDEX_4, 0xffffffff, 0xC060000C,
+	mmSMC_IND_DATA_4, 0xc0000fff, 0x00000100,
+	mmCGTT_DRM_CLK_CTRL0, 0xff000fff, 0x00000100,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+};
+
+static const u32 fiji_mgcg_cgcg_init[] =
+{
+	mmCGTT_DRM_CLK_CTRL0, 0xffffffff, 0x00600100,
+	mmPCIE_INDEX, 0xffffffff, 0x0140001c,
+	mmPCIE_DATA, 0x000f0000, 0x00000000,
+	mmSMC_IND_INDEX_4, 0xffffffff, 0xC060000C,
+	mmSMC_IND_DATA_4, 0xc0000fff, 0x00000100,
+	mmCGTT_DRM_CLK_CTRL0, 0xff000fff, 0x00000100,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+};
+
+static const u32 iceland_mgcg_cgcg_init[] =
+{
+	mmPCIE_INDEX, 0xffffffff, ixPCIE_CNTL2,
+	mmPCIE_DATA, 0x000f0000, 0x00000000,
+	mmSMC_IND_INDEX_4, 0xffffffff, ixCGTT_ROM_CLK_CTRL0,
+	mmSMC_IND_DATA_4, 0xc0000fff, 0x00000100,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+};
+
+static const u32 cz_mgcg_cgcg_init[] =
+{
+	mmCGTT_DRM_CLK_CTRL0, 0xffffffff, 0x00600100,
+	mmPCIE_INDEX, 0xffffffff, 0x0140001c,
+	mmPCIE_DATA, 0x000f0000, 0x00000000,
+	mmCGTT_DRM_CLK_CTRL0, 0xff000fff, 0x00000100,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+};
+
+static const u32 stoney_mgcg_cgcg_init[] =
+{
+	mmCGTT_DRM_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xffffffff, 0x00000104,
+	mmHDP_HOST_PATH_CNTL, 0xffffffff, 0x0f000027,
+};
+
+static void vi_init_golden_registers(struct amdgpu_device *adev)
+{
+	/* Some of the registers might be dependent on GRBM_GFX_INDEX */
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	if (amdgpu_sriov_vf(adev)) {
+		xgpu_vi_init_golden_registers(adev);
+		mutex_unlock(&adev->grbm_idx_mutex);
+		return;
+	}
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		amdgpu_device_program_register_sequence(adev,
+							iceland_mgcg_cgcg_init,
+							ARRAY_SIZE(iceland_mgcg_cgcg_init));
+		break;
+	case CHIP_FIJI:
+		amdgpu_device_program_register_sequence(adev,
+							fiji_mgcg_cgcg_init,
+							ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		break;
+	case CHIP_TONGA:
+		amdgpu_device_program_register_sequence(adev,
+							tonga_mgcg_cgcg_init,
+							ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		break;
+	case CHIP_CARRIZO:
+		amdgpu_device_program_register_sequence(adev,
+							cz_mgcg_cgcg_init,
+							ARRAY_SIZE(cz_mgcg_cgcg_init));
+		break;
+	case CHIP_STONEY:
+		amdgpu_device_program_register_sequence(adev,
+							stoney_mgcg_cgcg_init,
+							ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		break;
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+	default:
+		break;
+	}
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+/**
+ * vi_get_xclk - get the xclk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns the reference clock used by the gfx engine
+ * (VI).
+ */
+static u32 vi_get_xclk(struct amdgpu_device *adev)
+{
+	u32 reference_clock = adev->clock.spll.reference_freq;
+	u32 tmp;
+
+	if (adev->flags & AMD_IS_APU)
+		return reference_clock;
+
+	tmp = RREG32_SMC(ixCG_CLKPIN_CNTL_2);
+	if (REG_GET_FIELD(tmp, CG_CLKPIN_CNTL_2, MUX_TCLK_TO_XCLK))
+		return 1000;
+
+	tmp = RREG32_SMC(ixCG_CLKPIN_CNTL);
+	if (REG_GET_FIELD(tmp, CG_CLKPIN_CNTL, XTALIN_DIVIDE))
+		return reference_clock / 4;
+
+	return reference_clock;
+}
+
+/**
+ * vi_srbm_select - select specific register instances
+ *
+ * @adev: amdgpu_device pointer
+ * @me: selected ME (micro engine)
+ * @pipe: pipe
+ * @queue: queue
+ * @vmid: VMID
+ *
+ * Switches the currently active registers instances.  Some
+ * registers are instanced per VMID, others are instanced per
+ * me/pipe/queue combination.
+ */
+void vi_srbm_select(struct amdgpu_device *adev,
+		     u32 me, u32 pipe, u32 queue, u32 vmid)
+{
+	u32 srbm_gfx_cntl = 0;
+	srbm_gfx_cntl = REG_SET_FIELD(srbm_gfx_cntl, SRBM_GFX_CNTL, PIPEID, pipe);
+	srbm_gfx_cntl = REG_SET_FIELD(srbm_gfx_cntl, SRBM_GFX_CNTL, MEID, me);
+	srbm_gfx_cntl = REG_SET_FIELD(srbm_gfx_cntl, SRBM_GFX_CNTL, VMID, vmid);
+	srbm_gfx_cntl = REG_SET_FIELD(srbm_gfx_cntl, SRBM_GFX_CNTL, QUEUEID, queue);
+	WREG32(mmSRBM_GFX_CNTL, srbm_gfx_cntl);
+}
+
+static void vi_vga_set_state(struct amdgpu_device *adev, bool state)
+{
+	/* todo */
+}
+
+static bool vi_read_disabled_bios(struct amdgpu_device *adev)
+{
+	u32 bus_cntl;
+	u32 d1vga_control = 0;
+	u32 d2vga_control = 0;
+	u32 vga_render_control = 0;
+	u32 rom_cntl;
+	bool r;
+
+	bus_cntl = RREG32(mmBUS_CNTL);
+	if (adev->mode_info.num_crtc) {
+		d1vga_control = RREG32(mmD1VGA_CONTROL);
+		d2vga_control = RREG32(mmD2VGA_CONTROL);
+		vga_render_control = RREG32(mmVGA_RENDER_CONTROL);
+	}
+	rom_cntl = RREG32_SMC(ixROM_CNTL);
+
+	/* enable the rom */
+	WREG32(mmBUS_CNTL, (bus_cntl & ~BUS_CNTL__BIOS_ROM_DIS_MASK));
+	if (adev->mode_info.num_crtc) {
+		/* Disable VGA mode */
+		WREG32(mmD1VGA_CONTROL,
+		       (d1vga_control & ~(D1VGA_CONTROL__D1VGA_MODE_ENABLE_MASK |
+					  D1VGA_CONTROL__D1VGA_TIMING_SELECT_MASK)));
+		WREG32(mmD2VGA_CONTROL,
+		       (d2vga_control & ~(D2VGA_CONTROL__D2VGA_MODE_ENABLE_MASK |
+					  D2VGA_CONTROL__D2VGA_TIMING_SELECT_MASK)));
+		WREG32(mmVGA_RENDER_CONTROL,
+		       (vga_render_control & ~VGA_RENDER_CONTROL__VGA_VSTATUS_CNTL_MASK));
+	}
+	WREG32_SMC(ixROM_CNTL, rom_cntl | ROM_CNTL__SCK_OVERWRITE_MASK);
+
+	r = amdgpu_read_bios(adev);
+
+	/* restore regs */
+	WREG32(mmBUS_CNTL, bus_cntl);
+	if (adev->mode_info.num_crtc) {
+		WREG32(mmD1VGA_CONTROL, d1vga_control);
+		WREG32(mmD2VGA_CONTROL, d2vga_control);
+		WREG32(mmVGA_RENDER_CONTROL, vga_render_control);
+	}
+	WREG32_SMC(ixROM_CNTL, rom_cntl);
+	return r;
+}
+
+static bool vi_read_bios_from_rom(struct amdgpu_device *adev,
+				  u8 *bios, u32 length_bytes)
+{
+	u32 *dw_ptr;
+	unsigned long flags;
+	u32 i, length_dw;
+
+	if (bios == NULL)
+		return false;
+	if (length_bytes == 0)
+		return false;
+	/* APU vbios image is part of sbios image */
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	dw_ptr = (u32 *)bios;
+	length_dw = roundup2(length_bytes, 4) / 4;
+	/* take the smc lock since we are using the smc index */
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	/* set rom index to 0 */
+	WREG32(mmSMC_IND_INDEX_11, ixROM_INDEX);
+	WREG32(mmSMC_IND_DATA_11, 0);
+	/* set index to data for continous read */
+	WREG32(mmSMC_IND_INDEX_11, ixROM_DATA);
+	for (i = 0; i < length_dw; i++)
+		dw_ptr[i] = RREG32(mmSMC_IND_DATA_11);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return true;
+}
+
+static void vi_detect_hw_virtualization(struct amdgpu_device *adev)
+{
+	uint32_t reg = 0;
+
+	if (adev->asic_type == CHIP_TONGA ||
+	    adev->asic_type == CHIP_FIJI) {
+	       reg = RREG32(mmBIF_IOV_FUNC_IDENTIFIER);
+	       /* bit0: 0 means pf and 1 means vf */
+	       if (REG_GET_FIELD(reg, BIF_IOV_FUNC_IDENTIFIER, FUNC_IDENTIFIER))
+		       adev->virt.caps |= AMDGPU_SRIOV_CAPS_IS_VF;
+	       /* bit31: 0 means disable IOV and 1 means enable */
+	       if (REG_GET_FIELD(reg, BIF_IOV_FUNC_IDENTIFIER, IOV_ENABLE))
+		       adev->virt.caps |= AMDGPU_SRIOV_CAPS_ENABLE_IOV;
+	}
+
+	if (reg == 0) {
+		if (is_virtual_machine()) /* passthrough mode exclus sr-iov mode */
+			adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
+	}
+}
+
+static const struct amdgpu_allowed_register_entry vi_allowed_read_registers[] = {
+	{mmGRBM_STATUS},
+	{mmGRBM_STATUS2},
+	{mmGRBM_STATUS_SE0},
+	{mmGRBM_STATUS_SE1},
+	{mmGRBM_STATUS_SE2},
+	{mmGRBM_STATUS_SE3},
+	{mmSRBM_STATUS},
+	{mmSRBM_STATUS2},
+	{mmSRBM_STATUS3},
+	{mmSDMA0_STATUS_REG + SDMA0_REGISTER_OFFSET},
+	{mmSDMA0_STATUS_REG + SDMA1_REGISTER_OFFSET},
+	{mmCP_STAT},
+	{mmCP_STALLED_STAT1},
+	{mmCP_STALLED_STAT2},
+	{mmCP_STALLED_STAT3},
+	{mmCP_CPF_BUSY_STAT},
+	{mmCP_CPF_STALLED_STAT1},
+	{mmCP_CPF_STATUS},
+	{mmCP_CPC_BUSY_STAT},
+	{mmCP_CPC_STALLED_STAT1},
+	{mmCP_CPC_STATUS},
+	{mmGB_ADDR_CONFIG},
+	{mmMC_ARB_RAMCFG},
+	{mmGB_TILE_MODE0},
+	{mmGB_TILE_MODE1},
+	{mmGB_TILE_MODE2},
+	{mmGB_TILE_MODE3},
+	{mmGB_TILE_MODE4},
+	{mmGB_TILE_MODE5},
+	{mmGB_TILE_MODE6},
+	{mmGB_TILE_MODE7},
+	{mmGB_TILE_MODE8},
+	{mmGB_TILE_MODE9},
+	{mmGB_TILE_MODE10},
+	{mmGB_TILE_MODE11},
+	{mmGB_TILE_MODE12},
+	{mmGB_TILE_MODE13},
+	{mmGB_TILE_MODE14},
+	{mmGB_TILE_MODE15},
+	{mmGB_TILE_MODE16},
+	{mmGB_TILE_MODE17},
+	{mmGB_TILE_MODE18},
+	{mmGB_TILE_MODE19},
+	{mmGB_TILE_MODE20},
+	{mmGB_TILE_MODE21},
+	{mmGB_TILE_MODE22},
+	{mmGB_TILE_MODE23},
+	{mmGB_TILE_MODE24},
+	{mmGB_TILE_MODE25},
+	{mmGB_TILE_MODE26},
+	{mmGB_TILE_MODE27},
+	{mmGB_TILE_MODE28},
+	{mmGB_TILE_MODE29},
+	{mmGB_TILE_MODE30},
+	{mmGB_TILE_MODE31},
+	{mmGB_MACROTILE_MODE0},
+	{mmGB_MACROTILE_MODE1},
+	{mmGB_MACROTILE_MODE2},
+	{mmGB_MACROTILE_MODE3},
+	{mmGB_MACROTILE_MODE4},
+	{mmGB_MACROTILE_MODE5},
+	{mmGB_MACROTILE_MODE6},
+	{mmGB_MACROTILE_MODE7},
+	{mmGB_MACROTILE_MODE8},
+	{mmGB_MACROTILE_MODE9},
+	{mmGB_MACROTILE_MODE10},
+	{mmGB_MACROTILE_MODE11},
+	{mmGB_MACROTILE_MODE12},
+	{mmGB_MACROTILE_MODE13},
+	{mmGB_MACROTILE_MODE14},
+	{mmGB_MACROTILE_MODE15},
+	{mmCC_RB_BACKEND_DISABLE, true},
+	{mmGC_USER_RB_BACKEND_DISABLE, true},
+	{mmGB_BACKEND_MAP, false},
+	{mmPA_SC_RASTER_CONFIG, true},
+	{mmPA_SC_RASTER_CONFIG_1, true},
+};
+
+static uint32_t vi_get_register_value(struct amdgpu_device *adev,
+				      bool indexed, u32 se_num,
+				      u32 sh_num, u32 reg_offset)
+{
+	if (indexed) {
+		uint32_t val;
+		unsigned se_idx = (se_num == 0xffffffff) ? 0 : se_num;
+		unsigned sh_idx = (sh_num == 0xffffffff) ? 0 : sh_num;
+
+		switch (reg_offset) {
+		case mmCC_RB_BACKEND_DISABLE:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].rb_backend_disable;
+		case mmGC_USER_RB_BACKEND_DISABLE:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].user_rb_backend_disable;
+		case mmPA_SC_RASTER_CONFIG:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].raster_config;
+		case mmPA_SC_RASTER_CONFIG_1:
+			return adev->gfx.config.rb_config[se_idx][sh_idx].raster_config_1;
+		}
+
+		mutex_lock(&adev->grbm_idx_mutex);
+		if (se_num != 0xffffffff || sh_num != 0xffffffff)
+			amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+
+		val = RREG32(reg_offset);
+
+		if (se_num != 0xffffffff || sh_num != 0xffffffff)
+			amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		mutex_unlock(&adev->grbm_idx_mutex);
+		return val;
+	} else {
+		unsigned idx;
+
+		switch (reg_offset) {
+		case mmGB_ADDR_CONFIG:
+			return adev->gfx.config.gb_addr_config;
+		case mmMC_ARB_RAMCFG:
+			return adev->gfx.config.mc_arb_ramcfg;
+		case mmGB_TILE_MODE0:
+		case mmGB_TILE_MODE1:
+		case mmGB_TILE_MODE2:
+		case mmGB_TILE_MODE3:
+		case mmGB_TILE_MODE4:
+		case mmGB_TILE_MODE5:
+		case mmGB_TILE_MODE6:
+		case mmGB_TILE_MODE7:
+		case mmGB_TILE_MODE8:
+		case mmGB_TILE_MODE9:
+		case mmGB_TILE_MODE10:
+		case mmGB_TILE_MODE11:
+		case mmGB_TILE_MODE12:
+		case mmGB_TILE_MODE13:
+		case mmGB_TILE_MODE14:
+		case mmGB_TILE_MODE15:
+		case mmGB_TILE_MODE16:
+		case mmGB_TILE_MODE17:
+		case mmGB_TILE_MODE18:
+		case mmGB_TILE_MODE19:
+		case mmGB_TILE_MODE20:
+		case mmGB_TILE_MODE21:
+		case mmGB_TILE_MODE22:
+		case mmGB_TILE_MODE23:
+		case mmGB_TILE_MODE24:
+		case mmGB_TILE_MODE25:
+		case mmGB_TILE_MODE26:
+		case mmGB_TILE_MODE27:
+		case mmGB_TILE_MODE28:
+		case mmGB_TILE_MODE29:
+		case mmGB_TILE_MODE30:
+		case mmGB_TILE_MODE31:
+			idx = (reg_offset - mmGB_TILE_MODE0);
+			return adev->gfx.config.tile_mode_array[idx];
+		case mmGB_MACROTILE_MODE0:
+		case mmGB_MACROTILE_MODE1:
+		case mmGB_MACROTILE_MODE2:
+		case mmGB_MACROTILE_MODE3:
+		case mmGB_MACROTILE_MODE4:
+		case mmGB_MACROTILE_MODE5:
+		case mmGB_MACROTILE_MODE6:
+		case mmGB_MACROTILE_MODE7:
+		case mmGB_MACROTILE_MODE8:
+		case mmGB_MACROTILE_MODE9:
+		case mmGB_MACROTILE_MODE10:
+		case mmGB_MACROTILE_MODE11:
+		case mmGB_MACROTILE_MODE12:
+		case mmGB_MACROTILE_MODE13:
+		case mmGB_MACROTILE_MODE14:
+		case mmGB_MACROTILE_MODE15:
+			idx = (reg_offset - mmGB_MACROTILE_MODE0);
+			return adev->gfx.config.macrotile_mode_array[idx];
+		default:
+			return RREG32(reg_offset);
+		}
+	}
+}
+
+static int vi_read_register(struct amdgpu_device *adev, u32 se_num,
+			    u32 sh_num, u32 reg_offset, u32 *value)
+{
+	uint32_t i;
+
+	*value = 0;
+	for (i = 0; i < ARRAY_SIZE(vi_allowed_read_registers); i++) {
+		bool indexed = vi_allowed_read_registers[i].grbm_indexed;
+
+		if (reg_offset != vi_allowed_read_registers[i].reg_offset)
+			continue;
+
+		*value = vi_get_register_value(adev, indexed, se_num, sh_num,
+					       reg_offset);
+		return 0;
+	}
+	return -EINVAL;
+}
+
+static int vi_gpu_pci_config_reset(struct amdgpu_device *adev)
+{
+	u32 i;
+
+	dev_info(adev->dev, "GPU pci config reset\n");
+
+	/* disable BM */
+	pci_clear_master(adev->pdev);
+	/* reset */
+	amdgpu_device_pci_config_reset(adev);
+
+	udelay(100);
+
+	/* wait for asic to come out of reset */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmCONFIG_MEMSIZE) != 0xffffffff) {
+			/* enable BM */
+			pci_set_master(adev->pdev);
+			adev->has_hw_reset = true;
+			return 0;
+		}
+		udelay(1);
+	}
+	return -EINVAL;
+}
+
+/**
+ * vi_asic_reset - soft reset GPU
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up which blocks are hung and attempt
+ * to reset them.
+ * Returns 0 for success.
+ */
+static int vi_asic_reset(struct amdgpu_device *adev)
+{
+	int r;
+
+	amdgpu_atombios_scratch_regs_engine_hung(adev, true);
+
+	r = vi_gpu_pci_config_reset(adev);
+
+	amdgpu_atombios_scratch_regs_engine_hung(adev, false);
+
+	return r;
+}
+
+static u32 vi_get_config_memsize(struct amdgpu_device *adev)
+{
+	return RREG32(mmCONFIG_MEMSIZE);
+}
+
+static int vi_set_uvd_clock(struct amdgpu_device *adev, u32 clock,
+			u32 cntl_reg, u32 status_reg)
+{
+	int r, i;
+	struct atom_clock_dividers dividers;
+	uint32_t tmp;
+
+	r = amdgpu_atombios_get_clock_dividers(adev,
+					       COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+					       clock, false, &dividers);
+	if (r)
+		return r;
+
+	tmp = RREG32_SMC(cntl_reg);
+
+	if (adev->flags & AMD_IS_APU)
+		tmp &= ~CG_DCLK_CNTL__DCLK_DIVIDER_MASK;
+	else
+		tmp &= ~(CG_DCLK_CNTL__DCLK_DIR_CNTL_EN_MASK |
+				CG_DCLK_CNTL__DCLK_DIVIDER_MASK);
+	tmp |= dividers.post_divider;
+	WREG32_SMC(cntl_reg, tmp);
+
+	for (i = 0; i < 100; i++) {
+		tmp = RREG32_SMC(status_reg);
+		if (adev->flags & AMD_IS_APU) {
+			if (tmp & 0x10000)
+				break;
+		} else {
+			if (tmp & CG_DCLK_STATUS__DCLK_STATUS_MASK)
+				break;
+		}
+		mdelay(10);
+	}
+	if (i == 100)
+		return -ETIMEDOUT;
+	return 0;
+}
+
+#define ixGNB_CLK1_DFS_CNTL 0xD82200F0
+#define ixGNB_CLK1_STATUS   0xD822010C
+#define ixGNB_CLK2_DFS_CNTL 0xD8220110
+#define ixGNB_CLK2_STATUS   0xD822012C
+#define ixGNB_CLK3_DFS_CNTL 0xD8220130
+#define ixGNB_CLK3_STATUS   0xD822014C
+
+static int vi_set_uvd_clocks(struct amdgpu_device *adev, u32 vclk, u32 dclk)
+{
+	int r;
+
+	if (adev->flags & AMD_IS_APU) {
+		r = vi_set_uvd_clock(adev, vclk, ixGNB_CLK2_DFS_CNTL, ixGNB_CLK2_STATUS);
+		if (r)
+			return r;
+
+		r = vi_set_uvd_clock(adev, dclk, ixGNB_CLK1_DFS_CNTL, ixGNB_CLK1_STATUS);
+		if (r)
+			return r;
+	} else {
+		r = vi_set_uvd_clock(adev, vclk, ixCG_VCLK_CNTL, ixCG_VCLK_STATUS);
+		if (r)
+			return r;
+
+		r = vi_set_uvd_clock(adev, dclk, ixCG_DCLK_CNTL, ixCG_DCLK_STATUS);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int vi_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk)
+{
+	int r, i;
+	struct atom_clock_dividers dividers;
+	u32 tmp;
+	u32 reg_ctrl;
+	u32 reg_status;
+	u32 status_mask;
+	u32 reg_mask;
+
+	if (adev->flags & AMD_IS_APU) {
+		reg_ctrl = ixGNB_CLK3_DFS_CNTL;
+		reg_status = ixGNB_CLK3_STATUS;
+		status_mask = 0x00010000;
+		reg_mask = CG_ECLK_CNTL__ECLK_DIVIDER_MASK;
+	} else {
+		reg_ctrl = ixCG_ECLK_CNTL;
+		reg_status = ixCG_ECLK_STATUS;
+		status_mask = CG_ECLK_STATUS__ECLK_STATUS_MASK;
+		reg_mask = CG_ECLK_CNTL__ECLK_DIR_CNTL_EN_MASK | CG_ECLK_CNTL__ECLK_DIVIDER_MASK;
+	}
+
+	r = amdgpu_atombios_get_clock_dividers(adev,
+					       COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+					       ecclk, false, &dividers);
+	if (r)
+		return r;
+
+	for (i = 0; i < 100; i++) {
+		if (RREG32_SMC(reg_status) & status_mask)
+			break;
+		mdelay(10);
+	}
+
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	tmp = RREG32_SMC(reg_ctrl);
+	tmp &= ~reg_mask;
+	tmp |= dividers.post_divider;
+	WREG32_SMC(reg_ctrl, tmp);
+
+	for (i = 0; i < 100; i++) {
+		if (RREG32_SMC(reg_status) & status_mask)
+			break;
+		mdelay(10);
+	}
+
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	return 0;
+}
+
+static void vi_pcie_gen3_enable(struct amdgpu_device *adev)
+{
+	if (pci_is_root_bus(adev->pdev->bus))
+		return;
+
+	if (amdgpu_pcie_gen2 == 0)
+		return;
+
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	if (!(adev->pm.pcie_gen_mask & (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
+					CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)))
+		return;
+
+	/* todo */
+}
+
+static void vi_program_aspm(struct amdgpu_device *adev)
+{
+
+	if (amdgpu_aspm == 0)
+		return;
+
+	/* todo */
+}
+
+static void vi_enable_doorbell_aperture(struct amdgpu_device *adev,
+					bool enable)
+{
+	u32 tmp;
+
+	/* not necessary on CZ */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	tmp = RREG32(mmBIF_DOORBELL_APER_EN);
+	if (enable)
+		tmp = REG_SET_FIELD(tmp, BIF_DOORBELL_APER_EN, BIF_DOORBELL_APER_EN, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, BIF_DOORBELL_APER_EN, BIF_DOORBELL_APER_EN, 0);
+
+	WREG32(mmBIF_DOORBELL_APER_EN, tmp);
+}
+
+#define ATI_REV_ID_FUSE_MACRO__ADDRESS      0xC0014044
+#define ATI_REV_ID_FUSE_MACRO__SHIFT        9
+#define ATI_REV_ID_FUSE_MACRO__MASK         0x00001E00
+
+static uint32_t vi_get_rev_id(struct amdgpu_device *adev)
+{
+	if (adev->flags & AMD_IS_APU)
+		return (RREG32_SMC(ATI_REV_ID_FUSE_MACRO__ADDRESS) & ATI_REV_ID_FUSE_MACRO__MASK)
+			>> ATI_REV_ID_FUSE_MACRO__SHIFT;
+	else
+		return (RREG32(mmPCIE_EFUSE4) & PCIE_EFUSE4__STRAP_BIF_ATI_REV_ID_MASK)
+			>> PCIE_EFUSE4__STRAP_BIF_ATI_REV_ID__SHIFT;
+}
+
+static void vi_flush_hdp(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 1);
+		RREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL);
+	} else {
+		amdgpu_ring_emit_wreg(ring, mmHDP_MEM_COHERENCY_FLUSH_CNTL, 1);
+	}
+}
+
+static void vi_invalidate_hdp(struct amdgpu_device *adev,
+			      struct amdgpu_ring *ring)
+{
+	if (!ring || !ring->funcs->emit_wreg) {
+		WREG32(mmHDP_DEBUG0, 1);
+		RREG32(mmHDP_DEBUG0);
+	} else {
+		amdgpu_ring_emit_wreg(ring, mmHDP_DEBUG0, 1);
+	}
+}
+
+static bool vi_need_full_reset(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		/* CZ has hang issues with full reset at the moment */
+		return false;
+	case CHIP_FIJI:
+	case CHIP_TONGA:
+		/* XXX: soft reset should work on fiji and tonga */
+		return true;
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_TOPAZ:
+	default:
+		/* change this when we support soft reset */
+		return true;
+	}
+}
+
+static const struct amdgpu_asic_funcs vi_asic_funcs =
+{
+	.read_disabled_bios = &vi_read_disabled_bios,
+	.read_bios_from_rom = &vi_read_bios_from_rom,
+	.read_register = &vi_read_register,
+	.reset = &vi_asic_reset,
+	.set_vga_state = &vi_vga_set_state,
+	.get_xclk = &vi_get_xclk,
+	.set_uvd_clocks = &vi_set_uvd_clocks,
+	.set_vce_clocks = &vi_set_vce_clocks,
+	.get_config_memsize = &vi_get_config_memsize,
+	.flush_hdp = &vi_flush_hdp,
+	.invalidate_hdp = &vi_invalidate_hdp,
+	.need_full_reset = &vi_need_full_reset,
+};
+
+#define CZ_REV_BRISTOL(rev)	 \
+	((rev >= 0xC8 && rev <= 0xCE) || (rev >= 0xE1 && rev <= 0xE6))
+
+static int vi_common_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->flags & AMD_IS_APU) {
+		adev->smc_rreg = &cz_smc_rreg;
+		adev->smc_wreg = &cz_smc_wreg;
+	} else {
+		adev->smc_rreg = &vi_smc_rreg;
+		adev->smc_wreg = &vi_smc_wreg;
+	}
+	adev->pcie_rreg = &vi_pcie_rreg;
+	adev->pcie_wreg = &vi_pcie_wreg;
+	adev->uvd_ctx_rreg = &vi_uvd_ctx_rreg;
+	adev->uvd_ctx_wreg = &vi_uvd_ctx_wreg;
+	adev->didt_rreg = &vi_didt_rreg;
+	adev->didt_wreg = &vi_didt_wreg;
+	adev->gc_cac_rreg = &vi_gc_cac_rreg;
+	adev->gc_cac_wreg = &vi_gc_cac_wreg;
+
+	adev->asic_funcs = &vi_asic_funcs;
+
+	adev->rev_id = vi_get_rev_id(adev);
+	adev->external_rev_id = 0xFF;
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		adev->cg_flags = 0;
+		adev->pg_flags = 0;
+		adev->external_rev_id = 0x1;
+		break;
+	case CHIP_FIJI:
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_CGTS_LS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_UVD_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x3c;
+		break;
+	case CHIP_TONGA:
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_DRM_LS |
+			AMD_CG_SUPPORT_UVD_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x14;
+		break;
+	case CHIP_POLARIS11:
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_DRM_LS |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_VCE_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x5A;
+		break;
+	case CHIP_POLARIS10:
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_DRM_LS |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_VCE_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x50;
+		break;
+	case CHIP_POLARIS12:
+		adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_DRM_LS |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_VCE_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x64;
+		break;
+	case CHIP_VEGAM:
+		adev->cg_flags = 0;
+			/*AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_GFX_3D_CGCG |
+			AMD_CG_SUPPORT_GFX_3D_CGLS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_BIF_MGCG |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_ROM_MGCG |
+			AMD_CG_SUPPORT_MC_MGCG |
+			AMD_CG_SUPPORT_MC_LS |
+			AMD_CG_SUPPORT_DRM_LS |
+			AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_VCE_MGCG;*/
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x6E;
+		break;
+	case CHIP_CARRIZO:
+		adev->cg_flags = AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_CGTS_LS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_VCE_MGCG;
+		/* rev0 hardware requires workarounds to support PG */
+		adev->pg_flags = 0;
+		if (adev->rev_id != 0x00 || CZ_REV_BRISTOL(adev->pdev->revision)) {
+			adev->pg_flags |= AMD_PG_SUPPORT_GFX_SMG |
+				AMD_PG_SUPPORT_GFX_PIPELINE |
+				AMD_PG_SUPPORT_CP |
+				AMD_PG_SUPPORT_UVD |
+				AMD_PG_SUPPORT_VCE;
+		}
+		adev->external_rev_id = adev->rev_id + 0x1;
+		break;
+	case CHIP_STONEY:
+		adev->cg_flags = AMD_CG_SUPPORT_UVD_MGCG |
+			AMD_CG_SUPPORT_GFX_MGCG |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_CGTS_LS |
+			AMD_CG_SUPPORT_GFX_CGLS |
+			AMD_CG_SUPPORT_BIF_LS |
+			AMD_CG_SUPPORT_HDP_MGCG |
+			AMD_CG_SUPPORT_HDP_LS |
+			AMD_CG_SUPPORT_SDMA_MGCG |
+			AMD_CG_SUPPORT_SDMA_LS |
+			AMD_CG_SUPPORT_VCE_MGCG;
+		adev->pg_flags = AMD_PG_SUPPORT_GFX_PG |
+			AMD_PG_SUPPORT_GFX_SMG |
+			AMD_PG_SUPPORT_GFX_PIPELINE |
+			AMD_PG_SUPPORT_CP |
+			AMD_PG_SUPPORT_UVD |
+			AMD_PG_SUPPORT_VCE;
+		adev->external_rev_id = adev->rev_id + 0x61;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	if (amdgpu_sriov_vf(adev)) {
+		amdgpu_virt_init_setting(adev);
+		xgpu_vi_mailbox_set_irq_funcs(adev);
+	}
+
+	return 0;
+}
+
+static int vi_common_late_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		xgpu_vi_mailbox_get_irq(adev);
+
+	return 0;
+}
+
+static int vi_common_sw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		xgpu_vi_mailbox_add_irq_id(adev);
+
+	return 0;
+}
+
+static int vi_common_sw_fini(void *handle)
+{
+	return 0;
+}
+
+static int vi_common_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* move the golden regs per IP block */
+	vi_init_golden_registers(adev);
+	/* enable pcie gen2/3 link */
+	vi_pcie_gen3_enable(adev);
+	/* enable aspm */
+	vi_program_aspm(adev);
+	/* enable the doorbell aperture */
+	vi_enable_doorbell_aperture(adev, true);
+
+	return 0;
+}
+
+static int vi_common_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* enable the doorbell aperture */
+	vi_enable_doorbell_aperture(adev, false);
+
+	if (amdgpu_sriov_vf(adev))
+		xgpu_vi_mailbox_put_irq(adev);
+
+	return 0;
+}
+
+static int vi_common_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return vi_common_hw_fini(adev);
+}
+
+static int vi_common_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return vi_common_hw_init(adev);
+}
+
+static bool vi_common_is_idle(void *handle)
+{
+	return true;
+}
+
+static int vi_common_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static int vi_common_soft_reset(void *handle)
+{
+	return 0;
+}
+
+static void vi_update_bif_medium_grain_light_sleep(struct amdgpu_device *adev,
+						   bool enable)
+{
+	uint32_t temp, data;
+
+	temp = data = RREG32_PCIE(ixPCIE_CNTL2);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS))
+		data |= PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+				PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+				PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK;
+	else
+		data &= ~(PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+				PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+				PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
+
+	if (temp != data)
+		WREG32_PCIE(ixPCIE_CNTL2, data);
+}
+
+static void vi_update_hdp_medium_grain_clock_gating(struct amdgpu_device *adev,
+						    bool enable)
+{
+	uint32_t temp, data;
+
+	temp = data = RREG32(mmHDP_HOST_PATH_CNTL);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_MGCG))
+		data &= ~HDP_HOST_PATH_CNTL__CLOCK_GATING_DIS_MASK;
+	else
+		data |= HDP_HOST_PATH_CNTL__CLOCK_GATING_DIS_MASK;
+
+	if (temp != data)
+		WREG32(mmHDP_HOST_PATH_CNTL, data);
+}
+
+static void vi_update_hdp_light_sleep(struct amdgpu_device *adev,
+				      bool enable)
+{
+	uint32_t temp, data;
+
+	temp = data = RREG32(mmHDP_MEM_POWER_LS);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
+		data |= HDP_MEM_POWER_LS__LS_ENABLE_MASK;
+	else
+		data &= ~HDP_MEM_POWER_LS__LS_ENABLE_MASK;
+
+	if (temp != data)
+		WREG32(mmHDP_MEM_POWER_LS, data);
+}
+
+static void vi_update_drm_light_sleep(struct amdgpu_device *adev,
+				      bool enable)
+{
+	uint32_t temp, data;
+
+	temp = data = RREG32(0x157a);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_DRM_LS))
+		data |= 1;
+	else
+		data &= ~1;
+
+	if (temp != data)
+		WREG32(0x157a, data);
+}
+
+
+static void vi_update_rom_medium_grain_clock_gating(struct amdgpu_device *adev,
+						    bool enable)
+{
+	uint32_t temp, data;
+
+	temp = data = RREG32_SMC(ixCGTT_ROM_CLK_CTRL0);
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ROM_MGCG))
+		data &= ~(CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
+				CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK);
+	else
+		data |= CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK |
+				CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK;
+
+	if (temp != data)
+		WREG32_SMC(ixCGTT_ROM_CLK_CTRL0, data);
+}
+
+static int vi_common_set_clockgating_state_by_smu(void *handle,
+					   enum amd_clockgating_state state)
+{
+	uint32_t msg_id, pp_state = 0;
+	uint32_t pp_support_state = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->cg_flags & (AMD_CG_SUPPORT_MC_LS | AMD_CG_SUPPORT_MC_MGCG)) {
+		if (adev->cg_flags & AMD_CG_SUPPORT_MC_LS) {
+			pp_support_state = PP_STATE_SUPPORT_LS;
+			pp_state = PP_STATE_LS;
+		}
+		if (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG) {
+			pp_support_state |= PP_STATE_SUPPORT_CG;
+			pp_state |= PP_STATE_CG;
+		}
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+		msg_id = PP_CG_MSG_ID(PP_GROUP_SYS,
+			       PP_BLOCK_SYS_MC,
+			       pp_support_state,
+			       pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+
+	if (adev->cg_flags & (AMD_CG_SUPPORT_SDMA_LS | AMD_CG_SUPPORT_SDMA_MGCG)) {
+		if (adev->cg_flags & AMD_CG_SUPPORT_SDMA_LS) {
+			pp_support_state = PP_STATE_SUPPORT_LS;
+			pp_state = PP_STATE_LS;
+		}
+		if (adev->cg_flags & AMD_CG_SUPPORT_SDMA_MGCG) {
+			pp_support_state |= PP_STATE_SUPPORT_CG;
+			pp_state |= PP_STATE_CG;
+		}
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+		msg_id = PP_CG_MSG_ID(PP_GROUP_SYS,
+			       PP_BLOCK_SYS_SDMA,
+			       pp_support_state,
+			       pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+
+	if (adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS | AMD_CG_SUPPORT_HDP_MGCG)) {
+		if (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS) {
+			pp_support_state = PP_STATE_SUPPORT_LS;
+			pp_state = PP_STATE_LS;
+		}
+		if (adev->cg_flags & AMD_CG_SUPPORT_HDP_MGCG) {
+			pp_support_state |= PP_STATE_SUPPORT_CG;
+			pp_state |= PP_STATE_CG;
+		}
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+		msg_id = PP_CG_MSG_ID(PP_GROUP_SYS,
+			       PP_BLOCK_SYS_HDP,
+			       pp_support_state,
+			       pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS) {
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+		else
+			pp_state = PP_STATE_LS;
+
+		msg_id = PP_CG_MSG_ID(PP_GROUP_SYS,
+			       PP_BLOCK_SYS_BIF,
+			       PP_STATE_SUPPORT_LS,
+			        pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+	if (adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG) {
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+		else
+			pp_state = PP_STATE_CG;
+
+		msg_id = PP_CG_MSG_ID(PP_GROUP_SYS,
+			       PP_BLOCK_SYS_BIF,
+			       PP_STATE_SUPPORT_CG,
+			       pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_DRM_LS) {
+
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+		else
+			pp_state = PP_STATE_LS;
+
+		msg_id = PP_CG_MSG_ID(PP_GROUP_SYS,
+			       PP_BLOCK_SYS_DRM,
+			       PP_STATE_SUPPORT_LS,
+			       pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+
+	if (adev->cg_flags & AMD_CG_SUPPORT_ROM_MGCG) {
+
+		if (state == AMD_CG_STATE_UNGATE)
+			pp_state = 0;
+		else
+			pp_state = PP_STATE_CG;
+
+		msg_id = PP_CG_MSG_ID(PP_GROUP_SYS,
+			       PP_BLOCK_SYS_ROM,
+			       PP_STATE_SUPPORT_CG,
+			       pp_state);
+		if (adev->powerplay.pp_funcs->set_clockgating_by_smu)
+			amdgpu_dpm_set_clockgating_by_smu(adev, msg_id);
+	}
+	return 0;
+}
+
+static int vi_common_set_clockgating_state(void *handle,
+					   enum amd_clockgating_state state)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+		vi_update_bif_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		vi_update_hdp_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		vi_update_hdp_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		vi_update_rom_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		vi_update_bif_medium_grain_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		vi_update_hdp_medium_grain_clock_gating(adev,
+				state == AMD_CG_STATE_GATE);
+		vi_update_hdp_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		vi_update_drm_light_sleep(adev,
+				state == AMD_CG_STATE_GATE);
+		break;
+	case CHIP_TONGA:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		vi_common_set_clockgating_state_by_smu(adev, state);
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int vi_common_set_powergating_state(void *handle,
+					    enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static void vi_common_get_clockgating_state(void *handle, u32 *flags)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int data;
+
+	if (amdgpu_sriov_vf(adev))
+		*flags = 0;
+
+	/* AMD_CG_SUPPORT_BIF_LS */
+	data = RREG32_PCIE(ixPCIE_CNTL2);
+	if (data & PCIE_CNTL2__SLV_MEM_LS_EN_MASK)
+		*flags |= AMD_CG_SUPPORT_BIF_LS;
+
+	/* AMD_CG_SUPPORT_HDP_LS */
+	data = RREG32(mmHDP_MEM_POWER_LS);
+	if (data & HDP_MEM_POWER_LS__LS_ENABLE_MASK)
+		*flags |= AMD_CG_SUPPORT_HDP_LS;
+
+	/* AMD_CG_SUPPORT_HDP_MGCG */
+	data = RREG32(mmHDP_HOST_PATH_CNTL);
+	if (!(data & HDP_HOST_PATH_CNTL__CLOCK_GATING_DIS_MASK))
+		*flags |= AMD_CG_SUPPORT_HDP_MGCG;
+
+	/* AMD_CG_SUPPORT_ROM_MGCG */
+	data = RREG32_SMC(ixCGTT_ROM_CLK_CTRL0);
+	if (!(data & CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK))
+		*flags |= AMD_CG_SUPPORT_ROM_MGCG;
+}
+
+static const struct amd_ip_funcs vi_common_ip_funcs = {
+	.name = "vi_common",
+	.early_init = vi_common_early_init,
+	.late_init = vi_common_late_init,
+	.sw_init = vi_common_sw_init,
+	.sw_fini = vi_common_sw_fini,
+	.hw_init = vi_common_hw_init,
+	.hw_fini = vi_common_hw_fini,
+	.suspend = vi_common_suspend,
+	.resume = vi_common_resume,
+	.is_idle = vi_common_is_idle,
+	.wait_for_idle = vi_common_wait_for_idle,
+	.soft_reset = vi_common_soft_reset,
+	.set_clockgating_state = vi_common_set_clockgating_state,
+	.set_powergating_state = vi_common_set_powergating_state,
+	.get_clockgating_state = vi_common_get_clockgating_state,
+};
+
+static const struct amdgpu_ip_block_version vi_common_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_COMMON,
+	.major = 1,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &vi_common_ip_funcs,
+};
+
+int vi_set_ip_blocks(struct amdgpu_device *adev)
+{
+	/* in early init stage, vbios code won't work */
+	vi_detect_hw_virtualization(adev);
+
+	if (amdgpu_sriov_vf(adev))
+		adev->virt.ops = &xgpu_vi_virt_ops;
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		/* topaz has no DCE, UVD, VCE */
+		amdgpu_device_ip_block_add(adev, &vi_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v7_4_ip_block);
+		amdgpu_device_ip_block_add(adev, &iceland_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v2_4_ip_block);
+		break;
+	case CHIP_FIJI:
+		amdgpu_device_ip_block_add(adev, &vi_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v8_5_ip_block);
+		amdgpu_device_ip_block_add(adev, &tonga_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		if (adev->enable_virtual_display || amdgpu_sriov_vf(adev))
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
+		else
+			amdgpu_device_ip_block_add(adev, &dce_v10_1_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v3_0_ip_block);
+		if (!amdgpu_sriov_vf(adev)) {
+			amdgpu_device_ip_block_add(adev, &uvd_v6_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &vce_v3_0_ip_block);
+		}
+		break;
+	case CHIP_TONGA:
+		amdgpu_device_ip_block_add(adev, &vi_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &tonga_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		if (adev->enable_virtual_display || amdgpu_sriov_vf(adev))
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
+		else
+			amdgpu_device_ip_block_add(adev, &dce_v10_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v3_0_ip_block);
+		if (!amdgpu_sriov_vf(adev)) {
+			amdgpu_device_ip_block_add(adev, &uvd_v5_0_ip_block);
+			amdgpu_device_ip_block_add(adev, &vce_v3_0_ip_block);
+		}
+		break;
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
+	case CHIP_POLARIS12:
+	case CHIP_VEGAM:
+		amdgpu_device_ip_block_add(adev, &vi_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v8_1_ip_block);
+		amdgpu_device_ip_block_add(adev, &tonga_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
+		else
+			amdgpu_device_ip_block_add(adev, &dce_v11_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v3_1_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v6_3_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v3_4_ip_block);
+		break;
+	case CHIP_CARRIZO:
+		amdgpu_device_ip_block_add(adev, &vi_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &cz_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
+		else
+			amdgpu_device_ip_block_add(adev, &dce_v11_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v3_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v6_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v3_1_ip_block);
+#if defined(CONFIG_DRM_AMD_ACP)
+		amdgpu_device_ip_block_add(adev, &acp_ip_block);
+#endif
+		break;
+	case CHIP_STONEY:
+		amdgpu_device_ip_block_add(adev, &vi_common_ip_block);
+		amdgpu_device_ip_block_add(adev, &gmc_v8_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &cz_ih_ip_block);
+		amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
+		if (adev->enable_virtual_display)
+			amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
+#if defined(CONFIG_DRM_AMD_DC)
+		else if (amdgpu_device_has_dc_support(adev))
+			amdgpu_device_ip_block_add(adev, &dm_ip_block);
+#endif
+		else
+			amdgpu_device_ip_block_add(adev, &dce_v11_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &gfx_v8_1_ip_block);
+		amdgpu_device_ip_block_add(adev, &sdma_v3_0_ip_block);
+		amdgpu_device_ip_block_add(adev, &uvd_v6_2_ip_block);
+		amdgpu_device_ip_block_add(adev, &vce_v3_4_ip_block);
+#if defined(CONFIG_DRM_AMD_ACP)
+		amdgpu_device_ip_block_add(adev, &acp_ip_block);
+#endif
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	return 0;
+}
diff --git a/sys/dev/pci/drm/amd/amdgpu/vi.h b/sys/dev/pci/drm/amd/amdgpu/vi.h
new file mode 100644
index 00000000000..0429fe33226
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vi.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VI_H__
+#define __VI_H__
+
+#define VI_FLUSH_GPU_TLB_NUM_WREG	3
+
+void vi_srbm_select(struct amdgpu_device *adev,
+		    u32 me, u32 pipe, u32 queue, u32 vmid);
+int vi_set_ip_blocks(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/vi_dpm.h b/sys/dev/pci/drm/amd/amdgpu/vi_dpm.h
new file mode 100644
index 00000000000..c43e03fddfb
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vi_dpm.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VI_DPM_H__
+#define __VI_DPM_H__
+
+extern const struct amd_ip_funcs cz_dpm_ip_funcs;
+int cz_smu_init(struct amdgpu_device *adev);
+int cz_smu_start(struct amdgpu_device *adev);
+int cz_smu_fini(struct amdgpu_device *adev);
+
+#endif
diff --git a/sys/dev/pci/drm/amd/amdgpu/vid.h b/sys/dev/pci/drm/amd/amdgpu/vid.h
new file mode 100644
index 00000000000..19ddd2312e0
--- /dev/null
+++ b/sys/dev/pci/drm/amd/amdgpu/vid.h
@@ -0,0 +1,516 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef VI_H
+#define VI_H
+
+#define SDMA0_REGISTER_OFFSET                             0x0 /* not a register */
+#define SDMA1_REGISTER_OFFSET                             0x200 /* not a register */
+#define SDMA_MAX_INSTANCE 2
+
+#define KFD_VI_SDMA_QUEUE_OFFSET                      0x80 /* not a register */
+
+/* crtc instance offsets */
+#define CRTC0_REGISTER_OFFSET                 (0x1b9c - 0x1b9c)
+#define CRTC1_REGISTER_OFFSET                 (0x1d9c - 0x1b9c)
+#define CRTC2_REGISTER_OFFSET                 (0x1f9c - 0x1b9c)
+#define CRTC3_REGISTER_OFFSET                 (0x419c - 0x1b9c)
+#define CRTC4_REGISTER_OFFSET                 (0x439c - 0x1b9c)
+#define CRTC5_REGISTER_OFFSET                 (0x459c - 0x1b9c)
+#define CRTC6_REGISTER_OFFSET                 (0x479c - 0x1b9c)
+
+/* dig instance offsets */
+#define DIG0_REGISTER_OFFSET                 (0x4a00 - 0x4a00)
+#define DIG1_REGISTER_OFFSET                 (0x4b00 - 0x4a00)
+#define DIG2_REGISTER_OFFSET                 (0x4c00 - 0x4a00)
+#define DIG3_REGISTER_OFFSET                 (0x4d00 - 0x4a00)
+#define DIG4_REGISTER_OFFSET                 (0x4e00 - 0x4a00)
+#define DIG5_REGISTER_OFFSET                 (0x4f00 - 0x4a00)
+#define DIG6_REGISTER_OFFSET                 (0x5400 - 0x4a00)
+#define DIG7_REGISTER_OFFSET                 (0x5600 - 0x4a00)
+#define DIG8_REGISTER_OFFSET                 (0x5700 - 0x4a00)
+
+/* audio endpt instance offsets */
+#define AUD0_REGISTER_OFFSET                 (0x17a8 - 0x17a8)
+#define AUD1_REGISTER_OFFSET                 (0x17ac - 0x17a8)
+#define AUD2_REGISTER_OFFSET                 (0x17b0 - 0x17a8)
+#define AUD3_REGISTER_OFFSET                 (0x17b4 - 0x17a8)
+#define AUD4_REGISTER_OFFSET                 (0x17b8 - 0x17a8)
+#define AUD5_REGISTER_OFFSET                 (0x17bc - 0x17a8)
+#define AUD6_REGISTER_OFFSET                 (0x17c0 - 0x17a8)
+#define AUD7_REGISTER_OFFSET                 (0x17c4 - 0x17a8)
+
+/* hpd instance offsets */
+#define HPD0_REGISTER_OFFSET                 (0x1898 - 0x1898)
+#define HPD1_REGISTER_OFFSET                 (0x18a0 - 0x1898)
+#define HPD2_REGISTER_OFFSET                 (0x18a8 - 0x1898)
+#define HPD3_REGISTER_OFFSET                 (0x18b0 - 0x1898)
+#define HPD4_REGISTER_OFFSET                 (0x18b8 - 0x1898)
+#define HPD5_REGISTER_OFFSET                 (0x18c0 - 0x1898)
+
+#define AMDGPU_NUM_OF_VMIDS			8
+
+#define		PIPEID(x)					((x) << 0)
+#define		MEID(x)						((x) << 2)
+#define		VMID(x)						((x) << 4)
+#define		QUEUEID(x)					((x) << 8)
+
+#define MC_SEQ_MISC0__MT__MASK	0xf0000000
+#define MC_SEQ_MISC0__MT__GDDR1  0x10000000
+#define MC_SEQ_MISC0__MT__DDR2   0x20000000
+#define MC_SEQ_MISC0__MT__GDDR3  0x30000000
+#define MC_SEQ_MISC0__MT__GDDR4  0x40000000
+#define MC_SEQ_MISC0__MT__GDDR5  0x50000000
+#define MC_SEQ_MISC0__MT__HBM    0x60000000
+#define MC_SEQ_MISC0__MT__DDR3   0xB0000000
+
+/*
+ * PM4
+ */
+#define	PACKET_TYPE0	0
+#define	PACKET_TYPE1	1
+#define	PACKET_TYPE2	2
+#define	PACKET_TYPE3	3
+
+#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3)
+#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF)
+#define CP_PACKET0_GET_REG(h) ((h) & 0xFFFF)
+#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF)
+#define PACKET0(reg, n)	((PACKET_TYPE0 << 30) |				\
+			 ((reg) & 0xFFFF) |			\
+			 ((n) & 0x3FFF) << 16)
+#define CP_PACKET2			0x80000000
+#define		PACKET2_PAD_SHIFT		0
+#define		PACKET2_PAD_MASK		(0x3fffffff << 0)
+
+#define PACKET2(v)	(CP_PACKET2 | REG_SET(PACKET2_PAD, (v)))
+
+#define PACKET3(op, n)	((PACKET_TYPE3 << 30) |				\
+			 (((op) & 0xFF) << 8) |				\
+			 ((n) & 0x3FFF) << 16)
+
+#define PACKET3_COMPUTE(op, n) (PACKET3(op, n) | 1 << 1)
+
+/* Packet 3 types */
+#define	PACKET3_NOP					0x10
+#define	PACKET3_SET_BASE				0x11
+#define		PACKET3_BASE_INDEX(x)                  ((x) << 0)
+#define			CE_PARTITION_BASE		3
+#define	PACKET3_CLEAR_STATE				0x12
+#define	PACKET3_INDEX_BUFFER_SIZE			0x13
+#define	PACKET3_DISPATCH_DIRECT				0x15
+#define	PACKET3_DISPATCH_INDIRECT			0x16
+#define	PACKET3_ATOMIC_GDS				0x1D
+#define	PACKET3_ATOMIC_MEM				0x1E
+#define	PACKET3_OCCLUSION_QUERY				0x1F
+#define	PACKET3_SET_PREDICATION				0x20
+#define	PACKET3_REG_RMW					0x21
+#define	PACKET3_COND_EXEC				0x22
+#define	PACKET3_PRED_EXEC				0x23
+#define	PACKET3_DRAW_INDIRECT				0x24
+#define	PACKET3_DRAW_INDEX_INDIRECT			0x25
+#define	PACKET3_INDEX_BASE				0x26
+#define	PACKET3_DRAW_INDEX_2				0x27
+#define	PACKET3_CONTEXT_CONTROL				0x28
+#define	PACKET3_INDEX_TYPE				0x2A
+#define	PACKET3_DRAW_INDIRECT_MULTI			0x2C
+#define	PACKET3_DRAW_INDEX_AUTO				0x2D
+#define	PACKET3_NUM_INSTANCES				0x2F
+#define	PACKET3_DRAW_INDEX_MULTI_AUTO			0x30
+#define	PACKET3_INDIRECT_BUFFER_CONST			0x33
+#define	PACKET3_STRMOUT_BUFFER_UPDATE			0x34
+#define	PACKET3_DRAW_INDEX_OFFSET_2			0x35
+#define	PACKET3_DRAW_PREAMBLE				0x36
+#define	PACKET3_WRITE_DATA				0x37
+#define		WRITE_DATA_DST_SEL(x)                   ((x) << 8)
+		/* 0 - register
+		 * 1 - memory (sync - via GRBM)
+		 * 2 - gl2
+		 * 3 - gds
+		 * 4 - reserved
+		 * 5 - memory (async - direct)
+		 */
+#define		WR_ONE_ADDR                             (1 << 16)
+#define		WR_CONFIRM                              (1 << 20)
+#define		WRITE_DATA_CACHE_POLICY(x)              ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 */
+#define		WRITE_DATA_ENGINE_SEL(x)                ((x) << 30)
+		/* 0 - me
+		 * 1 - pfp
+		 * 2 - ce
+		 */
+#define	PACKET3_DRAW_INDEX_INDIRECT_MULTI		0x38
+#define	PACKET3_MEM_SEMAPHORE				0x39
+#              define PACKET3_SEM_USE_MAILBOX       (0x1 << 16)
+#              define PACKET3_SEM_SEL_SIGNAL_TYPE   (0x1 << 20) /* 0 = increment, 1 = write 1 */
+#              define PACKET3_SEM_CLIENT_CODE	    ((x) << 24) /* 0 = CP, 1 = CB, 2 = DB */
+#              define PACKET3_SEM_SEL_SIGNAL	    (0x6 << 29)
+#              define PACKET3_SEM_SEL_WAIT	    (0x7 << 29)
+#define	PACKET3_WAIT_REG_MEM				0x3C
+#define		WAIT_REG_MEM_FUNCTION(x)                ((x) << 0)
+		/* 0 - always
+		 * 1 - <
+		 * 2 - <=
+		 * 3 - ==
+		 * 4 - !=
+		 * 5 - >=
+		 * 6 - >
+		 */
+#define		WAIT_REG_MEM_MEM_SPACE(x)               ((x) << 4)
+		/* 0 - reg
+		 * 1 - mem
+		 */
+#define		WAIT_REG_MEM_OPERATION(x)               ((x) << 6)
+		/* 0 - wait_reg_mem
+		 * 1 - wr_wait_wr_reg
+		 */
+#define		WAIT_REG_MEM_ENGINE(x)                  ((x) << 8)
+		/* 0 - me
+		 * 1 - pfp
+		 */
+#define	PACKET3_INDIRECT_BUFFER				0x3F
+#define		INDIRECT_BUFFER_TCL2_VOLATILE           (1 << 22)
+#define		INDIRECT_BUFFER_VALID                   (1 << 23)
+#define		INDIRECT_BUFFER_CACHE_POLICY(x)         ((x) << 28)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#define     INDIRECT_BUFFER_PRE_ENB(x)		 ((x) << 21)
+#define	PACKET3_COPY_DATA				0x40
+#define	PACKET3_PFP_SYNC_ME				0x42
+#define	PACKET3_SURFACE_SYNC				0x43
+#              define PACKET3_DEST_BASE_0_ENA      (1 << 0)
+#              define PACKET3_DEST_BASE_1_ENA      (1 << 1)
+#              define PACKET3_CB0_DEST_BASE_ENA    (1 << 6)
+#              define PACKET3_CB1_DEST_BASE_ENA    (1 << 7)
+#              define PACKET3_CB2_DEST_BASE_ENA    (1 << 8)
+#              define PACKET3_CB3_DEST_BASE_ENA    (1 << 9)
+#              define PACKET3_CB4_DEST_BASE_ENA    (1 << 10)
+#              define PACKET3_CB5_DEST_BASE_ENA    (1 << 11)
+#              define PACKET3_CB6_DEST_BASE_ENA    (1 << 12)
+#              define PACKET3_CB7_DEST_BASE_ENA    (1 << 13)
+#              define PACKET3_DB_DEST_BASE_ENA     (1 << 14)
+#              define PACKET3_TCL1_VOL_ACTION_ENA  (1 << 15)
+#              define PACKET3_TC_VOL_ACTION_ENA    (1 << 16) /* L2 */
+#              define PACKET3_TC_WB_ACTION_ENA     (1 << 18) /* L2 */
+#              define PACKET3_DEST_BASE_2_ENA      (1 << 19)
+#              define PACKET3_DEST_BASE_3_ENA      (1 << 21)
+#              define PACKET3_TCL1_ACTION_ENA      (1 << 22)
+#              define PACKET3_TC_ACTION_ENA        (1 << 23) /* L2 */
+#              define PACKET3_CB_ACTION_ENA        (1 << 25)
+#              define PACKET3_DB_ACTION_ENA        (1 << 26)
+#              define PACKET3_SH_KCACHE_ACTION_ENA (1 << 27)
+#              define PACKET3_SH_KCACHE_VOL_ACTION_ENA (1 << 28)
+#              define PACKET3_SH_ICACHE_ACTION_ENA (1 << 29)
+#define	PACKET3_COND_WRITE				0x45
+#define	PACKET3_EVENT_WRITE				0x46
+#define		EVENT_TYPE(x)                           ((x) << 0)
+#define		EVENT_INDEX(x)                          ((x) << 8)
+		/* 0 - any non-TS event
+		 * 1 - ZPASS_DONE, PIXEL_PIPE_STAT_*
+		 * 2 - SAMPLE_PIPELINESTAT
+		 * 3 - SAMPLE_STREAMOUTSTAT*
+		 * 4 - *S_PARTIAL_FLUSH
+		 * 5 - EOP events
+		 * 6 - EOS events
+		 */
+#define	PACKET3_EVENT_WRITE_EOP				0x47
+#define		EOP_TCL1_VOL_ACTION_EN                  (1 << 12)
+#define		EOP_TC_VOL_ACTION_EN                    (1 << 13) /* L2 */
+#define		EOP_TC_WB_ACTION_EN                     (1 << 15) /* L2 */
+#define		EOP_TCL1_ACTION_EN                      (1 << 16)
+#define		EOP_TC_ACTION_EN                        (1 << 17) /* L2 */
+#define		EOP_TCL2_VOLATILE                       (1 << 24)
+#define		EOP_CACHE_POLICY(x)                     ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#define		DATA_SEL(x)                             ((x) << 29)
+		/* 0 - discard
+		 * 1 - send low 32bit data
+		 * 2 - send 64bit data
+		 * 3 - send 64bit GPU counter value
+		 * 4 - send 64bit sys counter value
+		 */
+#define		INT_SEL(x)                              ((x) << 24)
+		/* 0 - none
+		 * 1 - interrupt only (DATA_SEL = 0)
+		 * 2 - interrupt when data write is confirmed
+		 */
+#define		DST_SEL(x)                              ((x) << 16)
+		/* 0 - MC
+		 * 1 - TC/L2
+		 */
+#define	PACKET3_EVENT_WRITE_EOS				0x48
+#define	PACKET3_RELEASE_MEM				0x49
+#define	PACKET3_PREAMBLE_CNTL				0x4A
+#              define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE     (2 << 28)
+#              define PACKET3_PREAMBLE_END_CLEAR_STATE       (3 << 28)
+#define	PACKET3_DMA_DATA				0x50
+/* 1. header
+ * 2. CONTROL
+ * 3. SRC_ADDR_LO or DATA [31:0]
+ * 4. SRC_ADDR_HI [31:0]
+ * 5. DST_ADDR_LO [31:0]
+ * 6. DST_ADDR_HI [7:0]
+ * 7. COMMAND [30:21] | BYTE_COUNT [20:0]
+ */
+/* CONTROL */
+#              define PACKET3_DMA_DATA_ENGINE(x)     ((x) << 0)
+		/* 0 - ME
+		 * 1 - PFP
+		 */
+#              define PACKET3_DMA_DATA_SRC_CACHE_POLICY(x) ((x) << 13)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#              define PACKET3_DMA_DATA_SRC_VOLATILE (1 << 15)
+#              define PACKET3_DMA_DATA_DST_SEL(x)  ((x) << 20)
+		/* 0 - DST_ADDR using DAS
+		 * 1 - GDS
+		 * 3 - DST_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_DST_CACHE_POLICY(x) ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#              define PACKET3_DMA_DATA_DST_VOLATILE (1 << 27)
+#              define PACKET3_DMA_DATA_SRC_SEL(x)  ((x) << 29)
+		/* 0 - SRC_ADDR using SAS
+		 * 1 - GDS
+		 * 2 - DATA
+		 * 3 - SRC_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_CP_SYNC     (1 << 31)
+/* COMMAND */
+#              define PACKET3_DMA_DATA_DIS_WC      (1 << 21)
+#              define PACKET3_DMA_DATA_CMD_SRC_SWAP(x) ((x) << 22)
+		/* 0 - none
+		 * 1 - 8 in 16
+		 * 2 - 8 in 32
+		 * 3 - 8 in 64
+		 */
+#              define PACKET3_DMA_DATA_CMD_DST_SWAP(x) ((x) << 24)
+		/* 0 - none
+		 * 1 - 8 in 16
+		 * 2 - 8 in 32
+		 * 3 - 8 in 64
+		 */
+#              define PACKET3_DMA_DATA_CMD_SAS     (1 << 26)
+		/* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_DAS     (1 << 27)
+		/* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_SAIC    (1 << 28)
+#              define PACKET3_DMA_DATA_CMD_DAIC    (1 << 29)
+#              define PACKET3_DMA_DATA_CMD_RAW_WAIT  (1 << 30)
+#define	PACKET3_AQUIRE_MEM				0x58
+#define	PACKET3_REWIND					0x59
+#define	PACKET3_LOAD_UCONFIG_REG			0x5E
+#define	PACKET3_LOAD_SH_REG				0x5F
+#define	PACKET3_LOAD_CONFIG_REG				0x60
+#define	PACKET3_LOAD_CONTEXT_REG			0x61
+#define	PACKET3_SET_CONFIG_REG				0x68
+#define		PACKET3_SET_CONFIG_REG_START			0x00002000
+#define		PACKET3_SET_CONFIG_REG_END			0x00002c00
+#define	PACKET3_SET_CONTEXT_REG				0x69
+#define		PACKET3_SET_CONTEXT_REG_START			0x0000a000
+#define		PACKET3_SET_CONTEXT_REG_END			0x0000a400
+#define	PACKET3_SET_CONTEXT_REG_INDIRECT		0x73
+#define	PACKET3_SET_SH_REG				0x76
+#define		PACKET3_SET_SH_REG_START			0x00002c00
+#define		PACKET3_SET_SH_REG_END				0x00003000
+#define	PACKET3_SET_SH_REG_OFFSET			0x77
+#define	PACKET3_SET_QUEUE_REG				0x78
+#define	PACKET3_SET_UCONFIG_REG				0x79
+#define		PACKET3_SET_UCONFIG_REG_START			0x0000c000
+#define		PACKET3_SET_UCONFIG_REG_END			0x0000c400
+#define	PACKET3_SCRATCH_RAM_WRITE			0x7D
+#define	PACKET3_SCRATCH_RAM_READ			0x7E
+#define	PACKET3_LOAD_CONST_RAM				0x80
+#define	PACKET3_WRITE_CONST_RAM				0x81
+#define	PACKET3_DUMP_CONST_RAM				0x83
+#define	PACKET3_INCREMENT_CE_COUNTER			0x84
+#define	PACKET3_INCREMENT_DE_COUNTER			0x85
+#define	PACKET3_WAIT_ON_CE_COUNTER			0x86
+#define	PACKET3_WAIT_ON_DE_COUNTER_DIFF			0x88
+#define	PACKET3_SWITCH_BUFFER				0x8B
+#define PACKET3_FRAME_CONTROL				0x90
+#			define FRAME_CMD(x) ((x) << 28)
+			/*
+			 * x=0: tmz_begin
+			 * x=1: tmz_end
+			 */
+#define	PACKET3_SET_RESOURCES				0xA0
+/* 1. header
+ * 2. CONTROL
+ * 3. QUEUE_MASK_LO [31:0]
+ * 4. QUEUE_MASK_HI [31:0]
+ * 5. GWS_MASK_LO [31:0]
+ * 6. GWS_MASK_HI [31:0]
+ * 7. OAC_MASK [15:0]
+ * 8. GDS_HEAP_SIZE [16:11] | GDS_HEAP_BASE [5:0]
+ */
+#              define PACKET3_SET_RESOURCES_VMID_MASK(x)     ((x) << 0)
+#              define PACKET3_SET_RESOURCES_UNMAP_LATENTY(x) ((x) << 16)
+#              define PACKET3_SET_RESOURCES_QUEUE_TYPE(x)    ((x) << 29)
+#define	PACKET3_MAP_QUEUES				0xA2
+/* 1. header
+ * 2. CONTROL
+ * 3. CONTROL2
+ * 4. MQD_ADDR_LO [31:0]
+ * 5. MQD_ADDR_HI [31:0]
+ * 6. WPTR_ADDR_LO [31:0]
+ * 7. WPTR_ADDR_HI [31:0]
+ */
+/* CONTROL */
+#              define PACKET3_MAP_QUEUES_QUEUE_SEL(x)       ((x) << 4)
+#              define PACKET3_MAP_QUEUES_VMID(x)            ((x) << 8)
+#              define PACKET3_MAP_QUEUES_QUEUE_TYPE(x)      ((x) << 21)
+#              define PACKET3_MAP_QUEUES_ALLOC_FORMAT(x)    ((x) << 24)
+#              define PACKET3_MAP_QUEUES_ENGINE_SEL(x)      ((x) << 26)
+#              define PACKET3_MAP_QUEUES_NUM_QUEUES(x)      ((x) << 29)
+/* CONTROL2 */
+#              define PACKET3_MAP_QUEUES_CHECK_DISABLE(x)   ((x) << 1)
+#              define PACKET3_MAP_QUEUES_DOORBELL_OFFSET(x) ((x) << 2)
+#              define PACKET3_MAP_QUEUES_QUEUE(x)           ((x) << 26)
+#              define PACKET3_MAP_QUEUES_PIPE(x)            ((x) << 29)
+#              define PACKET3_MAP_QUEUES_ME(x)              ((x) << 31)
+#define	PACKET3_UNMAP_QUEUES				0xA3
+/* 1. header
+ * 2. CONTROL
+ * 3. CONTROL2
+ * 4. CONTROL3
+ * 5. CONTROL4
+ * 6. CONTROL5
+ */
+/* CONTROL */
+#              define PACKET3_UNMAP_QUEUES_ACTION(x)           ((x) << 0)
+		/* 0 - PREEMPT_QUEUES
+		 * 1 - RESET_QUEUES
+		 * 2 - DISABLE_PROCESS_QUEUES
+		 * 3 - PREEMPT_QUEUES_NO_UNMAP
+		 */
+#              define PACKET3_UNMAP_QUEUES_QUEUE_SEL(x)        ((x) << 4)
+#              define PACKET3_UNMAP_QUEUES_ENGINE_SEL(x)       ((x) << 26)
+#              define PACKET3_UNMAP_QUEUES_NUM_QUEUES(x)       ((x) << 29)
+/* CONTROL2a */
+#              define PACKET3_UNMAP_QUEUES_PASID(x)            ((x) << 0)
+/* CONTROL2b */
+#              define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(x) ((x) << 2)
+/* CONTROL3a */
+#              define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET1(x) ((x) << 2)
+/* CONTROL3b */
+#              define PACKET3_UNMAP_QUEUES_RB_WPTR(x)          ((x) << 0)
+/* CONTROL4 */
+#              define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET2(x) ((x) << 2)
+/* CONTROL5 */
+#              define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET3(x) ((x) << 2)
+#define	PACKET3_QUERY_STATUS				0xA4
+/* 1. header
+ * 2. CONTROL
+ * 3. CONTROL2
+ * 4. ADDR_LO [31:0]
+ * 5. ADDR_HI [31:0]
+ * 6. DATA_LO [31:0]
+ * 7. DATA_HI [31:0]
+ */
+/* CONTROL */
+#              define PACKET3_QUERY_STATUS_CONTEXT_ID(x)       ((x) << 0)
+#              define PACKET3_QUERY_STATUS_INTERRUPT_SEL(x)    ((x) << 28)
+#              define PACKET3_QUERY_STATUS_COMMAND(x)          ((x) << 30)
+/* CONTROL2a */
+#              define PACKET3_QUERY_STATUS_PASID(x)            ((x) << 0)
+/* CONTROL2b */
+#              define PACKET3_QUERY_STATUS_DOORBELL_OFFSET(x)  ((x) << 2)
+#              define PACKET3_QUERY_STATUS_ENG_SEL(x)          ((x) << 25)
+
+
+#define VCE_CMD_NO_OP		0x00000000
+#define VCE_CMD_END		0x00000001
+#define VCE_CMD_IB		0x00000002
+#define VCE_CMD_FENCE		0x00000003
+#define VCE_CMD_TRAP		0x00000004
+#define VCE_CMD_IB_AUTO	0x00000005
+#define VCE_CMD_SEMAPHORE	0x00000006
+
+#define VCE_CMD_IB_VM           0x00000102
+#define VCE_CMD_WAIT_GE         0x00000106
+#define VCE_CMD_UPDATE_PTB      0x00000107
+#define VCE_CMD_FLUSH_TLB       0x00000108
+
+/* HEVC ENC */
+#define HEVC_ENC_CMD_NO_OP         0x00000000
+#define HEVC_ENC_CMD_END           0x00000001
+#define HEVC_ENC_CMD_FENCE         0x00000003
+#define HEVC_ENC_CMD_TRAP          0x00000004
+#define HEVC_ENC_CMD_IB_VM         0x00000102
+#define HEVC_ENC_CMD_WAIT_GE       0x00000106
+#define HEVC_ENC_CMD_UPDATE_PTB    0x00000107
+#define HEVC_ENC_CMD_FLUSH_TLB     0x00000108
+
+/* mmPA_SC_RASTER_CONFIG mask */
+#define RB_MAP_PKR0(x)				((x) << 0)
+#define RB_MAP_PKR0_MASK			(0x3 << 0)
+#define RB_MAP_PKR1(x)				((x) << 2)
+#define RB_MAP_PKR1_MASK			(0x3 << 2)
+#define RB_XSEL2(x)				((x) << 4)
+#define RB_XSEL2_MASK				(0x3 << 4)
+#define RB_XSEL					(1 << 6)
+#define RB_YSEL					(1 << 7)
+#define PKR_MAP(x)				((x) << 8)
+#define PKR_MAP_MASK				(0x3 << 8)
+#define PKR_XSEL(x)				((x) << 10)
+#define PKR_XSEL_MASK				(0x3 << 10)
+#define PKR_YSEL(x)				((x) << 12)
+#define PKR_YSEL_MASK				(0x3 << 12)
+#define SC_MAP(x)				((x) << 16)
+#define SC_MAP_MASK				(0x3 << 16)
+#define SC_XSEL(x)				((x) << 18)
+#define SC_XSEL_MASK				(0x3 << 18)
+#define SC_YSEL(x)				((x) << 20)
+#define SC_YSEL_MASK				(0x3 << 20)
+#define SE_MAP(x)				((x) << 24)
+#define SE_MAP_MASK				(0x3 << 24)
+#define SE_XSEL(x)				((x) << 26)
+#define SE_XSEL_MASK				(0x3 << 26)
+#define SE_YSEL(x)				((x) << 28)
+#define SE_YSEL_MASK				(0x3 << 28)
+
+/* mmPA_SC_RASTER_CONFIG_1 mask */
+#define SE_PAIR_MAP(x)				((x) << 0)
+#define SE_PAIR_MAP_MASK			(0x3 << 0)
+#define SE_PAIR_XSEL(x)				((x) << 2)
+#define SE_PAIR_XSEL_MASK			(0x3 << 2)
+#define SE_PAIR_YSEL(x)				((x) << 4)
+#define SE_PAIR_YSEL_MASK			(0x3 << 4)
+
+#endif